UCL Division of Psychology & Language Sciences

John Wells’s phonetic blog archive 1-15 July 2007



To see the phonetic symbols in the text, please ensure that you have installed a Unicode font that includes all the IPA symbols, for example Charis SIL (free download).


Browsers: I recommend Firefox (free) or, if you prefer, Opera (also free).


RSS feed for this site RSS feed

John Wells

Friday 13 July 2007

Stressed and unstressed schwa

Yesterday’s blog entry raises the further question: can [ə] on its own ever be stressed?

In a universal framework, yes of course it can. There are a fair number of languages around the world in which stressed schwa is attested, starting in Europe with Welsh, Romanian and Bulgarian. The Welsh for ‘mountain’ is mynydd [ˈmənɪð]; the Romanian for ‘apple tree’ is măr [mər]; the Bulgarian for ‘I am’ is съм [səm]. (No prizes for seeing the Indo-European connections here: Latin mons, malum, sum.)

You do occasionally get a phonetically stressed schwa in English through the restressing of a weak form, e.g. (be)cause [kəz], the adverb just [dʒəst] and going to, gonna [gənə]. But lexically there are no stressed schwas in English.


Tami Date reminds me that in Rebecca Dauer’s Accurate English she uses [ə] for the vowel in luck, cup, done etc. (= my lexical set STRUT). He asks,

Do you think Becky might have used the schwa in order to emphasize the point that the mouth should open only a little, not open a lot, for the sound in question?

— to which my answer was,

No. I think she was following many other American writers (notably Trager and Smith and their disciples) who claim that the STRUT vowel and the commA vowel belong to the same phoneme, being the stressed and unstressed allophones respectively. If they are co-allophones of the same phoneme, they should be written with the same symbol.
And in any case there are millions of Americans for whom there is no vowel quality difference between the two.

Just recently there have been two articles relating to English (weak, unstressed) schwa: in JIPA, Flemming and Johnson’s Rosa’s roses: Reduced vowels in American English, and now in Trans.Phil.Soc. Barry Heselwood’s Schwa and the phonotactics of RP English. (See here.)

Barry Heselwood’s ideas are particularly interesting, if you like to supplement phonetics by playing around with phonology. He sees most cases of English [ə] as being ‘anaptyctic’, i.e. absent from the underlying representation and inserted only to make a string pronounceable. Thus for example he thinks that abbot is /æbt/, but since [-bt] is not a possible cluster we break it up with [ə]. There are, however, many cases where this explanation does not work, e.g. in the case of support vs. sport, and for these he has a different explanation.

Thursday 12 July 2007

Oh, oh!

In the ‘quantitative-qualitative’ English transcription system introduced by Gimson in the sixties, which with minor adjustments remains our standard transcription today, there is just one change I sometimes wish he had not made. It concerns the symbol he introduced for the GOAT vowel, namely /əʊ/.

You can see why Gimson wanted to change the symbol that had been standard up until then, namely //. In mainstream RP the vowel no longer had a back rounded starting point. Rather, the starting point had become central and unrounded. It was this change that he wanted to symbolize: but there are certain drawbacks to his choice of symbol.

  • The letter ə had hitherto been used only for the weak vowel and the less prominent element of centring diphthongs. Using it for the more prominent part of a stressable diphthong confused matters, and still misleads beginners.
  • People tend to confuse the symbols əʊ (GOAT) and (MOUTH), both in reading and in writing, because they look too similar to one another.

Both of these problems would have been avoided if he had chosen to write the GOAT vowel as ɜʊ. This would also have appropriately signalled that the first element in GOAT is more or less identical in quality with the vowel in NURSE.

Even better, arguably, would have been to stick with . This symbol would have required some flexibility in interpretation, since the quality of the starting point is indeed nowadays some way from [o] (cardinal 7). (But we go on writing the STRUT vowel as /ʌ/, even though its modern quality is some considerable way from cardinal 14.)

If we had kept as the RP symbol, we would have been able to use the same GOAT symbol in both the accents taught to EFL learners, RP and GenAm. As it is, the use of different symbols implies a greater difference between BrE and AmE than really exists.

In my own teaching and authorship I have loyally continued to use Gimson’s system, including əʊ, because I feel strongly the importance of having a standard system that all reference books agree in using. Except for works edited by Clive Upton, that’s what we have, and I’m glad that we do. And even he uses the əʊ symbol.


Wednesday 11 July 2007

Any young U-RP speakers?

Jussi Wikström writes to quote from the Wikipedia article on RP:

Received Pronunciation must be distinguised [sic] from the Queen's (or King's) English, so named because it is spoken by the monarch. It is also sometimes referred to as BBC English, because it was traditionally used by the BBC. Yet, nowadays, this is slightly misleading. The queen, Elizabeth II, speaks an almost unique form of English.
... Today, overall, RP has three different forms: Conservative RP, Mainstream RP, Contemporary RP.
Conservative RP refers to a traditional accent which associated with older speakers and the aristocracy. This is known as "High British"...

As so often, Wikipedia is slightly wide of the mark. (I don’t like to criticize Wikipedia too strongly, because the obvious rejoinder is that I ought myself to set about revising the entries I find unsatisfactory.)

To me, ‘the Queen’s English’ is a somewhat dated synonym of Standard English: the standard dialect, not an accent. It is not the monarch’s personal idiolect, but a variety with millions of speakers.

I do not understand the proposed distinction between Mainstream RP and Contemporary RP (unless the writer thinks, wrongly, that ‘contemporary’ means ‘young’). And I have never heard anyone refer to High British (a Google search on '"High British" +pronunciation' throws up only some irrelevancies and the Wikipedia article itself.)

My correspondent goes on to ask,

Are there any younger speakers in the UK speaking the form of English that you would call U-RP? I have been trying to locate an answer to this question online, but have failed to find anything specific.

Personally, I incline towards a sociological definition of U-RP, as the accent typically used at any given time by those belonging to the English upper class (not the upper middle class) — the aristocracy, the peerage and landed gentry, exemplified (but not necessarily very typically) by the royal family. As such it changes over time, and has changed remarkably rapidly over my lifetime, as we can see by comparing the Queen’s pronunciation and Prince Harry’s.

So my answer to the question is ‘yes’. If we still have an identifiable upper class, then by definition we still have U-RP.

On the other hand, if my correspondent is really asking whether there are any young people today who sound like the Queen, the answer is obviously ‘no’. But then young people never do sound like their grandparents.

Tuesday 10 July 2007

A type of phonetics?

Dispina Kannaourou, a master’s student in Communication Design at Central Saint Martins, has sent me illustrations of her work, a piece entitled Fonotaip. She says,

My subject is British English accents. I work with typography and film and what I aim to do is illustrate accents typographically... I am constructing a typographic system that will illustrate the articulation through the letterforms. So that for instance GRas from Liverpool would look different from GRas from London but just using G, R, a, s.

Here are one or two excerpts.

You can see more here.

Dispina explains,

The title of this project consists of two words — phonetics and typography. I link the two disciplines and use the Roman alphabet to illustrate the sounds of the English language, in particular British accents. Fonotaip is a system that attempts a typographic interpretation of phonetics through illustrating and describing the theoretical assumption of speech production. The aesthetic quality of the letterforms derives from a grid system that represents a cross-section of the mouth. The system indicates the articulatory features that are believed to determine the quality of spoken vowels and consonants.

I don’t know what to make of all this. I fear I must be too much of a philistine to appreciate it properly.

Dispina Kannaourou is also one of the authors of a book called Lead Between the Lines, recently published by Gaffa Limited.

Now THERE’S something phonetically interesting. Is lead to be read as /led/ (metal) or /liːd/ (conduct)? If it’s about typography, I think it must be the first.

Monday 9 July 2007

American /nt/ reduction

Several correspondents have been in touch about /nt/ reduction (blog, 6 July) — or NT Simplification, as I called in Accents of English (vol. iii, p. 552).

Two of them were worried about my inclusion of the example gentle, wondering how this fits in with a rule that allows “the deletion of /t/ between /n/ and a weak vowel”.

Petr Roesel says,

The example gentle has a syllabic consonant after the nt-cluster and so the rule description should be expanded accordingly.

Masaki Taniguchi asks,

If you include gentle, then would it be between /n/ and a weak vowel or a syllabic l ...? Or do Americans have L-vocalization?

I regard syllabic [] as a realization of the underlying string /əl/. If this is right, then the rule holds as long as it applies before the syllabic consonant formation rule is applied.

underlying form /ˈdʒentəl/
nt reduction [ˈdʒenəl]
syllabic cons. formation [ˈdʒenl̩]
= surface form

But this is not without its problems. Before another syllabic consonant, [], Americans don't reduce /nt/ in this way. In words such as Clinton, accountant they tend instead to use a glottal stop: [ˈklɪnʔn̩]. (American /t/ tends to be glottal before a consonant in the next syllable.)

So the AmE nt reduction rule is a bit messy, because we have to allow it to operate before syllabic l (and syllabic r, as in winter), but not before syllabic n. But those are the facts.

Gentle giant: an Alaskan malamute

Masaki Taniguchi

The word continental /ˌkɑntəˈnentəl/ is a nice example. In the first /nt/ the /t/ becomes glottal, while in the second it disappears. So we get [ˌkɑnʔn̩ˈenl̩].

Here’s one way of doing the phonology, with some awkward rule ordering.

underlying form /ˌkɑntəˈnentəl/
syllabic n formation [ˌkɑntn̩ˈentəl]
t glottalling [ˌkɑnʔn̩ˈentəl]
nt reduction [ˌkɑnʔn̩ˈenəl]
syllabic l formation [ˌkɑnʔn̩ˈenl̩]
= surface form

I expect the Optimality Theory people have a better way of accounting for it. Any offers?

Friday 6 July 2007


Americans, as we all know, tend to pronounce twenty as /ˈtweni/ rather than /ˈtwenti/. That’s because they have an /nt/ reduction rule that allows the deletion of /t/ between /n/ and a weak vowel, as in winter, painting, center, counter, gentle, in front of and all other cases satisfying this structural description.

Londoners, too, often say twenty in this way. But they don’t lose the /t/ in winter, painting etc. (though they might make it a glottal stop). Londoners don’t say /ˈwɪnə/ rather than /ˈwɪntə/. In fact the only other cases for Londoners seem to be plenty, wanted, went (away), want to, trying to /ˈpleni, ˈwɒnɪd, ˈwen (əˈweɪ), ˈwɒnə, ˈtraɪɪnə/.

So we can say that for Londoners this is not a general phonological rule but a lexically specific one applying to just a few special cases.

And what about RP? In LPD I do mention the /ˈtweni/ form as a BrE possibility, but with a warning triangle against it. Yet I am aware that some people who qualify generally as RP speakers do use it, as well as (some of) the other London reductions mentioned. Ought I to remove the warning triangle?

Thursday 5 July 2007

Dumaresq is /dju'merık/ !

Isn’t the web wonderful? Within a few hours of my posting about Dumaresq (blog, 26 June), a reaction came from David Nash of the Australian National University. He says

Growing up in central NSW, I seem to recall "Dumaresq" in the daily recital of river heights on ABC Radio, and it was pronounced as you report. The stressed middle syllable is the same (in my "general Australian") as in the 2nd syllable in "America", contrasting with the lower vowel in "apparent".

Just to check, he also cc’d this message to Irene Poinkin, a linguist on the Australian Broadcasting Corporation’s Standing Committee on Spoken English, an expert on Australian placename pronunciation.

A day or so later she replied, confirming the pronunciation as /djuˈmerɪk/.

Then came a message from Jerome Poirrier, who after some research on Google tells me that Dumaresq seems to be a widespread name not only in Australia but also in Canada. He asks,

Might your 2nd Edition's transcription of "Dumaresq" with /er/ not have been reflecting precisely such a North American feature ?

Perhaps that is indeed the answer to the unexpected mismatch between spelling and pronunciation, now confirmed. Did the Dumaresq family emigrate to Australia from Canada? Historians?

David Nash

Irene Poinkin (left), and colleagues

Wednesday 4 July 2007

Medal for John Laver

Warm congratulations to John Laver, awarded a Gold Medal by the Royal Society of Edinburgh. It was presented to him in person by HRH the Duke of Edinburgh. Here is the citation.

Tuesday 3 July 2007


No, that isn’t a misprint for something else. Tokyo Sexwale is a leading South African politician.

In the Bantu languages of South Africa, the letter x can stand for at least three sounds, depending on the language: a lateral click, a palatoalveolar fricative or (as in IPA) a velar fricative. Conveniently for the phonetically unsophisticated speaker of English, the first and last of these map onto English /k/. So it would probably be safe to pronounce his name /seˈkwɑːleɪ/ (or perhaps /sɪˈkwɑːli/).

In the name of the language Xhosa the X stands for a voiceless alveolar lateral click, [ǁ] (or in old money [ʖ]). The following h shows that the accompanying velar component is aspirated, so in full we have [ˈk͡ǁʰɔːsa]. This spelling convention applies in Zulu, too.

In Tsonga, the letter x stands for [ʃ]. This is presumably due to Portuguese influence, since most speakers of Tsonga live in Mozambique.

But in Venda (aka TshiVenḓa) the letter x stands for the voiceless velar fricative [x].

I have no hard evidence, but I think Mr Sexwale’s background must be Venda (although according to Wikipedia he was born in Soweto). The name of the company he founded, Mvelaphanda, is certainly a Venda word.

And to confirm this speculation, the BBC Pronunciation Unit tells me, “Yes, the IPA for our entry [for Sexwale] indicates a velar fricative. The recommendation is based on the advice of colleagues in Focus on Africa, who, according to our history note from 1993, were adamant that the orthographic 'x' is pronounced as a velar fricative”.

So in English we should call him /seˈxwɑːleɪ/ or, failing that, /seˈkwɑːleɪ/.

Nelson Mandela and Tokyo Sexwale

Monday 2 July 2007


These days those who want to read their emails and browse the web on the move often do so by carrying with them a BlackBerry®. I suddenly started wondering: do we Brits pronounce the name of this device in the same way as we do the fruit blackberry, /ˈblækb(ə)ri/, or do we follow the Americans in keeping the -berry unreduced, and call the device a /ˈblækberi/?

As you would expect for someone of my age, I haven’t got a BlackBerry and don’t know any colleagues who have. So I consulted my youngest nephew, Robert, who is a corporate lawyer living and working in London. He confirms, “Yes, I've had one for several years, and most people I socialise with through work are as addicted to theirs as I am to mine!”

His first reaction was that “the kind of people who are most likely to use a BlackBerry in the UK are those who work in the international financial/legal/business world, and so national or regional accents tend to lose out generally to a more ‘mid-Atlantic’ accent that is more understandable and ‘acceptable’ to all.”

However he promised to carry out a quick preference poll among his colleagues. He offered them three possibilities:

  1. ˈblækberi
  2. ˈblækbəri
  3. ˈblækbri

— and reports, “Most people seemed to go for the second option for both the fruit and the hand-held device. A couple of people differentiated between the two, with the 2nd option for the hand-held, and the third option for the fruit. The only people to use the first option were non-native English speakers.”

So I needn’t have worried: in general we pronounce the hand-held in the same way as the fruit.


Rob Wells
Photo: Chris Weedon

current blog

Archived from previous months:

To search my web pages, use this Google search.

Google Custom Search

my home page