P is for Phonemic Chart

8 08 2010

(That’s phonEMIC, not phonETIC, by the way. There’s a big difference!)

Ever since I’ve been teaching in the US I’ve been challenged by the need to devise a chart of the phonemes of American English (General American or GA) that can be used in the same way as the original British English (RP) chart, both as a training and a teaching tool. (Incidentally, it’s an often overlooked fact that the layout of the original RP chart – along with lots of ways of exploiting it in class – is due to the work of Adrian Underhill).

Adrian Underhill's 'Sound Foundations' Chart (Macmillan)

In fact, the search for a GA equivalent goes back even earlier, to 1995, when I was assessing a CELTA course here in New York and was surprised to find that the language analysis trainer was trying to knock the round peg of GA sounds into the square hole of the RP chart. Fifteen years later I discover that not much has changed: another large training organisation here is using an “Americanized” version of the original RP chart, but one which not only includes five more vowel sounds than GA is normally credited with having, but adds two diphthongs ( /ʌɪ/ and /ɔʊ/) that, as far as I know, belong to no known variety of English!

Of course, the problem of devising a GA chart is complicated by the fact that – unlike the case of RP – there is no single, agreed upon, system of transcribing American vowels. (Compare any two American learners’ dictionaries, for instance). This is probably due to the fact that, while there is less accent variation across North America than there is within the British Isles, there is no single variety that can (or is allowed to) claim the prestigious status that RP enjoys.

In 2007, while teaching at SIT in Brattleboro, Vermont, I came up with a chart that was based closely on the description in Celce-Murcia et al. (1996) – see inset below (click to expand).

GA chart (after Celce-Murcia et. al., 1996)

The layout of the chart attempts to reflect the elegance of Adrian’s RP chart, with the consonants ranged from front-of-mouth to back-of-mouth obstruction, and the vowels roughly mapped on to the classic (Daniel Jones?) vowel quadrant. In terms of the symbols, the consonants were not a problem: the only change involved changing the symbol /j/ for a /y/. The vowels were another story.

First of all the layout had to be reconfigured to accommodate the fewer vowel sounds of GA (16 vs 20 in RP). While the three ‘heterogenous’ diphthongs are separated out and colour-coded, no attempt was made to distinguish the simple vowels from the vowels with an adjacent glide (/iy/, /ey/, /ow/) since the latter, technically, are not diphthongs.  Nor were combinations with /r/ (such as /ır/ and /or/) included, since, technically, these are not individual phonemes but are attempts to represent the way certain vowel sounds are “colored” by the consonants that follow them (which may be /r/, /l/ or /rl/). The only exception I made was the case of /ɜr/ which, as Celce-Murcia et al. point out, is used “to capture a significant difference in quality between the /ʌ/ in bud and the /ɜr/ in bird” (p. 105) and which they include as their “15th phoneme” of North American English (the 16th being the schwa).  Finally, an optional superscript /r/ was added to the schwa, because the combination of schwa and post-vocalic /r/ is often distinguished from schwa, phonetically, by being transcribed with a different symbol (ɚ). This represents the (phonemic) difference in GA between the final vowels in cheeta and cheater, for example. Note also that both /ɔ/ and /ɑ/ are represented in the chart, in deference to those varieties of GA that do distinguish between caught and cot.

This chart has served OK over the years, but I’ve not been entirely happy with it – not least because of the use of the consonant symbols /y/ and /w/ to flag lengthening and lip rounding, as well as the clumsy superscript [r]s. So I revisted the literature, and came up with a new one, based on the description in Roca and Johnson (1999). The consonants remain as they were. The main differences to the vowels is that I’ve abandoned the /y/ and /w/ add-ons, susbtituting symbols that more accurately realise the phonetic qualities of the homogeneous (adjacent glide) and heterogeneous (non-adjacent glide) diphthongs, colour-coding these respectively, as well as substituting the symbol ɚ for the r-coloured schwa alternative, and /ɝ/ for the r-coloured vowel in bird. I’ve also re-positioned /ʌ/ so that its central and back quality is more accurately represented, and turned the division between /ɔ/ and /ɑ/ into a dotted line to flag that, in some varieties, these two sounds are not distinguished. You can view a pdf version of the revised chart here: AmE phonemic chart v.5

All comments will be gratefully received and acknowledged.


Celce-Murcia, M., Brinton, D.M., and Goodwin, J.M. (1996) Teaching Pronunciation. Cambridge University Press.

Roca, I., and Johnson, W. (1999) A Course in Phonology. Oxford: Blackwell.


Click here ( US phonemic chart ) to see a pdf version of Adrian Underhill’s GA Chart – mentioned in his comments below. (Thanks, Adrian!)

About these ads



28 responses

8 08 2010
Alex Case

Interesting. I’d love to see a combined RP/ GA chart so that I could run my (knowing American pron and getting used to British pron) students through all the possible sounds and variations, but as that would entail me having to put on both a fake RP accent and a fake American accent, think I might be a bit too shy to really try it. Having said that, although my sounds are different and I disagree on some words, the RP one does seem to completely map onto my own Estuary English sound system.

Interested to see that you made up your own symbols. Are the RP ones not just taken from the IPA (genuine question)? How much are your own based on it? For example, isn’t /y/ a vowel sounds something like /u/?

8 08 2010
Scott Thornbury

Thanks Alex – there’s defintiely scope for a multi-variety phonemic chart, but I think it would have to be computer-based, with the ‘charts’ overlaid one upon the other, ideally with audio files embedded, so that you could click through the options and hear – and compare – the different sounds. More than my feeble skills allow I’m afraid. Any takers?

And, no, I didn’t invent any symbols, the r-coloured schwa and r-coloured /3/ symbols do exist in IPA, as do all the other symbols in the chart. (And you can find them all in your font index).

9 08 2010
Alex Case

How about the y/j thing? Sure I could find that out with a Google search in minutes, but easier just to ask here…

9 08 2010
Scott Thornbury

It’s true, the IPA has [j], while [y] represents the vowel sound in French ‘rue’ or German ‘früh’. The /y/ symbol for the consonant in ‘you’ is probably a legacy of American dictionary writers. But now that I’ve eliminated /y/ from the vowel section, I’m going to leave [j] as /y/, for the time being. /j/ has always been a little counterintuitive for trainees – an unnecessary oddity in the otherwise pretty transparent consonant section.

12 08 2010
Alex Case

I’ve had the same problems with “and this section is all pretty self evident, apart from this one, which is there to trick you” over the years with both students and trainees, but recently I’ve settled on “Think of it as the first letter of yes in German” and that seems to work for most

8 08 2010
Alex Case

PS, have had trainees on Cert courses absolutely refusing to believe me when I tell them phonemic isn’t the same as phonetic (and certainly nothing much like phonics), so I feel the pain in your first sentence

8 08 2010
Scott Thornbury

Alex, in my researches I found teachers’ websites that implied the terms were interchangeable.

9 08 2010
Alex Case

I had a feeling their high school English teachers had told them that too (so they must’ve been American, because my English teachers just told us to turn the page and keep reading…)

8 08 2010
Simon Greenall

This is extremely useful, thank you Scott. I really like Alex’s suggestion of a combined RP/GA chart.
It’s still surprising to me how state schools and universities in some countries opt for one or the other, without considering a comparative approach, even if, as in East Asia, there’s generally a greater tendency towards GA. At the planning stages of textbook publishing (boo hiss! :) it involves some really important discussions and decisions.
Many thanks.

9 08 2010

Thanks for the updated chart, Scott. It is good to see an actual American English version that attempts to both apply the logic of Underhill’s for RP and use only IPA symbols as opposed to most American English transcriptions. I was reminded of the struggles I endured during my Trinity Diploma course in trying to transcribe my Canadian accent using the RP symbols. It just never quite fit. (My colleague from Northern Ireland was having similar problems.)
A couple of points that immediately come to mind (difficult without Phonmap on WordPress):

The a sound for pot is the same as bought, at least in my version of Canadian English. I think this difference is noticed in certain varieties of American English on the East Coast but is not prevalent throughout. I’m not sure which should be categorized as Standard.
RP diphthongs for words such as ear, pure and air are not generally recognized in Standard American English. Yet, I can hear a slight glide from the strong vowel to the schwa before the rhotic r is pronounced in these words in American English. I wonder if I am only hearing this because the RP chart I have been working with for so long has been acting as a guide.
I am sure some others have some thoughts on this.

9 08 2010
Scott Thornbury

Thanks Jeff, for your comment. It’s true that (apparently) in the non-rhotic American accents (i.e. those that don’t pronounce the ‘r’s in ‘fourth floor’, for example, the centring diphthong (i.e. the glide to the schwa position) is used instead, as in non-rhotic British accents (such as RP). But my source (Roca and Johnson) doesn’t mention a glide in rhotic accents. That doesn’t mean to say it doesn’t happen, though. Maybe this is one to ask John Wells (on his very excellent phonetic blog (at http://phonetic-blog.blogspot.com/)

9 08 2010

Hi Scott,

I think this is a great chart! Thanks for sharing it with us! Here are the things I particularly liked about it:
- the fact that you used /e/ for the (homogeneous) diphthong /eɪ/ and /ɛ/ for the single vowel. Many dictionaries and textbooks use /e/ for both the diphthong and the single vowel, but in most varieties of AmE the vowel in “egg” is higher than in “bay”, so they need different symbols. (I know there might be some phonetic realizations of “egg” with the lower [e], but in fewer regions of the US, I suppose);
- The fact that /ɝ/ already comes with the diacritic for the rhotic sound, for it always (and only) occurs with the rhotic /r/ (as opposed to /ʌ/);
- The fact that you added both /ɔ/ and /ɑ/, but with a dotted line dividing them. It is true that most Americans don’t distinguish this sound in most words (like bought, cot, hot, dog, law, etc), but adding both phonemes is important not only because some people o distinguish those two phonemes in common words, but also because in words in which these phonemes are followed by /r/, they are always distinguished (“bar”, “garden” and “car” are always /ɑ/ whereas “cord”, “Gordon” and “port” are always /ɔ/). Besides these followed by /r/, there are also a few other words that are also always distinguished for no specific rule (not that I know of), like “father” and “sergeant”, which are always /ɑ/;

The only thing I will adapt from you chart when I use it (may I?) is the /y/ phoneme. I prefer the IPA /j/ just to keep consistency with the IPA. All symbols you are using are IPA, so why not use this one too? Also, if you are teaching someone who speaks German or French (or who is also learning French-L2 or German-L2), the /y/ symbol could be saved for the high front rounded vowel.

Now, here is a genuine question: did you deliberately decide not to use the long diacritic (:) for the high vowels /i/ and /u/? I know the fact that they are long is not their distinctive feature, but isn’t it a fact that these sounds are long? (perhaps this aspect is important only phonetically?)

ronaldojunior on twitter

9 08 2010
Scott Thornbury

Thanks, once again, Ronaldo for your comprehensive comment. With regard to the /y/ vs /j/ issue – see my earlier reply to Alex. And, since there is no high front rounded vowel in English, the [y] has only phonetic (not phonemic) significance, hence there is no need for it in the chart. But, I agree, it is a little inconsistent (and potentially misleading) to use use /y/ for yes. Maybe this is something I should change.

As for the diacritics to signal vowel length, it seems that the length of a vowel depends, not on its phonemic value, but on its environment. Thus, the vowel in key is longer than the vowel in keep or keen, whereas the difference between keen and kin is less to do with length than with tension. This means that the difference between keen and kin is that in the former the vowel “involves considerable tensing of the vocal apparatus” (Roca and Johnson p. 182). Lack of tension (or ‘laxing’) results in a certain degree of lowering and centralization, hence the difference between /ɪ/ (kin) and /i:/ (keen) is not one of length but of position. So – the diacritics are a feature of phonetics (not phonemics) and of tension, not lengthening. We should really get rid of them in the RP chart too!

11 08 2010
Adrian Underhill

Thanks for this Scott and all of you. This is really fascinating. In the mid 90’s I devised an attempt at a GA chart in Brazil, and this was tried out and experimented with in a number of language schools there. Two of the problems that you mention, are lack of a GA that does the focussing job of RP, and the disarray of phonetic, alphabetic and diacritic systems used by different US publishers in their dictionaries. For a chart to be fully functional it needs to be supported by (some) dictionaries. This has changed in recent years with the introduction of US versions of originally UK style advanced learner dictionaries from the UK, employing IPA symbols mapped onto a provisional GA. In order to maintain some of the original usability of the Sound Foundations chart and also a family resemblance, I have used this to design a US/GA chart, which I will put up here as soon as I’ve resolved a technical problem, with a commentary relating it to the very useful points you’ve raised. And I don’t have an answer for some of them either. More later…

12 08 2010
Scott Thornbury

Thanks for your comment, Adrian. Good point about the chart needing to be compatible with dictionaries. And coursebooks, too, I guess. I’m looking forward to seeing your version! (It should be easiy uploadable if it’s in pdf format, by the way).

12 08 2010
Adrian Underhill

Hi Scott and chart people

Thanks for your V5 chart Scott. Since my RP chart is consistent with the Macmillan Eng Dictionary I finally decided to use the same symbols as the US version of the Macmillan Dictionary, and this almost completely agrees with your V5 Scott

As you say consonants are not a problem, though I decided to keep with /j/ rather than /y/. Admittedly there are a few seconds of surprise from some though not all learners (mostly NESTs?) but once they’ve got it they get it and there’s no problem. And the price of /y/ is it has already been nabbed for something else.

Vowels are indeed another matter, and I agree with your choice of symbols for the glides, which I think are becoming standard as the UK publisher preference for IPA symbols over letters/diacritics takes hold.
As you say combinations with /r/ (such as /ır/ and /or/) are not individual phonemes and there is greater flexibility in adding the /r/ as a separate selection when needed (though the drawback is that it might suggest a sequence of two sounds rather than one ‘coloured’ sound. But that is a matter of convention and learners quickly adapt).
I felt that the optional /r/ was not needed for schwa and can likewise be added when needed. I have however used /ɜr/ as that is predictable and I think always the case (?) ie there is no need for /ɜ/ on its own. Happily the Macmillan dictionary comes to the same conclusions.

It is worth remembering that the chart is for teachers and learners, not for phoneticians. Thus all sorts of slight (phonetic) variations do not need to show up on the chart. My approach is to propose that (nearly) all (acceptable) variations of a phoneme (allophones) are to be found in the space inside the box belonging to that phoneme. And teachers and students can point off centre within a box if they wish to focus on finely tuned variations. Also, the symbol no longer prescribes an absolute sound value, but a family of allophones, with the current desirable model being in the centre of the box for that school or teacher or class or setting. Thus each variety of English (Nigerian or Scottish or …..) can and does impose its phonetic set values on the symbol. Each reinterprets the phoneme box to put its own sound at the centre. Thus symbols are relative for us teachers and learners and not absolute. There is negotiation involved in deciding what is the model that will be attributed to the symbol in that setting. This relativity is very much more important in the current ELF/EIL climate than it was in the 30’s when Daniel Jones modelled absolute sounds on recording. That was and is fine, but it is no longer the only way we are choosing to use symbols today.

I should also say that my prime use of the chart is with a pointer, where students are pointing out sequences of sounds, and running them together to make words and phrases. If you are not doing this activity, you might view the role of the chart in a different way, and prefer a more precise / wider set of choices.

The layout: I’ve tried to keep the same design criteria, so that there is information embedded in the layout, making the chart something of a map and more than a list

I have puzzled for a long time over differences between /ɔ/ and /ɑ/ and /D/
(the sound in RP pot) and pestered many native US and Canadian speaker for examples. I have met some who claim to distinguish /D/ from /ɔ/ and /ɑ/ but I think the choice to omit /D/ is helpful, though I have left that as a space which can be negotiated by teachers / learners locally, and also to draw attention to that difference with RP.

As you know I like to use handwritten phonemes (to reduce the scare factor and to imply learner usability) but the GA chart I have is for some reason slightly out of focus. It is a jpg and I will try to upload it now. But I hope to have a better one tomorrow and will post that too

12 08 2010
Adrian Underhill

Cannot upload my jpg. Will email in case you have a way. the reproduction is not perfect but it illustrates the layout

13 08 2010
Scott Thornbury

Thanks Adrian – the comments windows don’t support attachments, so I’ve added it as an appendix to the original posting (P is for Phonemic Chart).

13 08 2010
Scott Thornbury

Adrian, it’s gratifying (and a huge relief!) that we’ve reached more or less the same conclusions, coming at this task separately and perhaps with slightly different motives (yours being more classroom-focused, perhaps, compared to my narrower interest in training). As you say, apart from the /j/ (and I’m coming round to that) we differ only on my/3/for your /e/, and some slight differences in the arrangement. I can live with that!

Equally gratifying is to hear you spell out the rationale in these terms:
“It is worth remembering that the chart is for teachers and learners, not for phoneticians. Thus all sorts of slight (phonetic) variations do not need to show up on the chart. …” This and the rest of the paragraph should be read by all teachers and trainers alike. I will certainly make sure my future MA students read it and take it to heart.

Thanks a /bʌnʧ/!!!

15 08 2010
Adrian Underhill

That’s great! Thanks for getting my chart up there Scott.
Yes I quite agree, it is amazing we have the same conclusions especially after (in my case) years of puzzlement and head scratching, apparent lack of a GA as a reference, and US dictionaries that prefer anything to IPA. I think what has helped me is the US versions of UK ALDs, and their willingness to make a pitch for an IPA based ‘standard’. Hopefully US TESOL people can push against this and critique it, and perhaps come up with something more usable or learnable or generalisable.
By the way I can see the point of your dotted line division between /ɔ/ and /ɑ/ and in a related manner I’m still toying with the idea of sharing my blank box on the US chart (= RP sound /D/ ) between /ɔ/ and /ɑ/, but at the moment I prefer to leave the option open for /D/ . Again, it will be interesting to know what speakers and learners of US/GA pron think.

1 09 2010
Scott Thornbury

As a footnote to the discussion on the differences between BrE (RP) and AmE/GenAm pronunciation, and how this might affect the design of a phonemic chart, I’ve now had a chance to consult John Well’s comments on the subject, in The Longman Pronunciation Dictionary (LPD) (2000 edition).

First of all, Wells confirms my initial suspicion, i.e. “GenAm is not as tightly codified for EFL purposes as RP” (p. xv) and he outlines the following differences:

“There is considerable variability in GenAm vowels in the open back area. LPD follows tradition in continuing to distinguish the vowel of lot lɑːt from that of thought θɔːt. (Note, though, that books by American scholars generally do not use length marks.) However, fewer and fewer American distinguish these two vowel sounds from one another.” (ibid.)

“LPD distinguishes between the vowels ʌ and ə, although in AmE they can generally be regarded as allophones of the same phoneme, and for some speakers are more or less identical phonetically too. … The actual quality used by Americans for ə varies considerably, being typically more ɪ-like when followed by a consonant but more ʌ-like when at the end of a word.”

Wells transcribes the vowel in the AmE pronunciation of nurse and stir as ɝː (compared to RP ɜː ), which is what I’ve done (although without the length marks).

For the diphthongs, he has oʊ for RP əʊ but otherwise follows RP (although without ɪə, eə, ʊə of course).

If you don’t already, you should follow Wells’ excellent blog at:

8 09 2010

Would it not be easier to try and teach the Americans to speak English properly?

[Please...I'm joking]

15 09 2010

Hi Scott! I seem to have anticipated you in this discussion of yours. Look at my weblog:

By the way, I’m really happy now you have decided to modify your chart symbols for GA: these are much clearer and straightforward!

Also note that Celce-Murcia et al are about to release the 2nd edition of their “Teaching Pronunciation”:


As far as glides in rhotic accents are concerned, Jeff is right in saying that some words in GA may be pronounced with an [ə]-glide before a liquid. This effect is called “pre-r breaking” and arises in GA when /r/ followes the vowel phonemes /ɪ, e, æ/. Take the example of the term “fear”: especially if said slowly, this word may actually sound more like [fɪər] or [fiər]. See Wells’s LPD3 under “breaking”, p.103.

16 09 2010
Scott Thornbury

Thanks for that, Alex. I wish I’d seen your blog post before I wrote mine – but it does seem that our two ‘systems’ are converging – or that mine is showing a little more consistency across the two representations (i.e. BrE and GA). Any advice on how to improve it further would be welcome – because I do think that the BrE chart is of limited usefulness in an American English context, in that it displays more vowel symbols than are strictly necessary to represent the standard phonemic distinctions. (And I also need to get rid of the /y/ and use IPA (j/).

6 04 2012
Ndrew Allmark

I couldn’t help but notice your notice your comment about the extra diphthong /ʌɪ/, as I have noticed that this is used in the online Oxford dictionary and the 2002 concise paper edition instead of /aɪ/.
I’m not sure why it is used, but I had assumed it may be because it more closely resembles the mouth positioning of a minority RP pronunciaion of this sound. Is this the only dictionary to do this? And does anybody know if they have justified their choice?
The dicitonary also include it as the first two elements in a triphthong /ʌɪə/ as in ‘fire’. However in contrast neither Underhill and Kelly include any triphthongs in their phonology guides. To what extent do you feel triphthongs could be a useful pedagogical concept?

6 04 2012
Scott Thornbury

Thanks for the comment.

I wasn’t aware that the Oxford dictionaries prefer /ʌɪ/ to /aɪ/, and I imagine there are good reasons for this. The point is that there are not two separate phonemes for the PRICE phoneme, but that its phonetic realization is shifting and perhaps becoming more central. One to watch.

Regarding triphthongs, I’ll refer you to John Wells, who argues that English doesn’t have them because the vowel glides in words like FIRE and POWER occur beyond the boundaries of the syllable: http://phonetic-blog.blogspot.com.es/2009/12/triphthongs-anyone.html

6 04 2012
Ndrew Allmark

THanks very much, I’ll /trʌɪ/ or /traɪ/ to have a read through Well’s blog when I get a chance. :-)

7 05 2012
James H. Kanzelmeyer

I found your “Phonemic” chart to be in conflict with your exhortation that phonemic is definitely not phonetic. The chart below contains but 1 symbol that resembles IPA characters and no phonetic pronunciation reference. The names of the characters are the sounds from the appropriate range of phonetic allophones the reader is urged to utter based upon individual speech habits. American English speakers will have no trouble making the associations between their spoke phoneme word-elements and written word-elements. ESL students, unless they have had prior positive experience interacting with English-speakers, will require separate guidance in changing their native sound-letter associations for those that permit them to be understood. Case in point: Spanish pronunciation of the word pup will invariably be an embarassing “poop” until they acquire the hitherto unknown sound of short u.

Phonemic Chart from http://www.akses.org/amws03.htm

Sorry, the response form doesn’t allow pasting pictures or attaching doc files to display the chart in my response, but that URL should lead directly to it.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


Get every new post delivered to your Inbox.

Join 1,928 other followers