Thursday 30 September 2010

more on Russian ж

An anonymous commentator on yesterday’s blog (why don’t people identify themselves?) wrote
Russian ж is a retroflex, so shouldn't we use IPA ʐ?
This suggestion risks opening a whole can of worms.

If you compare a typical French ʃ or ʒ with a typical English ʃ or ʒ, you may notice that the French ones sound slightly ‘darker’ than the English. As Armstrong and Ward put it,
The palatal (i.e. j-like) quality which is often heard in English ʃ and ʒ is absent from the French sounds. [The Phonetics of French, 1932]

If you compare the Russian sounds written ш and ж, transliterated sh and zh respectively, you will notice that they are darker still. Here is Daniel Jones.
English ʃ … is a somewhat palatalized sound … in comparison with Russian ʃ, and does not have the characteristic ‘dark’ or ‘hollow’ quality of the latter. […] English ʒ differs from Russian ʒ in exactly the same way that English ʃ differs from Russian ʃ. [Jones and Ward, The Phonetics of Russian, 1969]

All three ʒ sounds (English, French, and Russian) are ‘darker’ than the ‘alveolopalatal’ ʑ of Polish, Japanese and other languages. On the other hand none of them are as ‘dark’ as the ‘retroflex’ ʐ of Standard Chinese (Pinyin r, as in 人 rén).

So what we are dealing with is a continuum of possibilities. I used to make my students first produce with their usual ʃ (or ʒ) and then prolong it while modifying the articulation so as to make it sound first clearer then darker, then slur backwards and forwards between the extreme points of ɕ ʑ and ʂ ʐ, passing through a range of subtly different varieties of ʃ and ʒ.

Providing a language has no phonemic contrasts of place within this range, it is entirely acceptable, indeed recommended, to use the symbols ʃ and ʒ. The precise coloration does not matter: we write ʒ in transcriptions of French and of English without causing any confusion (even though the sounds are not exactly identical).
In cases where a lanɡuaɡe does have a contrast within the ranɡe, we can discuss which symbols to use. So for Polish it is usual in the IPA tradition to write ɕ ʑ for the clearer pair and ʃ ʒ for the darker pair, though some prefer ʂ ʐ for the darker.

Russian is an interesting case. There is no simple direct contrast of place, but the fricatives spelt ш and ж are dark (non-palatalized) and very different from the palatalized fricative spelt щ, which tends to be longer and can also be interrupted by a plosive element, so ɕɕ or ɕtɕ. (Jones and Ward use the obsolete IPA symbols ʆ, ʆʆ, ʆtʆ.) But for the non-palatalized ones J&W write ʃ and ʒ.

Ladefoged and Maddieson, in The Sounds of the World’s Languages (1996), have a long discussion of the articulatory postures involved in sounds of this general type. They categorize the Polish sound spelt rz, which I equate with Russian ж, as ‘flat post-alveolar (retroflex)’. Yes, somewhat retroflex. But they are nowhere near as retroflex as the “genuinely retroflex gesture” L & M report in the Toda language, or in other Dravidian languages.

Perhaps we would do best to confine the use of the retroflex symbols to those languages where there are not only fricatives but also plosives and nasals that are distinctively retroflex. For Russian ж I’m sticking with ʒ.

Wednesday 29 September 2010

the digraph zh

In transliterating Russian names, the question arises what to do about Cyrillic ж, IPA ʒ, the voiced palatoalveolar fricative. The usual convention is to write it zh.

Today’s Guardian carries a full page devoted to the abrupt sacking of the mayor of Moscow, Yuri Luzhkov (Юрий Лужков). The feature comprises three articles by three named authors. And whereas Luke Harding and Simon Tisdall consistently spell his name correctly as Luzhkov, David Hearst spells it correctly twice, but incorrectly, ‘Luzkhov’, four times. Let’s hope this is just due to his inaccurate typing. It would be too depressing to think that Hearst, who spent several years in Moscow as the paper’s foreign correspondent there, is not sure how the name is written in Cyrillic and thus how it should be spelt in English.

(Phonetic footnote: since Russian assimilates voicing in consonant clusters, Luzhkov is actually pronounced luʃˈkɔf. So it would be phonetically accurate to write it as Lushkov or indeed Lushkof. The usual Luzhkov is not a transcription but a transliteration.)

How familiar is this zh convention to a general readership? We all know about Dr Zhivago, or at least all of us who are over a certain age. (Pasternak’s novel was published in 1957, and David Lean’s film of it came out in 1965.) Some will have heard of General Zhukov or other figures from Russian history whose name includes this sound/digraph. But there don’t appear to be any familiar Russian placenames that include it [PS: but see comment below!], nor any Ukrainian, Bulgarian, or Serbian names.

The logic behind the use of zh for this sound is transparent. As the letter s stands for the sound s and the digraph sh for the sound ʃ, so given that the letter z stands for the sound z the digraph zh must stand for the sound ʒ.

Respelling systems deployed to show pronunciation in some monoglot English dictionaries (notably those published in the USA) represent ʒ as zh pretty much without exception. So one can say that writing zh is a well-established convention, despite the claim in Wikipedia that it is ‘ad hoc’.

The only European language that uses zh for ʒ (or for anything else) in standard orthography appears to be Albanian — not a language often learned by outsiders. I can’t think of any non-European languages that use it, either.

But this digraph is indeed used in the Chinese romanization known generally as (Hanyu) Pinyin. So we are all becoming familiar with names such as Zhang, Zhou and Zhu. But here zh stands for a different sound! The corresponding sound in Mandarin is not ʒ but an affricate , or more precisely unaspirated . I notice that sports commentators who know very little about foreign languages often pronounce it z in the names of Chinese competitors. Those who are somewhat more sophisticated say ʒ. Only those who have bothered to find out the facts say , which is the closest English equivalent.

Tuesday 28 September 2010

prusiking around

Today’s entry is not so much about pronunciation as about spelling. There are several English verbs that we can spell confidently in their base form, but which may throw us into uncertainty when we want to write the 3sg, the past tense, or the present participle.

Take to rendezvous ˈrɒnd(e)ɪvuː. What’s the 3rd person singular? Clearly, it’s pronounced ˈrɒnd(e)ɪvuːz. The Concise Oxford wants us to spell it when he rendezvouses with us, but I cannot be the only one who feels very awkward with this. But then it would also be awkward to write when he rendezvous with us. The other two inflected forms, rendezvoused ˈrɒnd(e)ɪvuːd and rendezvousing ˈrɒnd(e)ɪvuːɪŋ, are not quite so bad.

Then there’s to ricochet, another French word. On the assumption that we pronounce it ˈrɪkəʃeɪ or ˌrɪkəˈʃeɪ, it would feel quite wrong to double the t when adding -ed or -ing. I certainly prefer ricocheted, ricocheting.

A former Russian correspondent got very excited about the verb to prusik, a technical term in mountaineering. When forming the present participle, he wanted to know, is it correct or not to double the k? He was disappointed when I declined to deliver an authoritative answer. I still don’t know: in comparison I can only adduce to frolic, frolicking, which doesn’t really help. Americans, who write worshiping, may be confident with a single k in prusiking. But what about us Brits, who write worshipping?

What stimulated all this was catching myself writing stymieing, only be struck by doubts whether it should not perhaps be stymying. (The COD gives both possibilities.) The word stymie ˈstaɪmi originated as a technical term in golfing, though I imagine most non-golfers know only its general meaning ‘thwart, obstruct’.

Monday 27 September 2010

rh and rrh

Can you spot the repeated spelling mistake in this, from Saturday’s Guardian magazine section?Yes, it ought to be mycorrhizae and mycorrhizal. The h goes after the rr.
How do you know the words must be misspelt, even if you’ve never come across them before? Answer: because they’re of Greek etymology, and Greek double rr is always followed by h. Think of catarrh, myrrh and diarrh(o)ea.

So, of course, is Greek word-initial r: we write rhapsody, rheostat, rhesus, rhetoric, rheumatism, rhinoceros, rhizome, rhododendron, rhombus, rhubarb and rhythm, all being words derived from Greek; and there are no Greek-derived words in English that begin with plain r.

In Wikipedia we read that a mycorrhiza is a symbiotic association between a fungus and the roots of a vascular plant.

Why do Greek words have these apparently superfluous h? It can be traced back to an allophonic fact about Ancient Greek /r/, namely that it was voiceless in word-initial position and when geminated.
Generally speaking [r] is a voiced sound, but in certain environments in classical Attic it seems to have been voiceless. What we are actually told by the grammarians is that ρ was ‘aspirated’ at the beginning of a word, and that when a double ρρ occurred in the middle of a word the first element was unaspirated and the second aspirated. … These descriptions are followed in the Byzantine practice of writing initial ῥ and medial ῤῥ, and are supported at an earlier period by Latin transcriptions such as rhetor, Pyrrhus… [W.S. Allen, Vox Graeca, CUP 1968]

Allen goes on to say that it can be shown to have been a matter of voiceless (rather than a cluster rh) by the fact that it was borrowed into Latin on the one hand as rh but into Armenian on the other hand as hռ hṙ, thus hռետոր hṙetor, and similarly in Coptic.

When I did Greek at school, the transition from the fourth form into the fifth was marked by requiring us thenceforth to write the proper polytonic accents on our Greek, which we had until then been allowed to ignore. So every initial rho had to be written with a rough breathing, and every doubled rho with first a smooth breathing and then a rough.

The Greeks themselves did not bother with breathings and accent marks until the Hellenistic period. They first appear in papyruses in the second century AD, and then only sporadically.

It seems a bit silly that now, well over a thousand years later, we have to go on including this unnecessary letter h after r in English spelling: but that’s the power of tradition.

The Greeks themselves officially abandoned polytonic orthography, including the breathing marks, in 1982. The rough breathing, in the shape of the letter h, lives on only in English, French and German.

If the second part of a compound word in Greek began with r, the consonant was automatically doubled. That’s why we have one r in rheostat but two in diarrhoea. And one in rhizome but two in mycorrhiza.

Friday 24 September 2010

never been there

On behalf of Akiko Ohkita, one of his students, Masaki Taniguchi asks what would be an appropriate intonation pattern for the phrase I have never been there in this passage, taken from a junior high school textbook.
The place I want to visit is Korea. My email friend, Mina, lives in Seoul. She writes to me about schools, movies, and music in Korea. I write to her about life in Japan. Korea is close to Japan, but I have never been there.

The textbook says that this been should be pronounced with a weak form, implying that it would not be accented.

Masaki, however, rightly thinks that been can be stressed and may even carry a nucleus in this context. As he says, visit was mentioned at the beginning, but is too distant to be counted as given/old information. He wonders if it would be possible to deaccent been in this context and place a nucleus on the first syllable of never. Are both options possible, he asks, depending on the speaker's mind?

The first thing to say is that the textbook is clearly wrong. It seems to me that native speakers would virtually always choose to place a nuclear accent on been.
kərɪəz \/kləʊs tə dʒəpæn | bət aɪv nevə \biːn ðeə

(Or the nuclear tone might be a fall-rise. In idiomatic English, we’d probably include the word actually, too.) The typical Japanese error, I suspect, would be to accent never but not been:

…* \never been there.

It would be possible to give never a nuclear accent, but only, I think, if there were another one on been.
…\/never | \been there.

Alternatively, it would be possible to place a contrastive accent on I’ve, to convey an implicit or explicit implication.

...but \/I’ve never been there (even though you may have).

This is one of the cases where been seems to function almost as the past participle of go. We say I go there often, I’ve been there often. In this context it is a content word.

Been is not a word for which it is useful to speak of strong and weak forms. Although there may be people who use biːn as the strong form, bɪn as the weak, Americans usually have bɪn as the only form, used in strong position as well as weak, while conversely in BrE there is nothing odd about using biːn in weak positions.

Judging by the spelling mistakes one sees, many people must pronounce been and being as homophones. Not me!

Thursday 23 September 2010

the jewelry is out?

Paul Reed, writing from South Carolina, says
I heard on the radio today an interesting statement. The DJ said,'The jury is out, and I don't mean the diamonds and pearls, but 12 of your peers.' I thought that this was intriguing that the DJ felt it necessary to clarify his meaning.
Is there an AmE variety where jewelry/jury are homophones? I have queried several of my friends (all students like me), and we all tend to have the /l/ or a different vowel, GOOSE for jewelry and FOOT for jury.

I wouldn’t have thought there were any speakers for whom jewelry and jury are categorical homophones (i.e. always pronounced identically), though I suppose it’s possible that there are speakers who sometimes pronounce them so similarly that a listener might be momentarily confused. Even if you vocalize the l, turning ˈdʒuːəlri jewelry into ˈdʒuːori or something of the sort, the vocalic element between the and the r is still going to be different in AmE from that of ˈdʒʊri jury.

The same applies, mutatis mutandis, in BrE. We usually spell the first word differently, as jewellery, but this doesn’t really imply any significant difference in pronunciation. In RP the can be smoothed prevocalically to ʊ, producing ˈdʒʊəlri. I suppose that if you vocalize the l in that version you might get something containing some sort of indeterminate ʊəo or even ʊːː that could be confused with the ʊə of jury, nominally ˈdʒʊəri but in practice often ˈdʒʊːri. However with the decline of the phoneme ʊə jury can also sometimes be ˈdʒɔːri or ˈdʒɜːri. In non-RP you can also get the historically conservative version ˈdʒuːri (lumped in with the majority form in this LPD graphic). None of these are likely to be confused with jewellery.

Furthermore, jewellery also has a BrE version ˈdʒuːləri, which I mark in LPD with the siɡn §, non-RP. Since the l in this form is prevocalic, it would not be a candidate for l vocalization.

Wednesday 22 September 2010

room, now and then

For you, does room rhyme with tomb, boom and zoom, or not?

In one of my preference surveys (1988), 81% of my British respondents said ‘yes’. They are the people who prefer ruːm, with the GOOSE vowel. The remaining 19%, the ones who said ‘no’, prefer rʊm, with FOOT. The 81% must also include the people from Scotland and Ulster who have no GOOSE-FOOT contrast, just an undifferentiated u or ʉ vowel.

Yuko Shitara-Matsuo’s 1993 survey of AmE preferences showed an even larger majority, 93%, for the GOOSE vowel.

A few days ago Stevie Rickard wrote
I'm teaching 1920s RP to some actors and I've advised them to pronounce 'room' rʊm on the basis of almost nothing at all. Shame on me. One of the actors thinks he was told to use ru:m for the same accent in another play. Would one of these pronunciations have been more acceptable than the other in this accent 90 years ago, do you think?

Fortunately I can quite easily answer this question. In the 1956 edition of EPD (and perhaps earlier — I haven’t checked), Daniel Jones, while still prioritizing the FOOT vowel on this word, commented
Note.—The use of the variant ruːm appears to be much on the increase.
This suggests that previously it had been unusual. I conclude that in the 1920s rʊm would have been the more usual pronunciation in RP.

Returning to the present day, the pronunciation of room as an independent word is not necessarily a guide to its pronunciation in compounds. Some who say ruːm for this word on its own nevertheless say rʊm in bedroom.

Personally, I’m with the minority who say rʊm in all environments.

We also get some fluctuation between GOOSE and FOOT in broom and groom. But as far as I know boom, doom, loom, and zoom always have . So does bloom, although in the old-fashioned mild BrE expletive blooming some use ʊ.

Tuesday 21 September 2010


Dictionaries tell us that the interjection/exclamation hallelujah is pronounced ˌhælɪˈluːjə (or -lə-). However, this is a word probably more often sung than spoken, and we all know that in singing strange things happen to pronunciation.

From the point of view of English, hallelujah is a strange word anyway. Apart from foreign proper names, it is the only word in which the spelling <j> corresponds to the pronunciation j (palatal semivowel). It has an alternative form, alleluia, which comes to us via Greek rather than direct from Hebrew and lacks this spelling oddity.

In the choir I belong to we are already gearing up for our Christmas show, and one of the works we will be performing is Handel’s Hallelujah Chorus, which is why I am thinking about this word. (The title itself is phonetically interesting, since it normally undergoes so-called stress shift to ˈHallelujah ˈChorus, in accordance with the usual rule.)

One thing that happens in singing is that vowels that would be weak in ordinary speech tend to become strong. So the second syllable of hallelujah is sung with (or an italianate monophthongal e), the final syllable with ɑː (or a).

The other thing that happens is that suprasegmental features (length, stress) have to be modified to fit in with the music. In Handel’s masterpiece we mostly get the usual (ˈ)halleˈlujah, but sometimes the special stressing halˈleluˈjah. In the fragment below, from the score as adapted for male voices only, you can see that at one point, while the tenor 1s and the basses are singing reigneth and the baritones have rapid normally-stressed hallelujahs, the tenor 2s stress the le syllable.It all sounds magnificent. If you’re in or near London, do come and hear us in December.

Monday 20 September 2010

are two heads better than one?

When I was writing my intonation book (English Intonation, CUP 2006), one point that I tried to make clear is that some aspects of intonation are more important than others. This is particularly relevant to EFL learners, who would do best to concentrate on the major issues and ignore, or at least leave till later, various minor ones.
So I organized the book in such a way that — after a brief introductory chapter — one chapter is devoted to each in turn of the “three Ts”: tone (falls, rises, and fall-rises), tonicity (which words do we accent?), and tonality (how do we divide the material up, where do we place boundaries?). The matters I regard as “the less crucial choices” (p. 10) I relegate to chapter five, “Beyond the Three Ts”. Among them are the prenuclear part of the intonation phrase (preheads, different kinds of head), finer analysis of tone (e.g. high fall vs. low fall), non-nuclear accenting, major and minor focus, and a discussion of which function words are (against the general rule) typically accented.

Now I read in David Deterding’s blog an account of a presentation given in China by my colleague Francis Nolan, who made an additional point.
However, it is not so important to imitate the finer distinctions of the intonational tunes of native speakers, partly because there is a huge amount of variation in tone usage in Britain and elsewhere, so listeners are accustomed to hearing substantial differences among the people they talk to. To support this, he played lots of data from speakers from around the UK and Ireland.
I think this is exactly right. Deep down, nearly all native-speaker varieties agree very substantially in the way they use intonation. Superficially, there are considerable differences in the details of pitch movement.

However, a questioner in the audience disagreed, saying that differences in head types must be important, since they are “shown” in my book. David says, rightly,
While [the account of different types of head] is almost certainly an accurate description of the intonational patterns of native speakers of RP British English, there is no way that listeners will misunderstand the message if a non-native speaker uses a rising head rather than a high head. But the questioner was adamant that the distinction is absolutely vital. It is in the book by John Wells, she insisted, so it must be important.

You can see my dilemma. If I hadn’t included the possibility of rising heads in my account, I would have been rightly criticized for lack of completeness. If I had followed the O’Connor and Arnold (1973) model, and presented rising head plus high fall as one of ten apparently equally important tunes — this is their number 6, the “Long Jump” — I would have failed to make the point that the distinction between this pattern and high head plus high fall (their number 2, the “High Drop”) is not actually terribly important.

I have assured David that I agree with him and with Francis. Perhaps in the book I ought to have made my point more clearly, so that all my Chinese readers could grasp it more readily.

Friday 17 September 2010


The word plantain has two quite distinct meanings. In the England of my boyhood it referred only to a garden weed belonging to the genus Plantago. Plantains grew on our lawn, amongst other places, to the annoyance of my father who wanted to have a grass-only green sward.

When I started to visit the West Indies I discovered its other meaning: a fruit of the genus Musa, similar to the banana. Scientifically, there is apparently no sharp difference between bananas and plantains: we tend to call them plantains if they are relatively large and eaten cooked, as a vegetable, and bananas otherwise. (The leftmost items in my picture are plantains. Don’t ask me where green bananas, eaten as a cooked vegetable in Jamaica, fit in.) Wikipedia says
Plantains tend to be firmer and lower in sugar content than dessert bananas. Bananas are most often eaten raw, while plantains usually require cooking or other processing, and are used either when green or unripe (and therefore starchy) or overripe (and therefore sweet).

According to the OED, plantain also has a third meaning (“now rare”), a plane tree.

Anyhow, this note is not about the meaning(s) of the word but about its pronunciation.

Although in LPD I prioritize ˈplæntɪn, I’ve always actually said ˈplɑːntɪn myself. So I would put the word in that subset of words that hover between the TRAP and BATH categories, with æ or ɑː in BrE, æ in AmE. (In AofE, page 135, this is set 59´, which includes other cases such as alas, exasperate and plaque.) In LPD I give the AmE pronunciation correspondingly as ˈplæntən (where ən stands for either syllabic or ən).

But in an American television program(me) I recently heard several people refer to it as a plænˈteɪn. I don’t know how prevalent this stressing is in the US: it’s not shown in the AmE dictionaries I have to hand. I suspect it’s a spelling pronunciation invented rather recently.

Intriguingly, the current Wikipedia entry not only gives this pronunciation for AmE but also, wrongly, attributes it to me!

Thursday 16 September 2010

instances of incidence

Spelling mistakes can be an interesting indication of facts of pronunciation. Here is something I saw in Saturday’s Guardian.They are of course not incidences of cold-calling but instances of it.

Many people do not, or do not consistently, distinguish in pronunciation between instance and incidence.

An incident is “an event, especially one that is unusual, important, or violent” (LDOCE); incidence [usually singular] is “the number of times something happens, especially crime, disease etc”. Despite these dictionary definitions, a quick Google search throws up 35,000 instances (yes!) of isolated incidences.

I wrote about this on 9 May 2008. Here’s what I said then.
The confusion arises because instance and incidence may be pronounced identically in rapid speech. This is because of the possible disappearance of the vowel of the middle syllable of incidence. We can, and often do, go straight from the s to the d, omitting the weak vowel (ə, or a conservative ɪ) that would otherwise stand between them. Since under these circumstances the d, now abutting on a voiceless consonant, gets devoiced, the result is that the -sd̥- of incidence ends up very similar to the -st- of instance.

In moderately-paced speech the deleted vowel seems to leave some compensatory lengthening of the preceding consonant: ˈɪn(t)sːd̥ən(t)s.
In rapid speech, however, I think this subtlety of timing can be lost, making incidence as good as homophonous with instance.

At UCL some of us at one time called this phenomenon ‘pseudo-elision’, as opposed to true elision where the deleted segment supposedly leaves no trace at all.

We find the same thing in words such as trinity, comedy, Cassidy, quality, university (can it rhyme with thirsty?). In trinity the tongue tip may remain in place on the alveolar ridge as we pass from the (lengthened?) [n] to the [t], with no intervening vowel.

At the time I wrote that, dear readers, you could not comment on it. Now (if you wish) you can.

Wednesday 15 September 2010


Judging by the comments on yesterday’s posting, it looks as if I sometimes don’t bang the drum loudly enough about what I have published. Although it was somewhat off-topic, people commenting started discussing the pronunciation of one. And some people’s perception of reality appeared to be very, shall we say, partial.
"One" has the LOT vowel, not the STRUT one, for an awful lot of people.In fact, the the STRUT vowel there sounds very archaic/dialectal to me
To me, pronouncing "one" with STRUT is part of a cruel parody of ultra-conservative RP ("where hez wun's het gawn?").

So last night I felt obliged to point out that in my Longman Pronunciation Dictionary I made known the results of the preference poll I conducted into the pronunciation of this word. It’s in the second edition (2000) onwards.
Here is the graphic from the current, third, edition (2008).
You will see that although there is a trend towards a preference for wɒn over wʌn (or for some perhaps wʊn, wən — anyhow, STRUT) in BrE, in no age group does it reach 50%. Putting it another way, for all age groups there is a majority reporting a preference for the STRUT vowel, not the LOT vowel.

In my lectures about the research underlying LPD I have for many years now been reporting this finding as illustrating a gradual trend of reduced deference towards RP. The fact that more younger people than older report a preference for ɒ in one and for æ in chance can be seen as a greater willingness on the part of northern respondents to report a preference for their own pronunciation in cases where it is known to deviate from the perceived norm (RP: wʌn, tʃɑːns).

Let’s hear no more nonsense about wʌn being “archaic”, “dialectal”, or “a cruel parody”.

On the other hand perhaps LPD does deserve credit for being the first (I think) pronunciation dictionary to mention the existence of the variant wɒn. The Cambridge EPD has now followed, but ODP ignores it, as does the OED. As far as I am aware, so do all published general dictionaries.

Tuesday 14 September 2010

the luv u giv

In my teens I spent many happy hours in public libraries, so became pretty familiar with the Dewey Decimal Classification that they used.
This subject classification system was devised by an American, Melvil Dewey (1851-1931). He was christened Melville, but being an advocate of a rationalized spelling system preferred to spell his name Melvil.
One of his publications (1912) was entitled Abridged Decimal classification and relativ index for libraries, clippings, notes, etc.. Note the spelling of relativ. Dewey believed in eliminating redundant and misleading vowel letters. He founded a Spelling Reform Association in 1886. You can see a sample of his rationalized spelling here. (I’m surprised, really, that he didn’t change the spelling of his surname to Dewy.)

I was reminded of the issue of spelling reform by reading this morning of the memorial tributes left by people mourning the tragic death of a teenager who fell from a tower block in London. According to my newspaper, one tribute read
RIP my darling, luv u 2 bitz. Gone but never forgotten.

This is not a bit of private informal txtng but a kind of public document, displayed in circumstances that can only be characterized as formal. In the absence of any official rationalization of spelling, people are more and more doing their own thing. Luv represents the pronunciation directly; love doesn’t.
Consistency might also demand gon in place of gone.

The indefatigable Masha Bell, author of Understanding English Spelling (Pegasus, 2004), has recently been discussing “surplus -e endings” in her blog (1 August).
As she points out, the use of final silent -e to signal a long vowel (as in rate, compare rat, and hope as against hop) is undermined by cases of final silent -e that have no such function: determine, famine, medicine etc.
She also compares the ‘redundant’ -e in such words as accurate, adequate, delicate, private, with the ‘phonic reliability’ of words such as accelerate, assassinate, calculate. (She doesn’t use phonetic transcription, which would perhaps bring the point out more clearly: ˈækjʊrət, ˈædɪkwət, ˈdelɪkət but əkˈseləreɪt, əˈsæsɪneɪt, ˈkælkjuleɪt.) This implies, interestingly, that perhaps we ought to complicate our spelling by introducing a difference between moderate (verb, ˈmɒdəreɪt) and moderat (adjective, ˈmɒd(ə)rət), or separate (verb, ˈsepəreɪt) and separat (adjective, ˈsep(ə)rət) and other similar pairs.
It would certainly, I think, be advantageous to abolish the -e not only in love but also in have and give. These three very basic words immediately undermine the ‘magic e’ rule that reading beginners are taught. It would be nice, too, to be able to distinguish liv (verb, lɪv) from live (adjective, laɪv).

Monday 13 September 2010


Lynne Murphy sent me a message about the
BrE and AmE pronunciations of Nokia, rendered as KNOCK-ee-ah and KNOW-kee-ah, respectively.

…the single consonant should signal an /o/ in my spelling world, so why the 'knock' vowel?

It’s true: in Britain we (all?) say ˈnɒkiə, with the LOT vowel, whereas Americans say ˈnoʊkiə, with the GOAT vowel. And it’s also true that the spelling o, when followed by a single consonant plus ia, otherwise corresponds to GOAT: phobia, Cambodia, magnolia, begonia, utopia, ambrosia, Scotia. (OK, anoxia is an exception, but x is pronounced as two consonant sounds, not one.) So the BrE pronunciation of Nokia does seem to be exceptional from a spelling-to-sound point of view.

(The rule in question does not apply to all vowel letters in this context: cf. Lydia, familiar, Abyssinia etc., with short ɪ. In bulimia we get not only ɪ but also unexpectedly . The same is true of memorabilia.)

It seems far-fetched to imagine any influence from gnocchi ˈn(j)ɒki. Anyhow, I believe Americans mostly say ˈnɑki if they know the word at all.

There is, however, a parallel in yog(h)urt, which is ˈjɒɡət in BrE but ˈjoʊɡɚt in AmE, and it may be relevant that in BrE we are happy to map ‘foreign’ o onto our rounded LOT, ɒ, whereas Americans are less ready to map it onto their corresponding unrounded and therefore phonetically more distant ɑ. However this plausible line of reasoning falls down when we consider the Australians, who use their GOAT vowel in yoghurt despite having a LOT vowel that is rounded, just as in BrE. (The word comes from Turkish yoğurt joˈuɾt.)

I don’t know what Australians do with Nokia.

In Finnish it’s ˈnokia, just as written. Finnish has contrastive vowel length, and these vowels are short. The company is named after a town near Tampere.

Friday 10 September 2010

a cat that malts?

As far as I know, malting is an activity performed only by brewers, distillers, and manufacturers of malted milk. So what kind of beverage is the cat in this cartoon preparing?

What the cat is doing, of course, is meant to be moulting (BrE) / molting (AmE): tiresomely shedding hair all over the carpet, the furniture, and people’s laps.

This is an interesting spelling mistake, because it seems to imply that the cartoonist, Nick Edwards, pronounces malt and moult identically. If, like me, he pronounced malt mɔːlt, i.e. with the THOUGHT vowel, and moult məʊlt, with the GOAT vowel, he would be unlikely to confuse them in spelling. The same applies, if like many other British people, he pronounces malt as mɒlt, with the LOT vowel, just as long as it’s still distinct from moult. And the point still applies if, like many people, he has a noticeable pre-dark-l allophone for GOAT, so pronounces moult as something like mɒʊlt.

Sorting out the possible mergers of vowels in the context of a following preconsonantal or final l is tricky. Even in London, let alone the rest of the southeast of England, there seems to be considerable variability. A trainman on Wimbledon station regularly announces trains as calling next at ˈeosfɪod Earlsfield, implying the existence of possible homophones such as well–whirl and fill–feel. One of the coaches at my running club, who would never do anything so vulgar as to drop an h, nevertheless momentarily confuses some of us by having hill and heel as homophones hɪo.

This is by no means only a working-class phenomenon: I remember the eminent (and rather posh) phonetician Gillian Brown telling me she had several of these contextual mergers of the back vowels before l in her own speech.

There is one group of southeasterners who, unlike most, merge LOT and GOAT in this environment, so that doll and dole are homophonous. Furthermore, rather over half of the English have LOT rather than THOUGHT in the set fault-malt-halt-salt (see LPD charts for salt). If the cartoonist belongs to both these groups, there is our explanation of confusion over the spelling of moult.

The cartoonist is aware of the misspelling, and says he will change it eventually. Meanwhile, let me thank him for providing me with an apt example.

Thursday 9 September 2010

polyglot text-to-talking head

Here’s a nice bit of software for everyone to play around with. It’s a multilingual text-to-speech device. You input a piece of text, select the appropriate language from the 27 that are available, choose a ‘voice’ (male or female; there are nearly 30 different named talkers available for English, from assorted geographical locations), add an ‘effect’ if you wish, press the “Say It” button, and hey presto, listen to the result.

I did some quick testing.

The device seems to be sensitive to British-American differences. I tested it on the word tomato, and the BrE voice I had chosen duly came up with -ˈmɑː-, the AmE one with -ˈmeɪ-. Same with centralization: BrE -laɪˈz-, AmE -lɪˈz-.

I tried it with short phrases in German and French, with good results.

And all with an animated talking head!
Text-to-speech synthesis seems to be making excellent progress.

Click here, and have fun. (Sorry, blogspot doesn’t allow me to create clickable graphics.)

Wednesday 8 September 2010


A linguolabial is a consonant articulated by the tip or blade of the tongue against the upper lip.

Linguolabials are found in the consonant inventories of very few languages. Ladefoged and Maddieson, in their book The Sounds of the World’s Languages (Blackwell, 1996), give some examples from Tangoa, a language spoken on an island belonging to Vanuatu, in which linguolabials contrast with both bilabials and alveolars for plosives, nasals and fricatives, e.g. t̼et̼e ‘butterfly’.

So much for linguolabials realizing separate phonemes. A different matter is the style-governed articulation of alveolars or dentals as linguolabial, something occasionally encountered in English. People on the VASTA discussion list have recently been discussing the articulatory habits of Britney Spears, the singer, who sometimes does precisely this with the English consonants l, θ and ð. I suppose she thinks it’s sexy. See this video clip.

Apparently Britney lip-syncs all her songs.

Tuesday 7 September 2010

antinomy, antimony

I heard ‘renumeration’ (remuneration) again on the BBC R4 Today programme this morning (blog, 14 July).

I was surprised, though, to find a similar error in a recently published paperback I am reading. It is The Man who Knew Too Much, by David Leavitt, an account of the life and work of the mathematician and cryptographer Alan Turing. (I recommend it, by the way.)

Not once but twice Leavitt refers to a logical paradox which he calls Russell’s antimony.

But even those of us who didn’t do chemistry at school probably know that antimony is the name of a chemical element, the one whose atomic number is 51 and whose symbol is Sb. It is now “increasingly being used in the semiconductor industry as a dopant for ultra-high conductivity n-type silicon wafers, in the production of diodes, infrared detectors, and Hall-effect devices”.

What Russell gave his name to was properly an antinomy, a contradiction, paradox, or conflict.

I’m just a little shocked, not so much that the author of the book got this wrong (anyone can make a mistake), but that no one at Weidenfeld and Nicolson, publishers of the hardback in 2006, noticed it, no readers of the hardback brought it to the publishers’ attention, and no one at Orion/Phoenix, publishers of the 2007 paperback that I am reading, noticed it either.

The stress pattern of antinomy is antepenultimate, ænˈtɪnəmi, like that of other compounds of Greek -nomy: autonomy, economy, astronomy, taxonomy. (It’s also the usual pattern for the book of Deuteronomy in the Bible, though when I was at school our headmaster made us call it ˈdjuːtərəˌnɒmi, to bring out our awareness that this was the Second book of the Law.)

The stress pattern of antimony, on the other hand, is initial: BrE ˈæntɪməni, AmE ˈæntɪmoʊni. So it belongs in this respect with ceremony and testimony.

The Greek word for law, νόμος nómos, has a short vowel between the nasals, whereas the Latin suffix -mōni-, in caerĭmōnia and testĭmōnium, has a long one. Although the etymology of antimony is obscure, the OED traces it back as far as a medieval Latin form antimōnium, evidently assimilated to this pattern. That is ultimately the reason why these two words of such similar appearance, and which are evidently confusable, have different stressings.

An interesting footnote: the OED mentions that the French name of the element, antimoine, has been interpreted by popular etymology as meaning ‘monks’-bane’.

Monday 6 September 2010


‘Lynneguist’ (Lynne Murphy), in her always-entertaining blog separated by a common language, currently discusses British and American words such as squidgy podgy pudgy splodgy dodgy.

It turns out that squidgy, which is apparently British but not American, dates back only to 1973 in its current meaning of ‘moist and pliant; squashy, soggy. Esp. of food’. In a different meaning, ‘short and plump; podgy’, it has been around a little longer, since 1891, and is first attributed to Kipling. (Why did Princess Diana’s lover give her the nickname Squidgy? She was not short and plump, but rather tall. ‘Moist and pliant' would seem to fit the bill better.)

Kipling also supplies the first OED citation of squdgy [sic], which Lynne aptly characterizes as “a word that looks like a typo”.
Why does it look like a spelling mistake? Because in English the letters <qu> are a digraph representing kw or sometimes k, and are always followed by a further vowel letter: quantity, question, quick, quorum, obloquy. The vowel can be silent, as in antique, but it’s got to be there. But in squdgy it isn’t.

Excursus: the third word in the name of the village Stow cum Quy in Cambridgeshire looks odd, too, but actually fits in with the usual pattern. (Most sources say Quy is pronounced kwaɪ, though Wikipedia currently disagrees and goes for kaɪ.)

Since squdgy is pronounced (by those who use it, of whom I am not one) as ˈskwʌdʒi, logically it ought to be spelt squudgy. But the doubled u looks odd (despite the familiar Latin words equus and vacuum) and is hence avoided.
There seem to be no other words that contain the sequence of sounds kwʌ.
The historical explanation is no doubt to be found in the fact that ʌ used to be ʊ. The sounds kw can be followed by most vowel sounds (queen, quick, quest, quack, qualm, quantity, quorum, quirk, equate, quote, quite, (quoit), queer, square), but never by ʊ uː aʊ ʊə. One of these, , may be an accidental gap, but the other three seem to show that you can’t have a close back vowel after kw.

The only other word spelt with qu followed immediately by a consonant is Qu’ran, Quran, alternative spellings for Koran. Here of course the letter q constitutes a transliteration of Arabic ق, as in the u-less Arabic borrowings qat, Qatar, qibla.

My favourite Chinese beer, Tsingtao / Qingdao 青岛 1tɕʰɪŋ 3tɑu, English ˌtʃɪŋˈdaʊ, has nothing to do with the case.

Friday 3 September 2010

a nice ice

An anonymous commentator on yesterday’s blog (meant to be for the previous day) expressed not only disbelief at the suggestion that native speakers of English make a distinction between an aim ən eɪm and a name ə neɪm but also dismay when native speakers (including me) insisted that we do.
The picture above is taken from John Trim’s EFL practice book English Pronunciation Illustrated (CUP, 2nd edition 1975).
Another commentator, going by the name of army1987, supplied a link to a Language Log posting (actually a follow-up to this one). Here you can find sound clips taken from a large spoken (telephone) corpus and test yourself on your ability to hear the contrast.

Measurements of segment duration show that final n, as in an ice (cream) is typically shorter in duration than initial n, as in a nice (day).

Impressionistically, I find the distinction to be relatively robust. OK, like many contrasts it may be obscured in rapid speech, particularly under less than ideal noise conditions; but, as I mentioned in a comment yesterday, I do remember being struck (and irritated) by a French colleague’s repeated mispronunciation of an L as ə ˈnel, which sounds like a knell rather than like what was intended. He was carrying over into English the French habit whereby final liaison consonants are regularly resyllabified with a following initial vowel: my standard example is les États-Unis

Historically, the distinction may presumably once have been less robust, judging by the etymology of adder and, in the other direction, of newt. (In the case of orange, from Sanskrit nāraṅga via Arabic nāranj, the reanalysis seems to have taken place in Italian, and the word then came to us via French already lacking the initial nasal.)

The existence of the distinction in contemporary English doesn’t stop us punning over the similarity.

Thursday 2 September 2010


Thanks to Ludwig Tan for a link to this BBC news article about one Heather Quinlan, who is “on a quest to record [the] full variety [of the New York accent] for posterity”.

I have to say that in my opinion displaying a sign reading “Do you have a NY Accent? Then talk to me” might produce a rather unbalanced sample of speakers. But I suppose it’s better than nothing.

Despite its publication on a British website, the article betrays an American point of view. Declaring that New Yorkers pronounce “Wa-ta” not “Water” is pretty meaningless for an English readership. But then declaring that they say “Awe-ful” not “Awful”, and calling this “’A’ broadened” is pretty meaningless from every point of view.
(The latter assertion is presumably about the THOUGHT vowel, which in NYC can be an opening diphthong, almost [], rather than the [ɔ, ɒ, ɑ] monophthong that we hear from other Americans. But for a British readership a “broadened” A would be taken as referring to the backed and lengthened vowel of BATH in RP etc, bɑːθ as against northern English baθ or American bæ(ː)θ, bɛəθ — a quite different matter.)

Fortunately the author of the article consulted Bill Labov, who introduced some scientific objectivity to the discussion.

The embedded clip, a compilation of (presumably) New Yorkers saying this and that, starts with a young lady who refers to them as niˈjɔrkɚz. No one comments on the apparent contradiction between this rhotic pronunciation and the assertion that New Yorkers drop r. (As we know, rhoticity in NYC is actually variable: some rs are dropped, some aren’t. It partly depends on the speaker’s socioeconomic status, their level of education, and the degree of formality. See Labov’s Social stratification of English in New York City.)
"A young person growing up in a college-educated household might, for example, complete the 'r' pronunciation even in his everyday speech," says Prof Labov.

Americans who know nothing about phonetics can nevertheless readily identify a New York accent. The same is not generally true of the British, to whom New Yorkers just sound American.

Over to you, Amy Stoller.

Wednesday 1 September 2010

linking semivowels (ii)

I think we’d better continue yesterday’s discussion for a further day.

Adam Brown very reasonably compares so-called linking /j, w/ with linking /r/, asking and answering his own question.
Is it worth teaching linking /j, w/? Well, one benefit is to avoid an overabundance of "linking" glottal stops in such situations.
I assume you would say the same thing about linking /r/. Just as 'two evils' and 'two weevils' are not the same, so Mark Anthony did not say "Friends, Romans, countrymen, lend me your rears."

Yes, and likewise more ice mɔːr aɪs sounds different from more rice mɔː raɪs — in the same way as an aim ən eɪm sounds different from a name ə neɪm. Syllable-final sonorants are shorter, more lightly pronounced, than the longer, more deliberate, syllable-initial ones.

The difference between linking r and the postulated linking semivowels is that native speakers (of RP-type accents) actually do typically insert an extra segment in the linking-r position. In non-rhotic accents the word more, pronounced in isolation, has a straightforward CV structure, m plus ɔː. Before a word beginning with a vowel, on the other hand, it can contain an extra segment, CVC, m plus ɔː plus r.

The analogy for my would be that in isolation it is CV, m plus , but in prevocalic position (my arms) it becomes CVC, m plus plus j. Since English j by definition is a non-syllabic palatal glide, and since already contains a non-syllabic palatal glide, it is difficult to see what the realization of j might actually consist of in the supposed homosyllabic sequence maɪj. (The case of I yearn or “my yarm” is different, with the palatal articulation now partly in the second syllable.) See vp’s comment on yesterday’s blog made early this morning.

I am confirmed in my view that “linking /j/ and /w/” are figments of the imagination. That does not necessarily imply that they are pedagogically valueless. I am willing to recognize that sometimes teaching something that is not strictly true may nevertheless be justified if it leads to better results than not teaching it. (That is the justification for some of the nonsense I encountered from the voice teacher whose classes on Accents and Dialects I attended anonymously when researching my own book on the subject — telling people to “place” the voice in the middle of the forehead, and suchlike.)

If talk of linking semivowels is a useful technique for getting EFL students to avoid prevocalic glottal stops (“hard attack”), then so be it. That might justify its use by teachers of learners whose first language is, for example, German.
But not of those whose first language is Spanish or Italian.

As for the claim that it is useful to help learners “avoid strong glottal attacks that can cause vocal issues”, I feel I need to see some kind of evidence that glottal stops cause voice problems. If the use of glottal stops really does trigger pathological conditions, you have to wonder how monoglot native speakers of (northern) German manage. In that language all initial vowels normally have a reinforcing glottal stop (“hard attack”). Given that, how do the Germans ever manage to remain in vocal health?