UCL Division of Psychology & Language Sciences
john Wells

John Wells’s phonetic blog archive 1-15 April 2008

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

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

RSS feed for this site RSS feed



Tuesday 15 April 2008

A cab, innit?

There was a nice story in Friday’s London freesheet Metro.

[A] girl hurriedly dialled directory inq­uiries to book a taxi from her home in London to Bristol airport, using the cockney rhyming slang Joe Baxi.
But the operator told her they were unable to find anyone by that name. Seething, the youngster snapped back: 'It ain't a person, it's a cab, innit.'
The operator duly gave her what she asked for and put her through to the nearest supplier of cabinets.
[The] cabinet saleswoman seemed equally confused.
'Look love, how hard is it?' she fumed. 'All I want is your cheapest cab, innit. I need it for 10am. How much is it?' The sales adviser told her £180. The tantrum-throwing teenager quickly left her address details before ringing off.
The next morning, rather than being picked up by a cab, the young woman had the cabinet dropped off.

The segments would have been the same in each case: kæbɪnɪʔ. Theoretically the two possibilities ought to have been disambiguated by prosody.

(1) It’s a \cab, | \innit?
(2) It’s a \cabinet.

(1) would be expected to have a longer æ than (2), and to have a second accent on ɪn of innit (= isn’t it). The æ in (2) would be expected to have shorter duration because it is subject to rhythmic clipping, aka foot-level shortening, due to the following unstressed syllables.

As we know, you can’t always rely on prosody.

Actually, though, the more I think about this story the less plausible I find it. It was probably invented by PR people working for the cabinet-makers. The only person actually named in the story is their marketing manager.

This is a cab, innit?

This is a cabinet.


Monday 14 April 2008


I met an old friend at IATEFL in Exeter: my eminent colleague David Crystal, whom I’ve known for over forty years, and his son Ben, whom I have met just once before. Prompted by this encounter, I was reading an on-line interview with them that I found on the internet.

In answer to a question about what it’s like writing and performing with other members of the family, David is quoted as saying

Ben trained as an actor, I've been in an amateur repartee company for many, many years...

For someone as verbally dexterous and quick-witted as David, perhaps that explains a lot. Repertory, repartee.

Actually, repertory is phonetically quite an interesting word. It’s OK for the Americans, since they maintain a strong ɔː vowel in the -ory ending. But we Brits weaken it to schwa, which leaves weak vowels in three successive syllables: ˈrepətəri. As usual, the penultimate schwa is subject to possible disappearance through compression, giving just ˈrepətri.

So perhaps it’s not surprising that it sometimes gets misheard.

Ben and David Crystal


Friday 11 April 2008

With some hesitation

I hope you don’t think it too vain of me if I show you a picture of the signing session yesterday at the IATEFL meeting in Exeter, where I am sitting at a table by the Pearson Longman stand in the book exhibition autographing a copy of LPD3.

* * *

Jürgen Trouvain writes

Reading [yesterday’s] blog on Richard Cauldwell's talk on fluency and filler words I wonder whether you have a collection of English orthographies for "um, uhm, er" and so on.
Interestingly, in other languages than English and German (and Dutch and French as well) other words got a filler function, in Japanese it is ano and eto (I think - this information should be confirmed by a native speaker). In Hungarian it is said to be tehát and ilyen. Maybe you are aware of other "ehm-words" in the languages of the world?

Well, not really, but I’m sure some of my readers are.

According to my Japanese dictionary, in that language it’s not ano but anoo, i.e. with a long second vowel.

According to my German dictionary, in German it’s äm or hm.

We do say hm in English, too, (or hmm), but it is not exactly a hesitation noise. As a longish m̥mm, a voiceless then voiced bilabial nasal with a falling tone, it shows doubt or disagreement, meaning something like ‘I understand what you say, but I don’t think I agree with it’. As m̥m, with shorter duration and an abrupt fall, it shows mild surprise and means much the same as ‘Oh!’ or ‘Well, well!’.

Jürgen also sends me another excruciating pun.

Q: What’s the best time to go to the dentist?
A: Two thirty.

Perhaps that should be a (stereotypical) Chinese dentist.


Jürgen Trouvain


Thursday 10 April 2008

Um... er...

In his paper at the 2007 Phonetics Teaching and Learning Conference at UCL, Richard Cauldwell discussed the factors making for fluency in the language of trainee air traffic controllers.

His aim was to isolate the factors that led candidates to be judged as relatively fluent or non-fluent in their use of English. He found that to progress from Level 3 (non-passing, ‘pre-operational’) to level 4 (passing, ‘operational’) you would need to

  • increase the average speed of your speech by 30% (from 100 to 130 wpm);
  • make your speech units longer (words and syllables),
  • increase the percentages of medium and fast speech units by 20%;
  • reduce silent pauses to 15% of duration;
  • use fillers to fill pauses.

Your fillers should not be distracting. (The term ‘fillers’ here is applied not so much to expressions such as you know or sort of as rather to hesitation noises.)

To progress to level 5 (‘extended’) or 6 (‘expert’) his data suggests that you would need to further reduce silent pauses, to 10%, and eliminate or nearly eliminate ums and uhs.

Phonetically, um is most often əm. (To what extent do people say ʌm as use rather than mention?) The variant that Cauldwell spells uh is more usually written er in British English, since it is phonetically like a long schwa, ɜː. To my way of thinking, the spelling uh is appropriate only for American ʌ.

In BrE we also have erm ɜːm.

At the BAAP Symposium in Sheffield last week there were, I am sorry to say, several speakers whose fillers were definitely distracting and indeed annoying. I think public speakers, including lecturers and conference presenters, ought to train themselves not to um and er.

Richard Cauldwell


Wednesday 9 April 2008


Today sees the launch of the third edition of my Longman Pronunciation Dictionary at the IATEFL Conference in Exeter. This afternoon I give a presentation about it (with co-presenter Gerald Kelly), and tomorrow we have an author signing session.

Stocks do not yet seem to have reached Waterstone’s bookshop near UCL — and their website shows it as available only for pre-order. Amazon seems better: when I checked yesterday, had it for immediate delivery. On the other hand said ‘in three to five weeks’, would take an advance order with no delivery date, and (the American site) had no mention of it at all. Be careful not to order an earlier edition by mistake!

I hope the new edition will be well received. I spent all of last year working on it, not only adding three thousand new words and carrying out a new pronunciation preference poll, but also spending close on 200 hours sitting in a recording studio monitoring actors, British and American, recording spoken versions.

Whereas the first edition (1990) started life as a WordStar file, and the second (2000) as a Word 6 file, this new edition is based on a properly structured Unicode XML file, with every element appropriately tagged (MAINPRON, VARPRON, AMEPRON, FORPRON etc.). To work on it I had first to familiarize myself with editing software supplied by the publisher, Pearson Education.

The CD-ROM supplied with the dictionary not only contains the entire text of the dictionary and spoken versions of each headword and most compounds, but also allows you to record and play back your own pronunciations. And there are a Self-Study Lab (interactive self-testing exercises) and a Teachers’ Resource Centre (worksheets for teachers to use and suggestions for classroom activities). These were written by Sue Maingay.

By clicking a button marked ASK PROFESSOR WELLS, users can send me an email query. I hope this doesn’t mean I shall be utterly swamped with questions.


Tuesday 8 April 2008

Changes in English vowels

Another interesting presentation at BAAP was an update on the DyVis project at the University of Cambridge. This project concerns voice identification in forensic cases. One line the researchers are pursuing is between-speaker differences in segment transitions, another is differences in sounds “undergoing a rapid change in pronunciation over time”. The Sheffield talk, given by Francis Nolan, was entitled ‘Sound change and within-speaker variability’.

Here is their formant plot showing changes in certain vowels of ‘Standard Southern British English’, comparing the speech of elderly people, born 1928-1936, with that of twenty-somethings born in the 1980s. (You can find this chart in an article by Gea de Jong, Kirsty McDougall, Toby Hudson, and Francis Nolan entitled The speaker discriminating power of sounds undergoing historical change: a formant-based study. It appeared in the Proceedings of ICPhS Saarbrücken, 2007, and can also be found on the DyVis website.)

Francis Nolan

You will see it confirms the vowel changes that people have been commenting on for some time:

  • the opening/backing of æ (TRAP), and
  • the fronting of (GOOSE) and ʊ (FOOT)
  • — and also to some slight extent a fronting of ɔː (THOUGHT).

What a lot can happen in fifty years!

The vowels (FLEECE) and ɑː (START), on the other hand, have remained static.


Monday 7 April 2008

A voice from the past

At the BAAP Colloquium in Sheffield last week Michael Ashby played us a fascinating set of recordings of phoneticians from the early part of the last century. He has now kindly made one of them available to me: a short recording of the voice of Ida Ward, made in 1926 or soon after.

You can hear part of it here.

'Once upon a /time |
there was a /king |
who 'failed to 'please his \subjects |
and 'was in \/consequence |
in 'instant \peril. ||

Ida Ward was born in 1880 in Bradford, the daughter of a Yorkshire wool merchant. She joined the staff of Daniel Jones’s Phonetics Department at UCL in 1919. Her Phonetics of English (Ward 1929) includes numerous phonetic examples not only from RP but also from Yorkshire and Cockney pronunciation, which she claimed to know well. (These details are taken from Collins and Mees, The real Professor Higgins, Mouton 1999.)

She is further known academically for her work on speech defects (as they were then called) and also on the West African languages Efik, Igbo and Yoruba. As a student I learnt a lot from the book she wrote jointly with Diedrich Westermann, Practical phonetics for students of African languages (OUP 1933). She died in 1949.

On the recording she sounds quite incredibly old-fashioned by today’s standards. Listen out for the striking lack of aspiration in time and its very front diphthong. Listen too for the tapped r, the ɪ (rather than the ə we should probably use now) and rather clear ‘dark’ l at the end of peril.

Given her background, she may well have been a speaker of ‘adoptive’ RP rather than a native speaker like Jones. Perhaps she overdid things a little. On the other hand recordings Michael played of her colleagues Lilias Armstrong and J.R. Firth are just as remarkable, and very different from anything one would hear nowadays.

Michael comments that

by the standards of the time, [she and Firth were] lower middle class at best — and probably adoptive speakers of RP. But where, in that period before radio, did adoptive speakers hear their models? In the theatre? In church? In elocution lessons? And surely both were far too acute observers to adopt silly hammed-up versions of RP? And if this kind of RP was an aberration, why didn't Jones notice?

Even about quite recent history we can but speculate.

Michael Ashby

Ida Ward


Friday 4 April 2008

Vista fonts: one final instalment

Having completed my investigation of the fonts that came with Windows Vista on my new computer, I can now report on the expanded versions of Courier New and Tahoma. Like Arial, Times New Roman, and Microsoft Sans Serif, they have been extended to cover all the IPA symbols, including diacritics.

Here is what the phonetic symbols look like in the extended Courier New.

Not bad at all! Of the IPA symbols, only the upsilon (U+028A) is unsatisfactory, being wrongly shaped as an inverted small cap omega, just as is the corresponding character in Times New Roman (blog, 12 March).

And here is Tahoma.

The only symbol wrong here is the iota (U+0269), which is missing its curly bottom, just as in Arial (blog, 12 March). This is the synonym for ɪ (U+026A) that the IPA adopted in 1943 but discontinued in 1989.

There must be something about Latin versions of Greek letters.


Thursday 3 April 2008

Marcel Berlins and Sarkozy

The columnist Marcel Berlins pronounces his name ˈbɜːlɪnz, and would like everyone else to do so too.

In yesterday’s Guardian he makes various disparaging remarks about the BBC Pronunciation Unit. (By now you must have noticed which daily paper I get.)

He says he complained to the Pronunciation Unit about the

widespread mangling of the president of France’s surname. It is very definitely not Sar-coe-zy, with the stress on the middle syllable. Indeed some reporters hardly bothered with the Sar bit, making him sound like the thing you put over a teapot to keep it hot. The three syllables of his name should be stressed pretty much equally. [... ] I assume that the pronunciation unit provided the right advice. But what’s the point, if those speaking disregard it?

So the poor old pronunciation unit is condemned whether or not it provides advice Berlins approves of, blamed for the supposed iniquities of the announcers.

It is ridiculous to suggest that we should pronounce all three syllables of Sarkozy with equal stress: we’re speaking English, a language in which no non-compound three-syllable word has three stresses. (In compounds, perhaps OK: deep-sea-blue.) I assume Berlins is really calling for final stress, which is what with our English ears we tend to hear in French words: saʁkoˈzi.

I expect American announcers pronounce it with final stress. But the BBC is not American.

The name is originally Hungarian, and Hungarian has initial stress. So that could be seen as justifying stress not on the last syllable but on the first. But only an out-of-touch pedant would demand retention of the Hungarian stress.

In British English we tend to give French words penultimate stress. That’s how we mostly say cliché, cachet, idée (fixe), Calais, Orly, and many others. It’s how many of us would stress Chirac.

So I think it’s fine in BrE to give M. Sarkozy stress on the ko.

Now, Mr Berlins, what about the neglect of the r-sound in the French pronunciation? Do you think we ought to go against our non-rhotic customs and make an effort to pronounce it? As a uvular? I bet you don’t even pronounce it yourself.

* * *

P.S. Harry Campbell points out that in Hungarian the name is actually Sárközy, so that the middle vowel “ought” to be ø, as if spelt eu in French.

Since [this vowel sound] would present no problems to a French speaker, the question is, when and how did his dots fall off?

In English English this Hungarian vowel would normally map onto ɜː (the NURSE vowel), not əʊ (the GOAT vowel). So should it be sɑːˈkɜːzi? I think not.

The Hungarian spelling also implies that the initial consonant should be not s but ʃ.

Marcel Berlins

Nicolas Sarkozy


Wednesday 2 April 2008

If...the Queen

Here is Steve Bell’s If... cartoon from yesterday’s Guardian.

As the diminutive President Sarkozy and his ultra-chic wife came to dinner with the Queen and Prince Philip, our Prime Minister Gordon Brown, also invited, arrived late, allegedly because of getting lost on the way to the Palace.

The Queen may well pronounce lost in the old-fashioned U-RP way as lɔːst, shown by the spelling lorst, but she surely doesn’t say got as gɔːt instead of gɒt.

Gordon Brown in turn is always caricatured as speaking broad Scots, as shown here in the spelling tae, i.e. teː, for to. It also seems to be implied that he misheard morning (dress), in which as a rhotic Scot he would have an r, as maroon.

But the point of interest is the representation of the Queen’s vowels. She has the now obsolescent variant of the GOAT vowel involving a front starting point, so (or something similar but with less rounding on the second element). This is caricatured here by the spellings hellay (hello), lair (lower) and dane’t knay (don’t know). Like Prince Charles, she seems to have little backness and lip rounding at the end of her MOUTH vowel, hence Brine for Brown. Furthermore she uses (or is perceived by the cartoonist to use) a very open endpoint in her SQUARE vowel, so wah (wear).


Tuesday 1 April 2008


NEW! Thanks to new software I picked up at BAAP, as from today you will be able to download this blog as a podcast for your iPod, to study at your leisure.

* * *

Here’s a joke for your EFL class (or your seven-year-old).

How do you weigh a whale?
— You go to a whale-weigh station!

* * *

And here’s a nice mispronunciation recently heard from a BBC R4 newsreader. He wanted to refer to the newspaper City AM, given away free every weekday morning to London commuters. Its name derives from City (= financial centre) and AM (a.m., = morning). But the newsreader didn’t realize this. He called it ˌsɪti ˈæm.

current blog


To search my web pages, use this Google search.

Google Custom Search

my home page