I'm interested in phonetics in order to speak as properly as possible. And here's the thing, there are four vowels with ambiguous symbols:

  1. The first problem is the sound [ɛ] like in dress: /drɛs/ according to WordReference and Oxford dictionary but [e] like in dress: /dres/ according to Cambridge dictionary. So I've made some research and the only reason I've found so far is from teflpedia:

    The English phonemes /e/ and /eə/ use "e" instead of "ɛ" only for historical reasons, and several dictionaries prefer "ɛ" over "e".

    What are these historical reasons? And why do they continue to use [e] instead of [ɛ]? Because first in French we use [e] for the sound 'é' like in fiancé which is a different sound. It is a weird behavior because their table shows that the majority of dictionaries tend to use [ɛ] so why do they keep going on this way? TEFL stands for Teaching English as a Foreign Language not Confusing Foreigner, doesn't it?

  2. The second problem is [a] instead of [æ]. In the word cat: /kæt/ according to Cambridge and wordreference but /kat/ according to Oxford. Is there another historical reason here?

  3. The third problem is between [ə:] and [ɜ:]. In the word bird Oxford says /bəːd/ and Cambridge and wordreference say /bɜːd/. Which one is the correct one?

  4. The fourth problem is between [eə], [ɛː] and [ɛə]. In the word bear Cambridge writes /beər/, Oxford writes /bɛː/ and WordReference writes /bɛə/. It is seriously confusing!

Thank you if you have any explanation.

  • 7
    Trust me, I have the same trouble with those symbols. I'm pretty sure that [ɛ] for dress and [æ] for cat are correct, and that the others are not.
    – tchrist
    Apr 4, 2018 at 10:30
  • 8
    I don't think "ambiguous" is really the word you intend to describe these. Something is ambiguous when it has multiple meanings. What you have here are multiple notations for the same thing. This is redundant, not ambiguous.
    – Barmar
    Apr 4, 2018 at 19:08
  • 2
    Consider that for instance in parts of the US, "bear" is quite often pronounced more like "bar". So how do you have a consistent phonetic encoding when the language is not consistent?
    – jamesqf
    Apr 4, 2018 at 19:58
  • 6
    @jamesqf: This question is about encodings of pronunciations, not words, I believe. Apr 4, 2018 at 20:28
  • 2
    @MooingDuck James is saying that pronounciation is not consistent. Sure, he did pick a word to make that point, but...
    – Beanluc
    Apr 4, 2018 at 22:54

5 Answers 5


Actually, it's "worse" than that. Nearly all the vowels of English have more than one possible representation in IPA. For example:

  • The vowel sound of the word "kit" can be written as [ɪ] or [i]
  • The vowel sound in "lot" in British English can be written as [ɒ] or [ɔ]
  • The vowel sound in "fleece" can be written as [i], [iː], [ij] or [ɪj]
  • The vowel sound in "goose" can be written as [u], [uː], [uw], [ʉ], [ʉː], [ʉw] or [ɵw]

  • The vowel sound in "choice" can be written as [ɔɪ], [oɪ], [ɔj] or [oj]

  • The vowel sound in "face" can be written as [eɪ], [e], [ɛɪ], [ej] or [ɛj]

It's not really a matter of something special about the four cases that you mention. Quite simply, the IPA is not precise enough, and the phonetic positions of English vowels are not specific enough (variation in the realization of any particular vowel sound exists both between speakers and within the output of any individual speaker), for there to be a single unambiguous one-to-one mapping between English vowels and possible IPA representations. Therefore, consistency between different transcriptions is just a product of convention, and in fact different people have used different conventions for various reasons.

Some of the possible criteria that people have used to judge phonemic transcription conventions:

  • phonetic accuracy: how close are the characters to the phonetic definitions for the IPA symbols?

  • simplicity: how many unique characters are used in the transcription system? How many characters are used that are expected to be unfamiliar to people who are just starting to learn the system?

  • symmetry: how well does the transcription reflect the phonological relationships between sounds, and the way sounds behave phonologically?

  • redundancy: does the system distinguish different vowels in only one way (e.g. ɪ vs. i, or i vs. iː) or in two (e.g. ɪ vs. iː)? Are the vowels that occur in unstressed vowels identified with any of the vowels that occur in stressed syllables (e.g. is the vowel in "strut" transcribed the same way as the vowel at the end of the word "comma", since for most speakers, these do not contrast) or are reduced vowels transcribed with special symbols reserved for vowels in unstressed syllables?

  • conservatism: how similar is the transcription system to transcription systems that have been used in the past?

Actually, the issue of "conservative" phonemic transcriptions that don't use the IPA letter that is closest to the usual modern phonetic realization of a phoneme doesn't only come up with regard to English. The transcription of Danish vowels is similarly problematic (with the symbol ɛ being used to represent a vowel that many speakers pronounce more like [e], and the symbol æ being used to represent a vowel that many speakers pronounce more like [ɛ]). There is at least one similar case in the transcription of French: the nasal vowel found in words like vin is conventionally transcribed as /ɛ̃/, even though for many speakers it has a noticeably different quality from non-nasal /ɛ/.

The following blog posts may be illuminating:

I would say the best way to learn more about English phonetics is to

  • practice listening to the way various speakers pronounce the sounds

  • get feedback about your own pronunciation from a good teacher, if you can,

  • to learn about phonetic details, read* explanations from phoneticians. For example, the web sites I linked to have a number of additional posts about various aspects of pronunciation, and there are other good sites like Alex Rotatori's blog.

    *(Or listen, I suppose--there seem to be a number of Youtube videos and so on nowadays that cover these topics, although I don't have much experience with using videos to learn about English pronunciation so I can't give any specific suggestions.)

As Windsor Lewis says in the post I linked to above, the kind of transcriptions you find in places like dictionaries are mainly for telling you the distribution of phonemes in a word--they won't tell you exactly how to pronounce a word in terms of phonetics.

  • 1
    I was going to make a comment in my answer that IPA was probably the most consistent and scientific coding still with some variability, but not as much as the others (like Kenyon/Knott or OED), but your answer shows that maybe IPA is crazier
    – Mitch
    Apr 4, 2018 at 16:25
  • "Quite simply, the IPA is not precise enough" In my experience as a dialect speaker and British emigré living in the US, I would argue the IPA is too precise, and there needs to be a representation that encodes vowels like those in "bus, cup", "poppy, common" that accounts for standard shifts in both pitch and length across multiple accents. There is a lot of commonality: so RP /bʌs/ /kʌp/ becomes US /bəs/ /kəp/ but in Northern British accents is /bus/ /kup/; RP /ˈpɒpi/ /ˈkɒmən/; US /ˈpɑpi/ /ˈkɑmən/ (or /ɑ:/) NB /'pɔpi/ /ˈkɔmən/. Maybe I'm just talking about standardized spellings.
    – Rich
    Apr 5, 2018 at 19:17
  • 1
    @Rich: Well, it's sort of both. There are a fair number of IPA symbols--certainly more than the number of phonologically distinct vowel sounds in any language--but certain IPA symbols like e, a, u, ə are commonly used for vowels that would be more accurately described with another symbol (such as ɛ, ɑ, ʉ, ɐ) just because the first symbol is more familiar.
    – herisson
    Apr 5, 2018 at 19:28
  • I defer to you :-)
    – Rich
    Apr 6, 2018 at 17:13

Within one language community, the IPA may be simplified for dictionary entries. The /r/ is a classic example. In strict IPA usage, it is the sign for an r sound with a short trill, as in Italian Roma, but English sources routinely use this sign for any standard pronunciation of r.

In this recording from the late 1920s of John Gielgud delivering a speech from Othello, the actor uses the tapped /ɾ/ in very, the retroflex standard /ɹ/ in dearest, and initial, trilled /r/ in several words. These would all likely be transcribed as /r/ because the differences are either positional variants or dictated by the stage pronunciation of the day.

In contrast to German and French, English does not need to distinguish between two open, unrounded, middle vowels, one of which generations of schoolchildren on both sides of the Atlantic have learned (and learnt) to call short e. In other words, your “historical reasons” boil down to habit, custom, and a reluctance to use symbols outside the Latin alphabet, including standing one on its head like /ɹ/.

When Oxford Press took on Clive Upton as a pronunciation consultant for its dictionaries for native speakers, he introduced a more truly international method of transcription by insisting that English short e be rendered as ɛ. This was adopted by the Concise Oxford Dictionary in 1995. Moreover, both sources for dictionary.com, the American Random House and the British Collins, both render shred as /ʃrɛd/.

John Wells of University College London comes up with a rather odd objection:

In some languages, notably French and German, one needs to distinguish two e-type vowels, a closer one (IPA [e]) and an opener one (IPA [ɛ]). The English bet vowel lies between them, but is more similar to [ɛ], which is why Upton prefers that symbol. However, from the point of view of an EFL learner whose native language is, say, Japanese or Greek — languages that have no such distinction — it is quite unnecessary to distinguish the "[e]" at the starting point of the face diphthong from the "[ɛ]" of bet. And following IPA principles, if we are to choose just one of the two symbols we should prefer the simpler one.

Wells does not mention why he would favor EFL learners from Japan and Greece over those from Germany or France except that it is “simpler,” i.e., using the standard Latin e rather than an epsilon so it looks like the short e of old.

  • 1
    Technically, the tapped sound of Italian cara or vero is [ɾ]. The trilled sound is [r]. Those are different rhotic phonemes in Spanish, with minimal pairs like pero–perro. They aren't distinct phonemes in Italian, though.
    – tchrist
    Apr 4, 2018 at 13:07
  • @tchrist: corrected.
    – KarlG
    Apr 4, 2018 at 13:50
  • 6
    My reading of John Wells' objection to using "[ɛ]" is not that we should favor EFL learners from Japan and Greece over those from Germany and France, but rather that we should favor ease of comprehension by native speakers over EFL learners from Germany and France, especially since there's no way to make things easier for all EFL learners. As a native speaker without a deep learning of IPA, I find dictionary entries that use /e/ easier to use, intuitively, than those that use /ɛ/, so that makes sense to me for dictionaries mainly aimed at native speakers.
    – 1006a
    Apr 4, 2018 at 14:09
  • 3
    That's the "no way to make things easier for all EFL learners" part: Using /ɛ/ might help EFL users from countries X and Y, but it leaves learners from countries A and B cold, so might as well do what works for native speakers. In other words, if using /ɛ/ was especially helpful to all EFL learners, maybe it would be worth switching, but it's not, so it's not.
    – 1006a
    Apr 4, 2018 at 14:27
  • 3
    @1006a: You're simply crafting a better argument on the same premises.
    – KarlG
    Apr 4, 2018 at 14:33

What you are seeing is not variation in pronunciation by different varieties. Most all dictionaries (OED, M-W, Collins, online dictionaries) will give one pronunciation for British English (RP) or General American English (GenAmE) (possibly with a very common variant that is not dialectical).

What you are seeing is a difference in phonetic encoding.

English dictionaries tend to have their own idiosyncratic pronunciation symbols. They are consistent within a dictionary, but between dictionaries may differ (as you note).

Somewhere in the introduction or FAQ of these dictionaries is an explanation of the pronunciation symbols. To your explicit question of the history of the symbols, there may well be given a history of the choice, or more likely not.


Vowels change. A hundred years ago, the standard southern British pronunciations of bear, cat, code, and cut were [bɛə], [kæt], [koʊd], and [kʌt]. Now, they're [bɛ:], [kat], [kəʊd], and [kɐt].

Why do the vowels change in English while they don't change in, say, Italian? English has enough vowels that they keep jostling each other to get more personal space, and they end up moving each other in various directions. Italian only has seven vowels, they're all relatively far from the other vowels, and they're perfectly happy to stay where they are.

You can argue over whether people should keep the English IPA vowel symbols the same for consistency (but then they wouldn't match what the symbols mean in other languages) or change them to match other languages, but be inconsistent with previous notations for the same language (as well as with American English, where the pronunciation hasn't changed in the same way).

Neither solution is perfect. And the dictionaries seem to be using different strategies for the two vowels /æ/ and /ʌ/.

I don't know whether that's the problem with /ɛ/ and /ɜ:/ as well, or whether there's some different reason for the ambiguity in these symbols. I actually suspect that the reason that /ɜ:/ was originally chosen rather than /ə:/ is that at that time, these were the only two vowels where the only difference was in length. So they chose /ɜ:/ rather than /ə:/ so as to get a different symbol for each vowel (and they believed that they were justified in doing this because speakers used both /ə:/ and /ɜ:/ for the vowel in NURSE).

  • Surely cat and cat are the other way round?
    – as4s4hetic
    Apr 4, 2018 at 11:11
  • @as4s4hetic: No, the vowel in pat has been moving away from the one in pet and towards the one in part, which is the direction from [æ] to [a], although it probably hasn't reached [a] yet in all speakers. See this blog post. And meanwhile the vowel in pet has been moving in the same direction—the direction from [e] towards [ɛ]. Apr 5, 2018 at 12:13
  • Hmm. I think cat needs a new diphthong like æ but with 'a' and 'ə'. Apr 5, 2018 at 17:08
  • @WillCrawford: I believe I've read that any of the "short" vowel phonemes (kit, dress, trap, lot, strut, foot) may have centering diphthongs as allophones, at least in certain contexts and for some speakers, but this is usually considered to be a phonetic detail, not part of the underlying phonological specification of the vowel.
    – herisson
    Apr 5, 2018 at 17:56
  • @sumelic it's more of an anatomic detail (how you get your tongue from the "main" vowel to the consonant :o) Apr 6, 2018 at 4:24

Vowels in English are notoriously varied. It very much depends on your accent as to which sound you produce. And that in turn depends on your region and class.

I have always learned that /ɛ/ is short and open, as in pen, whereas dress would be /e/.

The Oxford dictionaries would describe one particular (prestigious) accent, but I don't quite understand why Cambridge would be much different from that. I am not aware that there are different interpretations on the IPA between different publishers.

  • 10
    But ... lots of people use the same vowel for pen and dress. Apr 4, 2018 at 11:11
  • 1
    Indeed. It's very complicated! Apr 4, 2018 at 12:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.