Friday, 29 July 2011

allegedly aged

How do you pronounce aged?

It depends on where it stands in the sentence. If it is used attributively, before a noun, meaning ‘(very) old’, then it has two syllables, ˈeɪdʒɪd (or, for some, ˈeɪdʒəd). This is what you say in an aged woman or my aged parents. It is also the pronunciation we use in, for example, care of the aged, where you could argue that it is used attributively before a deleted (understood) noun people.

But when it is used predicatively, meaning ‘having a specific age’, then it is pronounced as a monosyllable, eɪdʒd. This is what you say in children aged 5 or over or a man aged between 30 and 35.

There are one or two other -ed words that vary in the same fashion, notably blessed. Attributively, disyllabic: a moment of blessed silence; where’s my blessed notebook? Predicatively, monosyllabic: we’re both blessed with good health; the couple had their marriage blessed (by the vicar). Against this rule, however, in the hymn Our blessed redeemer, ere he breathed | his tender last farewell the word has to be pronounced as a monosyllable despite being attributive.

There’s also accursed, which for me is always əˈkɜːsɪd. But others always say əˈkɜːst. Others again may vary as with aged and blessed. In modern English, though, the word is really only used attributively: we say all this accursed mud, but not ?all this mud is accursed.

As we all know, there are a few adjectives in -ed in which the ending is irregularly pronounced as a separate syllable, i.e. against the rule that this pronunciation belongs only after stems ending in t or d. Examples are crooked, learned, naked, rugged, wicked, wretched, all disyllabic. (But as verb forms, crooked and learned are monosyllabic: she crooked krʊkt her finger, he learned lɜːnd what had happened.)

Derived forms in -edly and -edness seem mostly to have the syllabic pronunciation. So markedly and markedness have three syllables each, while supposedly and allegedly have four. But there are also those that have a nonsyllabic ed: determinedly, ill-favouredness, good-naturedness. I think shamefacedly can go either way.

I heard Judge Judy on television speaking of someone’s əˈledʒɪd crime. But I think we normally say əˈledʒd, and that her pronunciation (possibly a one-off slip) was a back-formation from allegedly. So when she said your alleged striking of the defendant with trisyllabic alleged she was thinking of the underlying unnominalized you allegedly struck the defendant.
_ _ _

This blog will be suspended for the month of August. During this time I may possibly see some of you face-to-face at the UCL Summer Course in English Phonetics in London or at the International Congress of Phonetic Sciences in Hong Kong.

Next posting: 1 September.

Thursday, 28 July 2011

Unicode 6.0

Phonetic-symbol anoraks/nerds/geeks can have hours of fun browsing the Unicode Standard, the repository of all the characters that can be displayed on a modern computer screen (blog, 22 Jan 2007). If you haven’t got the book (which is hefty), browse online.

Now there’s a new version of the Standard, 6.0 (well, it came out last October, actually). Unlike previous versions, it has not been published as a printed book, but is available only online.

So what’s new in version 6.0? In brief: there are 2,088 new characters, including (I quote)
• over 1,000 additional symbols—chief among them the additional emoji symbols, which are especially important for mobile phones
• the new official Indian currency symbol: the Indian Rupee Sign
• 222 additional CJK Unified Ideographs in common use in China, Taiwan, and Japan
• 603 additional characters for African language support, including extensions to the Tifinagh, Ethiopic, and Bamum scripts
• three additional scripts: Mandaic, Batak, and Brahmi

There are also extensive technical changes to do with character properties and format specifications.

Two new Cyrillic characters cater for Azerbaijani. Two new Arabic characters and ten new Devanagari characters cater for Kashmiri. Thirty-two new Ethiopic characters cater for Gamo-Gofa-Dawro, Basketo, and Gumuz. Complete new blocks of letters cater for Mandaic, for Batak, and for Brāhmī.

Is there anything of particular interest to phoneticians and IPA users?

How about a symbol for a voiceless retroflex lateral fricative? A sort of combination of ɬ and ɭ? It’s not (yet) an official IPA symbol, but it’s a logical combination of two. Here it is, U+A78E. (Unicode numbers are given in hexadecimal and prefixed with the identifier U+.)
If you’ve always wanted a COMBINING DOUBLE INVERTED BREVE BELOW, it’s now available. But unless you’re a Uralic Phonetic Alphabet aficionado, you’ll have managed without. Do you have a use for subscript h k l m n p s t? I doubt it. Even if you do, you’d probably simply use the subscripting tag <sub> </sub>, as I have just done. In Unicode 6.0 they’re ready-made at U+2095 to U+209C.

Students of the minority languages of China may welcome three new Bopomofo characters to cater for Hmu and Ge. (Bopomofo is a phonetic notation system based on Chinese characters.)

It’s one thing to have a symbol recognized in Unicode and assigned a U+ number. It’s something else for the new symbol to become available in an available font. We’ll just have to wait and see if and when these new characters make an appearance in documents on our display screens.

Don’t hold your breath.

Wednesday, 27 July 2011

our cake

As we were chatting at a summer party last weekend, someone brought up the possible minimal pair archaic vs our cake. We were trying to think of a context in which there might be a plausible confusion between the two. Finally someone came up with a scenario involving two daughters visiting their elderly mother. They brought with them a cake as a gift, which their mother promptly stored in a tin. But she had numerous cake tins, all looking much the same, and tended to accumulate old cakes or pieces of cake and keep them in the tins for months. When it’s time for tea, one of the daughters looks at all the cake tins, picks one, and asks her mother “Is this one archaic/our cake?”

Non-native speakers may be surprised that this is even considered a minimal pair. Surely aʊə ˈkeɪk is rather different from ɑːˈkeɪɪk. Well, yes, if that’s what you say. But if you are one of the many NSs who pronounce our as ɑː(r) (rather than as aʊə(r)), then the difference is only a matter of the vocalic material between the two velar plosives, keɪk vs keɪɪk, which comes down to a subtle question of timing.

There are two reasons why our might be monophthongal ɑː(r). It might be through the operation of the optional process of smoothing, which deletes the second part of the diphthong when followed by another vowel [and, I should have added, compression, which makes two syllables one]. This is what gives us RP pɑː power, ˈɡɑː striːt Gower St, etc. (The quality of the resultant monophthong may or may not be identical to that of the ordinary ɑː of START words.)

But it might also be simply that ɑː is the default pronunciation of our. Not everyone has our as a homophone of hour. That is true for me, and for an unknown number of other NSs. The two words make a possible minimal pair, ɑː our vs ˈaʊə hour. And ɑː is not just a weak form: it’s the strong form too. (I ought to do a preference survey for this.)

When I was taught the Lord’s prayer as a child, it began ˈɑː ˈfɑːðə, hu ˈɑːt ɪn ˈhevn̩.

I might ask you “Did your bus come on time? We had to wait for ˈa(ʊ)əz for ˈɑːz.”

I don’t think there are many NNSs who pronounce our, ours as ɑː(r), ɑː(r)z. On the other hand there may well even be a majority of NSs who do. No one knows.

Kenyon & Knott included ɑr as a possibility for AmE as long ago as 1953 (possibly even in 1944 — I haven’t got the first edition to hand). For BrE priority goes, I think, to Jack Windsor Lewis, in whose Concise Pronouncing Dictionary (1972) ɑː is included just as a weak form.

When still edited by Daniel Jones, EPD did not recognize the ɑː variant. It was only when Gimson and Ramsaran took over that it was acknowledged as a possibility. Now the OED, too, has caught up.

Tuesday, 26 July 2011

hearing plosives

It’s an old truism of acoustic phonetics that you can’t hear the hold stage of a voiceless plosive (as such).

What you get in the middle of a word such as happy ˈhæpi or lucky ˈlʌki is a short period of silence, as the airstream is for a moment prevented from moving through the vocal tract and out of the body. No air movement means no sound.

How, then, can we identify the place of articulation? How do we know that in the first word we have a bilabial p but in the second a velar k?

We know because of the formant transitions created as the organs of speech move into place for the complete closure (the ‘approach’ phase) and then again as they separate (the ‘release’ phase). You identify the p in happy through its effect on the end of the æ and on the beginning of the i. You identify the k in lucky by what you hear in the course of the ʌ and the i.

In the case of a fully voiced plosive, all you can hear during the hold phase is voicing. Again, you identify the place of articulation through the information contained in the formant transitions before and after, i.e. in the approach and the release. That’s how you know that abbey has a b, ladder a d, and lagging a g.

Let’s get back to the plosive clusters we were discussing on Friday apropos of Gdynia. In an English word such as acting ˈæktɪŋ we normally have the same ‘masking’ phenomenon we observed for the voiced plosives in hugged and Ogden.

In the English kt or gd the plosives typically overlap, in that we make the approach for the second plosive before releasing the first. The sequence of events is velar approach – velar hold – alveolar approach (inaudible, because the velar hold is maintained) – double hold (velar and alveolar) – velar release (inaudible, because the alveolar hold is maintained) – alveolar hold – alveolar release. The only audible phases are the velar approach and the alveolar release. In these the formant transitions supply the clues to the places of articulation.

To identify the place of a plosive it is sufficient to hear either the approach or the release. You do not need both. That is how we can tell that the word is ˈæktɪŋ rather than, say, ˈæptɪŋ or ˈætkɪŋ or ˈættɪŋ.

My impression from my first visit to Poland was, then, that for Polish gd- the two plosives did not overlap in the English way. Rather, the sequence of events was straightforwardly velar approach – velar hold – velar release – dental approach – dental hold – dental release. Rather than taking place during the velar hold, the dental approach was delayed until after the velar release. The tiny transitional nonsyllabic schwa between the plosives is created in the tiny interval of time between the two articulatory gestures, velar and dental.

We can leave the native speakers of Polish to debate whether this non-overlapping is usual, as I supposed, or only found in careful or overenunciated pronunciation, as some seem to claim. My impression is that if we compare English actor with Polish aktor it is typical for the English plosives to overlap but for the Polish ones not to. Similarly with the name Magda.

In strongly Japanese-accented English, on the other hand, a word such as actor tends to have a greater interval between the release of the k and the approach of the t. This space might be identified as a Japanese voiceless ɯ̥ (thus “ アクター”). Typically, it seems to be much longer than the momentary mini-voiceless-schwa of the Polish kt. It reflects the Japanese mora-based timing in which equal time is allotted to each of a, k(u), ta, a.

Monday, 25 July 2011

nuh-nuh (3)

In response to the postings about the taunting tune there has been no one from China, Japan or Korea who acknowledges it as a tune used by children in those countries in the way it is used in Europe. Bernstein claimed it was universal, found in all cultures; but perhaps he was wrong.

From Japan, Masaki Taniguchi writes
I have been thinking about the teasing melody for a week. I have traced my memory of childhood. Now I clearly remember a teasing melody that my friends and I used to use in the early 1960's, which went |so so mi mi | so mi mi |. I also remember two versions of lyrics to go with this melody…
I have just contacted my childhood friend who used to live in the same town in Nagasaki Prefecture. He says that he vaguely remembers such a melody. He also says that today bullying is considered very inhumane and probably such a song has disappeared.
Nobody else that I have asked knows or remembers such a melody. I have asked a professor of music, and he recommended another who, he said, would be knowledgeable in such matters. I asked her, but she said she knew nothing of the sort, except for /ja:i ja:i/ (pitch: HL HL). This, I am sure, is known to all Japanese. It is certainly used for teasing.
I have also conducted a small survey. In a class with 12 students, I sang the melody you introduced in your blog, using the sounds used there, and told them that it was used among English children. I asked them what kind of impression they perceived and what kind of feeling they thought children would mean to express. Out of the 12, seven said it sounded "lively", four said, "teasing", and one said it sounded like a "lullaby". After that, I showed them your blog and we discussed it.

I think this negative evidence is important, because it tends to show that the claim of universality is false.

Meanwhile, from England, Jill House confirms that the phenomenon is extensively discussed in Mark Liberman’s 1978 dissertation ‘The intonational system of English’ (Indiana University Linguistics Club). She also points out that the tune underlies not only ‘Bye baby bunting’ but also the song for the children’s game ‘Ring-a-ring-a-roses’. That’s true for the version I sang as a child, but — as Wikipedia reveals — there are or were other variants of the tune which differ.
Anyhow, thanks to all correspondents.

Friday, 22 July 2011

Gdynia unmasked

I was in my teens when I first became aware of places in Poland called Gdynia and Gdańsk. I remember wondering how they could be pronounced without omitting the initial g. The nearest thing in English seemed to be the ɡəd- sequence in Lady Godiva ɡəˈdaɪvə and Godolphin (nowadays we have another example in Gaddafi), but I knew that wasn’t right. Yet as an initial cluster gd- seemed impossible to pronounce.

Then I realized that English does have the articulatory sequence gd, but in final position. We get it in the past tense of verbs ending in g, thus for example bagged bægd, hugged hʌgd. We get it medially, too, in Ogden ˈɒɡdən. All I needed to do was to transfer this gd to syllable-initial position.

That still seemed very difficult to do. The reason (I know now) is that in English we normally pronounce these plosive sequences as overlapping articulatory gestures. You can’t hear the release of the g in hugged, or for that matter in Ogden, because it is ‘masked’ by the concurrent hold phase of the d. What you hear is a velar approach, a long hold, and an alveolar release (if you’re lucky). And we don’t ever have this sort of thing at the beginning of a syllable.

It wasn’t until I first visited Poland, when I was twenty, that I discovered that in Polish the initial plosives in Gdańsk ɡdaj̃sk and Gdynia ˈɡdɨɲa are not like the English gd in hugged. Rather, when my Polish friends demonstrated the pronunciation to me, they released the velar plosive BEFORE completing the approach for the dental. This inevitably gave rise to a tiny transitional vocoid between the two hold phases, but it was not long enough to count as a separate schwa segment (compare the definite schwa in English Godiva).

If you can play .ogg sound files, Wikipedia has one of Gdynia here.

Despite my A level in Greek, somehow I’d failed to realize that classical Greek has an exactly parallel, but voiceless, cluster in words such as κτείς kteís ‘comb’ and κτίσις ktísis ‘foundation’. (What we did to pronounce those words in the Classical Sixth I can no longer remember.)

The stem of κτείς is κτεν- kten-, and from this is derived the modern zoological Latin name Ctenophora, the phylum of marine animals also known as ‘comb jellies’. In English we abandon any attempt at the initial plosive cluster, and pronounce them simply as tɪˈnɒfərə.

Modern Greek has dissimilated the plosive sequence, making the first element fricative. The modern word for ‘comb’ is χτένι ˈxteni.

So there are two attested escape routes: deletion or dissimilation.

But the Poles succumb to neither of these tempting articulatory simplifications, and persist with a sequence of plosives. Unmasked.

Thursday, 21 July 2011


Watching the goings-on in the House of Commons yesterday, I was reminded of another unusual British surname with a pronunciation over which you might hesitate if you came across it in writing. Andrew Selous is the MP for Southwest Bedfordshire, and when the Speaker called on him by name we heard it spoken: səˈluː.

Older readers will recall the Selous Scouts, the special forces of the Rhodesian Army. Nowadays there is a Selous Game Reserve in Tanzania. Both are named after Frederick Selous (1851-1917), a British explorer and big-game hunter.

The Oxford Names Companion suggests that this surname is English and ‘of uncertain origin, perhaps a habitation name from an unidentified place named with the OE elements s(e)alh willow + hūs house’. I would have thought that a name with that etymology would be more likely to be pronounced ˈseləs (cf. Backus from bǣc ‘bake’ + hūs , Malthus from mealt ‘malt’ + hūs ) — which it isn’t.

But maybe that etymology is not correct. Wikipedia states here that Selous is the Anglicized form of the Dutch name Slous. If that were the case, I would expect it to have been pronounced in Dutch as slɔus, which might give English slaʊs. Wrong again.

Very much more plausibly, Wikipedia tells us in its article on the big game hunter that the name was originally French.
Frederick Courteney Selous was born on 31 December 1851 at Regents Park, London, as one of the five children of an aristocratic family, third generation of French-Huguenot heritage. His father, Frederick Lokes Slous (original spelling) (1802–1892), was notably Chairman of the London Stock Exchange.

As a French name, Slous would be expected to be pronounced slu. Equally, the spelling Selous implies səlu. And French schwa is a very variable entity: cela can be sla or səla. This origin would satisfactorily account for the English pronunciation səˈluː.

Wednesday, 20 July 2011

a gay accent?

The current issue of the online Economist has a piece entitled “Gay pitch, vowels… and lisp?”. (Thanks, Jo-Anne Ferreira, for the alert.) It seems to be based wholly on this piece in Dialectblog, which in turn reports on a recent journal article (Podesva, Robert J (2011). The California Vowel Shift and Gay Identity. American Speech, 86.1, 32-49.).

The question is, is there such a thing as a “gay accent”? Quite rightly, the author begins with a disclaimer.
…before going further, let me state that I believe gay men speak with as wide an array of voices as heterosexual men. I don’t give credence to the idea of a universal “gay voice.”

I would add from my own experience that not only do most gay men not speak in a way that indicates their sexual orientation, but that some men who do “sound gay” are — as far as one can tell — heterosexual. (We could perhaps agree that a good example of this would be the BBC television performer Laurence Llewelyn-Bowen.)

Neither article mentions women’s voices. There is the equal question, whether there exists a recognizable “lesbian accent”. If there is, then similar reservations would apply.

It appears that Podesva’s article is based on the speech of a single gay man living in San Francisco. It turns out that the whole Spring 2011 issue of American Speech is devoted to “Sociophonetics and Sexuality”. So one would hope that it contains further relevant research based on the speech of more than one single individual. Unfortunately the UCL Library has discontinued its subscription to the journal, so I have not yet read this issue.

Let’s get down to the nitty-gritty. As Podesva’s informant’s speech shifts from the formal to the informal end of the formality scale, it is claimed that his accent becomes more “Californian”. This is supposed to involve a number of related vowel changes. Chatting with friends, the subject
exhibits three markers of California English in this latter situation: the word “bad” is pronounced with a vowel closer to the vowel in “bod,” and the vowels in “boot” and “boat” are both pronounced fronter.

An opener TRAP vowel? Fronted GOOSE and GOAT? Does that remind you of anything? Yes, these are characteristic features of British English as compared to American.

Thirty years ago, in Accents of English (CUP 1982, p. 21-22), I wrote
…it is of interest to ask what speech characteristics are perceived as effeminate or mannish, respectively. I suspect that many of them are prosodic matters — intonation, pitch range, rhythm, tempo. […] Many gay men can certainly switch ‘camp’ voice quality and vocal mannerisms on and off at will.

— to which I would now add “and so can many people who are not gay”.

After further discussion, I continued
…it may frequently happen that a pronunciation which would be entirely usual in one locality may sound effeminate in another. This appears to be the case, for example, with the use of a voiceless intervocalic [t] in words such as better, party — normal in England, but in America widely perceived as unmasculine. The same applies, I suspect, to the use of [ɑː] in BATH words.

To which we evidently may be able to add the qualities of TRAP, GOOSE, and GOAT.

If Americans perceive a British accent as sounding gay, do we Brits perceive an American accent as sounding butch?

On the other hand, what about a northern Ireland accent? Voiced intervocalic /t/, back GOAT, harsh voice quality (all = butch), but even opener (and backer) TRAP and even fronter GOOSE (= camp). How confusing.

I think there’s a lot more to it than this.

Tuesday, 19 July 2011


Most readers will be aware that the UK Prime Minister, David Cameron, has appointed Lord Justice Leveson to oversee a public inquiry into phone hacking and media regulation. Sir Brian Henry Leveson QC is also known as Lord Leveson. His surname is one of those tricky British names not necessarily pronounced in the way the spelling would lead you to expect.

According to the BBC Pronouncing Dictionary of British Names (1990), the Christian name (forename) Leveson is ˈluːs(ə)n, and the surname Leveson-Gower, the family name of the Earls of Granville, is ˈluːs(ə)n ˈɡɔː(r).

I suspect that as a forename Leveson is now obsolete, and I don't believe there is any Leveson-Gower now active in British public life, though several members of the family were active in politics in former centuries. Furthermore, Sir H.D.G. Leveson-Gower was a well-known cricketer, captaining the English test side in 1909/10. His brother Frederick was also a county cricketer, and in 1895 played for the Gentlemen of England against Oxford and Cambridge. (Oh, long-lost era of gentleman amateurs!)
At one end stocky Jessop frowned,
The human catapult
Who wrecks the roofs of distant towns
When set in his assault.
His mate was that perplexing man
We know as "Looshun-Gore",
It isn’t spelt at all that way,
We don’t know what it’s for.
But as with Cholmondeley and St. John
The alphabet is mixed,
And Yankees cannot help but ask -
"Why don't you get it fixed?"

EPD/CPD confirms ˈluːs(ə)n in Leveson-Gower, but does not tell us anything about a forename. It does, however, say that the simple surname Leveson is ˈlevɪs(ə)n.

Back to Sir Brian. He does not appear to be related to the Leveson-Gowers. The BBC Pronunciation Unit tell me that some fifteen years ago his office gave the pronunciation of his name as ˈlevəsən, which we can take as a minor variant of ˈlevɪs(ə)n. Jo Kim tells me “I checked this again with the Sentencing Council press office today (Lord Justice Leveson is the Chairman) and they confirmed that ˈlevəsən reflects his own pronunciation“. (Thanks, Jo.)

That didn’t stop the Home Secretary, Theresa May, referring to him as Lord ˈliːvɪsən in the House of Commons yesterday.

Monday, 18 July 2011

nuh-nuh (2)

Friday’s blog garnered a very welcome wealth of comments. It appears that this defiant/mocking/taunting tune is indeed, if not universal, then at least very widespread. It seems to be found in all European languages (and their worldwide offshoots), though for example no one has yet reported it here for Chinese, Japanese, Korean, or any other non-European language.

How did I come to overlook the obviously related tune of Bye baby bunting? (Though the tune given in Wikipedia starts off with a narrower interval than the minor third we were discussing.)

I thought the most thought-provoking comment was one by MKR.
"Research seems to indicate that this exact constellation of two notes [viz., a falling minor third] (and its three-note variant) is the same all over the world, wherever children tease each other, on every continent and in every culture" (Leonard Bernstein, The Unanswered Question, lecture 1, "Musical Phonology"; starting at 27:00 in this video). Bernstein returns to the example in lecture 3, "The Delights and Dangers of Ambiguity" (at 08:45 in this video), where he attributes the prevalence of the pattern to its tonal ambiguity.

I watched the first hour of the first Bernstein lecture.

It seems to me that Bernstein’s discussion leading up to 27:00 involves some rather amateurish linguistics. The universals m and ɑ are not ‘phonemes’ found in all languages: rather, they are sound-types or phonetic segments that are supposedly universal.

Actually, though, and contrary to Bernstein’s claim, WALS reports thirteen languages with no nasals in their consonant inventory. (Some of these, it is true, may make use of m as a positional variant of some consonant that is otherwise not nasal.)
A total of 13 languages in the sample are listed as having no nasals in their consonant inventories. Some of these languages, such as Quileute (Chimakuan; Washington State), Rotokas (West Bougainville; Papua New Guinea) and Pirahã (Mura; Brazil), make no systematic use of nasality in their sound system at all; the last two have especially small phoneme inventories overall. The majority of these languages, however, do make use of nasality, but it patterns in such a way that simple nasal consonants do not need to be considered contrastive segments.

The Quileute language “is famous for its lack of nasal sounds, such as [m]”.

In Central Rotokas, we read,“nasals are rarely heard except when a native speaker is trying to imitate a foreigner’s attempt to speak Rotokas. In this case the nasals are used in the mimicry whether they were pronounced by the foreign speaker or not”.

In Pirahã, while there are no nasal phonemes, /b/, i.e. the consonant that is distinctively voiced and labial, is realized as [m] after a pause.

So Bernstein seems to be wrong about m, at least for Quileute and Central Rotokas. Likewise I am left wondering what evidence there is that, as he claims, “this exact constellation of two notes [viz., a falling minor third] (and its three-note variant) is the same all over the world, wherever children tease each other, on every continent and in every culture”.

It’s certainly a widespread, but is it a universal?

Friday, 15 July 2011


Children in England have a paralinguistic way of crowing ‘you can’t catch me’ and showing defiance or provocation. It is to sing ˈnɜːˈnɜːnəˈnɜːˈnɜː to this tune. (Midi sound clip here.)

The vowel may also be rather opener, ranging over or nɑː.

I don’t know of any established way of spelling this interjection, but nuh-nuh-nuh-nuh-nuh or na-na-na-na-na would perhaps do.

I have never seen any discussion of this item in the literature of paralanguage (but perhaps a reader knows better).

It was in my active repertoire when I was a child and — as far as I know — children in England still use it today. Is it also to be heard in other parts of the UK? And in north America and elsewhere? Is it international? Do French children and Japanese children do the same thing?

I raise this issue to celebrate my having at last learnt how to produce musical notation at will for the computer screen. (Some of you may have been familiar with this method for ages, but it is new to me.) The ABC system makes it very easy to write down the score you want. My source file for the above read simply
X:1reference number
L:1/16unit note length
G3 E2A G3 E3tune body
Then free software converts it into standard musical notation and generates a sound clip. (I used the convenient facility on the website.)

Thursday, 14 July 2011


Speaking in the House of Commons yesterday about setting up an official inquiry into the phone hacking scandal, the Prime Minister David Cameron mentioned that the Culture Secretary, Jeremy Hunt, would have a ‘quasi-judicial’ role in the matter of the attempt by News International, now withdrawn, to purchase the shares of BSkyB that it does not already own.
Downing Street has said the government, with the exception of Mr Hunt who has a quasi-judicial role in the final decision, will back the motion. The prime minister's spokesman said the motion essentially reflected what Mr Cameron had said on Monday.
He pronounced the prefix quasi- as ˈkwɑːzaɪ.

This pronunciation, while by no means unusual, is interesting in that it combines in the same morpheme two different ways of treating Latin words taken into English.

One is to follow the usual English reading rules (spelling-to-sound rules), which treat long vowels as having undergone the Great Vowel Shift. This is what is done for Latin words that are well integrated into English. Thus we have for example creator and major with -eɪ-, and aquarium with -eə- deriving from an earlier -eɪ-. Likewise, we have for example appendicitis and minor with -aɪ-.

The other is to give the long vowels ‘continental’ values, rendering Latin ā as ɑː and ī as . This is what we do with words perceived as being less thoroughly integrated. We usually pronounce errata with -ɑː- nowadays, though that was not always the case. Similarly, in vivo usually has -iː-. (But viva meaning ‘oral exam’ is ˈvaɪvə.)

The word quasi ‘as if, as it were’ actually has short vowels in Latin, quăsĭ. Latin stressed short ă is normally mapped onto English æ, eg aquifer, per capita, so this prefix ought to be ˈkwæsi. But English speakers are often pretty cavalier with Latin vowel quantities in English, and even those who have studied Latin (such as Mr Cameron, who must have done it at Eton) often get them wrong. If, as is usual, the vowels in this prefix are treated as long, we should get either GV-shifted ˈkweɪzaɪ or ‘continental’ ˈkwɑːsi(ː). Dave’s ˈkwɑːzaɪ is a combination of the two.

Given the further uncertainty over whether or not to voice the intervocalic s — is it s or z? — we end up with quite a combinatorial explosion of possibilities for this humble prefix.

Wednesday, 13 July 2011


Steve Bell seems to be confusing an Australian accent with a New Zealand one.

Here, in yesterday’s Guardian cartoon, the Australian Rupert Murdoch is ordering his tame policeman to arrest the Queen. His accent is caricatured by jocular respelling.

While both Australians and New Zealanders — but particularly the latter — can give the rest of us the impression that they are pronouncing DRESS words with the KIT vowel (Rebekah → Ribikah) and TRAP words with the DRESS vowel (shag → sheg), it is only New Zealanders whose KIT vowel is so centralized as to lead us to perceive it as the STRUT vowel. Australians, on the other hand, make it closer and fronter than in many other accents, nearer to [i].

That is the easy way to tell Aussies and Kiwis apart. Get them to say fish and chips. The ones who seem to say “fush and chups” are the Kiwis.

But in Steve Bell’s cartoon Murdoch is represented as pronouncing the Windsor bitch as “the Wundsor Buttch”. Since Murdoch is an Aussie, it would have been better shown as “the Weendsor Beetch” (though of course that too is an exaggeration).

The last respelling, throne as “thrine”, is an interesting attempt to reflect the characteristic Australian GOAT vowel variant in the region of [æ̈ʏ]. (This is the variant that Paul Kerswill observed in Milton Keynes a few years ago, leading to a flurry of media claims about the supposed influence of Australian soaps on British popular accents.)

Tuesday, 12 July 2011

distributive expletives

I was watching Judge Judy on TV the other day when she used the familiar expression expletive deleted. She stressed the first syllable of expletive, as I think Americans usually do: ˈeksplət̬ɪv. But in Britain I think we usually stress the second syllable, ɪkˈspliːtɪv, ek-, ək-.

I have no survey data to back up my impression that this is typically a BrE-AmE difference. Perhaps I ought to include it in my next pronunciation preference survey. Dictionaries tend to imply that both stressings are found on both sides of the Atlantic. Here’s the on-line OED:The only other familiar adjectives ending in -etive seem to be secretive and interpretive, which follow the stressing of the stem, ˈsecret or inˈterpret respectively, i.e. are stressed on the antepenultimate. There’s also suppletive, a term probably known only to grammarians, which I think most people on both sides of the Atlantic pronounce with penultimate stress, səˈpliːtɪv (the OED has an improbable sʌ- for the first syllable). This reflects the stress pattern of supˈpletion. However in LPD I do also give a variant with initial stress, though I really don’t know to what extent, if any, it is in use.

Words ending in -ative are notoriously idiosyncratic. The best I could manage in LPD was this.
In words of three syllables, the first receives the stress, and the suffix vowel is weak (ˈfricative, ˈvocative, ˈlaxative, ˈnarrative; exception creˈative). In longer words, the stress usually falls on the same syllable as in the underlying stem: acˈcusative, conˈsultative, preˈservative; ˈoperative, ˈqualitative, agˈglutinative, ˌarguˈmentative; adˈministrative. There is sometimes a vowel change (deˈrive — deˈrivative), and there are several exceptional cases (comˈbine —ˈcombinative, ˈalternate — alˈternative, inˈterrogate —ˌinterˈrogative, ˈdemonstrate — deˈmonstrative). Where the primary stress is on the last syllable of the stem, the suffix has a reduced vowel (ˌinterˈrogative); but otherwise in these longer words (ˈcumulative, ˈlegislative) the choice between weak-vowelled ət ɪv ǁ ət̬ ɪv and strong-vowelled eɪt ɪv ǁ eɪt̬ ɪv depends partly on social or regional factors, with British English RP tending to prefer ət ɪv, American English eɪt̬ ɪv: see individual entries.

Those in -itive, on the other hand, are straightforward. They are stressed on the antepenultimate, eg comˈpetitive, deˈfinitive, proˈhibitive, ˈsensitive, ˈpositive, inˈtuitive. I am tempted to say the same about those in -utive. We can certainly agree on conˈsecutive, eˈxecutive, diˈminutive. I think most people also say conˈstitutive, though there may be some who go for ˈconstitutive following the model of ˈconstitute. What about attributive, contributive, distributive, retributive? I use antepenultimate stress in these, but then I also stress the -trib- in conˈtribute, disˈtribute. I wonder about those many Brits who prefer initial stress in these two words. How many of them carry that over into ˈcontributive, ˈdistributive?

Monday, 11 July 2011

a rud druss?

In yesterday’s (London) Sunday Times, on the comment page, Rod Liddle had a short piece about a tape allegedly implicating “cuddly, bearded, man of peace” Gerry Adams as boss of a former IRA death squad. The tape is currently in the custody of (‘in the clutches of’) an unnamed Massachusetts university. According to Liddle, the taped interviews were recorded by students, ‘presumably as part of a degree in one thousand years of “Bruddish upprussion”’.

Forget the politics. What’s going on with this jocular respelling? It’s intended, of course, to convey an American accent to a British readership. We all know that Americans voice intervocalic t, making it sound like d, and that their KIT vowel tends to be rather laxer or opener than the BrE mainstream. So “Bruddish” for British is fair enough.

But what about “upprussion” for oppression?

This is the first time that I remember having seen a jocular respelling relating to AmE DRESS.

There’s a complex set of vowel sound changes currently in progress in part of the United States, known as the Northern Cities Shift. It was first sketched out by Labov over thirty years ago. There’s an account written by Labov here, or a simpler one in Wikipedia.
What is happening in places like Buffalo, Detroit and Chicago is that the TRAP vowel is getting tenser and closer (but that’s happening in other places, too). This encroachment on the territory of the DRESS vowel is compensated by the lowering and backing of DRESS. The LOT(-PALM) vowel moves forward to take over some of the space vacated by TRAP.

The consequence is that NCS LOT may sound like other people’s TRAP, while NCS DRESS may sound like other people’s STRUT. Hence Liddle’s respelling upprussion.

lexical set example RP etc other AmE NCS
LOT a bottle of Scotch ə bɒtl̩ əv skɒtʃ ə bɑdl̩ əv skɑtʃ ə badl̩ əv skatʃ
DRESS a red dress ə red dres ə rɛd drɛs ə rɐd drɐs

Friday, 8 July 2011

Welsh respellings

Christy MacHale wrote
Your recent article on Latvian spelling, together with the observation that Latvian generally insists on respelling loanwords according to its own conventions, set me thinking about the practice of Welsh in this regard, particularly with reference to geographical names. Being a Hispanist and a lover of Welsh, I remember tuning in a couple of decades ago to a series of BBC Cymru programs, about different aspects of the Hispanic world. It featured such spellings as Mecsico and Portiwgal […]

He (or is it she? sorry...) was also particularly dismayed to find karaoke rendered in Welsh as carioci, despite the fact that word-final e is perfectly at home in Welsh (as in bore ‘morning’) though not in English. But the item that most concerned him/her was the name of the Basque country (Gwlad y Basg), Basque Euskadi, rendered as Ewscadi. Even as a reflection of the Spanish or Basque pronunciation this is not accurate: s/he feels it ought to have -addi, reflecting the Iberian -aði pronunciation.

Christy wanted to know who was responsible.
I said
I suppose if anybody can make rulings on this sort of thing, it must be the Welsh Academy (Yr Academi Gymreig), with whose Dictionary (1995) you may be familiar.

Although Christy deplores the fact, it is very clear that loans in modern Welsh regularly come via English (rather than direct from, say, Japanese or Spanish) and that they are then respelt in accordance with Welsh orthographic conventions. Modern Welsh spelling does not use the letters k, q, v, x or z. The spellings complained of are very normal in Welsh.

In the Academy's Dictionary, under words spelt in English with ka-, you'll find Cabwci, cafftan, cacemono, Calahari, caleidosgop, Calefala, camicasi etc.

On the other hand the same dictionary gives Kafkaésg and kalmia, so evidently personal and Linnaean names are treated differently.

Under English Q- we find cwadrîl, cwaga, cwarts, cwasar, Cetshwa, cworum, cwoca. But Quaker is properly cymricized as Crynwr (crynu ‘tremble’).

Christy would be pleased to find that the dictionary also actually gives caraoce rather than carioci. (As readers will know, the Japanese word karaoke カラオケ is formed from kara ‘empty’ plus oke from English orchestra. It is usually pronounced in English as ˌkæriˈəʊki.)

Thursday, 7 July 2011

silent final consonants

Commenting on bombardier yesterday, djbcjk said
I've always said ˌbɒməˈdɪə. I'd no idea that the second b was pronounced, having never heard it actually said.
This reaction is understandable, given that the second b is silent not only in bomb bɒm || bɑːm but also in bombing and bomber. However I suppose we all agree that bombard is bɒmˈbɑːd || bɑːmˈbɑːrd (or possibly bəm-), and bombardier is formed from that rather than directly from bomb.

Historically and etymologically the relationship is like that of climb – clamber, thumb – thimble, crumb – crumble. In modern English most people do not feel the items in these pairs to have an obvious relationship to one another.

Other words spelt with a final silent b include lamb, comb, dumb, plumb, limb, numb, and (generally) jamb and iamb. Again, the b remains silent in inflected forms such as combing, dumbest. This gives us the interesting pair of homographs written number: the one to do with counting, ˈnʌmbə(r), and the comparative of the adjective numb, ˈnʌmə(r).

Interestingly, the handful of words with final mn, in which the n is silent, all have obviously related forms in which there is a pronounced n. Thus we have damn dæm but damnation dæmˈneɪʃn̩, and similarly autumn – autumnal, solemn – solemnity, column – columnist, condemn – condemnation, hymn – hymnal.

Phonologists in the Chomsky-Halle tradition see these stems as ending in a final /b/ or /n/, obligatorily deleted after /m/ unless a vowel follows across no boundary or just an internal boundary. But they would also see a similar relationship in pairs such as sign – signal, (im)pugn – pugn(acious), (con)dign – dignity. Let’s not go there: in my view it’s orthographic, etymological, and pretty obvious to classicists, but is not part of contemporary English phonetics.

Wednesday, 6 July 2011


While other branches of the British army have corporals and lance-corporals among their non-commissioned officers, artillery units have bombardiers and lance-bombardiers. As an NCO rank, bombardier is pronounced ˌbɒmbəˈdɪə.

A company called Bombardier is in the news at the moment. They are “Britain’s last train makers”, but have just failed to win a contract to build new trains for Thameslink.

As those of you in the UK may have noticed, the newsreaders use a different pronunciation for the name of the company. They call it bɒmˈbɑːdieɪ or something similar. They are correct to do so.

The reason is that — despite its supposed Britishness and its manufacturing works in Derby — Bombardier Inc. is actually a Canadian conglomerate, named after its founder Joseph-Armand Bombardier, a Québécois. He was the inventor of the snowmobile. In French his name is pronounced bɔ̃baʁdje, so the English name of the company is an anglicized version of this.

Tuesday, 5 July 2011

women's tennis

Two correspondents wrote last week complaining about the way commentators for the Wimbledon tennis tournaments were pronouncing Maria Sharapova’s name. As they pointed out, in Russian her surname bears antepenultimate stress: she is Шарапова ʃəˈrapəvə, so in English we ought to call her ʃəˈræpəvə. But we don’t, we call her ˌʃærəˈpəʊvə with penultimate stress.

Dismayed as purists and my correspondents may be, there’s not much we can do about this. I am told that the tennis player herself is quite content to be given penultimate stress in English and to be known as BrE ˌʃærəˈpəʊvə, AmE ˌʃɑrəˈpoʊvə.

There were two other woman tennis players last week the pronunciation of whose names perhaps deserves comment. One is Sabine Lisicki. She is German, born in Troisdorf, although her name must be of Czech (or some other Slavonic) origin. Neither of my German pronunciation dictionaries ɡives the pronunciation of her name. In Czech it would presumably be ˈlisitski. The English commentators called her lɪˈzɪki, lə-.

The other is the new women’s single champion, Petra Kvitova. She is Czech, and in Czech her name is written Kvitová (with the obligatory unstressed feminine ending -ová borne by all Czech females) and pronounced ˈkvitovaː. Our commentators all had a problem with the cluster kv-, which they solved by inserting an anaptyctic schwa, giving kəˈvɪtəvə.

I am not sure why kv- presents such a problem to English speakers. We seem to manage to produce sv- without anaptyxis in Svengali, svelte, Svalbard and, um, svarabhakti. We manage ʃv- in nazi-era mock-German Schweinhund as well as in Schweitzer and Schwarzwald. And kw- is an everyday cluster for us and not so very different from kv-.

Monday, 4 July 2011

denasalized nasals

Congratulations to Young Shin Kim, who has just successfully defended her doctoral thesis (dissertation) at UCL. It is entitled “An acoustic, aerodynamic and perceptual investigation of word-initial denasalization in Korean”. Here she is with her supervisor, Michael Ashby.
Young Shin had noticed that when she played a recording of a Korean word beginning with m, English listeners often perceived it as beginning with b. She has gone on to demonstrate, in great detail, that in many cases Korean initial “nasals” are indeed pronounced as plosives.

Here’s an example. Listen to the sound clip of 그런데메밀, which “ought” to be kɯɾʌnde memil, taken from the phrase 그런데메밀꽃 kɯɾʌnde memilkkot meaning ‘then buckwheat flowers’. In the second word, compare the two consonants written m. The first, in me-, is, as you can hear, denasalized. It is a fully voiced b. The second, word-medial, in -mil-, is an ordinary nasal. Compare the two consonants on this spectrogram: nasal formant bars for the second, but not for the first.
For her dissertation, Young Shin recorded word-initial nasals from a relatively large number of Korean informants, and carried out listening tests with English and Korean listeners. She also ran auditory and spectrographic tests demonstrating that they are indeed denasalized, often even having plosive-like release bursts. Nevertheless, they “remain somewhat different from [Korean] voiced plosives in the low and high frequency regions”.

On the face of it, this is an improbable phonetic development in Korean, given that the language already has three sets of contrastive plosives. At the bilabial place, for example, there are an aspirated fortis , an unaspirated fortis p=, and a (relatively) unaspirated lenis p, the latter having a voiced allophone in intervocalic position. Not many languages have three contrasting sets of plosives, fewer still four or five.

Young Shin deserves particular praise for having noticed this development when generations of phoneticians working on Korean had failed to do so. An exception, intriguingly enough, was Daniel Jones, who as long ago as 1924 noticed that there was something funny about Korean initial nasals. In the ‘spesimɛn’ of ‘korɪən’ he published in the m.f. that year (p. 14), jointly written with K. Minn, he analysed the initial nasals as mb, nd, commenting as follows. (The restriction to pre-u position, noted by Jones, has not been maintained.) But in the specimen of Korean he published in the Principles of the IPA there is no mention of this, and no one seems to have followed it up — until now.

Friday, 1 July 2011


I wrote this yesterday, but was unable to post it because of problems at
_ _ _

The newsreader on BBC R4 this morning mentioned that when the newlyweds William and Kate visit Calgary they will attend a ˈrəʊdiəʊ. (That’s the Calgary Stampede, “the greatest outdoor show on earth”.)

My ears pricked up, because I’m aware that in LPD I prioritized a differently stressed version of this word, rəˈdeɪəʊ.

I know that both stressings are possible in English, but I had the impression that — despite our English tendency to go for initial stress (cf video, Romeo, stereo) — the penultimate stress was more correct and therefore to be expected from BBC newsreaders. The reason it was (I thought) considered correct is its origin as a Spanish word, roˈðeo.

Our pronunciation dictionaries give both possibilities, as you would expect. So does the OED. But they prioritize them differently.

BrE -ˈdeɪ- -ˈdeɪ- ˈrəʊ-
AmE -ˈdeɪ- ˈroʊ- ˈroʊ-

The OED has an interesting note.
The stress of the Spanish word is on the penultimate syllable. In English, pronunciations closely resembling the Spanish pronunciation are frequent in areas of former Spanish settlement in the western United States, especially in California and the southwest. Alongside these, Dict. Amer. Regional Eng. (at cited word) records various naturalized pronunciations which show shift of stress to the first syllable (so especially in Midwestern and eastern states) and/or substitution of /i/ for Spanish /e/ or its naturalized equivalent /eɪ/ in the second syllable.

I notice that Romeo similarly has penultimate stress in Italian, roˈmɛːo. And in Latin the word vidēō (‘I see’) has a penultimate long vowel and therefore penultimate stress. But no one dreams of reproducing this penultimate stress in English. These two are anglicized as ˈrəʊmiəʊ, ˈvɪdiəʊ, with the penultimate vowel weakened as you would expect to i.

Ought I to change my priorities for rodeo?