Friday, 30 September 2011

what the L?

Every now and again someone asks why pronunciation dictionaries do not show dark l explicitly. If milk is pronounced mɪɫk, why do we write it as mɪlk?

There are various kinds of answer one can give.

1. The distinction between clear and dark l is not particularly important for EFL. And anyhow there are plenty of native speakers who do not make the distinction.

2. There are no pairs of words in English distinguished by the clear-dark distinction. Writing the clearer variant as (which exaggerates its palatalization somewhat), we can say that [] and [ɫ] are allophones of the same phoneme /l/. Their distribution is conditioned by the phonetic context: in RP and similar accents, is used before a vowel or j, ɫ is used elsewhere (including before a major boundary). As with other allophonic variation, we ignore it in dictionary transcription because it is more economical to state it once rather than mention it on every occasion. The learner needs to learn the general rule rather than memorize the appropriate variant for each word. Dictionary transcription, and EFL transcription in general, is phonemic (or, if you prefer, broad).

3. Every word or stem that ends in the lateral is sometimes pronounced with ɫ, sometimes with , depending on what follows. Although the citation form of kill is kɪɫ, and killed is always kɪɫd, in killing we have ˈkɪlʲɪŋ. While kill them is ˈkɪɫ ðəm, kill it is ˈkɪlʲ ɪt. And likewise for thousands of other items.

4. Even among NSs who follow the rule given, there is some disagreement in the precise definition of ‘boundary’. Whereas mainstream RP of my generation has in phrases such as Isle of Man and Middle East, there are other NSs who use ɫ in these phrases. For stylize I say ˈstaɪlʲaɪz, just as in island ˈaɪlʲənd, but there are many others for whom the morpheme boundary triggers pre-l breaking and/or dark l, thus ˈstaɪəlʲaɪz, ˈstaɪ(ə)ɫaɪz.

Whereas my kind of speech applies the rule to syllabic l just as to non-syllabic, there are other speakers who claim to make all syllabic laterals dark. How this works out in cases of potential compression (positional loss of syllabicity), e.g. fiddling ˈfɪdl̩ɪŋ ~ ˈfɪdlɪŋ, I am not sure.

Abercrombie adduced a nice example. In I feel ill he (like me) would use a clear l at the end of feel: aɪ fiːlʲ ˈɪɫ. But in I may not look ill, but I do feel ill it switches to dark: aɪ ˈmeɪ nɒt ˈlʊk ɪɫ | bət aɪ ˈduː ˈfiːɫ ɪɫ. I'd do just the same in not the Far East, but the Midd[ɫ̩] East. Why?

Thursday, 29 September 2011


Robin Walker has been canvassing opinions.
The citation form of 'twelfth' in all the dictionaries I've checked is /twelfθ/, but the other day I thought I caught myself eliding the /f/. Was that me being 'sloppy' or is this something that we tend to do in colloquial speech?
Given that Robin is the newsletter editor of the IATEFL Pronunciation SIG, one can only assume that the question is somewhat faux-naïf. You’d think he would not just THINK but KNOW whether he ‘caught himself’ performing a common casual-speech reduction. Whether or not one calls such reductions ‘sloppy’ is not a phonetic question but rather a reflection of how far we have shaken off (or otherwise) popular attitudes to language and acquired a degree of scientific objectivity. Anyone interested in English pronunciation must surely know that this reduction IS something we tend to do in colloquial speech. So why ask us?

Nit-picking, I might further object that you can’t elide phonemes. Phonemes are mental representations, and I would prefer to say that it is a not a mental construct /f/ but a physical segment (speech sound) [f] that might or might not be elided.

Anyhow, I just replied

Jack Windsor Lewis gave a longer answer.
In my opinion anyone who pronounces twelfth in clearly the way dictionaries seem to suggest is normal actually produces something very likely to sound artificial and pedantic. I dou•t that anyone much notices if in naturally fluent speech [twelθ] is used. I'm uncomfortable that dictionaries generally suggest that [twelθ] is a less usual than versions with at least two simultaneous fricatives. Those who aim at saying /twelfθ/ and succeed in not sounding abnormally deliberate in the way they say it probably always have at least some overlap of the two fricatives and may sometimes produce a bilabial voiceless fricative at the same time. A phonetic notation [twelθ͡f] wdnt be far wrong.
I felt happiest recommending in my Concise Pronouncing Dictionary of British and American English (Oxford University Press 1972,1979) "twelfθ with f and θ usually made simultaneously".

I surely cannot be the only one whose ordinary slow pronunciation of this word is indeed twelfθ. OK, the labiodental and dental fricatives certainly overlap (though surely the labiodental turbulence starts before the dental). But segment overlap is nothing new or exceptional. The l in this word overlaps with the f, too. The w overlaps with the t. The mental representation underlying my articulation is clearly twelfθ, and that formulation faithfully represents the articulatory targets that my actual articulations may or may not achieve (depending, as usual, on speech rate, formality etc.).

Similarly, I have no hesitation in claiming that my basic pronunciation of sixth is sɪksθ. And that of clothes, kləʊðz.

True, there are speakers for whom reduced forms of various kinds have been lexicalized, so that the unreduced form is in some sense irrecoverable. Most (all?) of us have lexicalized the two-syllable reduction of every (i.e. ˈevri rather than ˈevəri). I’m aware that for victory I personally do not feel at all happy with the dictionary form ˈvɪktəri, since I feel I can naturally say only ˈvɪktri. On the other hand I cannot go along with people who claim that police is pliːs — for me, although I might sometimes reduce it this way in rapid speech, it is basically unquestionably pəˈliːs, i.e. comparable to polite and pollution rather than to pleat and playful.

Similarly, for me twelfth is not a good rhyme for health.

Wednesday, 28 September 2011

Duke of York sound changes

You’ll have heard of the grand old Duke of York. As you know, he had ten thousand men. He marched them up to the top of the hill, and he marched them down again.

It’s thirty-five years since Geoff Pullum wrote an article entitled ‘The Duke of York gambit’, about derivations of the general form A→B→A, that is derivations in which an underlying representation is mapped on to an intermediate form distinct from it, and then on to a surface representation which is identical with the earlier stage. Whether such derivations can be justified synchronically is an issue on which I express no opinion. But there certainly seem to be historical sound changes that proceed in just this way: something changes to something else, then changes back again.

Take popular London English. If Dickens is to be believed, London working-class speakers in the nineteenth century tended to confuse v and w: bevare of vidders! (beware of widows). They certainly don’t now. Historically, wvw.

Another well-known, nay stereotypical, Cockney feature is h-dropping. Remarkably, current London yoof — despite the supposed influence of Jamaican English, which shares this feature — generally don’t drop h. So, in the appropriate lexical contexts, hØh.

Londoners have diphthong shift, no? That is, the PRICE vowel has shifted in popular London speech from to something in the area of ɑɪ, ɒɪ? And the FACE vowel has gone from to ʌɪ, æɪ? Not any more. Paul Kerswill, Sue Fox and associates have shown that in inner-London Multicultural London English (blog, 2 July 2010) PRICE has reverted to and FACE to (or even ). So we have ɒɪ and æɪ.

If you’ve got the odd twenty minutes to spare, I’d like to recommend this brief talk by Paul Kerswill on just this topic of MLE, given at a TEDxEastEnd event in the wake of the recent rioting.

We like to think of sound changes as typically originating in the working-class speech of big cities, then spreading out socially and geographically. Many of the BrE sound changes of the last 500 years can be explained in this way, with working-class London as the point of departure. But that’s clearly far from the whole story.

Tuesday, 27 September 2011


At the running club these days I can, alas, do no more than jog a mile or two if neither my arrhythmic heart nor my arthritic hip are playing up. But it’s still a great place to socialize with friends old and new.

One of our coaches is something of a linguistic paradox: a highly educated man, with a degree from one of our oldest universities, established in his chosen profession. But he retains a strong working-class south London accent. As he was giving out the announcements recently, he told us in connection with some forthcoming event that further ˈdɪiʔɛ̈oz were available on the club website.

This pronunciation of details exemplifies, inter alia, intervocalic t-glottalling and l-vocalization, stigmatized features that one would not usually hear from someone with his educational background. Naturally he also has the usual British word-initial stress for this lexical item. (Compare AmE: in Yuko Shitara’s poll, 75% of Americans voted for final stress in detail.)

This set me musing about similarities and differences between the two major sound changes now vying for supremacy in the world of English non-initial t, namely voicing and glottalling. I am thinking above all of cases such as butter and a lot of, where AmE normally has a voiced tap ɾ and BrE may have any of t, ɾ, ʔ.

It is clear that an American-style voiced tap is by no means uncommon in Britain, particularly in high-frequency items such as a lot of. Conversely, a British-style glottal stop seems to be not unknown in north America.

But the two rival developments affecting t do not operate in identical environments. Yes, their environments overlap, as in the cases quoted. But t-voicing is blocked by a non-vowel right-hand environment (as in the two plosives in that’s right!), an environment in which glottalling is clearly more frequent than prevocalically. ˈðæʔs ˈraɪʔ

Furthermore, t-voicing is blocked (somewhat mysteriously, from my point of view) if the following vowel is unreduced and there is no word boundary. You get it in later ˈleɪɾə(r) and my late uncle maɪ ˈleɪɾ ˈʌŋkl̩, but not in latex ˈleɪteks.

For Americans, then, there is no chance of voicing the t in detail. For those who stress the second syllable, the t is like a word-initial t, always voiceless. For the 25% of Americans who prefer initial stress the same constraint as in latex comes into play, and again there is no chance of voicing the t. But for Brits who revel in glottalling — like our running coach — this is just one more candidate item for a glorious glottal stop.

Monday, 26 September 2011

an unexpected assimilation

I spent last week in Poznań, Poland, where I ɡave some Esperanto-medium lessons on general articulatory phonetics at Adam Mickiewicz University as part of their Interlinguistic Studies programme. The students were enthusiastic, mostly young, and from a variety of different language backgrounds, which makes for lively sessions.

These students had never previously experienced the dictation of nonsense words, so we had good fun working on them as a relief from the hard-core stuff. I find nonsense words a great way to revive the attention of those who are beginning to flag a little in class.

One topic that came up was assimilation in the place of articulation of nasals before various obstruents. We discussed whether the sound corresponding to the first n in dankon ‘thank you’ is, or ought to be, a dental/alveolar n, or whether it can, or should, get allophonically assimilated to the homorɡanic velar ŋ. (As you may know, this type of assimilation happens in many languages, but not — strangely — in Russian.)

When we turned to nasals before various other obstruents, an Italian participant remarked that in his own speech (in Esperanto as in Italian) he pronounces n before s as palatalized, (older IPA, ) : so for penso ‘thought’ he says penʲso. This is unexpected, because the following consonant, for him as for everyone else, is a plain common-or-garden s. Is this some kind of assimilation to the preceding vowel rather than to the following consonant? Or some kind of consonantal dissimilation?

We didn’t have time to pursue the matter then in class, but what I ought to have done — I realize now — is to investigate whether he does the same thing when the preceding vowel is back (e.g. monstro ‘monster’).

I have never seen any such phenomenon mentioned in phonetic descriptions of Italian. I wonder how widespread it is. Here is what Mioni says on the subject in the Italian section of Fonematica contrastiva (Bologna, 1973). Before s he reports a straightforward apico-dental n. The “lievemente palatalizzato” (lightly palatalized) he mentions as found only in the position before tʃ, dʒ, ʃ.
The English Wikipedia article on Italian phonology says baldly
Nasals assimilate to the point of articulation of whatever consonant they precede. For example, /nɡ/ is realized as [ŋɡ].
The Italian Wikipedia mentions a ‘mediopalatale’ variant of /n/, but does not elaborate.
esistono altri allofoni di /n/ come la dentale e la mediopalatale, di solito non riportati nella trascrizione larga. (Other allophones of /n/ are found, for example dental and mediopalatal, usually not reflected in broad transcription.)

Perhaps we’ve discovered something new.
_ _ _

I notice that above I used the expression ‘common or garden’. I’m aware that this is a British expression for which the AmE equivalent is ‘garden-variety’. And that gives me an opportunity to pass on Karen Chung’s recent discovery of an interesting article about Briticisms creeping into American English. Go here to read about run-up, go missing, snog, sort out, laddish etc.
_ _ _

European Day of Languages

Friday, 16 September 2011


At the time of writing we’re still waiting for news of the miners trapped underground in a drift mine in the Swansea Valley. We hope for the best but fear the worst.

The Gleision ˈɡlaɪʃɒn mine is at Cilybebyll ˌkɪləˈbebɪɬ, a village near Pontardawe ˌpɒntəˈdaʊi. I was impressed by the way the Sky News newsreader on TV yesterday evening handled the Welsh place names. He didn’t hesitate or stumble; he didn’t even break his rhythm. His ɬ was exemplary.

Gleision is a straightforward Welsh name, the plural of glas ‘blue, green’, so meaning just ‘blues’ or ‘greens’. The other two names involved are more interesting, because they bear witness to the influence of Latin on Welsh, dating from the time before the Anglo-Saxons arrived, when southern Britain was part of the Roman Empire.

Pontardawe means ‘bridge on the (river) Tawe’. The first element is pont, the Welsh for ‘bridge’, an obvious borrowing from the Latin pons, pont- of the same meaning. This word also gave us French pont and Spanish puente. No doubt the Romans introduced their bridge-building technology into Britain, and with it their word into the British (= Old Welsh) language.

Cilybebyll literally means ‘back of the tents’. The last element, bebyll, is the soft-mutated plural of the word pabell ‘tent, tabernacle’, from the Latin papilio, papilion-. In Latin this word primarily meant ‘butterfly’, but it was also a Roman army slang word for ‘military tent’, “probably from the similarity of shape when the ends of the covering are turned over at the entrance of the tent” (OED). The same word came into English via French as pavilion pəˈvɪliən, -ljən.
_ _ _

I shall be away again next week. Next blog: 26 September.

Thursday, 15 September 2011

names, names

Thanks to Karen Chung for pointing me towards a website called The Name Engine.
The Name Engine® provides the correct name pronunciations of athletes, entertainers, politicians, newsmakers, and more. Even well-known names are often pronounced in different ways, leaving you to wonder what the correct pronunciation is. You'll find the right answer here. Better yet, you'll actually hear the right answer.
Sounds as if the BBC Pronunciation Unit will be put out of business, not to mention pronunciation dictionaries.
All names are painstakingly researched for authenticity. Personal confirmation is the ultimate goal. At a minimum, they are confirmed by individuals with firsthand knowledge of the name in question. These individuals include team play-by-play announcers, public relations representatives, sports information directors, agents, etc.

Pronunciations are given in respelling (no IPA) and as sound clips. This is an American website, and both are strictly in AmE only.

At the moment the database of names included is very limited. It is divided into Sports (twelve subsections) and Miscellaneous (Companies/Brands, Entertainment, Newsmakers, Places, Politics).

Under Places there are just fifty or so geographical names, all of them places in the US except Abbotabad, Kyrgyzstan, Montevideo, and Qatar. (The “correct” pronunciation of the last-mentioned is given as ˈkɑːtɚ, a possibility I don’t countenance in LPD. In Arabic it’s ˈqɑtˁɑɾ.)

I looked in vain under Sports for Sharapova (blog, 1 and 5 July) and under Entertainment for Beyoncé. In the latter section I also learnt that Björk is “correctly” pronounced disyllabically, as biˈjɔːrk. H’m. (In Icelandic, monosyllabic bjœɾ̥k, and in BrE usually bjɔːk.) Under Companies/Brands, Bombardier (blog, 6 July) is given the respelling “bom-BAR-dee-ay”. But the associated sound clip is stressed differently, as bɑːmˌbɑːrdiˈeɪ.

The website looks very professional and clearly has great potential. There is no information given about who sponsors it or runs it. On the face of things, if it lives up to its grandiose claims, it might in time be more reliable than do-it-yourself Forvo (which currently boasts “1,108,951 pronunciations in 279 languages”).

I wonder whether The Name Engine® will get round to giving Americans guidance on the pronunciation of “difficult” British names such as Leicester.

Wednesday, 14 September 2011


In reaction to David Starkey’s egregious comments about last month’s rioting in London, Hubert Devonish, professor of linguistics at the University of the West Indies in Mona and Coordinator of the Jamaican Language Unit, wrote an interesting piece, Of riot and Rastamouse, that appeared first as a blog and then as an article in Jamaica’s leading newspaper, the Gleaner.

Rastamouse is a popular British children’s animated cartoon, featuring a cast of problem-solving mice musicians who play reggae and wear appropriate clothing and talk in an appropriately Jamaican way.

If you’ve never seen Rastamouse (perhaps because you’re not an under-10 in the UK), try this sample.

I say the characters “talk in an appropriately Jamaican way”. This covers not only accent (pronunciation) but also elements of Jamaican Creole grammar, e.g. the use of me as a subject pronoun. To a British ear they certainly sound Jamaican. But there are two interesting points to be made:

They don’t have the Multicultural London English we discussed recently. They sound definitely Caribbean.

Prof. Devonish mentions “the heavy anglicisation of the Jamaican Patois spoken by the characters”. So their language is what has been called “London Jamaican”, characteristically spoken by those of Jamaican birth or heritage who have lived for many years, or all their lives, in London.

And yet… It turns out that none of the principal actors who do the voices in the cartoon were born in Jamaica. They are native Londoners. The lead character is played by the voice actor Reggie Yates, who is actually not of Caribbean but of of Ghanaian descent.

And why not? I think his accent, even if it might not convince Jamaicans, is entirely appropriate for this cartoon mouse.

Tuesday, 13 September 2011

Donna and Benny

One of the songs our choir is rehearsing this season is a Donna Summer number called Last Dance. It will be familiar to any of you who were alert to the popular music of thirty-odd years ago (which I was not — the song was new to me).

As often happens in the lyrics of popular music, in this song dance is made to rhyme not only with chance but also with romance. This is something of a problem for speakers of RP and similar accents: we normally pronounce dɑːns and tʃɑːns but rə(ʊ)ˈmæns, which means the rhyme doesn’t work properly.

Our choirmaster has given us clear instructions that in this song we are to sing dæns and tʃæns, as if we were American.

(Well, sort of. Real Americans would be quite likely to do BATH Raising and come up with something like dɛːnts, tʃɛːnts or deənts, tʃeənts.)

Interestingly, the rehearsal tracks supplied to us show a very selective Americanization of this and other same-genre songs. We are expected to do t-voicing but not to add rhoticity. So the title words of another number, Get this party started, come out as ˈɡet ðɪs ˈpɑːdi ˈstɑːdɪd.

The choir is very good at supplying us with rehearsal tracks. For each piece of music we are offered a separate sound file for each of the four voice parts (tenor 1, tenor 2, baritone, bass). So we can listen repeatedly and practise on our own by singing along.

Another piece we are doing is a compilation of Christmas number ones down the years. Among them is Benny Hill’s comedy piece Ernie (The Fastest Milkman in the West). The soloist on the rehearsal track, doing the Benny Hill narration, adopts what is intended to be an appropriate West-of-England accent — Benny Hill was from Southampton, as is immediately revealed by the way he speaks. To English people from elsewhere the most obvious feature of a west country accent is its rhoticity. So we duly get ˈɝːni and ˈmɪlkkɑːrt. But for fastest, where the real Benny Hill would have said ˈfaːstɪst, the voice on our rehearsal track says ˈfɑːrstɪst, a lovely example of hypercorrection.

When imitating a rhotic accent we non-rhotic speakers find it quite difficult to sort out farther and father, larva and lava. It is hard to change scarf skɑːf, for example, to skɑːrf without at the same time changing half hɑːf to hɑːrf. The phrase fastest milkcart is particularly tricky, with a quick succession of ɑː’s, only one of which should properly be made r-coloured. Thinking of the spelling each time would solve the dilemma, but somehow we can’t do that in the middle of the flow of speech. It makes you admire British actors like Hugh Laurie, with his remarkably authentic-sounding American accent, even more.

Monday, 12 September 2011

tonicity again

My recent trip to China furnished further evidence, if evidence were needed, of the failure of many NNSs of English to master the part of English intonation that concerns tonicity (focus marking, the location of the nucleus — blog, 15 October 2009).

One of the poster papers at the ICPhS made the tentative claim that the focus in Mandarin Chinese is well marked, with ‘post-focus compression’ of the pitch range identifying it clearly despite all the lexical tone in the utterance; but that in Cantonese Chinese it is not.

This chimes in with the limited experience I have of intonation in Chinese English. Whereas those whose first language is Mandarin seem on the whole to locate the English nucleus correctly, those who are speakers of Cantonese often do not.

In Cantonese-speaking Hong Kong, as you might expect in a former British colony, there are many NNSs whose English is really excellent. But there are also many who struggle to a greater or lesser extent. Our tour guide on the afternoon excursion to the sights of Hong Kong island, whose English was very fluent but not very good, drew our attention to “the Jockey Club on your right-hand side | and a cemetery on your left-hand side”, so violating the rule about avoiding placing the nucleus on a repeated item. (He also pronounced number as ˈlɐmbɐ etc., which was very confusing at first.)

As we all know, it’s not just the southern Chinese who have this problem in English intonation. Our flight home was with Air France. The cabin attendant reciting the pre-takeoff drill reminded us that “the safety instructions are in the seat pocket in front of you”.

Does nobody teach French learners of English that they should not accent pronouns except for contrast? All NSs of English would say “…the seat pocket in front of you”.

Friday, 9 September 2011

reverse search

”What you ought to get your publishers to provide,” said one fan of LPD to me in Hong Kong three weeks ago, “is a way of carrying out a reverse search. Then the user could enter a pronunciation and find what word is pronounced that way.”

He was somewhat nonplussed when I told him that the CD-ROM bundled with the third edition of LPD provides precisely that facility. It’s called Sound Search, and you see the button for it when you fire up the on-screen LPD.

Press the Sound Search button, and up pops this dialogue box. Use the phonetic symbol buttons to enter your search term. Then, if what you have entered corresponds to the pronunciation of an English word, a Find button will appear. Press that, and you’ll get the orthographic version in a Results box.

(It’s a pity about the misspelling DIPHTONGS, and the misaligned diacritics at and æ̃, but no one’s perfect.)

I think this search facility works pretty well.

It can cope with optional sounds. So any of tʃɑːns, tʃɑːnts, tʃæns, tʃænts will find chance.

There’s an asterisk to give you a wild card. Input tʃɑːn* and you get offered a list of 18 possibilities.

It copes with both RP and GenAm. Input ʃɑːk and you get not only BrE shark but also AmE shock.

It delivers homophones efficiently. Input saɪt and you get cite, -cyte, sight, and site.

Shortcomings? As I say, no one’s perfect. The handling of run-on entries leaves a lot to be desired. Although siː correctly returns C, se, sea, see, si, entering siːz gives you only seise and seize — it fails to identify seas, sees and C’s, Cs, c’s.

You can’t even enter the modern BrE ɒʊ variant of əʊ (as in ɒʊld old).

If you enter saɪ, you get not only the correct Cy, psi, sigh, xi but also a rogue siamang — because alongside a main pronunciation ˈsiː‿ ə mæŋ this word also has a second pron ˈsaɪ‿ , i.e. ˈsaɪ‿ ə mæŋ, which the software has misinterpreted as a complete form.

Thursday, 8 September 2011


It’s not actually phonetics, but my colleague Bas Aarts asks me to tell you about iGE, the interactive Grammar of English that the Survey of English Usage at UCL has developed.

It is an app for the iPhone, iPod, and iPad. (Since my mobile is an Android phone and I have no Apple products, I haven’t been able to try it out myself.)

As you can see, it offers a “complete” grammar of English, incorporating an extensive glossary, a guided course of instruction, and interactive exercises and puzzles. There’s a cut-down version that is free and a full version that costs a modest £4.99/$6.99/€5.49.

There was a report in Tuesday’s Evening Standard.
So — who’s going to be the first to fill the gap in the market for something similar dealing with English phonetics? Currently there are various monolingual and bilingual dictionaries available for your mobile/cellphone/handy, some with spoken versions of the headwords, but no specialist pronunciation dictionary.

Wednesday, 7 September 2011

Shigeru Takebayashi, 22 Sept 1926 - 10 March 2011

I was sorry to hear that Shigeru Takebayashi (竹林滋), the eminent Japanese phonetician and lexicographer, had died earlier this year.

He was the author of a well-regarded Japanese book on English phonetics (1982, second edition with Hiroko Saito 1998). He was also the editor of a number of Japanese-English dictionaries, including the Pocket Kenkyusha Japanese Dictionary and the Kenkyusha Japanese-English Learner’s Dictionary. An autographed copy of the latter sits on the bookshelf next to my desk. It lists the headwords in romanized alphabetical order, which is exactly what I want. It also supplies the accent pattern for each headword, information not otherwise easily found in English.

My own dealings with Prof. Takebayashi date back to 1990. Shortly after the first edition of my LPD had come out, the publishers arranged for me to meet him in Tokyo. The idea was that he would write an introduction-cum-explanation in Japanese for my dictionary, to be bundled with it in a slip case. To do this he needed to ask me some questions about it.

Our meeting was almost like a PhD viva, with me as the candidate and Prof. Takebayashi as the examiner. He asked me a series of searching, nay penetrating, questions about how I had compiled the dictionary, how it was arranged, what was included and how the transcription system worked. I answered to the best of my ability.

The upshot of this interrogation was excellent. The acuity of his questioning and his understanding of the lexicographical and other problems involved gave me great confidence in him, and presumably the quality of my answers gave him confidence in me. His booklet duly appeared accompanying the version of LPD sold in Japan.

Issue 41 of Lexicon, the journal published in Tokyo by the Iwasaki Linguistic Circle, is dedicated to his memory. There is also an article about him in the Japanese Wikipedia.

Tuesday, 6 September 2011

warming up

Another September, another season for the choir I sing in, another series of weekly rehearsals.

We have just appointed a new Deputy Musical Director. His first duty was to conduct the warm-up with which we start each session. (I’m not clear why or how warm-ups help you sing better, but actors and singers all seem to believe in them. I can see that at least they fulfil the social function of getting everyone — about 160 people at last night's rehearsal — to start focussing on a joint activity carried out together.)

The DMD surprised us by choosing for the very first exercise a “fully rolled r” (a voiced alveolar trill). “If you can’t roll an r properly, do a lip roll instead.” (That is, ʙ.)He continued to get us to work on non-English sounds or sequences. The next exercise went ŋiː ŋiː ŋiː ŋiː ŋiː.

Then it was an easier mɑː mɑː mɑː mɑː mɑː. Intriguingly, he described ɑː as a ‘bright’ vowel. (I’d have thought it was comparatively dark, but then I’m no synaesthete; and perhaps he was referring to the voice quality he wanted rather than to vowel quality.)

And so it went on, with the tongue twister red ’n’ yellow lorry, red ’n’ yellow lorry, red ’n’ yellow lorry, red ’n’ yellow lorry… and then finally bumblebee, bumblebee, all on increasingly complicated practice riffs up and down a succession of scales.

Fortunately he did not give us too much of the pseudo-anatomical articulatory nonsense we often got from his predecessor. Nevertheless, when calling for hard attack to initial vowels in the lyrics of one song he did mention that he needed a ‘glottal’ (i.e. a glottal plosive, a glottal STOP, ʔ), which he described as being ‘when the vocal cords are coming together very quickly’. H’m.

Checking vocal warmup techniques on the web, I find plenty more of the same sort of thing. Perhaps you are better able than I am to understand what might be meant here (under Tongue Trills) by
The d is what will give you connection from chest to head voice. The d tends to hold your vocal cords together through your bridge….

Does anyone seriously believe that an alveolar plosive could “hold your vocal cords together”? Why do they say such things?

Monday, 5 September 2011


Look at this headline from Saturday’s on-line Liverpool Echo. Can you see anything wrong?
Yes, then ought to be than.

I was quite surprised the first time I came across this misspelling, which was a few years ago. But the explosion of material published on the internet without the attention of a copy editor (and sometimes even with such attention, as presumably here, the Echo being a reputable newspaper) has made me realize how very widespread it is. (Since there is also a perfectly good word then, it would not be trapped by a simple spellchecker.)

There are plenty of other examples to be found on the web.

And yet I don’t think people commonly misspell ran as (w)ren, or tan as ten. In all core native accents of English the words of the TRAP set are consistently distinguished from those of the DRESS set.

So what’s going on?

In speech, the word than is almost always pronounced in its weak form, ðən.

ˈbetə ðən ˈevə
ˈmɔː ðən ju kʊd biˈliːv
ˈmɔː ðən ˈʌðə dʒæbz

It is difficult to envisage a context in which one would want to accent it, thereby triggering the strong form. (I exclude the obvious one of naming the word rather than using it, as in ”How do you spell ‘than’?”.)

The only way to trigger an obligatory strong form in ordinary conversation seems to be by resorting to stranding (blog, 28 May 2008).

A mouse is something that an elephant is bigger than.

The syntax here involves the fronting of a mouse, with the consequence that than is stranded, deprived of the NP it governs. As with prepositions and indeed all other function words, such stranding in English calls for the use of the strong form of the stranded item. The word normally remains unaccented.

I would say
ə ˈmaʊs ɪz sʌmθɪŋ ðət ən ˈelɪfənt ɪz ˈbɪɡə ðæn

I wonder if anyone actually pronounces the strong form as ðen (to rhyme with ten). That is not inconceivable, given the extreme rarity of strong than and therefore the extreme rarity of opportunities for the language-acquiring child to hear how it is pronounced. (After all, the usual ðən might result from the weakening of any of putative ðen, ðæn, ðʌn, ðɑːn, ðɒn — compare the strong and weak forms of them, at, us, are, from.)

You’d think, though, that most children would have been exposed somewhere along the line to such utterances as Who are you bigger than? Who is Mary younger than? Which of your brothers are you older than?

Nine is one fewer ðən ten. Ten is what nine is one fewer ðæn.

Friday, 2 September 2011

vowel colour

What colour would you say ɛ was? And ɒ?

For many of us those questions may seem pretty fatuous. We’re used to the metaphorical use of the term “vowel colour” as a synonym of “vowel quality”: something to be described in terms of front/back, close/open (or high/low), and rounded/unrounded. But actual hues? Is this vowel pink, that one green? Meaningless questions, surely.

Not for everyone. Some people exhibit a neurological condition known as synaesthesia. For them, numbers or letters or days of the week are characterized by different hues. Read about it here.

(It’s not the same as phonaesthesia, which is to do with sound symbolism. Though I suppose the two are related.)

If there are synaesthetes who think that particular letters have particular colours (and apparently there are), what about speech sounds? Are they coloured, too? That’s the subject of a piece of research currently being carried out by Rob Drummond of Manchester Metropolitan University.

Anyone can take part in this research, even if they have never thought of speech sounds as being coloured. Just go to or visit the dedicated page on Facebook, and volunteer as a subject to answer an online questionnaire.
…we need as many people as possible to help us by completing a task online. Anyone at all can take part, as long as you are able to hear sounds on the computer you are using. The task itself is extremely straightforward and will take no longer than 10-15 minutes to complete.

Rob asks me to invite all my readers to take part, whether native speakers of English or not. So how about it?

Thursday, 1 September 2011


I greatly enjoyed the ICPhS XVII in Hong Kong last month. There were over seven hundred participants from all over the world. It was good to see so many colleagues again, and to listen to some excellent oral and poster presentations. I enjoyed the conference all the more perhaps in that I hadn’t offered a paper myself and so didn’t have to worry about performing.

There was an unexpectedly interesting opening plenary by Klaus Kohler, demonstrating among other things that German listeners needed no more than the palatalization of a single segment n to hear kann Ihnen rather than just kann, deeply buried in the middle of a rapidly spoken colloquial sentence.

Some assorted nuggets of interest:
• In the Berber language Tashlhiyt many words are vowelless, for example kk ‘cross’. Geminate consonants contrast with single ones even in word-initial (and utterance-initial) position, e.g. ttut ‘forget him’ vs. tut ‘she hit’.
• In Iraqi Arabic the voiceless ‘pharyngeal fricative’ ħ and its voiced counterpart the ‘ayn ʕ can actually be aryepiglottic trills, according to John Esling — who proposes to write them ʜ and ʢ respectively. If I understood him correctly, he also claims that the ‘glottal’ stop is actually epilaryngeal. My knowledge of anatomy is not sufficient to enable me to judge these claims.
• In the Wu Chinese of Qingtian there is a tonal depression feature reminiscent of that of Zulu. The triggering consonants are, however, now voiceless.
• In the Chinese of Qiyang there are complex contour tones that don’t fit the usual tone templates. They are high and low fall-rise-fall tones. On a five-point scale, where 5 is the highest, their pitch patterns are 4232 and 2142.
• The Swedish accent 2 (tone 2) is the marked one: it takes longer to process than does accent 1.
• Everyone now seems to call the intonation nucleus or tonic the focus. Well yes: but as I see it the nuclear syllable actually marks only the word at the end of the whole focus domain. Anyhow, among laboratory phoneticians the trendy term for the low, more or less level, pitch of the tail in intonation is now ‘post-focus compression’.
• In some Australian English el has become æl, making celery a homophone of salary and hell a homophone of Hal.

A keen young researcher reported her rediscovery of the wheel by revealing to us that the Polish affricate spelt cz is somewhat different from the Czech one spelt č, the first being retroflex and the second merely postalveolar. I’ve been teaching this for forty years and more, in the context of the range of ʃ-like sounds we can make and how they vary from one language to another. (Compare both the Polish and Czech sounds with the ch ち of Japanese.) No doubt my predecessors taught it for forty or more years before that. We’ve even covered it in this blog: see the sound file posted in the blog entry for 3 March 2008. Notwithstanding, the researcher portentously declared that in her paper “I revise the affricate inventories of Polish and Czech… This conclusion is supported by the results of an acoustic study of Polish and Czech affricates”. It’s also supported by the ear of any halfway decent practical phonetician.

The next ICPhS, in four years’ time, will be in Glasgow, 10-14 August 2015. You read it here first.