John Wells’s phonetic blog archive September 2007
To see the phonetic symbols in the text, please ensure that you have installed a Unicode font that includes all the IPA symbols, for example Lucida Sans Unicode or Charis SIL (click name for free download).
|Friday 28 September 2007|
/What was that?
Listen to this sound clip from the CD-ROM that accompanies my book English Intonation.
What do you hear the speaker say?
Since the publication of the book, several readers have kindly written to draw my attention to various misprints and so on in the text or errors on the CD. I am grateful to them.
Sometimes, though, the supposed error is not an error at all. In the sound clip (p. 109 in the book) Michael Ashby is actually saying
Do you object to dogs?
and that is what I hear if I listen.
But my Japanese correspondent heard
Do you get to dogs?
and listed this for me as an error on the CD.
You can see why the misapprehension might arise. The first three syllables of Do you object are spoken very fast. I can still hear (or think I hear) the segments [dəjuəbˈdʒek-]. But I can see how a Japanese listener might miss the second [ə] and the non-audibly released [b], then mishear [dʒ] as [g]. The usual contextual elision of the final [t] of object and the equally usual non-audible release of [k] accounts for the rest of the mishearing.
The moral I draw is that EFL leaners do need to know about contextual effects such as weakening, elision, assimilation, and plosive releases. Otherwise they are in danger of failing to understand ordinary spoken English.
And this wasn’t even sloppy colloquial speech with accompanying extraneous noise. It was scripted and then read aloud in soundproof studio conditions.
From [dəjuəbˈdʒektəˈdɒgz] (the whole thing lasting less than a second) you have to be able to recover Do you object to dogs?.
|Thursday 27 September 2007|
Reacting to John Maidment’s neologism yoghism (blog, 25 Sep.), Eric Armstrong writes about yogh:
Wikipedia says that it's "pronounced either [joʊk], [joʊg], [joʊ] or [joʊx]". But their entry makes no mention of the IPA's usage, does it...
I have good news for Eric.
Ezh and yogh are different letters. And the Unicode name for ʒ is ezh.
In Unicode, the symbol ʒ has the number U+0292. This means that to enter it in modern versions of Word you can type 0292, select this string of digits, and do Alt-X. Providing the current font allows, an ezh will appear on your screen. Writing HTML code for a web page you can insert
The Unicode number for yogh, on the other hand, is U+021D. It is a fairly recent addition to Unicode, and therefore still missing from many computer fonts, even phonetic ones. For example, it is not included in Lucida Sans Unicode. (This means that you may not see it properly displayed in the sidebar, depending on what fonts you have available.)
Yogh is not a phonetic symbol, but a letter used in Middle English and Middle Scots to stand for a velar fricative or palatal approximant. (By the way, I call it /jɒg/.) Norman scribes didn’t like letters not found in the Latin alphabet, so yogh was phased out and replaced by gh. When it was retained, it came to look just like z, which is why we have a spelling-to-sound mismatch in such names as Dalziel /diˈel/ and Menzies /ˈmɪŋɪs/.
Ezh has an angled top, whilst yogh has a curly top.
John Maidment ought really to have referred to creeping ezhism.
The confusion between the two names is found, I am sorry to say, to the Phonetic Symbol Guide by Geoff Pullum and Bill Laduslaw (University of Chicago Press, 19962), where the phonetic symbol is unfortunately labelled YOGH. But the IPA Handbook calls it Ezh or Tailed Z.
By the way, Geoff, welcome back to Blighty!
|Wednesday 26 September 2007|
As I read yesterday’s Guardian, I did a double take.
It was in the Conference diary, a parliamentary sketch writer’s impressions of the Labour Party conference currently being held in Bournemouth. The writer was talking about a group of former Conservatives who had crossed the floor to support Labour. Despite various others whom the conference might have especially applauded,
the conference reserved its enthusiasm for a shameless performance from the Europhile MP, Quentin Davies, as plummy and red-faced an ex-Tory as you could hope to find. However, many in the hall thought the standing ovation for Davies was nauseous. “I never thought I’d belong to a party in which people were called Quentin. What next, Rupert?” whispered one.
(I must explain for the non-Brits that Quentin and Rupert are stereotypically perceived as names found only among members of the upper class: certainly not working-class names such as one would traditionally expect to find among members of the Labour Party.)
Anyhow, at first I interpreted Rupert as a vocative:
'What \next, Rupert?
But that didn’t make sense, and I realized that the correct reading must be to interpret Rupert as “Is there next going to be someone called Rupert?”. That is, the intended intonation was
'What \next, | /Rupert?
Clearly, when reading silently, I do subconsciously select appropriate intonation patterns as I go. Here I had at first selected the wrong one.
All would have been clear if the sketch writer had punctuated differently:
“What next? Rupert?” whispered one.
|Tuesday 25 September 2007|
Predating /ʒ/ in liege
That’s predating in the OED Newsletter sense, finding or constituting an earlier attestation of, /ˌpriːˈdeɪtɪŋ/, not in the ecological sense of regularly preying on, /priˈdeɪtɪŋ/.
Obviously (blog, 20 Sept.) I should have checked in Jack Windsor Lewis’s Concise Pronouncing Dictionary, published by OUP in 1972. There it is: liːʒ as the second choice for liege. And CPD gives very few second choices. Anyhow, CPD clearly predates LDP by twenty years. Jack’s own blog has more thoughts on the subject.
Meanwhile Kensuke Nanjo writes to point out that Webster’s Third New International Unabridged (1961) gives this variant in the second place with the label “sometimes”. I’ve checked, and he’s quite right. But I was right in saying that it’s not in the more widely used Webster’s Collegiate.
So Merriam-Webster was there even before ODP, and long before LPD.
P.S.Er... the OED Newsletter term is actually antedate, not predate.
In a related email, John Maidment says
What annoys me slightly is the pronunciation of Beijing with /ʒ/ instead of /dʒ/. This creeping yoghism should be stamped out.
(His neologism refers to yogh, not to yoga.)
To which one might add: the same applies to raj, Taj Mahal, adagio and several others.
Yogh (missing in many fonts)
|Monday 24 September 2007|
Here is an email I received recently.
“Dear Sir, my name is XXX and I am the bureau chief here in London for [...] Australia publications [...].
— to which I replied,
“You really mustn’t believe everything you read in the papers. The Australian accent is not "well and truly creeping into the everyday [sic] of Britons".
I feel like Ben Goldacre with his Bad Science column in the Guardian.
“The archives at badscience.net are overflowing with just a small sample of the media’s crimes: preposterous cherry-picking, outrageous overextrapolation, startling ignorance or whitewashing of known methodological flaws and, worst of all, reporting the authors’ speculative conclusions, from the discussion section of a paper, as if they were the experimental results themselves.”
|Friday 21 September 2007|
The general rule for whether or not the verb to be is accented — which applies to all auxiliary verbs, too — is that we accent it only if it shows contrastive polarity (= positive vs. negative) or contrastive tense.
You 'thought he wasn’t 'ready, | but 'actually he 'was.
There is a further case where to be is accented, a case which seems to be illogical and so is perhaps best regarded as idiomatic: in the pattern wh word – be – pronoun, or syntactic manipulations thereof. I devote a separate section (3.18) to this in my intonation book.
'What 'is it?
However none of these explanations fit two sentences I have heard recently.
('Welcome 'back!) — We’re 'happy to 'be back!
(In the second example I picked up no contrast of tense or polarity.)
I think these fall under the heading of “Difficult cases of tonicity” (p. 182-184 in my book). As I comment there,
There is always a strong pressure not to accent repeated words. Yet the nucleus has to go \somewhere. This may lead to its being placed on a function word, even one that may appear to be utterly lacking in semantic content.
PS: Martin Barry adduces an apt Biblical quotation:
God said 'let there be 'light, | and there 'was light.
|Thursday 20 September 2007|
My liege, we are besieged
Vassals (and wives) in Shakespeare often address their lord as my liege. The pronunciation given in dictionaries is /liːdʒ/. But I keep hearing Shakespearian actors (of both sexes) pronounce it as /liːʒ/.
As far as I know, LPD was the first to record this variant, nearly twenty years ago. You will not find it in EPD or in standard dictionaries such as the COD or Webster’s Collegiate. The ODP has it as an American, but not British, variant. And Kensuke Nanjo, always hot on new variants, has included it in the new Genius English-Japanese Dictionary.
It’s not only liege. I have heard the /ʒ/ variant in siege and besiege, too. So what is it about /dʒ/ after high front vowels? Is this some kind of contamination from prestige /-iːʒ/ or beige /beɪʒ/?
Then there is the corresponding position after high back vowels. Most English people, or at least those with any kind of familiarity with French, pronounce rouge and luge with /ʒ/, as in French. I have heard it in deluge, too. Water-colour artists know about gamboge tint, often pronounced with /ʒ/.
So it looks like something that favours a fricative rather than an affricate after a high(ish) long vowel. Is this an incipient sound change?
Well, I’ve not heard anyone pronounce huge as /hjuːʒ/ ... yet.
|Wednesday 19 September 2007|
Ph, part two
Nigel Greenwood points out that although Classical Greek explains the ph in photograph (blog, yesterday), it doesn’t explain the ph in caliph — or, I might add, in nephew.
Caliph comes from Arabic khalīfah via medieval Latin calīpha. Probably it was felt that not only Greek words but all foreign borrowings with /f/ should have the spelling ph.
Nephew comes from Latin nepos, nepōtem via French neveu. For the first few centuries of its existence in English it was spelt with v; the spelling with ph is presumably a bit of learnèd Latinizing (as with the b in debt). The pronunciation was also traditionally /v/, which is what I say myself; but most people nowadays say it with /f/, which must have originated as a spelling pronunciation.
Most of the remaining words in which ph = /f/ are of Greek origin. This even may include typhoon, which has a complicated etymology involving the coming together of an Urdu-Persian-Arabic word of possible Greek origin (τῡϕῶν) with a Chinese (Cantonese?) expression tai fung ‘big wind’ (Mandarin: dà fēng).
In Macpherson the explanation is quite different. In Scottish Gaelic, as in other Celtic languages, some consonants are subject to lenition (mutation) in certain positions. The spelling reflects this: /p/ mutates to /f/, and the result is spelt ph. So the son of the parson is Mac an Phearsain, or in English Macpherson.
Why we write ph in Westphalia (German Westfalen) or indeed in Randolph and Humphrey (which are names of Germanic origin) I have no idea. It’s usually put down to “classicizing”.
* * *
Here in its entirety, except for the sender’s name, is an email I received recently. I have faithfully reproduced the spelling and punctuation.
dear MR Johon
It doesn’t appear to be an email scam: she doesn’t ask for details of my bank account or ask for my help in transferring a huge sum of stolen money.
But can this really represent the achievement of someone who has studied English at university for four years?
She certainly needs someone’s help. But I’m afraid it is not going to be mine.
|Tuesday 18 September 2007|
Why “ph” in photograph?
Ron Yorke wrote asking,
I have discussed with my Greek colleague, and others, the use of 'ph' in lieu of 'f' in a word or element of a word that is of Greek origin. As far as the Greek language is concerned, there is no reason for it.
Your Greek colleague is wrong. There is a reason for it, a historical reason. Although the letter phi (ϕ) of Modern Greek is pronounced [f], and usually transliterated into English as f, the phi of Classical Greek is transliterated as ph. This is because in classical times it was actually pronounced as an aspirated [ph]. The Romans represented this rather naturally as ph, e.g. Philippus, and we follow Latin despite the change in pronunciation that has occurred meanwhile. Hence photograph etc.
(The evidence in support of this claim is summarized in W. Sidney Allen’s Vox Graeca, CUP 19873, pages 16-27.)
In each case the classical voiceless aspirated plosive has since developed into a voiceless fricative.
The voiced plosives have become fricatives, too, so that classical Euboea is now called Evvia, and as a modern placename Delphi is sometimes transliterated Dhelfi. Unlike Classical Greek, Modern Greek megalos ‘big’ contains a voiced velar fricative.
Just occasionally, a classical education is really useful.
|Monday 17 September 2007|
Ten days ago (6-7 September) I discussed the syllable structure of Japanese. But speakers of Japanese do not usually think in terms of the syllable: they think in terms of the mora.
A mora corresponds either to the [(C(j))V] part of a syllable, or to one of the elements [N], [Q], or [:] (the second part of a long vowel).
The elements [N], [Q], and [:] cannot co-occur in the same syllable. It follows, then, that a syllable can comprise either one mora or two, but not more than two.
The word for ‘bread’, pan [pa-N], has two moras. The proper name Honshū [ho-N-ɕu-ː] has four. Hokkaidō [ho-Q-ka-i-do-ː] has six.
The English disyllabic word corner is borrowed into Japanese as the four-mora kōnā [ko-ː-na-ː].
The mora is claimed to be the basic unit of timing in Japanese, in the same way that the syllable is claimed to be in syllable-timed languages and the stress-group (foot) in stress-timed languages. Awareness of the mora is also doubtless reinforced for literate speakers by the kana writing systems, in which there is one symbol per mora.
The devoicing or elision of certain high vowels (7 Sep.) does not affect the mora count. Hajimemash(i)te still comprises six moras despite the disappearance of the [i].
A Japanese word or morpheme cannot begin with any of the elements [N], [Q] or [:]. They are found only within a word, following a [(C(j))V] element and making a syllable with it. That is why we still need to talk about syllables in Japanese. Moras alone are not enough.
|Friday 14 September 2007|
French and Frenglish
I flew back from Japan via Paris, with Air France.
The door of the aircraft lavatory bore the sign POUSSER ‘(to) push’. It occurred to me that it could just as well have read POUSSEZ ‘push!’. The infinitive and the imperative are pronounced identically, both being /puse/. It seems to be purely a spelling convention that for the sign on a door we write the first rather than the second.
The infinitive form has to be regarded as being preceded by an understood il faut ‘one must’ or veuillez ‘be so good as to’.
* * *
After the cabin doors were closed at Kansai International, as we were preparing for the rollback, there was a cabin announcement in three languages — French, Japanese, and English — about the need to check the number of passengers aboard. The English one went
Too 'meg this easier for /us, | would you 'please return to your /seat | and \seed down.
Ignore the inappropriate voicing assimilation. I wish teachers of English to the French would teach them one very simple rule: do not accent a pronoun unless it is contrastive (emphatic). This is much simpler than many grammatical rules which teachers (rightly) consider it their duty to teach, and would make an important contribution to improving the learner’s intellibility. Why is it that the teaching of grammar is considered essential, but the teaching of intonation a luxury that can be ignored?
What a native speaker would say is
To 'make this \/easier for us | ...
Phrasal verbs are more complicated, and perhaps one ought to be indulgent about sit down. Non-use of the weak form of to marks the speaker out as not being a native speaker. And the /iː - ɪ/ contrast is notoriously difficult for the French to master, so we must be indulgent about that too.
|Friday 7 September 2007|
Feedback: Uniqlo and devoicing
It is good to receive so many emails commenting on what I write here. This feedback is much appreciated. Otherwise I might feel I was just talking to myself.
Yoshihiro Kawabata (who writes to me in a mixture of English and Esperanto) offers some further information about the name Uniqlo (blog, 3 September). He tells me that according to the Japanese version of Wikipedia their abbreviated name and logo was originally written “UNI-CLO”. But when they were setting up their local company in Hong Kong in 1998 they inadvertently misspelled it as “UNI-QLO” on the registration form. Since then they have used the English trademark with QLO. Kawabata-san thinks they dropped the hyphen in the late 1990s.
He tells me further that in standard Japanese ユニクロ yunikuro is accentless, i.e. has the pattern LHHH. But in Kansai dialect it is LHLL, i.e. has an accent on the second mora. (L = low pitch, H = high pitch)
* * *
Several correspondents commented on yesterday’s blog about syllable structure, making the point that Japanese /i/ and /u/ are devoiced or deleted in certain phonetic contexts, thus giving rise to consonant clusters and coda consonants. Jerome Perrier mentions that sukiyaki begins [ski-], and ohayō gozaimasu ‘good morning!’ ends [-mas]. This is true, and I faithfully copy it in my attempts at elementary Japanese (hajimemash(i)te). But I regard it as a superficial detail of allophonic realization, a kind of moraic consonant formation, not affecting the underlying CV principle.
Nigel Greenwood says that we get the same kind of thing in Portuguese, so that the [u] in Lagos /ˈlaguʃ/ is often devoiced, too.
|Thursday 6 September 2007|
Yokohama and the like
Languages differ in the syllable structures that they allow. We can distinguish
Speakers of type (ii) languages, with syllable structure [CV], may have difficulty producing the consonant clusters ([CCV]) and final consonants ([CVC]) they need when learning type (i) languages. When I discuss this topic in lectures, with reference to the learning of English pronunciation, I usually choose Japanese as my example of a type (ii) language, taking the place name Yokohama to illustrate the canonical syllable structure [CV]. Yo-ko-ha-ma = [CV CV CV CV].
A few days ago one of the audience, who had heard me lecture on this before, complained that I always use this same word as an example. Why not use some other examples for a change?
OK, then, sticking with Japanese proper names: Nakamura, Watanabe, Taniguchi, Toyota, Nagasaki, Hakodate, Nagano, Shikoku, Haneda, Narita. And of course there are hundreds more.
More exactly, the canonical syllable structure is [(C)V]: the consonant at the beginning may be absent, leaving just [V]. That’s what we have in the first syllable of Okinawa, Ikebukuro, Akasaka and in the first two syllables of Ueno. Furthermore, the vowel may be double rather than single, yielding a long vowel or a diphthong.
In Hawai‘ian, all syllables comply with this [(C)V] structure: Honolulu, Waikiki, Kamehameha, Oahu, Kahului. The same applies to other (all other?) Polynesian languages: Samoa, Papeete, Rotorua, Fiji, Tonga [to-ŋa].
Japanese, on the other hand, is not such a strict type (ii) language. Not all Japanese syllables have the structure [(C)V]. Some have a slightly more complicated structure. First, the consonant can have an associated palatal glide, as in Kyoto, Ryuku, giving a very limited possibility of [CCV]. Secondly, the vowel may be followed within the syllable by [N], a ‘moraic’ nasal, or by [Q], an obstruent agreeing in identity with the initial C of the following syllable. Since both [N] and [Q] are consonants, we do have some possible [CVC] syllable structures, though again highly restricted.
We can see these slightly more elaborate syllable structures in cases such as Sendai, Namba and Hokkaido, Sapporo, Beppu.
So the structural formula for a Japanese syllable is [C(j)VX], where X = [N] or [Q)]. (The Japanese affricates romanized as ch, j, and ts count as single consonants.)
But that’s it. That’s why in Japanese milk gets transformed into [mi-ru-ku] and McDonalds into [ma-ku-do-na-ru-do]. And why Japanese EFL learners tend to want to do something similar in their English pronunciation.
|Wednesday 5 September 2007|
Continental and English values
As Brits do when away from home, I have been watching BBC World to keep up with the news on television.
I was struck by how one of the weather forecasters pronounced the name of the capital of Nigeria, Lagos. Its usual pronunciation in BrE is unquestionably /ˈleɪgɒs/, with the vowel of FACE. But what I heard was /ˈlɑːgɒs/, with the vowel of START. (Next day’s weather forecaster reverted to the usual pronunciation.)
I’ve also heard the occasional /ˈlɑːgoʊs/ from Americans.
This reflects a general difficulty we have in English when reading aloud proper names that look Spanish or Portuguese. What are the correct reading rules to use? Do we give stressed vowels open syllables their “continental” values or their “English” values? In the first case we’ll read the letter a aloud as /ɑː/, e as /eɪ/, and i as /iː/. In the second, a will be /eɪ/, e /iː/, and i /aɪ/. Our usual pronunciation of Lagos has the English values.
When we see the name Toledo, the continental values are appropriate for the place in Spain, but the English values for the place in Ohio. Another Spanish city is Vigo, which we nowadays pronounce in English with the vowel of league (= continental); but for Vigo Street in central London it was traditionally the vowel of tiger (= English). Granada in Spain has /ɑː/, but Grenada in the West Indies has /eɪ/.
The problem is particularly acute in the western USA, where there are large numbers of Spanish-derived place names. Does Coronado in California have /ɑː/ or /eɪ/? Compare on the one hand bravado and Colorado(= continental) and on the other hand tornado and Barbados (= English). [PS: However, locals pronounce /æ/ in Colorado, as Eric Armstrong reminds me.]
As well as the Lagos in Nigeria, there’s another Lagos in Portugal. Its local pronunciation is presumably [ˈlaguʃ] or [ˈlaɣuʃ], so I suppose in English it’ll be /ˈlɑːguːʃ/.
|Tuesday 4 September 2007|
The importance of being accented
In my intonation book there’s a very brief discussion of the possible relationship between accent and pronoun reference (anaphora):
In the following examples, the reference of the pronoun he depends on whether or not it is accented:
The point is that an ordinary unaccented subject pronoun (here, the he that is the subject of hit) has the same referent as the subject of the previous verb (here, threatened). But by accenting the subject pronoun we signal that its referent is not the same as the subject of the previous verb. Accenting he means that there is a change of subject, in this case from Bill to Jim. Likewise, accenting him indicates that the object of hit is different from the object of threatened; though since in this case logic means that a change of subject must also imply a change of object, it is not actually necessary to spell this out by accenting him as well as he.
So accent on a subject pronoun means that there is a change of grammatical subject. Lack of accent means that there isn’t.
Now comes an interesting example from Language Log.
If leadership never takes time off, people will be skeptical whether they can.
On his first reading of this, Arnold Zwicky says, he took the pronoun they to refer to the company’s leadership, i.e. assumed that they was unaccented. But then it became clear to him from the rest of the story that the intended referent of this pronoun was actually people. Reading aloud, this requires a contrastive accent on they.
The difficulty is that the distinction is not normally shown in writing. Hence the newspaper report that he quotes was ambiguous. Probably the writer of the report didn’t realize this. If he had realized, and had wished to avoid the ambiguity, he might have chosen one of various ways of resolving it: for example, he could have italicized they, or changed it to they themselves.
|Monday 3 September 2007|
I noticed recently that a new clothes shop had opened in Wimbledon, where I live, with the name Uniqlo — a name that obviously violates the well-formedness rules of English spelling. You just can’t have a q followed by a consonant letter. Even a following vowel letter (other than u) or a word boundary implies a foreign origin, usually Arabic.
How are we meant to pronounce this name?
I went into the shop to see what it was like, but decided it was not for me. The noisy pounding background music put me off. The company’s website is equally tiresome, with badly-designed Flash animation everywhere. (I must be getting old.)
I have never heard anyone talk about the shop and never heard an advertisement for it on radio or television. So I’ve never actually heard anyone pronounce its name.
Uniqlo is an international chain. Here in Japan (where I am teaching at Kansai Gaidai for a week) there are branches all over the place. It turns out that it is a Japanese company. According to Wikipedia, the name was coined in 1984. Rather prosaically, it is no more than a portmanteau of the English words unique and clothing.
The Japanese name is ユニクロ yunikuro. I think in English we’re probably meant to call it /ˈjuːn i kləʊ/. Or do they expect us to say /ju ˈniːk ləʊ/? Or even /juː ˌniːk ˈkləʊ/, as in unique clo(thing)?
Can anyone give us an authentic answer?
I bet the Japanese owners of the name never even considered its awkwardness as an English spelling.
PS: Paul Mills mentions that I could have checked on youtube.com, where there are a number of sound clips of people pronouncing the word. For example, here and here are TV reporters in New York, who say /ˈjuːn i kləʊ/. (By the way, whatever happened to the idea that Americans say [oʊ] where Brits say [əʊ]? 'Not \/these Americans.)
To search my web pages, use this Google search.