UCL Division of Psychology & Language Sciences

John Wells
Photo: Masaki Taniguchi

John Wells’s phonetic blog archive July 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


my home page

I was away from home for nearly three weeks, attending conferences.


Thursday 17 July 2008

Video clips for the English sounds

A correspondent writes that he is looking for a website that would offer video clips demonstrating each of the sounds of English.

Vague non-analytical, watch-my-mouth-say-a-word things abound... I still haven't found what I believe my students need. I would like them to have an on-line video reference on each symbol for the 44 [sounds of English].

I suggested that my correspondent try the BBC Learning English website. As far as I can see, this offers just what he wants.

I haven’t been through it all in detail, but spot checks suggest it’s pretty good.

There are clips for each individual phoneme and for assorted troublesome pairs.

The presenter for the video clips is Alex Bellem, who teaches general linguistics at SOAS and also has EFL experience. When the videos were made in early 2008 she was working in the BBC’s Pronunciation Unit.

A less satisfactory site is one from OUP China. Here you can listen to the sounds in isolation and in keywords. But there is no video support, so my correspondent’s requirements are not met. Some of the audio files are a bit flaky, too: try the one for θ to see what I mean.

Alex Bellem


Wednesday 16 July 2008


I did five BBC radio interviews on Monday, sparked off by a press release from the mail-order fashion company Littlewood’s Direct to the effect that their latest set of television advertisements features Esperanto.

The first was interview was with BBC R5Live, who phoned me before 07:30 and wanted immediate comment on something which at the time I knew nothing about. So they got it, for what it was worth.

The BBC World Service Newshour programme was more on the ball. Inviting me to go into the studio at Bush House around 13:30, they first obtained a copy of the ad from Littlewoods Direct and emailed it to me.

The ad features a strong-yet-sexy tribal princess from some distant island culture. A plane has crashed, or at least made an emergency landing, on the beach, and the princess seems to be summoning her (surprisingly white-looking) tribe to help themselves to the cargo windfall.

One of the BBC World Service’s assistant producers knows some Esperanto, and asked me whether the princess’s language was actually Esperanto, because she had her doubts. I listened, and agreed with her: it isn’t. It sounds like no language known to me. Duba!, the princess cries, and the subtitle reads “Behold!”. (In Esperanto that would be Vidu! or Rigardu!. Meanwhile duba means ‘doubtful’.) Can anyone out there identify the language? Or is it (as I suspect) something cobbled together ad hoc?

You can watch it for yourself.

The BBC asked the Littlewoods press officer to comment on this. For some hours, apparently, he couldn’t get hold of anyone in authority to explain the matter, but finally, just before I was due to be interviewed at Bush House, the company issued a further press release to the effect that the timing of the earlier press release was a mistake, and that the ad in question would not be aired till next week.

If you go to the company’s website page where you are supposed to be able to watch the ads, you find that two other ads are available for watching, but that the one we are interested in has become unavailable and is “coming soon”. (Fortunately it made it to YouTube before being withdrawn, which is why I can nevertheless show it to you above.)

Probably Littlewoods are now desperately casting around for an actor who can dub the stuff into proper Esperanto. I suppose it’s all good publicity.


Tuesday 15 July 2008


When I was a boy I voraciously consumed anything and everything that I could find that was to do with language. So you won’t be surprised when I tell you that around the age of ten, with my father’s help and encouragement, I learnt fingerspelling.

Also known as the ‘manual alphabet’, this is a conventional way of indicating letters of the alphabet with your hands. It is used to or by the hearing-impaired to spell out the names of people or places that do not have recognized sign names.

Unlike the American version, the British fingerspelling system uses both hands.

Decades later, when I was at UCL, I tried to learn a bit of British Sign Language. (Although fingerspelling is used alongside BSL, the full language is quite different.) But by that time my ability to learn new stuff had deteriorated so much that I didn’t really get anywhere with it. Now, decades later again, BSL is a respected academic discipline: at UCL we even have a professor specializing in sign language. Prof. Woll’s book The Linguistics of BSL: an Introduction (CUP), with Rachel Sutton-Spence, was the winner of the 1999 Deaf Nation Award and 2000 BAAL Book Prize.

The reason for this posting is that recently I was at a function held in a local authority hall in north London. On the wall, from some other function, was a fingerspelling chart. I found that I remembered it all perfectly (chiefly, as a muscular memory rather than an intellectual one), and took great pleasure in teaching it to a young relative who was there. We had fun signing slowly back and forth to one another.

One of my Facebook friends has a partner who is profoundly deaf and uses BSL. Facebook has an application that allows you to send messages in fingerspelling. So I have sent him a message and had a reply. (Of course, it’s a bit pointless really, because most deaf people can also process the ordinary alphabet perfectly well.)

I get the impression that the chart in the sidebar (from Wikipedia) is for right-handers, but the one below for left-handers. Do you see what I mean? Or perhaps it just doesn’t matter which hand you do the pointing with.

Prof. Bencie Woll


Monday 14 July 2008

A little while

Masaki Taniguchi has been reading the Bible, and asked me about the likely intonation of John 16:16.

In a little while you will see me no more, and then after a little while you will see me.

His suggestion, which he asked me to comment on, was

In a 'little \/while | you will 'see me no /more, | and then \/after a little while | you \will see me.

I told him that this was fine, except that I think I would use a fall rather than a rise on more, since that is the end of Christ’s first declaration.

In a 'little \/while | you will 'see me no \more, | and then \/after a little while | you \will see me.

The reason for the choice of nucleus placement in the second half is that a little while and see me are repeated (old, given) material, which in English means they are likely to get deaccented.

I noted that the wording he quoted from was the New International Version. In the Authorized Version (AmE: King James version) that I was brought up on it reads

A little while, and ye shall not see me: and again, a little while, and ye shall see me.

I then found an American audio clip of the chapter from which this verse is taken. The reader chooses the intonation pattern

A 'little \/while | and you will 'not \see me. || A\/gain a little while | and you \will see me.

— which nicely vindicates Masaki’s suggestion, as amended by me.

Interestingly, in the Greek original there are two different words translated into English as ‘see’.

μικρὸν καὶ οὐκέτι θεωρεῖτέ με, καὶ πάλιν μικρὸν καὶ ὄψεσθέ με.
mikron kai ouketi theōreite me, kai palin mikron kai opsesthe me.
little and no-longer you(pl.)-look-at me, and again little and you(pl.)-will-see me.

The first word, θεωρεῖτε theōreite, means ‘look (at)’ rather than ‘see’, and is in the present tense. The second, ὄψεσθε opsesthe, is the ordinary verb ‘see’, and is in the future tense. The first verb is from the stem that underlies our modern word theory, the second from the stem that underlies (syn)opsis. These subtleties are lost in the English versions. They are also lost in the Latin vulgate:

Modicum, et jam non videbitis me; et iterum modicum, et videbitis me.

— with videbitis (you will see) in both places.

A literal English translation would indeed be awkward.

A little, and you no longer look at me; and again a little, and you will see me.

I think it might affect the intonation, too.


Friday 11 July 2008

Pwnage for n00bs

Question: what word not to be found in standard dictionaries gets more Google hits than velar (3.4m) and over a hundred times as many as ejective (34k)? Answer: pwnage (3.6m). What other word not to be found in standard dictionaries gets three times as many hits as even that? Answer: n00b, usually spelt with two zeroes in the middle (9.6m).

You can visit the on-line noobstore, which offers for sale “official pure pwnage merchandise”.

Pure Pwnage is a Canadian TV show.

In the blog for 20 June we discussed the verb to pwn, noting that it originated as a mistyping of own, and is mostly pronounced just like own, or with an additional initial p.

But pwn has taken off on its own trajectory both of semantics and of morphology. In meaning it involves dominating or humiliating a rival, particularly in on-line gaming. In word formation it has spawned the abstract noun pwnage (where there is no comparable *ownage). [PS. Kilian Hekhuis points out that this is not true, and that ownage gets over 6 million Google hits. About a third of these actually relate to the two-word phrase own age, I think, but that still leaves 4 million.]

A n00b is a newbie, newcomer, beginner or neophyte. Obviously pronounced nuːb, this word betrays its north American origin (since Brits would have chosen a spelling newb or ny00b or something).

Here’s the text from the Pure Pwnage screenshot above, enlarged a bit so that you can test yourself on your knowledge of current trends in English spelling and usage. Don’t ask me for help, grab a passing teenager. Otherwise you’ll have been pwn3d, you n00b.



Thursday 10 July 2008

Assessing oral performance

Sergio Verdejo is an EFL teacher and phonetics instructor in Santiago, Chile.

On Facebook recently I read that

Sergio is trying to come up with a really systematic evaluation system of oral performance in the phonetics can't be arbitrary!

I sympathize with him. It’s certainly very difficult to be other than impressionistic when evaluating oral performance in a foreign language. I asked him

Do you test them on transcription? In my experience many Spanish-speaking students don't even know which sounds they’re meant to be aiming for, so of course don’t make the proper distinctions match-much-march, seat-sit, law-low etc.
Transcribing a passage of orthography into IPA can be marked objectively. It would also show whether the student is aware of weak forms and other characteristics of connected speech. On the other hand it might not evaluate skills in fluency, pausing, intonation etc.

Only yesterday I heard a radio interview with the new manager of Chelsea FC, Luiz Felipe Scolari. (For the next few days it’s available here.) He is Brazilian, but speaks English well enough to give a press conference in the language. You would expect him to know at least how to pronounce the name of the team he manages, Chelsea. But he thinks it is ˈtʃelsia, no doubt misled by the -ea in the spelling. He knows how to spell the name of his team, but not how to say it aloud.

I think that ignorance of the correct phonetic constitution of a word (i.e. which sounds it should contain) is one of the commonest errors in EFL, and ought to be one of the most treatable.

In his reply to me Sergio mentioned he had been thinking

about what kind of result we would actually get if, for instance, a number of phonetics instructors had to evaluate one single piece of utterance. Evaluation might be quite impressionistic, but what would you think if the very same recording of a NNS student yielded both good and bad results?

To which all I can say is yes, it’s fraught with difficulties.

Now I read that Pearson, the company that owns the publisher Longman, has acquired the company that produces Versant Tests, software which is claimed to

measure speaking and listening language skills. ... [It] helps organizations quickly, objectively, and accurately test spoken language.

I wonder how true this hype can be. Does anyone have any experience of these tests? If it’s so difficult for human beings to do it well, I would expect a computer program to do it worse.

Sergio Verdejo


Luiz Felipe Scolari



Wednesday 9 July 2008

International intonation (ii)

What, then, can we say about tonicity? By this term I mean the location of sentence accents and particularly their use to signal focus. Two principles that apply in English and seemingly in many languages are that the last sentence accent within a given stretch (the intonation phrase) is particularly important — which is why we call it the nucleus or tonic — and that it signals the material on which the speaker chooses to focus, i.e. mark as foreground as against background, new as against given, comment as against topic, rheme as against theme.

What is not clear is that either of these principles is actually universal. Unless contrastive focus is involved, it is not clear that all languages impose a focus at all (it’s been claimed that Danish is like that, not to mention French). And intonation is not the only way to signal focus, you can also do it syntactically.

This presumably is why nucleus placement is arguably the most complicated part of English intonation, and the hardest to learn.

It is not difficult, in simple cases, to understand the different meanings associated with switching narrow focus around.

A: What did you see? A red car?
B: (i) No, | a red bus.
B: (ii) No, | a green car.

But here the focus is obvious from the wording.

In English we readily exploit tonicity to mark focus in examples like this:

A: I want to buy a comb.
B: But you’ve already got a comb!

Yet this doesn’t seem to happen, or not as much, in African English.

In English we are so anxious to avoid accenting repeated items that we say

(Would you like your coffee) with milk | or without milk?

But in Spanish, I gather, this would be

...con leche | o sin leche?

In languages with a freer word order than English you can often use a change of word order to show focus instead of, or as well as, intonation.

In an interesting experiment that she has not yet written up for publication, Amanda Cheung, when an MA student at UCL, tested English people and speakers of Cantonese on whether they could correctly match up question and answer in pairs such as this (quoted from memory, to the best of my recall):

Q1. Have you any special dietary requirements?
Q2. Would you prefer beef or pork?
A1. I don’t eat meat.
A2. I don’t eat meat.

(A1 goes with Q1, A2 with Q2.)

The English score was near perfect, the Chinese no better than random. So no universal here.

Daniel Hirst and Albert di Cristo contributed an introductory article ‘A survey of intonation systems’ to the book they edited, Intonation Systems (CUP 1998). Their first sentence ends:

...intonation is paradoxically one of the most universal and one of the most language specific features of human language.

We can all agree with that.


Tuesday 8 July 2008

International intonation (i)

Most learners of English as an Additional Language (the latest expression for EFL/ESL) are not taught intonation and do not study intonation. Yet they do not speak English on a monotone. A few may be gifted mimics who succeed in imitating intonation along with everything else in the phonetics of the target language. For most, though, their intonation patterns are presumably those of their first language, transferred to English.

The same applies to English learners of foreign languages.

On the whole, even though this may make the speaker sound strange, typical of their origin, boring, or annoying, it seems not to cause much of an actual breakdown in communication. How can this be?

It must be because the principles of intonation in language are sufficiently universal for us to be able to rely on them even in a foreign-language situation.

However it may also be the case that many of the nuances of meaning carried by intonation in the speech of native speakers are overlooked or misinterpreted by some or most non-native speakers. But this may also apply between different native varieties of the same language: think only of Ulstermen interacting with Londoners.

This leads me to ask which parts of intonation are universal and which are language-specific. I am referring only to the linguistic (systemic) use of the pitch of the voice, not the paralinguistic (presumably non-systemic) factors such as pitch range, speech rate and voice quality.

Following Halliday, we can analyse intonation in terms of three systems: tonality (chunking, the signalling of syntactic boundaries); tone (e.g. fall vs. rise, signalling certain grammatical functions as well as such things as the speaker’s attitude to what they are saying); and tonicity (the location of sentence accents, particularly nuclear accents, mainly used to signal focus).

I think that the principles of chunking are probably pretty universal. No one needs to be taught to make or hear the difference between (1) and (2) in speech:

(1) I don’t know.
(2) I don’t, | no.

Clearly, tone varies wildly at the superficial level. The question is what underlying regularities there may be. We can dismiss obviously untrue claims such as that statements always have a fall, questions a rise. But there are other candidates for the status of universal that are worth considering.

How general is it that, as in English, wh questions tend to have the same pattern as statements, while yes-no questions are different? Pretty widespread, I think.

In most kinds of English, and perhaps in most languages, when we pronounce lists we signal that the list is incomplete by using a rise, and that it is complete by using a fall.

(3) /One, | /two, | \three.
(4) Do you want /coffee | or /tea | or \cola?

Is it a universal that the signal for list completion (if there is one) is always a falling tone?

Is it a universal that exclamations have a fall, pardon-questions a rise?

(5) \Marvellous!
(6) What lovely \flowers!
(7) A: I’ve just spifflicated them.
  — B: You’ve just /what? | /Splifflicated them?

I’ll leave the question of tonicity for tomorrow.

The reason I have been thinking about these matters is that I have agreed to give a talk on intonation at the forthcoming World Esperanto Congress in Rotterdam. What I said about intonation in EFL applies equally to intonation in Esperanto: somehow speakers manage to understand one another in the language very well despite the lack of any agreed, taught or described intonation system.

Malneto de mia prelego


Monday 7 July 2008

Jumieka langwij

On the 27th of this month I am due to fly to Cayenne in French Guiana (Guyane) for the biennial conference of the Society for Caribbean Linguistics.

So it seems a suitable moment to mention that a vigorous debate is currently in progress in Jamaica, sparked by the announcement that plans are afoot to translate the Bible into Jamaican Creole. Read about it here.

There’s a web page devoted to “Jumieka Langwij”.

Di hiem a dis sait a fi bring tigeda haxpek ahn suos a Jumiekan langwij fi chrai prizaabi fi paasteriti ahn fi di huoliip we lib a farin wid Jumiekan kanekshan uu maita hinchres iina dehn linguistik eritij. hUoliip a dem kuda gat pierans ar grampierans uu kiahn kot di patwa, bot dem siem wan kiaahn piik di mada-tong, ar els piik wahn luokalaiz verjan, laik Landan patwa.
The aim of this site is to bring together aspects and sources of Jamaican language in an effort to preserve it for posterity and for the many living abroad with Jamaican connections who may have an interest in their linguistic heritage. Many of these may have parents or grandparents who are or were patois speakers, but are themselves not fluent in the mother-tongue, or else speak a localized variant, such as London patois.

This spelling you see here is more or less the one used in Cassidy and Le Page’s Dictionary of Jamaican English (CUP 1967, second edition 1980). It differs by adding one or two more symbols: an optional italic h for the h that comes and goes before word-initial vowels, particularly after a final vowel in the preceding word, and hn to stand for what the web page author (not Cassidy and Le Page) calls a “soft, breathy n” (actually nasalization of the preceding vowel), as duohn duõ ‘don’t’.

This spelling is phonemic (‘ebri leta fi soun, aalwiez soun siem wie’) and uses the ordinary alphabet with no diacritics. Its main disadvantage for those unfamiliar with it and with IPA is the use of ai for the vowel of PRICE and of ie, uo (phonetically spot on) for FACE-NEAR and GOAT-FORCE respectively.

See the pronounsieshan & pelin gaid.

The late Fred Cassidy, the deviser of this spelling, was an enthusiast for the spelling reform of Standard English, too. When I was much younger, and not then involved in the Spelling Society, he invited me to dinner and urged me to campaign for the cause. His co-author, Bob Le Page, was the external examiner for my PhD.

You can read about the serious academic side of language in Jamaica — the Jamaican Language Unit of the UWI — here.

There’s a nice introductory video clip, too.

Fred Cassidy

Hubert Devonish
coordinator, Jamaican Language Unit


Friday 4 July 2008

Today’s puns

Our first pun comes from thelondonpaper.

Why was the biscuit crying?
Because its mother had been a wafer too long.

As we say in England, groan. If you must, you can use this joke to illustrate the importance of pre-fortis clipping, which does not operate across a word boundary (away#for). It also illustrates the importance of weak forms.

Jack Windsor Lewis will be delighted to know that one of his invented examples, illustrating the same point about pre-fortis clipping, has turned up in real life. BBC Radio Four has an assistant producer called Jo King. You must be Jo King.

David Deterding has a whole web page devoted to what he calls Phonetics through Jokes. This one was contributed by Karen Chung.

A man went to see a psychiatrist. "I keep on dreaming I'm a teepee or a wigwam," he said.
"I know the problem," said the psychiatrist. "You're two tents."

Our cartoon strip comes from the Washington Post via Language Log. It illustrates American t-voicing, and would probably leave many British readers at a loss.

© Copyright 2007 King Features Syndicate. All Rights Reserved.


Thursday 3 July 2008

Zheng Jie

Following the discussion of Serbian and Polish names at Wimbledon (blog, 25 and 27 June), David Deterding writes to say that he was

surprised you haven't mentioned the mangling of Chinese names, which seems particularly relevant at the moment with the success of Zheng Jie.

Why, he asks, do commentators

insist on saying something like ʒʌŋ ʒiː when something like dʒəŋ dʒeə is quite close to the proper pronunciation and perfectly easy in English?

In Chinese her name is written 郑洁, Pinyin Zhèng Jié, pronounced 4tʂəŋ 2tɕi̯ɛ (or you may find it easier to think of the distinctively unaspirated obstruents as devoiced dʐ, dʑ). Zheng is the family name, Jie her personal name.

A reasonable thorough anglicization would be ˈdʒʌŋ ˈdʒɪə, or as a halfway house perhaps ˈdʒəŋ ˈdʒiːe. (Those who have a monophthongal SQUARE vowel could pronounce the personal name dʒeə, i.e. dʒɛː, as David suggests.) So why do commentators not say something like that?

I’ve commented elsewhere on the problem of in foreign names. We are so brainwashed by the influence of French that we find it difficult to believe that any foreign language has an affricate rather than a fricative ʒ. Hence pronunciations such tɑːʒ in Taj Mahal instead of tɑːdʒ. Hence ʒ instead of in Zheng.

In the case of Chinese zh- this tendency is exacerbated by the prevalence of zh as a respelling for ʒ (see OBGP below) and as a romanization of Cyrillic ж (Zhukov, Zhivago).

David goes on to say

While we are thinking about it, the Beijing Olympics seems an excellent opportunity to persuade the BBC to get the pronunciation of Beijing right (ie with rather than ʒ in the middle). How difficult is it to use a perfectly ordinary sequence of English phonemes beɪdʒɪŋ in saying a name?

The BBC Pronunciation Unit people are well aware of how Beijing should be pronounced. Here’s what it says in their OBGP.

But it’s one thing to tell announcers and commentators what they ought to do, another thing to get them to actually do it.

Zheng Jie
Zheng Jie


Wednesday 2 July 2008


Neil Entwistle is an Englishman recently convicted of murder in the United States.

I was struck by the fact that two of the presenters on BBC television pronounced his name ˈenʔwɪsl̩. Others, like me, pronounce it ˈentwɪsl̩. Why the difference? Why my surprise? It᾿s not because I wouldn’t glottal a t between n and w: I am perfectly happy to say slantwise as ˈslɑːnʔwaɪz.

It’s all to do with syllabification. The medial consonant sequence ntw is one of the few such sequences that allow two phonotactically well-formed possible places for the syllable boundary, nt.w or

Like many English surnames, this one is derived from a placename. Entwistle is a village between Bolton and Blackburn in Lancashire, just a few miles from where my youngest brother’s family live. I’m also familiar with another place in Lancashire called Oswaldtwistle, and a Derbyshire village called Tintwistle. So it’s natural for me to see -twistle as a separate morpheme.

According to my syllabification rules, dubious consonants go with the more strongly stressed of the flanking syllables, unless there is a major morpheme boundary blocking this. In the first case we would have ˈent.wɪs.l, where the t is in an environment that favours glottalling. In the second case we would have ˈen#twɪs.l, where t cannot be glottalled because it is syllable-initial.

The announcer’s pronunciation of Entwistle is therefore what would be predicted in the absence of a morpheme boundary. Mine, on the other hand, is what is to be expected if there is a morpheme boundary between En- and -twistle.

I looked up the etymology of Entwistle in the Oxford Names Companion. Its origin is OE henna, water hen (or perhaps ened, duck) plus OE twisla tongue of land in a river fork. So the historical morpheme boundary corresponds to my contemporary perception. That explains my non-use of a glottal stop in this word (and also my non-use of pre-fortis clipping for the first vowel plus nasal). That’s why it’s ˈen.twɪsl̩.

PS. Judith Crompton tells me her mother was an Entwistle, and that her family pronounce the name as I do.




Tuesday 1 July 2008


The egregious Archbishop Akinola of Nigeria is reported as wanting to save the Anglican church from apostate leaders. I have not heard him personally say apostate aloud (and he now apparently denies that he said it); but when he was reported on BBC radio the reporter pronounced it as ˈæpəsteɪt.

However the only stressing of apostate given in dictionaries is əˈpɒsteɪt.

Historians know the word as the appellation of Julian the Apostate, Roman Emperor 355-363 AD. In the modern world, ‘apostate’ is a favourite word not only of intolerant evangelical Christians but also of Islamic jihadist extremists denouncing their political opponents.

Why, though, is it standardly pronounced a'postate, with the stress on the penultimate? Why doesn’t it follow the antepenultimate pattern of other trisyllables in -ate such as 'delegate, 'vertebrate, 'reprobate? I am not sure what the historical answer is, but it must be something to do with the fact that unlike them apostate does not contain the Latin participial ending -āt(us), in which the vowel is long. It comes from Greek, and consists of a prefix apo- and a stem -stăt- with a short vowel. Literally, it means one standing (stat) away (apo) from something. In classical Greek ἀποστάτης apostatēs meant a runaway slave or a deserter.

Presumably this Greek origin is also the explanation for the unexpected spelling of the corresponding abstract noun apostasy əˈpɒstəsi (Greek ἀποστᾰσία apostasia; compare stasis).

The same applies to ecstasy, fantasy and idiosyncrasy, which are also Greek-derived. Confusingly, it doesn’t apply to Greek-derived democracy, theocracy and other -cracy words, which have -κρᾰτία -kratia in Greek and follow the familiar spelling pattern of Latin-derived accuracy, celibacy, delicacy, candidacy, profligacy etc.

Julian the Apostate

Blogroll links:


To search my web pages, use this Google search.

Google Custom Search

my home page