Wednesday 30 June 2010

tittles and jots

It is only just over two months since I last discussed the question of foreign “accented letters” in ordinary spelling. We need them for loanwords (façade, cliché) and in particular for foreign names (congrats to the Guardian for getting Žižek right, see above). They are perceived as problematic not only because we don’t use them in writing ordinary English, but also because not all software can cope with them (certain email clients being notoriously unable to deal with anything beyond ASCII), and because documentation on how to input them is often inadequate.

The current issue of the Economist has an interesting article on the subject.
OVER at Gulliver, our correspondent reports on his trip through Tromso airport. Or, as a commenter, Lafayette, notes, shouldn't we write it Tromsø, as the Norwegians do?
Our style book rule is to use the diacritic marks on French, German, Spanish and Portuguese names and words. The rest have to do without. Why?

It’s not difficult to input diacritics for those languages nowadays. Users of Word for Windows ought to be conversant with the keyboard shortcuts (see table). They’re just as easy to input in a Mac.

We need to remember that in German Günther is a different name from Gunther, and Köhler a different name from Kohler.

The Economist article was a follow-up to an earlier one that dealt particularly with the diacritics used in eastern European languages.
Estonia[n] has the õ, Latvian the ķ, Lithuanian the ų, Polish the infernally similar ż and ź, not to mention the ł; the Czechs have the ů, the Slovaks the ŕ and the Hungarians the ő. There are dozens of other examples, but you get the point. They tend to get overlooked.

East Europeans will have regained their real place in the world once their names are spelled properly, not mutilated by an inadequate foreign character set.

...But they matter. Estonia’s national anthem, for example, starts: “Mu isamaa, mu õnn ja rõõm” (My fatherland, my happiness and joy). Written in the western character set, “onn ja room” becomes something quite different: the comical "small hut and crawl".

The õ of Estonian happens to be needed for Portuguese, too, so it is covered by the Word keyboard shortcuts. But if ctrl-apostrophe e gives us é and ctrl-comma c gives us ç, why can we not use ctrl-apostrophe s to input ś and ctrl-comma k to input ķ? Wake up, Word!

Meanwhile, you will be glad to know that you can include any Unicode character lacking on your keyboard in your blog comments here, but only if you first compose your message in a Unicode-compliant word processor (e.g. Word), then copy and paste. In Word itself, failing implementation of the improvements I have suggested, you can use Insert Character, or alternatively enter the Unicode number and then do Alt-X.

Tuesday 29 June 2010


…talking of which (yesterday’s blog), why is the dental fricative in brothel ˈbrɒθl̩ voiceless?

There are two reasons why we might expect to find ð rather than θ in this word.

First, the apparent cognates in other languages have a voiced consonant. The French word is bordel bɔʀdɛl, the Italian is bordello borˈdɛllo, the German is Bordell bɔrˈdɛl.

Second, the spelling th in this position (following a stressed vowel in a disyllable) regularly corresponds to the pronunciation ð: father, mother, brother, other, bother, feather, gather, whether, rather, slither, smithy, swarthy, worthy, Swithin. True, there are no other words in -thel except the names Bethel and Ethel, which both have θ. Even so, the pattern is very clear. Words spelt with medial th and pronounced θ are otherwise just those of Latin or Greek origin: author, cathode, lethal, method. But brothel looks and feels Germanic.

The OED reveals the reason for the exceptionality of brothel. It’s really not a cognate of bordel at all. The etymology goes back to Old English.
ME. broþel, f. OE. broðen ruined, degenerate, pa. pple. of bréoðan to go to ruin
The modern sense arises from confusion with an entirely different word BORDEL (q.v.); the brothel was originally a person, the bordel a place. But the combinations bordel-house and brothel's house ran together in the form brothel-house, which being shortened to brothel, the personal sense of this word became obs., and it remains only as the substitute of the original bordel.

Monday 28 June 2010

from bitch to beach

Native speakers of English have no difficulty in hearing and making the difference between the FLEECE vowel and the KIT vowel. We effortlessly distinguish green from grin, reed from rid, and reach from rich. But anyone who has been in contact with NNSs knows that for many EFL learners this distinction can be very problematic. For most people from southern Europe, and from many other parts of the world too, English and ɪ both sound like versions of the i of their own language. They find it really difficult to hear any difference between them, and even more difficult to consistently make the difference when speaking English.

It’s not for nothing that in the US spick is a derogatory term for ‘Hispanic’.

It is interesting that the potential confusion can extend into written English. You’d think that even if hearing the difference is difficult, nevertheless seeing it and spelling it would be straightforward. The letters in G-R-E-E-N look different from those in G-R-I-N. But there is plenty of evidence that people get the spellings of such minimal pairs confused, too. Sometimes it can lead to awkward or embarrassing results. This, from the Engrish website, was the frontage of a hotel in Jurmala, Latvia. With this name, you might take it for an up-market brothel.I’m glad to say that they have now corrected it. This is from the hotel’s current website.
An unrelated point of linguistic interest: on the website, the hotel gives its address with the street name in Latvian (in the Latin script), as 23 Jūras iela, but the name of the town in Russian (Cyrillic), as Юрмала. Presumably most of their clientele consists of Russians, who wouldn’t notice the difference between a bitch and a beach anyhow.

Even so, why don’t they get their English checked?

The unique location in the dune area lets You enjoy peaceful relax... Be sure, that Your day can have different taste, if You wake up with a sunrise at the sea shore and breathe in fresh pine-wood aroma, walking along sandy beach!

Friday 25 June 2010

yanks and kiwis

A collaborative US-NZ research project being carried out at the universities of Yale and Auckland is collecting data on different accents in American English.
We'd like recordings from anyone who has grown up speaking English in America or Canada. We would like to get information about as many different types of American English as possible. The more diverse our participants, the more representative it will be of the ways Americans speak.

Everything is being done over the internet, and recordings can be made by anyone who has access to a computer with a microphone and a sound card.
Your computer’s internal microphone is usually good enough, or you can use an external microphone or headset.

The researchers say
We want to record speech examples from as many people as possible, but it is not feasible for us to record participants in person. That's why we've set up this web site. When you click on the link below, a web page will come up with a recording box and a set of questions to enter into the comments box for the recording. When you're ready to begin the survey, there will be a "record" button to click on. You can then read aloud the words for the experiment.
When you've finished the words, press the "stop" button and the "send" button. The whole survey will take about 3 minutes to do.

So the recordings will be restricted to words read aloud. Sociolinguists have demonstrated that people’s pronunciation in this formal “word list” style can differ considerably from what they do in less formal styles, and in particular from what they do in spontaneous conversation.
We know that New Yorkers and Bostonians are more likely to articulate a nonprevocalic r when reading lists of words aloud than they are in other styles of speech. What might be ˈfaːmə (farmer) in a New Englander’s ‘natural’ speech may become ˈfɑːrmɚ when read aloud. A typical casual bẽəd for bad reverts to bæd in careful speech.
Nevertheless, despite this limitation, something is much better than nothing. Go for it, American and Canadian readers!

Thursday 24 June 2010

accent intolerance

The call centre for my bank is evidently located somewhere in Scotland. At any rate, every customer service officer I have spoken to there speaks with a Scottish accent. I’m ashamed to say that sometimes I have had difficulty understanding them, particularly over a bad phone line and when they speak fast, no doubt parroting a script they have uttered many times before. I sometimes have to ask them to repeat what they’ve said, more slowly.
But I acknowledge that my difficulties in coping with the Scottishness of their pronunciation is a shortcoming of my own rather than of theirs.
Not everyone takes that line.
Today’s Guardian has an amusing article poking fun at the letters page of the Daily Mail (a sensationalist right-wing paper).

Its criticism is directed towards the letters editor of the Mail for printing a xenophobic rant about TV coverage of the World Cup. The writer of the letter had to sit through “a match between Bongo Bongoland and the Former Soviet Republic of Bulimia and other meaningless events”. In mitigation (?), the Guardian points out that the paper also saw fit to print “the tragic tale of a man from the Home Counties who has been forced to listen to a Northern Irish accent”.
I can't understand the popularity of the woman on The One Show on BBC1 each evening [Christine Bleakley]. Her accent is almost indecipherable. Lord Reith must be turning in his grave.

I agree that we English people tend to feel that from the point of view of intelligibility a Belfast accent is a Glasgow accent in spades. But it does not follow that Ulster people should be banned from the airwaves.

There’s a lesson here for EFL, too. Learners should not be exposed only to an unvarying diet of RP (or General American). Like NSs, they need to practise their listening comprehension for a wide range of varieties of English.

Wednesday 23 June 2010

Fred Cassidy and Jamaican

Fred Cassidy (1907–2000) is best known among linguists as the moving force behind the Dictionary of American Regional English. A decade after his death, this enormous work is still not quite finished. Publication of the fifth and final volume, S–Z, is expected next year.

But Cassidy had another string to his bow. Creolists and people interested in West Indian English know him as the joint author of the Dictionary of Jamaican English (CUP: first edition 1967, second edition 1980), and also of an earlier popularization, Jamaica Talk (London: Macmillan, 1961). I studied both in great detail when I was working on my PhD on Jamaican pronunciation in London. Cassidy’s DJE co-author, Bob Le Page, was my external examiner.

As well as being a remarkable work of historical scholarship, the DJE also introduced the world to the orthography Cassidy invented for Jamaican Creole. This can actually be seen as an English spelling reform designed explicitly for JC. It is a phonemic notation, and uses some English letter combinations in ways that are logically chosen but at first can seem surprising. For example, the Jamaican FACE vowel, although a monophthongal [eː] in educated or ‘acrolectal’ Jamaican, is an opening diphthong of the [ɪɛ] type in popular or ‘basilectal’ Jamaican. Cassidy therefore writes it ie. But this means that the reader has to be able to interpret pie as the word we usually spell pay, and bied as either bathe or beard (homophonous for most Jamaicans).

When I was working on JC, people — Jamaicans, particularly — thought that to study it was a strange and eccentric thing to do. The intervening forty-odd years have seen a revolution in attitudes towards the local language/dialect/patwa, and we have now reached the stage where a local company, with the support of the Jamaican Language Unit of the University of the West Indies, is putting out video clips in which the language formerly used only on informal occasions is now used formally. Here is a serious-minded account of Haiti and its troubles, all narrated in JC. Key sentences are flashed on the screen in Cassidy’s orthography.
A Fi Di Piipl is produced by the Jamaican Language Company in association with the Jamaican Language Unit. This video, presented and narrated in Jamaican (Creole), takes its theme from the January 2010 disaster in Haiti, and from the perspective of a neighbouring country, seeks to understand the history of Haiti and its role in the world.

In ordinary spelling and standard grammar, this would be Because after the earthquake CARICOM was doing a large amount (a whole heap) to help out the country. (Don’t ask me why h is preserved in whole but dropped in heap and help.) The name of the series, A Fi Di Piipl, means “it’s for the people”.

Some years before his death, Cassidy was visiting London and invited me to dinner at his hotel. His main purpose was to try and persuade me to become active in the movement to reform English spelling. I suppose it is partly because of this arm-bending that I am now president of the Spelling Society.

Tuesday 22 June 2010

advocating carbonates

The pronunciation of the suffix -ate is quite tricky from the EFL point of view. (NSs never think about it, of course.) The general rule is that it is strong, eɪt, in verbs, but weakened to ət (or, for some, ɪt) in adjectives and nouns.

So on the one hand we have celebrate, deteriorate, imitate, lubricate with eɪt, and on the other considerate, obstinate, private; climate, consulate, palate, senate with ət.

There are no verbs in which the suffix is weakened. Not for me, at least — though I do occasionally hear someone saying that he ˈædvəkəts this or that policy, which makes me want to elicit the ing-form or the past tense, since I really can’t believe that anyone would say ?ˈædvəkətɪŋ or ?ˈædvəkətɪd.

EFL learners need to be aware of the ambiguity of the spelling in words like moderate, separate, deliberate, delegate, which are pronounced differently depending on whether they are verbs or adjectives/nouns. An adverb like separately ˈsep(ə)rətli, of course, has a weak suffix like the adjective from which it is derived. In phonetics, we speak of affric[]ting a plosive but of producing an affric[ət]e. We hope to be articul[ə]te as we articul[]te.

Adjectives and nouns are trickier, though. There are a few adjectives that categorically have a strong vowel in the suffix (innate, ornate, sedate). There are many nouns in which usage is not settled (candidate, magistrate). In “more technical words”, as I put it in LPD, the suffix is strong: thus in botany the adjectives describing shapes of leaves (cordate, lanceolate, peltate) and in chemistry the names of salts (nitrate, phosphate, sulfate).
But even this is not fixed. The other day I was slightly startled to hear someone speak of carbon[ɪt] of soda. Perhaps I need to get out more.

Monday 21 June 2010


Rick Bidlack sent a message to Jill House and to me
I would like to bring some recent work of mine to your attention, because it involves a recording you made in 1995 called The Sounds of the IPA. I purchased a copy of that disk about 18 months ago, and that gave me the raw materials needed to realize an idea I'd been contemplating for almost two decades. I am a musician, an audio engineer and a computer programmer, and I've had a lifelong interest in language and linguistics. The result of that intersection of circumstance and events is now embodied in a program called Clang, which is available at, and yes, those are your voices that we hear in Voice mode.
So please download Clang and have a listen. For me this is a work in
progress. Version 2.0 has just been released (this is the first version with the IPA recordings), and I'm starting to work on the next version which involves orchestrated blocks of sound and long-term temporal structure.

Jill said
The way you have edited the sounds and generated the ever-changing stream of noise has resulted in some extraordinary sound worlds (reminiscent of everything from baby babble through heavy breathing to the sound of contented hens)! Now I've got to a bit that sounds like an echoing wind tunnel... It's quite fun trying to work out which sounds some of the bits are taken from. It's probably against the spirit of continuously-changing generation, but do you ever record stretches that you like because they achieve a particular effect?

What do you think?

Friday 18 June 2010

resolution of syllabic consonants

The other thing that Denis Lyons (blog, 16 June) told me he was concerned about is what he describes as the use of
a pronounced "t" sound for the more silent version (in words such as Britain, little, etc.) …
I think what he is referring to is actually a change in the treatment of the nasal or lateral that follows the t, namely the replacement of the syllabic n or l by a sequence ən or əl respectively.

Hence instead of a nasally released plosive in Britain ˈbrɪtn̩ (the ‘more silent’ version) we get an orally released plosive (the ‘pronounced’ version) followed by a schwa, thus ˈbrɪtən. Correspondingly, instead of a laterally released plosive in little ˈlɪtl̩ we get a centrally released plosive likewise followed by a schwa, thus ˈlɪtəl (or further development to ˈlɪto and the like).

The same thing happens with d in this environment, as in garden and middle, where alongside traditional ˈɡɑːdn̩ and ˈmɪdl̩ you can also hear ˈɡɑːdən and ˈmɪdəl, ˈmɪdo. It’s not primarily a change in the plosive as such but a change from a syllabic consonant to a schwa plus nonsyllabic consonant.

This tendency to avoid nasal/lateral release has indeed often been commented on. People have sometimes characterized the result as sounding ‘childish’.

There is also a third possibility with the t in words such as Britain, little. That is the use of a glottal stop ʔ rather than an alveolar t before the syllabic consonant, giving ˈbrɪʔn̩, ˈlɪʔl̩. The former often seems to pass unnoticed; the latter may attract attention.
Cruttenden says (Gimson’s Pronunciation of English, seventh edition, p. 180):
Use of [ʔ] to replace /t/ [sic] … before syllabic [n̩] … was until recently stigmatized as non-RP but … [is] now acceptable in London Regional RP.
He refers to Anne Fabricius’s article ‘Ongoing change in modern RP: evidence for the disappearing stigma of t-glottalling’, English Worldwide 23: 115-136.

Thursday 17 June 2010

Gösta Bruce 1947–2010

I was very sad to hear that Gösta Bruce, the President of the International Phonetic Association, died on Tuesday evening, after a short illness. He was only 63.

Gösta Bruce was professor of phonetics at Lund University, and was well known internationally for his research in the field of prosody, and in particular for his description of the intonation of Swedish. As is well known, this language has two word accents (tonemes). They interact with the sentence accents and boundary tones in interesting ways, which Gösta described in his 1977 doctoral dissertation Swedish Word Accents in Sentence Perspective.

I heard the news of Gösta’s death from Gunnel Melchers, who comments
He was a brilliant scholar and a lovely person.

Sidney Wood writes
We knew he was ill, and in hospital for cancer surgery, and had to miss this year's Swedish Phonetics Meeting last week, that it had been his and the Lund department's turn to organise. We were shocked to learn of his death just one week after the operation.

The funeral will be on the first of July.
We shall miss him.

Wednesday 16 June 2010

fronted GOOSE

Denis Lyons writes to express his worries about what he sees as recent unwelcome developments in English pronunciation.

One of them is the use of a close central vowel, [ʉː] or thereabouts rather than a fully back [uː], in words of the GOOSE set.

Long before this recent development, Daniel Jones pointed out, getting on for a century ago, that English [uː] is not as back as the cardinal-type quality that one hears in German. (Compare German du, or even French doux, with English do.) In my 1971 PhD thesis I commented on the backness of Jamaican [uː] as compared to what is usual in England. But it was Caroline Henton who first properly documented the new fronting of RP GOOSE, in her 1983 article ‘Changes in the vowels of Received Pronunciation’, JPhon 11:353-371.

Where does this development come from? As with most other sound changes, nobody really knows. But Jones pointed out in his Outline, nearly a century ago, that an ‘advanced’ variety of was used after j (see scan).
So perhaps what happened was essentially a takeover by this allophone, which became the default realization in all positions and thus displaced the backer variety used in words like spoon and food in conservative speech.

Since then things have gone further: today one can indeed hear unrounded mid or front varieties of GOOSE.
I’m fine. How are ˈjɨː?

Tuesday 15 June 2010


There’s one Zulu word we have all become familiar with over the last week or two: vuvuzela, the blaring plastic horn blown by South African football supporters.
Turn on your TV right now to whichever sports channel is showing the England's soccer game against the USA in the World Cup in South Africa. Turn the sound up. Why does it sound as if several dozen propeller-drived airplanes have started up their engines in the stadium? Has someone dropped one of the commentator's mikes into a huge beehive? No. It's just that South Africans love to bring annoying three-foot-long one-note plastic trumpets to every game and blow them continuously. (They all seem to be tuned roughly to A below middle C.) [Geoff Pullum in Language Log].

Harry Campbell draws my attention to the claim reported in the media.
A vuvuzela is tuned - to use the term loosely - at the B flat below middle C, and has a similar frequency to speech tones, says Trevor Cox, president of the Institute of Acoustics. This makes it particularly tricky for broadcasters to tune out, as to do so would dampen the commentators' voices - and not in a good way.
"I'm looking at its wave patterns and there are at least six very strong harmonics in there. It would sound really horrible to notch these out - if one coincides with the vowel sound e, you won't be able to hear the -es in the commentary. It would sound unnatural."

Er, well, yes. I’m not going to start messing around with spectrograms myself, but it is obvious that the vuvuzela racket covers quite a range of frequencies. It is not a ‘pure tone’ like a whistle, which would have a single frequency that one might filter out.
The fundamental note is reportedly “roughly [at] A below middle C” or “at the B flat below middle C” (accounts vary, as you see). Let’s say 230 Hz or thereabouts. That would imply harmonics (overtones) at integer multiples of that frequency: 460 Hz, 690 Hz, 920 Hz etc. This coincides with the range of vowel first formants (not just e, all vowels), which is approximately 250–1000 Hz. So if you filter out the vuvuzela noise, you’re going to lose information that could be important in keeping speech intelligible.
Nevertheless, here is a plugin that claims to do that. Feel free to try it.
I think a better way would be for broadcasters to separate the commentary feed from the crowd noise feed, and to suppress all or most of the latter. Except for people who like the noise of the vuvuzela.

Monday 14 June 2010

making clicks

In a response to Friday’s blog, Ryan asked
Do you have any tricks for teaching people to pronounce clicks? I'm okay with pulmonic and glottalic consonants, but with the velarics I can only seem to do simple, singly articulated dental, postalveolar, and lateral clicks, usually nasalized unless I try really hard. Forget adding a coarticulation like aspiration or voicing. I'm wondering if you have any insight into pedagogical techniques for acquiring such sounds.

Here is how I would teach people to do the Zulu clicks. (Given the font problem, I’ll use the old IPA symbols.)

1. Take the “simple, singly articulated” clicks as the starting point. Make sure that you can produce each of them on its own.

Zulu cIPA ʇdentallike tut-tut, tsk-tsk
Zulu qIPA ʗpostalveolara single, hollow sound
Zulu xIPA ʖlateralgee-up

2. Try to become conscious of the fact that while you were making these sounds you had a velar closure, as in the hold phase of k. Say ʇ, and after releasing the tip of the tongue think about the back of the tongue. You should find that it’s in contact with the velum. Likewise for ʗ and ʖ.
3. Now work on that simultaneous velar articulation. Take a word such as English bookcasekkeɪs, with a long (geminated) velar plosive. (If you’re Italian, try secco; if you’re Japanese, sekken せっけん (石鹸), etc.). Pause halfway through the long k plosive. Make a few clicks as you hold on to the velar closure.
4. Now, repeat the nonsense disyllable ɑkːɑ, first as it is, then with a click in the middle of the velar hold. Start slowly, then speed up. Reduce the duration of the velar hold until it is like an ordinary single k, but with a simultaneous click in the middle.
5. Lose the initial ɑ. Try to repeat with each of your three clicks. You have produced the Zulu syllables ca ʇa, qa ʗa, xa ʖa.
6. Pronounce the Zulu words cacá ‘be clear’, qaqá ‘undo’, xoxá ‘tell’. (Each is the imperative of a verb. The first syllable is low-pitched, the second high. Vowels in penultimate syllables are long and stressed.) You have now overcome the first hurdle, which is to make clicks in words, in running speech, rather than just in isolation.

7. Next, tackle the aspirated clicks. You have to release the velar k with aspiration. If you can already produce unaspirated k= and aspirated , this should be straightforward. Practise your three clicks with an aspirated velar release as you finish. You have produced the Zulu syllables cha ʇʰa, qha ʗʰa, xha ʖʰa.
8. Pronounce the Zulu words chachá ‘shell peas’, qhaqhá ‘cut open’, xhoxhá ‘jab’.

9. Next, take the voiced clicks. You have to pronounce these in the middle of a g instead of a k. Practise ɑɡːɑ (compare big girl), adding a click in the middle. Then reduce the length of the ɡ. Keep the voicing going throughout. When you have achieved this with each of the three clicks, you will have produced the Zulu syllables gca gʇa, gqa gʗa, gxa gʖa.
10. Pronounce the Zulu words gcagcá ‘marry’, gqâgqa ‘scatter’, gxugxúma ‘be nervous’. (á = high tone, â = falling tone, unmarked = low.)

11. Next, the nasal (“nasalized”) clicks. (Some people, like Ryan, find these easier than than the plain ones.) They are accompanied not by k or g but by ŋ. Make a long ŋː, then put a click in the middle. (Keep the voicing going). I’m assuming your previous phonetic training has taught you to produce syllable-initial ŋ — just check that you can do ŋa. Add a simultaneous click, and you have produced the Zulu syllables nca ŋʇa, nqa ŋʗa, nxa ŋʖa.
12. Pronounce the Zulu words ncencéza ‘jingle’, nqenqéza ‘ring’, nxenxéza ‘urge on horse’.

13. You know how to make breathy voice (murmur). Make a breathy-voiced ŋ̤. (The subscript-diaeresis diacritic is problematic in this font, so I’ll write this as ŋʱ.) Add a simultaneous click and a following vowel, and you have the Zulu syllables ngca ŋʱʇa, ngqa ŋʱʗa, ngxa ŋʱʖa.
14. Pronounce the Zulu words ngcengcá ‘suckle’, ngqangcqá ‘tremble with rage’, íngxângxa ‘green-striped frog’.

15. You have now learnt all fifteen Zulu clicks: three basic clicks multiplied by five kinds of accompanying velar articulation.
16. Reinforce your knowledge by working through Ladefoged’s examples. If you feel strong, look at this video clip, which purports to make the very similar Xhosa clicks “easy to remember”.

(My examples are all taken from Say it in Zulu by D.C. Rycroft and A.B. Ngcobo, classroom materials made available to me at SOAS, 1976.)

Friday 11 June 2010

anglicizing clicks

Yesterday I promised that today I would discuss the anglicization of clicks in borrowed words.

Perhaps someone from southern Africa can answer this question better than I can, at least as far as South African English is involved.

Let’s start with the voiceless clicks, the ones spelt in Xhosa, Zulu, and other Nguni languages as c, q, and x. They are dental, (post)alveolar/retroflex, and alveolar lateral respectively. In current IPA they are written ǀ ǃ ǁ; in earlier IPA they were ʇ ʗ ʖ. My own impression is that the usual anglicization of these sounds is k. (Recall that clicks involve a simultaneous velar closure. A voiceless click is articulated during the hold phase of a voiceless velar plosive, k. Some people like to show this explicitly in the phonetic notation, thus k͡ǀ etc.)

Thus the name of the Xhosa people and language is pronounced in English with a simple k in initial position, ˈkɔːsə or ˈkəʊsə. In the Xhosa language itself the initial consonant is ǁʰ (old ʖʰ), an aspirated alveolar lateral click. This name seems to be the only anglicized word with an original click that is reasonably familiar outside South Africa.

Some historians will know about an event referred to as the Mfecane (also Difaqane, Lifaqane), the “scattering” of non-Zulus fleeing the Zulus in the early nineteenth century. In Zulu that’s mfɛˈ|aːnɛ (I don’t know what the tones are). I have heard it called əmfeˈkɑːneɪ, though I’m not sure whether that is the usual anglicization — the word is not even mentioned in Branford’s Dictionary of South African English (OUP 1980).

Anthropologists and linguists may know about the !Kung (or !Xũ) people and language. They live in the Kalahari/Kgalagadi (pictured above). As Wikipedia tells us,
To pronounce "ǃKung" one must make a click sound before the 'k' sound, often represented in texts as an exclamation mark.
Strictly that should be a (post)alveolar click, old IPA ʗ, with affrication of the pulmonic-air velar component. But I think people (or non-phoneticians at any rate) usually give up and just say kʊŋ.

What about voiced clicks? These are the ones spelt in Xhosa and Zulu as gc, gq, gx. They are articulated during the hold phase of a g. I can’t think of any words or names with a voiced click that are used in English.

Then there are the nasal clicks, made simultaneously with a ŋ. These are the ones spelt in Xhosa/Zulu as nc, nq, nx or (breathy-voiced) ngc, ngq, ngx. The only relevant example I can think of is the Ndebele surname Ncube, which British newsreaders pronounce as ˈnuːbeɪ, ignoring the click. The OBGP calls this the ‘established anglicization’ of this name, and comments as follows.
I have never heard anyone speak of the lawyer Bulelani Ngcuka, or of his politician wife Phumzile Mlambo-Ngcuka. Perhaps some South African reader has? (That’s Xhosa ˈŋǀʱuːkʼa, old notation ˈŋʇʱuːkʼa, tones not known.) He was a member of the law firm founded by Griffiths Mxenge (presumably mˈǁɛŋɛ, old notation mˈʖɛŋɛ), who was assassinated by the apartheid police in 1981. I haven’t heard that name pronounced in English either.

So, summing up, the best tentative rule I can offer Jongseong is: to anglicize a click, ignore the velaric (click) component, and pronounce the rest, which is necessarily some kind of velar pulmonic consonant. Voiceless clicks become k, voiced ones ɡ, and nasal ones ŋ. But in initial position (which is where clicks are usually found), the last-mentioned obviously becomes n.

Thursday 10 June 2010


Brian Jongseong Park, who wants to offer some advice to the National Institute of the Korean Language on recommended Korean spellings of names of World Cup squad members, asked about the pronunciation of Kagisho Dikgacoi in the original language. This is the name of a South African footballer, who also plays for Fulham FC in the UK.
A quick internet search seems to indicate that it’s a Tswana name, although the ‘c’ in the spelling suggests a non-Tswana origin, if it does indeed stand for the dental click found in Zulu and other languages. My guess for the original pronunciation would be [kaˈxisʰʊ dɪ(k)xaˈǀoi] based on the scant information I've been able to glean on Tswana (or with [χ] instead of [x]), but there are a few puzzling areas. […]
Also, although this is irrelevant to my question, how is the dental click [ǀ] usually approximated in an anglicized pronunciation? Or are words that have this sound that have found their way into English so rare that there is no usual approximation?

Being no expert on Tswana, I approached the BBC Pronunciation Unit. My former student Jo Kim, now working for the Unit, told me
Our own recommendation for this name is:
kuh-HEE-soh dick-ACK-oy (-h as in hot, -oh as in no, -oy as in boy, stressed syllables in upper case)
(That is, kəˈhiːsəʊ dɪˈkækɔɪ.)
This is established anglicisation and the pronunciation given to us by Fulham FC Press Office, who consulted the player himself and wrangled up this English pronunciation between them. This was the anglicisation preferred by Mr. Dikgacoi according to the press office but I wasn't able to speak to the player himself.

A South African contact at the World Service told us that this name is Tswana. I spoke to a Tswana speaker on the phone (so I was unable to see anything!) but I certainly heard this: ka'xisʰo dik'xaǀoi

Wikipedia gives a quite different anglicization, namely diːˈxɑːtʃwɑː. Jo continued:
The 'dee-KHAA-chwaa' pronunciation was actually suggested to us by a colleague at ESPN [an American cable TV network, JCW] at one point as the “correct” pronunciation. I did hear audio of some Anglophone sports commentators based in Africa calling him by that pronunciation but I suspect that 'oi' being pronounced as 'waa' is a gallicization, perhaps due to the abundance of French names in South Africa.

So: in Tswana ka'xisʰo dik'xaǀoi; in English kəˈhiːsəʊ dɪˈkækɔɪ. (Americans would probably prefer kəˈhiːsoʊ dɪˈkɑːkɔɪ.)

Jongseong also wanted to know about the usual anglicization of clicks in borrowed words. I’ll look at that issue tomorrow.

Wednesday 9 June 2010

opera and beatbox

Here is an interesting MRI video of the organs of speech at work, produced by scientists at the Magnetic Resonance Engineering Laboratory of the University of South California.

It shows two performers in action.

One is an opera singer doing her thing (O mio babbino caro) and then demonstrating different vowel sounds at different pitches. We see how the vowel space appears to shrink with rising pitch.

The other is a beatbox emcee making what the commentary calls percussive sounds. These seem to be more accurately stops made mainly with various non-pulmonic airstream mechanisms. It is difficult to identify them with certainty, but they mostly seem to be clicks and reverse-clicks, i.e. made with an ingressive or egressive velaric airstream; some are probably implosives or ejectives, i.e. made with an ingressive or egressive glottalic airstream.

Listen and repeat after me. (The account given here really doesn’t help much.)

Tuesday 8 June 2010

nouns and verbs

Tami Date wrote to me about a textbook he was reviewing. In one exercise, students were given a printed dialogue that read in part as follows.
John: There are many students from different countries in our school.
Su Chol: What language do you speak to each other?
John: We all speak English.
Yong Sil: Do you speak English at home, Salma?
Salma: No, I don't. I speak my mother tongue, Swahili. Are there any Korean high schools in Tokyo?
Su Chol: Yes, there's one. There's a Korean university, too...
Salma: By the way, when did Koreans come to Japan?
Yong Sil: Our great-grandparents came over to Japan before 1945. Then Korea was under Japanese rule....

(He makes no comment, and neither do I, on the extreme implausibility of the wording in this invented dialogue. Can you imagine live human beings talking like this to one another? Nor can I.)

Tami asks about nucleus placement in the second line. He thinks, rightly, that the nucleus should go on language.
What language do you speak to each other?

Yet if we apply the rule of thumb that the nucleus goes on the last new lexical item, we would expect it to go on speak.
(?) What language do you speak to each other?

Tami defends his view by saying that there is a hierarchy of accentability among lexical words, in which nouns rank first because of their richer semantic value.

I’m not sure that I would know how to measure so nebulous a quality as ‘semantic value’, but Tami is certainly right that nouns tend to be preferred over other parts of speech when we assign the nucleus. I touch on this point in my English Intonation book, §3.29.
…a more general tendency: we put the nucleus on a noun where possible, in preference to other word classes.
This is seen in various constructions which involve having a verb at the end of a sentence or clause. A final verb is usually deaccented, and the nucleus goes on a preceding noun.
ˈHow’s the ˈhomework going?
I’ve ˈstill got an ˈessay to write.
ˈWhich ˈbook did you choose?
You can contrast the Which/what/whose N type with the case where there is no overt noun, so that the nucleus defaults to the verb:
ˈWhich did you ˈchoose?
ˈWhat ˈbooks are you reading?
ˈWhat are you ˈreading?

Going back to the example from the dialogue, we would have
ˈWhat ˈlanguage do you speak?
ˈWhat do you ˈspeak?

Monday 7 June 2010

American “dialect” areas

Thanks to Amy Stoller for alerting me (and others) to a remarkable website about North American English pronunciation. It was created by one Rick Aschmann, who says
This is just a little hobby of mine, that I thought might be interesting to a lot of people. Some people collect stamps. Others collect coins. I collect dialects.

[By “dialects” Aschmann means what I would call “ (native speaker) accents”, i.e. pronunciation varieties: as far as I can see there is nothing here about morphology or vocabulary.]

Eric Armstrong comments
his remarkable maps are quite amazing! Having studied the Atlas of North American English quite a lot, he's done a remarkable job of combining their materials into one huge, very strong resource. Really cool, especially for someone who considers himself as a hobbyist. Maybe this site was old to others, but it was a real find for me!
— with which I would concur.

Aschmann’s site is based on Labov’s work but supplemented with several hundred audio recordings, mostly from YouTube, keyed into their appropriate geographical locations.

This site is clearly full of interesting material, and Aschmann is to be congratulated on it. I haven’t yet had time to explore it more than very superficially, so what follows is a criticism that some may consider superficial.

Regrettably, Aschmann eschews IPA notation in favour of the impenetrable “respelling” system found in the American Heritage Dictionary (but not used by any phoneticians or dialectologists, unknown outside North America, and not even the same as the Trager-type notation used by Labov). In this day and age, what good reason can there be for writing “ä” instead of ɑ or “ô” instead of ɔ?

Aschmann is under the mistaken impression that IPA notation is phonetic rather than phonemic.
This pronunciation system has the advantage that it is phonemic, rather than phonetic (like the IPA), and thus allows different dialects to use the same pronunciation key and get the right result for each dialect.
IPA notation can, of course, be either or neither. By far its largest consumer group is learners of foreign languages and particularly of English, who use it virtually only as a phonemic system.

Personally, I think my system of lexical sets (blog, 1 Feb 2010) performs the function referred to more transparently and helpfully than any alphabetic notation.

[See also this discussion on Language Log.]

Friday 4 June 2010


The first victim of the shootings (yesterday’s blog) was the gunman’s twin brother, at his home in Lamplugh.
The name of this village is pronounced ˈlæmpluː. According to the BBC Pronouncinɡ Dictionary of British Names it can also be ˈlæmplə.

But this phonetic transcription is ambiguous — certainly for me, and probably for most NSs. It does not specify whether the syllabification is ˈlæm.pluː (lam-ploo) or ˈlæmp.luː (lamp-loo). These two possibilities are rhythmically different. The second implies pre-fortis clipping, the first does not.

Compare the parallel case of Entwistle, which we discussed two years ago (blog, 2 July 2008).

The newsreader on Sky TV pronounced it ˈlæm.pluː, which I think is probably right.

The etymology of the name is … wait for it! … Welsh. (Sorry about that.) The Oxford Names Companion refers it to nant plus *bluch ‘bare valley’ (or is that ‘box stream’?), but I think we can ignore that. Here’s what Ekwall says, much more interestingly.So it’s from Welsh, and the second element was a loanword from Latin. Latin ē in loanwords regularly yields Welsh wy, as in ecclēsiaeglwys ‘church’, prophētaproffwyd ‘prophet’, venēnumgwenwyn ‘poison’, frēnumffrwyn ‘bridle’. This is an interesting historical fact, because from to ui̯ (and in the case of Lamplugh ultimately to English ) is a considerable phonetic step.

But so is its treatment in French, e.g. habēreavoir avwaʀ.

Thursday 3 June 2010


The tragic news of the Cumbria shootings introduced me to a place name I hadn’t previously come across. The gunman lived in the village of Rowrah ˈraʊrə.

This name is not in any of my reference books. It joins cowrie, along with Maori and perhaps for some Nauru, as an example of the possible sequence -aʊr- within a morpheme. I say ‘possible’ sequence, because I at least feel a very strong pressure to insert a schwa before the r and then smooth the result, giving ˈra(ʊ)ərə, ˈka(ʊ)əri.

This is because of the rareness or impossibility of having any of the stressed long vowels / diphthongs that end in or tend towards the close front or back areas immediately before r within the same morpheme. (The items I’m referring to are iː eɪ aɪ ɔɪ uː əʊ aʊ, i.e. FLEECE FACE PRICE CHOICE GOOSE GOAT MOUTH.)

For example, the only instances of -iːr- in RP and similar accents are cases such as key-ring, where a morpheme boundary intervenes. Otherwise, anything that would historically have had -iːr- (and still does in some accents) has -ɪər- instead, as in period ˈpɪəriəd and weary ˈwɪəri. (Obviously, in accents like south Walian, and for that matter Cumbrian, this historical development failed, so they still have the long monophthong.)

Similarly, what would otherwise have been -eɪr- has turned into -eər-, as in Mary, various. What would otherwise have been -uːr- yielded -ʊər- (fury, mural), and what would otherwise have been -əʊr- yields -ɔːr- (glory, moron). So far, so categorical.

When we come to the wide diphthongs, there seems to be more variability. In words like spiral and virus some speakers have ˈspaɪrəl, ˈvaɪrəs. Not me: I have to use aɪə here, or its smoothed reduction, thus ˈspa(ɪ)ərəl, ˈva(ɪ)ərəs.

Which is where we get back to cowrie and Rowrah.

Wednesday 2 June 2010


Among the new members of the House of Lords announced a few days ago is John Prescott, the former Labour cabinet minister.
Lord Prescott, as we must now call him, has long appalled and delighted journalists with his tortured relationship to the English language. His syntax gets muddled, his grasp of vocabulary is strained, every one pokes fun at his English, yet he is not ineffective as an orator and he has a loyal political following.
Two years ago, asked by The Scotsman newspaper whether he was looking forward to a place in the Lords, he memorably declared
I’m against too much flunkery and titles.

— thus producing an idiosyncratic coinage of his own making, a blend of flunkey and flummery.
Both of these words have interesting origins.

According to the OED a flunkey, a male servant in livery, a footman, is a word of Scottish origin (or, as the OED words it, “orig. Scotch”), and supposedly a corruption of flanker, literally a sidesman or an attendant at your flank. Is this a Scottish a misheard as an English ʌ?

Flummery, on the other hand, is one of the very few English words of Welsh origin. Although nowadays it usually means ‘empty compliments, humbug’, it was (and perhaps for some still is) a kind of pudding / dessert / sweet dish, in Welsh llymru ˈɬəmrɨ, ˈɬəmri. For this semantic development, we can compare waffle. There’s a recipe here. Welsh stressed ə is regularly mapped onto English ʌ and vice versa.

I haven’t been able to discover anything further about the origins of the Welsh word. I suppose it could be related somehow to llwm ‘bare, destitute’, though there can hardly be any connection with llymrïaid ‘sand-eels’.

So flummery is like Floyd or Fluellin. A Welsh voiceless alveolar lateral fricative is rendered into English as fl, as frequently happened up to the nineteenth century. We don’t do that nowadays: rather, we now tend to render ɬ as θl or, slightly more sophisticatedly, as xl.
As Jack Windsor Lewis insists, few non-Welsh speakers can get it correctly as ɬ. Present company excepted, of course.

Tuesday 1 June 2010


Yesterday, in a sudden outbreak of anorakery (you read it here first? — this word already has 259 Google hits), I travelled the length of the newly opened London Overground line from West Croydon to Dalston Junction.
This is the old East London Underground line from New Cross Gate / New Cross to Whitechapel, closed for refurbishment a few years ago, now reopened with significant extensions at either end. The southern extension is just a suburban route that was previously connected to London Bridge station. The northern extension is more interesting. There is an entirely new piece of track with a striking new bridge over Shoreditch High Street, linking to a stretch of what was once upon a time the Broad Street line (closed 1986), north to Dalston (that’s ˈdɔːlstən). There are sparkling new stations at Shoreditch High Street, Hoxton and Haggerston, as well as at Dalston Junction itself.

I thought I remembered having read somewhere that Hoxton and Haggerston, adjacent parts of Hackney borough, are etymological doublets, like frail and fragile. After all, what’s a dropped r, a lost schwa, a bit of voicing assimilation, and a switch between a front and back short open vowel, when we’re talking about a thousand years of onomastic development?

It turns out I was wrong. According to the Oxford Names Companion, Hoxton was the farmstead (tūn) of a man called Hōc, whereas Haggerston was the boundary stone (stān) of a man called Hærgod. So there’s a metathesis in the latter that has made the names more similar to one another, but they have quite distinct origins.

Dalston, on the other hand, does seem to be a doublet of Darlaston (in the West Midlands). Both were the tūn of a man called Dēorlāf.

The southern extension of the line passes through the once fashionable suburb of Penge. Lo and behold! More place name etymology reveals that here we have yet another Welsh- (or rather British-) derived name (blog, 6 May). Penge comes from the Celtic elements that in modern Welsh would be Pencoed, ‘top of the wood’. It might as well have been called Woodhead.