Sounds Interesting Observations On English and General Phonetics

Sounds Interesting
How do you pronounce omega, tortoise and sloth? – and why? Do

charted and chartered sound the same? How do people pronounce the
names Charon, Punjab and Sexwale?
In this engaging book, John Wells, a world-renowned phonetician and
phonologist, explores these questions and others. Each chapter consists of
carefully selected entries from Wells’s acclaimed phonetics blog, on
which he regularly posted on a range of current and widely researched
topics such as pronunciation, teaching, intonation, spelling and accents.
Based on sound scholarship and full of fascinating facts about the
pronunciation of Welsh, Swedish, Czech, Zulu, Icelandic and other
languages, this book will appeal to scholars and students in phonetics
and phonology, as well as general readers wanting to know more about
language.
Anyone interested in why a poster in Antigua invited cruise ship
visitors to enjoy a game of porker, or what hymns can tell us about
pronunciation, should read this book.
j.c. wells is Emeritus Professor of Phonetics at University College

London and a Fellow of the British Academy. His interests centre on
the phonetic and phonological description of languages but also extend to
lexicography and language teaching.
Sounds Interesting
Observations on English
and General Phonetics
J.C. WELLS
Emeritus Professor of Phonetics, University College London
Illustrations by Lhinton Davidson

University Printing House, Cambridge CB2 8BS, United Kingdom
Cambridge University Press is part of the University of Cambridge.
It furthers the University’s mission by disseminating knowledge in the pursuit of

education, learning and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107427105
© J. C. Wells 2014
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2014
Printed in the United Kingdom by Clays, St Ives plc
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data

Wells, J. C. (John Christopher) author.
Sounds interesting : observations on English and general phonetics / J.C. Wells.
pages cm
ISBN 978-1-107-07470-5 (Hardback) – ISBN 978-1-107-42710-5 (Paper back)
1. English language–Phonetics. 2. English language–Pronunciation. 3. English language–
Phonology. I. Title.
PE1135.W36 2014
414–dc23 2014022387
ISBN 978-1-107-07470-5 Hardback

ISBN 978-1-107-42710-5 Paperback
Cambridge University Press has no responsibility for the persistence or accuracy of

URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Contents
Preface xi
1 How do you say…? 1

1.1 sloth 1
1.2 nucular snitchel 2
1.3 artisanal 3
1.4 C. diff. 3
1.5 shih tzu 4
1.6 omega 5
1.7 plethora 5
1.8 diocese 6
1.9 West Indian islands 7
1.10 place names 8
1.11 Madejski 9
1.12 Charon 9
1.13 Judea 11
1.14 Chagos 11
1.15 proceed and precede 12
1.16 hypernymy and thrombosis 12
1.17 amiodarone 13
1.18 French, German or Welsh? 13
1.19 Loony what? 14
1.20 Richter 15
1.21 Punjab and feng shui 15
1.22 ylang-ylang 16
1.23 Mbabane 17
1.24 brand names 18
1.25 Sexwale 19
1.26 tortoise 20
1.27 continental and English values 21
1.28 my liege, we are besieged 21
1.29 Rothersthorpe 22
1.30 Campbell 23
1.31 malapropisms 24
1.32 ho, ho, ho 24
1.33 d’oh, doh. . . duh 25
1.34 ring in the old ear 25
1.35 /Who did you say were all liars? 26
1.36 anglicized Welsh place names 27
v
vi Contents
1.37 Jersey h 28
1.38 Friern Barnet 28
1.39 veterinary 29
1.40 Marcel Berlins and Sarkozy 29
1.41 repertory 30
1.42 Penwortham 30
1.43 hello? 31
1.44 Lucida 32
1.45 phonetic numerals 33
1.46 Llantwit Major 34
1.47 apostasy 34
1.48 pwn 35
1.49 Entwistle 36
1.50 Chinese into English 36
1.51 inter(n)ment 37
2 English phonetics: theory and practice 38

2.1 assimilation 38
2.2 internal intrusive r 38
2.3 royal toil 39
2.4 wronger 40
2.5 compression 40
2.6 compression in hymnody 41
2.7 GOAT compression 43
2.8 analogical decompression 44
2.9 compression anomalies 45
2.10 oh, oh! 45
2.11 stressing schwa 46
2.12 Latin stress in English 47
2.13 gemination and degemination 48
2.14 reductions in casual speech 50
2.15 abject haplologies 51
2.16 the happY vowel 52
2.17 crazes and crazies 53
2.18 banded or bandied? 54
2.19 the status of schwa 54
2.20 charted and chartered 55
2.21 singing strongly 56
2.22 happY prefixes 56
2.23 happY endings 57
2.24 some 58
2.25 the symbol for STRUT 60
2.26 the Teutonic Rule 60
2.27 problems with lexical stress 62
2.28 twenty-twenny 63
2.29 phantom r 64
2.30 Nadsat is non-rhotic! 65
2.31 non-rhotic loanwords 66
2.32 Gordon Brown 67
Contents vii
2.33 an albatross on the balcony 68

2.34 changes in English vowels 68
2.35 Christmas puzzles 69
2.36 data processing under stress 70
2.37 a cab, innit? 70
2.38 instances and incidences 71
2.39 settee 72
2.40 imma 73
2.41 terraced attacks 74
2.42 synthetic homographs 75
2.43 mine and my 75
2.44 um. . . er. . . 76
2.45 thou hast, he hath 77
2.46 rhotic 78
2.47 progressive assimilation 79
2.48 homographs 80
3 Teaching and examining 82

3.1 deceptive strings 82
3.2 practical tests 82
3.3 faulty abbreviations 83
3.4 exotic sounds 83
3.5 the Swedish sj-sound 84
3.6 pausing a problem 85
3.7 volcano 86
3.8 suddenly in Sydney 87
3.9 perceptions of /æ/ 87
3.10 misled by spelling 88
3.11 unaspirated /p t k/ 88
3.12 Yokohama and the like 89
3.13 struggling to teach phonetics 91
3.14 accents and actors 92
3.15 minimal pairs in action 94
3.16 a French user of English 95
3.17 Taiwan English 95
3.18 the IPA Certificate 96
3.19 dead letters 97
3.20 alveolopalatals 98
3.21 BrE or AmE for TEFL? 99
3.22 spot the mistake 101
3.23 excuse my excuse 101
3.24 miscellanea 102
4 Intonation 103
4.1 introduction 103
4.2 intonation notation 104
4.3 idiomatic intonation 105
4.4 systems 106
4.5 politeness 107
viii Contents
4.6 fall-rise or fall plus rise? 108

4.7 empty things 109
4.8 counterpresuppositional insists 110
4.9 that’s funny 112
4.10 where? wohin? 113
4.11 examining intonation 114
4.12 the importance of being accented 115
4.13 accented be 116
4.14 O lift: your voice 117
4.15 stating the bleeding obvious 118
4.16 chunking 119
4.17 des accents enfantins 119
4.18 intonation idioms in the Germanic languages 120
4.19 compound stress in English and German 122
4.20 accenting phrasal verbs 123
4.21 train times 125
4.22 the rule of three 126
4.23 are you asking me or telling me? 126
4.24 accent on a VP subordinator 127
4.25 on the train 129
4.26 idiomatic nucleus on a pronoun 130
4.27 not a drop 131
4.28 a little while 132
4.29 international intonation 133
5 Symbol shapes, fonts and spelling 136

5.1 labiodental flap 136
5.2 similar symbols 137
5.3 clicks 138
5.4 IPA capitals 139
5.5 dashes 139
5.6 linguolabial trill 140
5.7 the symbol ɮ 140
5.8 scenes from IPA history 141
5.9 affricates 143
5.10 orthographic schwa 143
5.11 placing the suprasegmental symbols 144
5.12 why Hawai’i? 145
5.13 funny letters 146
5.14 anomalous g 147
5.15 the spelling wor 148
5.16 practicing (sic) wot I preach 148
5.17 why ph? 149
5.18 yogh and ezh 150
5.19 bevy, bevvy 151
5.20 low and long 151
5.21 palatoalveolars and the like 152
5.22 IPA reforms, 1976 153
5.23 Ortuguese 154
Contents ix
5.24 Panglish, or unintelligibility? 154

5.25 definitely 155
5.26 lexicographers and spelling 156
5.27 respelling 157
5.28 pea’s, plea’s 158
5.29 sibilant genitives 158
5.30 greengrocers’ spellings 160
5.31 O tempora! O mores! 161
5.32 the cost of spelling 161
5.33 capital eszett 163
5.34 Jumieka 164
5.35 fame at last! 165
6 English accents 167

6.1 po(r)ker 167
6.2 holey ground 167
6.3 hair lair! 168
6.4 minor royalty 168
6.5 Prince Harry 169
6.6 fricative t 169
6.7 hypercorrection 170
6.8 Montserrat Creole 170
6.9 television newsreaders’ RP 171
6.10 accents change over time shock horror 172
6.11 bad sociophonetics 173
6.12 The Book of Dave 174
6.13 character, calendar 176
6.14 Grand Turk 176
6.15 popular phonetics 177
6.16 changing London speech 178
6.17 American T voicing and sentence 179
6.18 sexy accents 180
6.19 Maori names 181
6.20 rich man, poor man 183
7 Phonetics around the world 184

7.1 Scottish Gaelic 184
7.2 tangnefedd a thragwyddoldeb 184
7.3 Welsh letter names 186
7.4 Welsh ll 187
7.5 c what I mean 188
7.6 Leewards, ABCs and Virgins 189
7.7 clicks demonstrated 190
7.8 an impossible sound 191
7.9 fricative trill, trill fricative 192
7.10 all the speech sounds in the world 192
7.11 WALS 193
7.12 IPA in a logo 194
7.13 frozen in Italian 194
Preface
After my retirement in 2006 I started to write a phonetic blog discussing

everything to do with phonetics, seen from my personal and professional per-
spective. I did this partly to keep my mind active after retirement; in a sense the
interaction with a web-based readership served to replace the daily dialogue with
colleagues and students that I enjoyed while employed as professor at UCL. Now
I’m delighted to be able to share this compilation from it with a wider audience.
I’ve assumed that the reader is familiar with basic phonetic notions and with
the International Phonetic Alphabet: see the IPA Handbook (Cambridge Univer-
sity Press 1999) and the IPA chart (www.langsci.ucl.ac.uk/ipa/ipachart.html). If
you want to brush up on this, there are plenty of textbooks available. Try
Practical Phonetics and Phonology by Beverley Collins and Inger M. Mees
(Routledge 2013, third edition). Wikipedia is also a useful and generally reliable
resource. You may wish to refer to my own works: Accents of English (Cambridge
University Press 1982, three volumes), Longman Pronunciation Dictionary
(Pearson Education 2008, third edition – referenced below as LPD) and English
Intonation: an Introduction (Cambridge University Press 2006).
In this book I put phonetic symbols in bold, without slashes or brackets unless
it is relevant at that point to distinguish phonemes (in slashes //) from allophones
or general-phonetic sound-types (in square brackets []). To indicate letters as
opposed to sounds, I use italics. The prosodic conventions I use are a vertical
stroke (|) to show an intonation phrase boundary, underlining to show the
location of the nuclear (tonic) syllable, and the marks \, / and \/ to show a fall,
a rise and a fall-rise respectively: see fuller discussion in 4.1–2 below.
Words written in capitals, e.g. DRESS, are keywords standing for the entire
lexical set of words containing the vowel in question: see my Accents of English,
chapter 2.2, or the Wikipedia article on ‘lexical set’. Occasionally I use an
asterisk (*) to denote an unacceptable form, as is usual in linguistic work.
I will be delighted if this modest collection of observations inspires a few
readers to further explore traditional general phonetics for themselves and to
make their own observations on how both English and other languages are
pronounced. Linguistic diversity is fascinating, and pronunciation never
stands still.
John Wells
London, February 2014
xi
1 How do you say. . .?
The pronunciation of English words,
including proper names
1.1 sloth
I saw an American TV programme recently based on the theme of the

seven deadly sins. One of them is sloth, which the presenter pronounced slɔːθ, to
rhyme with cloth. But this is not my pronunciation: I call it sləʊθ, to rhyme with
both and growth.
The literal meaning of the name of both the deadly sin and the slow-moving
animal named after it is ‘slowness’. The word is cognate with the adjective slow:
as warm is to warmth, so slow is to sloth. There are some abstract nouns in -th
that involve a vowel change when compared with the adjective they are based on:
deep – depth, wide – width. But historically sloth was not one of them. Indeed, in
earlier times it was sometimes spelt slowth or sloath. So the form rhyming with
cloth must have originated as a spelling pronunciation based on the now current
spelling, which according to the OED was first used in the sixteenth century.
British dictionaries generally give only the pronunciation rhyming with growth.
But in American English (AmE) the pronunciation rhyming with cloth appears to
predominate.
1
2 how do you say…?
1.2 nucular snitchel

The pronunciation of nuclear as if it were spelt nucular is well
known and widely condemned as a mispronunciation. President Bush, for
example, normally said ˈnuːkjəlɚ instead of the standard AmE ˈnuːkliɚ. This
(mis)pronunciation presumably arose through the influence of the large number
of familiar words ending in -kjəlɚ (British English [BrE] -kjʊlə or -kjələ),
among them circular, particular, spectacular, molecular, secular, perpendicular,
jocular.
Two other mispronunciations that one hears from time to time are percolator
as ˈpɜːkjəleɪtə instead of ˈpɜːkəleɪtə and escalator as ˈeskjəleɪtə instead of
ˈeskəleɪtə. In these, a yod has been inserted between k and əl under the influence
of words such as speculator and perhaps even our own technical term articula-
tor. And then we have prenuptial and nuptials pronounced -tʃuəl, as if spelt
prenuptual, nuptuals. I have also heard people referring to a defibrillator as if it
were spelt defribulator.
But my all-time favourite of this type of thing has to be the Wiener schnitzel
masquerading as a ˈsnɪtʃəl, thereby solving at a stroke what are from the English
point of view two phonotactic problems in the German-style pronunciation:
initial ʃn- and morpheme-internal -ts-.
Dictionaries perform various roles. The main users of pronunciation diction-
aries such as LPD are teachers or students of English as a foreign language
(EFL) who wish to know how to pronounce a word whose written form
they perhaps know well but about whose pronunciation they are uncertain.
A second group of users are native speakers who want to have their ignorance
remedied or their prejudices confirmed or disconfirmed by a lexicographer
they consider authoritative.
The lexicographer, though (me, for instance), if he is trained in linguistics,
will always be acutely conscious that there is no absolute standard of correctness
in language, particularly in a language with such a wide speakership and inad-
equate orthography as English.
That is why I have always striven to make LPD play a further role: that of
documenting the current actual state of the language. That is why I felt it
essential to include widespread though ‘incorrect’ pronunciations.
However, if I had merely listed these variants along with all the other variants
that warrant inclusion, I would be failing in my duty, because I would be
ignoring the attitude of careful, educated, literary-inclined, speech-conscious
users towards them. That’s why I use a special warning mark: the exclamation
mark enclosed in a triangle.
Not to include such variants would be to fail to document the language
properly – a charge that I think could reasonably be levelled against certain other
pronunciation dictionaries. To include them but without a distinguishing mark
would be to fail to indicate the attitude of other speakers towards them.
1.4 C. diff. 3
1.3 artisanal
How do you pronounce artisanal? That was a question a correspond-
ent asked me. At the time I had to reply that I did not know the word: it wasn’t
part of my vocabulary and I’d never heard anyone say it. However, despite the
fact that artisan is ˌɑːtɪˈzæn, I thought the regular stress effect of the suffix -al
ought to yield ɑːˈtɪzənəl. Adjectives in -al, of three or more syllables,
have penultimate stress if the penultimate vowel is long (archetypal, primeval,
universal) or followed by a consonant cluster (dialectal, incidental), but other-
wise have antepenultimate stress (personal, industrial, medicinal). (There are
one or two exceptions in which the penultimate vowel lengthens on adding -al,
as adjectival -ˈtaɪvəl.)
The word artisanal was not in the OED until added in 2008. It seems to have
come into use quite recently as a kind of antonym of industrial, referring to
small-scale production methods, agricultural or other.
Some would have us believe that it is ˈɑːtɪzənəl or ˌɑːtɪˈzænəl. I don’t think
so. That looks like someone’s guess – someone who hasn’t fully absorbed the
English stress rules.
Early one morning I was jolted out of my half-sleep by hearing someone on
my bedside radio, on the farming programme, use the word not once but twice.
I can’t remember what they were talking about, but my semi-conscious mind
did note that the pronunciation used was indeed ɑːˈtɪzənəl. Result! And the
OED now agrees.
1.4 C. diff.
The bacterium Clostridium difficile has been in the news recently.

It is a cause of diarrhoea, usually acquired in hospital. Like all organisms it
has a Latin binomial name, the first part referring to the genus and the second
to the species. There’s no difficulty about how to say the generic name: clearly,
klɒˈstrɪdiəm. This can also be abbreviated to C., and in this form is often
pronounced just as siː.
It’s the specific name that raises the interesting question. Apparently this
species of Clostridium is called difficile because when it was first discovered
it was difficult to grow in the laboratory. In Latin, difficile is the neuter
singular nominative of the adjective difficilis ‘difficult’ – neuter because it
agrees in gender with the neuter Clostridium. (With a masculine or feminine
noun it would be difficilis.) In accordance with the usual rules for pronouncing
Latin words in English, you would expect it to be dɪˈfɪsɪleɪ or perhaps
dɪˈfɪkɪleɪ.
But it so happens that difficile is also the French word for ‘difficult’. And
in French it is pronounced difisil, anglicized as ˌdɪfɪˈsiːl. Because doctors and
other health care professionals these days are much more likely to know French
than to know Latin, it is perhaps not surprising that the bacterium is generally
known in English as klɒˈstrɪdiəm ˌdɪfɪˈsiːl.
Or C. diff. ˌsiː ˈdɪf for short.
1.5 shih tzu

Here’s a dreadful pun: I went to the zoo the other day, and all they
had was a dog. It was a shih tzu. Apologies for the indelicacy.
This breed of dog can be spelt either shih tzu or shi tzu. Either way, it is an
old (Wade–Giles) romanization of the Chinese word for ‘lion’, which in Hanyu
Pinyin would be written shizi, or with tones shīzi. (See http://en.wikipedia.org/
wiki/Wade–Giles and http://en.wikipedia.org/wiki/Hanyu_Pinyin for further
details.) In Chinese characters it is 狮子 (simplified), 獅子 (traditional). The
Chinese pronunciation is ʂɨɻ d̥ z̥ ɯɹ, said on a high level pitch (the first syllable has
tone 1, the second is toneless).
This word is in LPD, where I gave its English pronunciation as ˌʃɪt ˈtsuː. On
reflection, having heard dog fanciers talking about the breed, I think that the
second syllable ought rather to be shown as ˈzuː or ˈsuː. (In English we can’t
really manage ts in syllable-initial position, so we transfer the plosive into the
preceding syllable. The spelling tz normally corresponds to the pronunciation ts,
as in quartz; but the presence of z in the spelling tends to lead people to use z
when the plosive and fricative are separated, as here.)
Given that, the pun reduces to ˌʃɪt ˈzuː vs. – what? Well, actually ˈʃɪt ˌzuː.
What stops the pun being perfect is intonation: the fact that in the latent
version zoo has already been mentioned, so that on the second mention it will
be deaccented, throwing the nucleus onto shit. The dog is a shih tzu, but the
rubbishy zoo is in this context a shit zoo.
I must thank Kwan-hin Cheung for the choice of symbols in the trans-
cription of Chinese here. In 1992 he published the following paper in
Chinese: ‘北京話 ‘知’ ‘資’二韻國際音標寫法商榷 [IPA transcription of the
so-called ‘apical vowels’ in Pekinese]’, Research on Chinese Linguistics in
Hong Kong (ed. T. Lee) (Linguistic Society of Hong Kong 1992, pp. 37–46).
He comments, ‘The collection itself was published bilingually, with a
majority of papers in English. This paper predates John Laver’s discussion
on apical vowels in his Phonetic Description of Voice Quality (Cambridge
University Press 1994, pp. 319–20).’
In his article Cheung mentions the symbols [ʅ] and [ɿ], which ‘linguists
within Chinese communities . . . often regard . . . as the IPA symbols for
the two sounds’, although in fact ‘not only are they not official IPA symbols
but they are hardly internationally known’. He would represent shi in IPA
as [ʂ͡ ɨ] and si as [s͡ ɯ]. The slur marks reflect the fact that friction may extend
into the vowel.
1.7 plethora 5
1.6 omega
‘This food contains valuable omega-3 fatty acids,’ says the nutrition
expert on The Food Programme on BBC Radio 4. But he pronounces it əʊˈmiːgə.
I’ve got this pronunciation of omega in LPD as an AmE variant, along
with -ˈmeɪgə and -ˈmegə. But at the time I drafted the first edition of LPD,
back in the late 1980s, I clearly assumed that the only British pronunciation
was ˈəʊmɪgə (or -məgə).
Not any more! The traditional BrE pronunciation of the name of the last letter
in the Greek alphabet is indeed ˈəʊmɪgə. Studying classical Greek from the
age of twelve on until I got my BA in Classics at age twenty-one I do not think
I ever heard any other way of pronouncing it. Yet given that the classical Greek
name was just ὦ (ō), this could be seen as a little surprising. Why don’t we
simply call the Greek letter əʊ? The name omega dates from the Byzantine
period. By then the distinction between classical short ŏ and long ō had been lost,
and the two letters ο (ŏ) and ω (ō) had to be distinguished by special names:
‘small o’ o micron and ‘big o’ o mega.
Modern Welsh does something similar with the letters i and u. In southern
Welsh they stand for the same vowel sound (ci ‘dog’ rhymes exactly with tu
‘side’), so when I was learning Welsh I was taught to call the first one i dot and
the second u bedol (i horseshoe).
OK, the medieval name of the Greek letter was ὦ μέγα (ō mega). That still
doesn’t explain why in English we traditionally stress it on the first syllable.
Indeed, given that the element mega is in implicit contrast with micron, you
might think that the syllable -meg- would bear fossilized contrastive stress. As
usual, the reason that it doesn’t is to be found in Latin. Most Greek words in
English have passed to us via Latin, and on the way have become subject to the
Latin stress rule (see also 1.3 above and 2.12 below). The Latin rule says look at
the penultimate syllable: if it is light (¼ short vowel, not more than one following
consonant), then stress the antepenultimate. In the mega element the e is indeed
short (cf. megabyte, mega store, megalomaniac) and it is followed by the single
consonant g. So stress goes onto the preceding syllable. The e itself undergoes
vowel reduction. Hey presto: ˈəʊmɪgə.
1.7 plethora
How do you pronounce plethora?

Not everyone has this word in their vocabulary. The OED defines it as first
having a medical meaning – a morbid condition – and then a figurative one,
‘over-fullness in any respect, superabundance; any unhealthy repletion or excess’.
The Longman Dictionary of Contemporary English (Longman 2009, fifth edition;
hereafter LDOCE), more in tune with current usage, defines it as ‘a very large
number of something, usually more than you need: a plethora of suggestions’.
I pronounce it ˈpleθərə. That’s also what’s recommended by every dictionary

I have to hand, and what I usually hear. I have also, however, heard plɪˈθɔːrə,
which you will find in LPD and some other dictionaries as an alternative.
Ha!, I thought to myself, this must be another of those words like omega, a
Greek/Latin word with a short penultimate vowel followed by a single conson-
ant: the Latin stress rule (see 2.12 below) therefore gives us antepenultimate
stress. People ignorant of the quantity of the o vowel mistakenly suppose it to be
long, and consequently accent it.
But I was wrong. It turns out that our modern word comes from a medieval
Latin plēthōra, from a Greek word πληθώρη (plēthōrē). The o in the penultimate
syllable was definitely long. The OED reports that the 1731 edition of Nathan
Bailey’s Dictionary had ‘the etymological pronunciation’ pleˈthōra, but that the
1742 and 1755 editions of the same book have ˈplethora. And that’s what most
of us say today, despite the etymology.
1.8 diocese
The area that an Anglican or Roman Catholic bishop controls is a

diocese. How do we pronounce this word? More trickily, what is its plural? In
England there are a number of dioceses. How do we say that? And the derived
adjective is diocesan, as in Diocesan Board of Finance. Where does the stress go?
If you had asked me that when I was aged nine or ten, I think I would have
been able to reply confidently: ˈdaɪəsɪs, ˈdaɪəsiːz, daɪˈɒsɪzn. But that’s because
my father was a vicar, and these were everyday words in our house. As far as
I know, those were also the pronunciations used by everyone else concerned with
ecclesiastical administration, at least in our diocese (Liverpool). It means that
diocese must have been attracted in its phonetics to the singular-plural alternation
we see in such Greek-via-Latin-derived words as crisis, thesis, basis, oasis,
emphasis, neurosis, ellipsis, analysis and axis. (Compare also this and these.)
singular plural
crisis ˈkraɪsɪs crises ˈkraɪsiːz
thesis ˈθiːsɪs theses ˈθiːsiːz
diocese ˈdaɪəsɪs dioceses ˈdaɪəsiːz
However, this obviously produces a mismatch between sound and spelling. More
to the point, most people are not from clergy families and may have no experi-
ence of ecclesiastical terminology. Not surprisingly, they tend to pronounce
both diocese and dioceses the way they are spelt, with a regular sibilant-stem
plural, making diocese perhaps ˈdaɪəsiːs or even ˈdaɪəsiːz and for dioceses
appending the usual additional-syllable ending ɪz (or əz).
Jack Windsor Lewis points out that this word was spelt diocess from
the sixteenth century to the end of the eighteenth; this was the only form
1.9 West Indian islands 7
recognized by Dr Johnson and the other eighteenth-century lexicographers, and

was retained by some (notably by The Times newspaper) into the nineteenth
century. During that century, however, diocese (as in French) became the
established spelling.
The general issue I faced when compiling LPD was: do I prioritize the
pronunciation used by those who use the word regularly, day in day out? Or
do I heed those speakers for whom the word is a relatively unfamiliar written
term? I chose the first, namely ˈdaɪəsɪs; other dictionaries, in this case at least,
tend to choose the second, -siːs or -siːz.
There’s a similar problem with cervical, where medical specialists generally
use penultimate stress and a long i but the general public uses initial stress and
a short i.
1.9 West Indian islands
Thoughts on returning from abroad: if you fly with Virgin Atlantic

or British Airways from London to Antigua in the West Indies, you find
everyone from passengers to check-in staff to the aircraft captain pronouncing
it ænˈtiːgə, just as the locals do. But if you fly there from San Juan or New
York on an American airline, as likely as not it will be referred to as ɑːnˈtiːgwə,
with a Spanish pronunciation (sort of). Probably this is because Americans
are much more likely than Brits to have learnt some Spanish at school, and
of course the name does look Spanish and was originally Spanish, so that
those unfamiliar with the name tend to assume it is Spanish, despite the
island’s three and a half centuries of speaking nothing but English. The
Brits, on the other hand, mostly have French, rather than Spanish, as their
main foreign language (if indeed they know a foreign language at all), and
perhaps anyhow retain some ancestral memory of how to pronounce the names
of former colonies.
Why, though, do we pronounce this Spanish-derived name without the w it
has in Spanish? For the town of Antigua in Spanish-speaking Guatemala we
indeed retain it. I don’t know the answer. As a Spanish word antigua is anˈtiɣwa
(it is the feminine form of the adjective antiguo anˈtiɣwo). We preserve this w
when we anglicize the names Paraguay, Uruguay and Guatemala: so why do we
treat Antigua differently? I don’t know why, but we do.
Antigua’s twin island is Barbuda, pronounced – by those who are familiar with
it – as bɑːˈbjuːdə. It tends to crop up mainly in the official name of the country,
Antigua and Barbuda. Non-locals therefore often need to pronounce this name,
and they do tend to say it without its j as bɑːˈbuːdə, which is what it would be if
it were Spanish-speaking and not just Spanish-derived.
You see something similar to Antigua with jaguar. In the UK we anglicize
it very thoroughly as ˈdʒægjuə, while the Americans are more likely to go for
a Spanish-style ˈdʒæɡwɑːr. In Nicaragua, however, do I detect a slight
movement in Britain away from the anglicized ˌnɪkəˈrægjuə in the direction

of -ˈrɑːgwə, as in AmE?
Another West Indian island with a tricky pronunciation is Dominica.
Locally, and locals would insist correctly, it is stressed on the penultimate:
ˌdɒmɪˈniːkə. Those not familiar with the island tend to assume that it has
antepenultimate stress, dəˈmɪnɪkə. Fans of Chomsky and Halle’s Sound
Pattern of English (Harper & Row 1968) will see that the disagreement relates
to differing views about the length of the penultimate vowel: if long, then
stress-attracting, but if short – as in Monica, Veronica, America, angelica and
many others – then stress-repelling. In the case of Dominica matters are made
worse by the existence of a similarly named but quite distinct West Indian
country, the Dominican Republic. Its name is normally pronounced with the
expected antepenultimate stress. (Dominica, also known as the Commonwealth
of Dominica, is the island located between Guadeloupe and Martinique in the
eastern Caribbean. Its official language is English, the unofficial language
French Creole. The Dominican Republic, aka Santo Domingo or Dominicana,
occupies the eastern half of the island of Hispaniola, the western half being
Haiti. Its language is Spanish.)
1.10 place names
Driving to Gatwick Airport a few days ago to meet an arriving

passenger, I passed through the village of Burgh Heath. As on previous occa-
sions when I have travelled that route, I wondered idly how it’s pronounced.
Is the first word bɜː or ˈbʌrə?
When I got back home I looked it up in the BBC Pronouncing Dictionary
of British Names (Oxford University Press 1990, second edition), which says
it can be either. Just not bɜːɡ.
I further learnt that Burgh in Norfolk is ˈbʌrə, but Burgh in adjoining
Suffolk is bɜːɡ. Things are different in the north of England: Burgh-by-Sands
in Cumbria is metathesized to brʌf, which must mean that for many locals it’s
more like [brʊf].
It’s worse than -ough.
One day I wanted to go to a funeral on the outskirts of Birmingham. To
reach my destination the map said I had to look for the road leading to Alcester.
Er. . . what was that? I checked with my brother, who lives not too far away, and
he said it was ˈɔːlstə. Then I looked in LPD and found that I agree.
And there’s no call for Americans to feel superior to the wacky British.
In the States you never know what will happen with Spanish names.
I remember passing through Salida, Colorado. That’s the Spanish for ‘exit’,
and it was at the mouth of a canyon, so I thought that in English it would be
səˈliːdə. But the local radio station announcers, who should know, pro-
nounced it səˈlaɪdə.
1.12 Charon 9
Even English-derived names can be surprising. I remember driving through

Placerville, California, and discovering to my surprise that it was not ˈpleɪsɚvɪl
but ˈplæsɚvɪl.
My correspondent John Cowan points out that ˈplæsɚ is the standard AmE
pronunciation of placer, a deposit of gravel which contains particles of gold or
other precious minerals. Placer mining is the kind you do with a pan in which
you wash the gravel, letting it flow over the lip of the pan along with most of the
water, leaving the heavier gold behind. Despite appearances, the word is derived
from Spanish placer, which may be of Portuguese or Catalan origin. Another
correspondent, William Pitfield, adds that Burgh Heath gave its name to the local
telephone exchange, covering a significantly larger area, including the place
where he grew up. He doesn’t recall any of the locals calling it anything other
than ˈbʌrə Heath, at least among themselves. The exception was in the days
before Subscriber Trunk Dialling, when it was wise to ask the operator for bɜː
Heath, as otherwise they were liable to look for it under Borough and then assure
you that the exchange didn’t exist.
1.11 Madejski
Any visitor to Reading, on the way from London to Bristol, will be

struck by the road signs everywhere pointing to the Madejski stadium. Why does
the stadium have a Polish name? And how do people pronounce it?
Reading Football Club’s stadium is named after John Madejski, one of the
wealthiest people in the UK, the club’s benefactor. He was born in Britain,
with an English mother, but his father was Polish, an airman who was stationed
here during the Second World War. As a Polish name, Madejski is pronounced
exactly as spelt. The Polish diphthong written ej is similar to the English
vowel in face, so the name would anglicize very easily as məˈdeɪski. How-
ever, English people do not know much about the pronunciation of foreign
languages, except to some extent French. So they do not expect the letter j
to stand for anything other than dʒ or (in foreign words) ʒ. And the resultant
*məˈdeʒski would violate English phonotactics and thus be impossible. In fact,
I am told that the usual pronunciation of the name in Reading is məˈdʒeski
or məˈdʒedski, as if spelt Madjeski or Majedski. The metathesis makes it
pronounceable.
1.12 Charon
A few years ago astronomers were discussing just what a planet is

and exactly which heavenly bodies qualify as pukka planets. One of the candi-
dates for planetary status is Charon, Pluto’s moon. One TV news presenter called
it ˈʃeərən. It is, however, traditionally pronounced in English with k, just as are
other Greek words and names spelt ch-: Charybdis, chemistry, chiropractic,
chlamydia, Chloë, chlorine, chlorophyll, cholesterol, choreography, Christ,
chromosome, chronograph and also (usually) chimera, chiropody. Pronunciation
dictionaries are unanimous for initial k in Charon: ˈkeərən, -ɒn.
Long before his name was given to the moon of Pluto, Charon was a figure
in classical mythology: the grim ferryman who ferried dead souls across the
river Acheron (or Styx) in Hades. I well remember Charon from Virgil’s
Aeneid, book six, line 299. As a teenager I was required to learn off by heart
20–30 lines of it:
portitor has horrendus aquas et flumina seruat
terribili squalore Charon, cui plurima mento
canities inculta iacet, stant lumina flamma,
sordidus ex umeris nodo dependet amictus.
‘A frightful ferryman serves these waters and streams, in terrible filth, Charon. On his
chin is an unkempt white beard, his eyes stand out with flame, a dirty tunic hangs knotted
from his shoulders.’
Virgil’s hexameters make it clear that the first vowel in Charon was actually
short in Greek and Latin:
terrĭbĭ|lī squā|lōrĕ Chă|rōn, cuī | plūrĭmă | mentō
The Greek form, Χάρων, makes it clear that his second vowel was long. So
perhaps we ought to be saying ˈkærən (or even ˈkærəʊn) rather than ˈkeərən.
I found most of the Latin literature I studied at school and university pretty
uninspiring. But the sixth book of the Aeneid was an exception. Virgil’s lines
about the lost souls yearning to be transported across the river really have that
tingle factor:
huc omnis turba ad ripas effusa ruebat,
matres atque uiri defunctaque corpora uita
magnanimum heroum, pueri innuptaeque puellae,
impositique rogis iuuenes ante ora parentum:
quam multa in siluis autumni frigore primo
lapsa cadunt folia, aut ad terram gurgite ab alto
quam multae glomerantur aues, ubi frigidus annus
trans pontum fugat et terris immittit apricis.
stabant orantes primi transmittere cursum
tendebantque manus ripae ulterioris amore.
‘To this place there rushed a whole crowd, pouring out onto the river bank: mothers
and men and the lifeless bodies of great-hearted heroes, boys and unwed girls, young
men placed on the funeral pyre before their parents: as many as the leaves that fall
in the woods at the first frost of autumn, as many as the birds that flock together onshore,
leaving the deep abyss, when the cold weather drives them across the sea to warmer
lands. They were standing there, praying to be carried across first, and holding out their
hands in yearning for the opposite bank.’
1.14 Chagos 11
1.13 Judea
BBC TV showed a rather well done series on Roman history, Ancient
Rome: The Rise and Fall of an Empire. One episode focused on the story of the
Jewish Revolt, which swept through Judea in AD 66, and a disgraced general’s
attempt to put it down. I noticed that every single actor pronounced Judea as
dʒuˈdeɪə. But when I was doing Ancient History at school it was most definitely
dʒuˈdiːə (or rather the derived-by-smoothing dʒuˈdɪə). As far as I know, that’s
still what people say in church, too.
Judea or Judaea is the Latin name, via Greek Ἰουδαία, of the southern part
of ancient Palestine. Two millennia ago, in those languages, it was pronounced
something like juːˈdaia – but in both languages subsequently became juːˈdeːa.
The regular development of this in English is dʒuˈdiːə or dʒuˈdɪə. Other names
with a written e deriving from Latin ae and Greek αι include Egypt and Ethiopia.
In English they still have iː. Other words with final stressed -ea include idea,
diarrh(o)ea, Thea and Korea. I have not heard eɪə in any of them. For compar-
able cases, in which a traditional iː now faces competition from an upstart eɪ, we
have to look to nucleic and the -eity words: spontaneity, homogeneity, simultan-
eity and, above all, deity. Indeed, for deity 80% of the BrE respondents in my
1988 pronunciation preference poll (reported in LPD) preferred the eɪ form.
Merriam-Webster gives both possibilities, while prioritizing iː – as it also does
for Judea, for which British dictionaries haven’t yet caught up with the possibil-
ity of eɪ.
1.14 Chagos
The Chagos islanders were forced from their homes at gunpoint in
the 1960s, as an entire British colony was handed to the US so that a monitoring
station could be built in the Indian Ocean. The exiles now live in the slums of
Mauritius where they belong neither socially nor economically. They’re cam-
paigning to return to the Chagos Islands.
On BBC Radio 4 I heard two contributors discussing the issue. One of them
pronounced the name of the archipelago ˈʃɑːɡɒs, the other ˈʃæɡɒs. This suggests
that neither of them had consulted the BBC Pronunciation Unit. Possibly, as
experts on the subject, they felt they didn’t need to. Because what it recommends
is ˈtʃɑːɡəʊs.
In view of the fact that the islands were discovered by Vasco da Gama and
presumably named in Portuguese, we should expect the spelling ch to correspond
to ʃ rather than tʃ (which is what it would imply in Spanish). The adjacent islands,
Mauritius and the Seychelles, are French-speaking, which would again favour
ch ¼ ʃ. Until 1965 the Chagos Islands were administered as part of Mauritius.
I can quite see that Americans (who now run the islands and who usually know
some Spanish) would in all likelihood pronounce the name ˈtʃɑːɡoʊs.
1.15 proceed and precede

Every now and again my students would reveal that they were
confused about the words precede and proceed. Accounts of phonetic processes
and allophonic rules often refer to a preceding consonant, or to something
being preceded by a vowel, and so on. In the sequence ABC, B precedes
C and is preceded by A.
But students sometimes write this as a ‘proceeding’ consonant, or being
‘proceeded’ by a vowel. Worse, since to proceed from can mean to follow,
they sometimes interpret my spoken precede, which they imagine to be pro-
ceed, as meaning ‘follow’, so that they also have the meaning the wrong
way round.
In theory, the distinction between the ɪ of precede and the ə (weakened
from əʊ) of proceed ought to be robust. After all, in my own speech and in that
of most of my students valid doesn’t rhyme with salad, nor rabbit with abbot;
the initial syllables of finesse and phonetics differ. But in practice, clearly the
distinction may not be robust: weak pre- and pro- can get confused.
And I have just detected one of my favourite authors, Jared Diamond, com-
mitting the reverse mistake. In his marvellous book Collapse (Penguin 2006),
on p. 501 of the UK paperback edition, we read that ‘LA smog generally [gets]
worse as one precedes inland’. Oh dear. What a precedent.
1.16 hypernymy and thrombolysis
One of our students chose hypernymy as a dissertation topic. How do

we pronounce this word? Where does the stress go? I would say that following
the pattern of synonymy sɪˈnɒnəmi it must clearly be haɪˈpɜːnəmi. But not
everyone finds this obvious. Yet if you go along with Chomsky and Halle in
The Sound Pattern of English (MIT Press 1968), native speakers of English
have as part of their grammar (¼ their implicit knowledge of the language) a
complex set of rules allowing them to produce the appropriate word stress and
vowel qualities for English words, while the words themselves are specified in
our mental lexicon without indication of stress placement and possible vowel
reduction.
Chomsky and Halle were wrong, of course. The fact that anyone was uncertain
about how to pronounce hypernymy bears this out. I have recently noticed
another counter-example to the SPE principle. Every adult native speaker knows
the words paralysis and analysis. Some also know one or more of catalysis,
dialysis, electrolysis or some fifty other technical words with the same ending.
All have antepenultimate stress: pəˈræləsɪs, əˈnæləsɪs, kəˈtæləsɪs, daɪˈæləsɪs,
ˌelekˈtrɒləsɪs, etc.
Some years ago I was diagnosed with a heart condition and underwent an
angioplasty procedure. Since then I have attended the cardiac support group
1.18 French, German or Welsh? 13
run by my local hospital. One of the techniques for treating a heart attack or
stroke is known as thrombolysis. But I have noticed that the cardiologists, cardiac
surgeons, paramedics and nurses don’t call this θrɒmˈbɒləsɪs, which is what you
would predict on the basis of all the other -lysis words. They call it ˌθrɒmbəʊ
ˈlaɪsɪs. Ah well, language doesn’t obey rules as much as some linguists would
perhaps like.
In my lectures I sometimes talk about allophony. Guess how I pronounce it.
Remember that everyone agrees that telephony is təˈlefəni and cacophony is
kəˈkɒfəni, and that I am someone who knows the rules and plays by them. But
millions of native speakers don’t.
1.17 amiodarone
We learn something new every day. When my cardiologist started me
on a drug called amiodarone, he pronounced it ˌæmiˈɒdərəʊn, and that’s what
my GP said, too. But shortly afterwards I went for a blood test. The phlebotomist
called it ˌæmiˈəʊdərəʊn.
So should it be ɒ or əʊ? A short o or a long one? Who cares? The spelling’s
the same, which is what matters for the pharmacist who has to dispense it.
As with so many learned, scientific or technical words, the spelling is fixed
while the pronunciation fluctuates. (Having boned up on heart disease, I am
almost inclined to say it fibrillates.) That’s because instead of hearing other
speakers and imitating what they say, we often create a pronunciation for
ourselves on the basis of the spelling, using the reading rules of English, which
are notorious for their uncertainty.
1.18 French, German or Welsh?

An announcer on Classic FM radio referred to the painter Gauguin
not as ˈgəʊgæ̃ , which is the usual semi-anglicized pronunciation used by edu-
cated English people, but as gaʊˈgæ̃ . It was not so much the American-style
end-stressing of the French name that caught my attention, as the interpretation
of orthographic au as if it were German rather than French. (Think Spandau
and Strauss.)
Mind you, the convention that maps French o (orthographically au, eau, ô,
etc.) onto English əʊ (GOAT) is rather odd. It would be phonetically more
accurate to equate it with English ɔː (THOUGHT). And in the cases spelt au,
including Gauguin, this would also coincide with English orthographic habits.
So ˈgɔːgæ̃ , anyone?
Evidently, the foreign-language spelling au is something of a trap. I also heard
another radio presenter, a BBC person this time, repeatedly pronounce the name
of the Welsh parliamentary constituency Blaenau Gwent as ˈblaɪnaʊ ˈgwent.
The Welsh plural ending -au (as in blaenau ‘tops, heights’, plural of blaen)
has various pronunciations in the varying regional forms and stylistic levels of
Welsh: in careful formal speech ai, or in the north aɨ, and in everyday speech
more usually e or a, or in Anglo-Welsh even ə. But nowhere is it au.
The only people who might say ˈblaɪnaʊ are those who do not know Welsh
and who make a false inference on the basis of what they know about the relation
between spelling and sound in German or some other irrelevant language.
Reading rules for au in non-English words:
in French words əʊ, reflecting French o (gauche, mauve, chauvinist);
in German, Spanish, etc. words aʊ (Frau, Spandau, gaucho);
but in Welsh words aɪ. So it’s Blaen[aɪ] or Blaen[ə] Gwent. The Welsh
count 1, 2, 3 as un, dau, tri, which sounds like English een, die, tree.
1.19 Loony what?
In 2006 the Guardian carried the news that Prince Charles is buying a
country estate in Wales. The name of the estate, Llwynywormwood, attracted
my attention because it is a mixture of Welsh and English that doesn’t really
seem to be fully at home in either language. The Welsh part Llwyn-y- might be
considered by the English to be unpronounceable, since in Welsh it is ɬʊɪn ə.
Not only is there the difficulty of ɬ, but also the ʊɪ is a diphthong, making ɬʊɪn
one syllable. The English part -wormwood won’t do in Welsh, since its second
syllable violates Welsh phonotactic rules, which do not allow w before ʊ.
In Welsh, llwyn means ‘bush’ or ‘grove’. The linking y is the Welsh definite
article, as in Pen-y-bont (head-the-bridge) ‘Bridgend’. So the whole name might be
anglicized as Wormwood Grove. Apparently its Welsh name is actually Llwyny-
wermwd, with the proper Welsh equivalent of wormwood as the second element.
The Welsh spelling ll stands for a voiceless alveolar lateral fricative ɬ. It comes
in many Welsh names (Llanelli, Llangollen, Llandudno, Llywelyn). Non-Welsh
speakers in attempting this sound may well produce xl, θl, kl or, at best, ɬl
instead. However, in Welsh consonants at the beginning of a word are subject
to mutation in particular positions, which means that after most Welsh prepos-
itions (among other things) any word with initial ɬ is said with plain l instead,
e.g. o Landudno ‘from Llandudno’. This provides a justification, if any were
needed, for non-Welsh people to pronounce word-initial Ll- simply as l: which
indeed we do in the name Lloyd (Welsh llwyd ‘grey’). And I heard somewhere
that the Llwyni (‘Groves’) Estate in one south Wales town is generally known as
the Loony Estate.
For those who don’t want to attempt the Welsh version ˌɬʊɪnəˈwermud,
I think – failing a recommendation from the BBC Pronunciation Unit –
I would recommend the anglicization ˌluːɪnəˈwɜːmwʊd, though this does make
it sound unfortunately like a loo in some sort of forest.
1.21 Punjab and feng shui 15
Worse, the nearby village is Myddfai. That’s Welsh ˈməðvai. What’s the
betting we hear ˈmɪdfeɪ?
1.20 Richter
Whenever there’s an earthquake, the newsreaders make refer-

ences to the Richter scale. They mostly pronounce it either ˈrɪxtə or
ˈrɪçtə. But Charles Richter, the creator of the earthquake magnitude scale,
was not from a German-speaking country, and the newsreaders’ otherwise
admirable familiarity with German pronunciation is here misapplied. He was
an American, born in Ohio. Being American, he naturally pronounced his
name ˈrɪktɚ, and we should do the same (or, in the case of us non-rhotic
speakers, as ˈrɪktə).
This information is readily available from the BBC Pronunciation Unit
and in the Oxford BBC Guide to Pronunciation (Oxford University Press
2006). It is explained in some detail in LPD. But newsreaders don’t always
heed offered advice.
1.21 Punjab and feng shui
The British seem to have a kind of unstated but deep feeling that
foreign words shouldn’t contain ʌ. I’m not sure whether Americans and speakers
of other non-British accents share this feeling.
You can see the effect I am referring to in the name of the Indian province
of Punjab. Older readers may remember Peter Sellers as an Indian doctor singing
to the patient Sophia Loren:
I remember how with one jab
Of my needle in the Punjab
I cured the beri-beri and the dreaded dysentery. . .
. . .correctly making the first syllable of Punjab rhyme with one wʌn (though not
correctly rendering the length of the vowel in the second syllable).
But more often these days I hear people saying things like ˌpʊnˈdʒɑːb,
pronouncing the first vowel not as ʌ but as ʊ.
Yet in Hindi, Urdu and Punjabi/Panjabi the name of this province is Panjāb
(in Hindi, phonetically pɐn ̃ dʒɑb). The Hindi/Urdu vowel transliterated as short
a (i.e. the first vowel in Panjāb) is regularly mapped onto English ʌ. Compare
the word pandit, pronounced in Hindi as pɐ̃ɳɖɪt̪ and in English as ˈpʌndɪt (and
usually spelt correspondingly as pundit). Hindi pakkā becomes BrE pukka ˈpʌkə.
But pundit and pukka are so well integrated that people don’t realize that they are
of Hindi origin. Evidently we don’t feel them to be foreign in the way that
Punjab is perceived to be.
It’s the same with the name of the Welsh nationalist party Plaid Cymru.
The word for Wales is ˈkəmri in Welsh and ought to be ˈkʌmri in English.
But English people persistently say ˈkʊmri.
And now from Chinese along comes feng shui, Mandarin 1fʌŋ 3ʂuei. Aficion-
ados know that the approved English pronunciation is not the *feŋ ʃuːi suggested
by the Hanyu Pinyin spelling but rather fʌŋ ʃweɪ. Yet. . . foreign words don’t
have ʌ. So we get people saying fʊŋ ʃweɪ.
1.22 ylang-ylang
One word that I keep hearing (mis)pronounced on television is not

a proper name, but a vocabulary word: ylang-ylang. One commercial current a
few years back proclaimed that Air Wick could fill your house with the delightful
fragrance of jəˌlæŋ jəˈlæŋ.
Oh dear. Dictionaries tell us that ylang-ylang, also written ilang-ilang, is a
Malaysian and Filipino tree, Cananga odorata, with fragrant yellow flowers
from which a perfume of the same name is distilled. The word comes from
Tagalog, where it takes the form álang-ílang. (Just how this got converted into
its modern English form is not clear.) Reference books are unanimous that this
is to be pronounced ˌiːlæŋ ˈiːlæŋ (or with final stress, and with the possibility
of ɑ rather than æ in AmE).
I think this error may come from children’s reading schemes in which the
letter A is called æ, B is bə and so on, so that the letter Y is called jə. And the
voice-over person can never have read Chaucer, where words such as yclept
might have helped.
1.23 Mbabane 17
Those who follow the Olympics will have noticed that one of the classes
of sailing boat competing there is an yngling. This was one of the words I added
to LPD for its new (third) edition. You won’t find it in many other dictionaries.
The original Ynglings were the oldest known Scandinavian dynasty, part histor-
ical but part mythical. As a Norwegian word it would presumably be pronounced
ˈʏŋlɪŋ, i.e. starting with a close front rounded vowel.
But now the name is applied to sailing boats. There is an International Yngling
Association, which according to Wikipedia describes the boat as an ‘agreeable
cross between a planing dinghy and a keelboat’. Apparently the designer,
Jan Herman Linge, wanted to build a keelboat for his young son, and named it
Yngling, the Norwegian (and Swedish) word for ‘youth, young man’; the name
is unrelated to the dynasty, the House of Yngling. Etymologically, the word
would correspond to English young plus the -ling of weakling, duckling.
And how do we pronounce it in English? In LPD I show it as ˈɪŋlɪŋ, and
I believe this to be the usual pronunciation among those who sail that kind
of boat.
But I heard an enthusiastic TV commentator call it a ˈjɪŋlɪŋ. You can see how
this pronunciation might suggest itself to an English speaker unaccustomed to the
idea that y at the beginning of a word could stand for a vowel.
This was no doubt the inspiration for the facetious Guardian letter-writer
yesterday who said he was ‘delighted to see that the ancient art of yngling has
at last been recognised as an Olympic sport. I can’t wait to see how we Brits fare
in the tongling and tiddle I pogling events’.
That will make sense only to Brits of a certain age who remember the Goons
and their Ying Tong song.
1.23 Mbabane
I’m sure we all know that Mbabane is the capital of Swaziland in

southern Africa.
This name is extraordinarily difficult to anglicize. In the local language,
siSwati, it is pronounced mbaˈbaːne. The initial nasal is non-syllabic, part of
the prenasalized voiced bilabial plosive mb. You find this kind of thing in all the
Nguni languages (the term Nguni covers Zulu and Xhosa and one or two others
as well as siSwati, all more or less mutually intelligible).
We can’t cope with this sound in English. Nor can we cope with an initial
cluster mb-. So, as with Mbeki, Ndola and Nguni, we have to insert a schwa
before the nasal to make it pronounceable in English. (An alternative tactic
would be to insert the schwa after the nasal, thus məbæˈbɑːni.)
Whereas the resulting əmˈbeki, ənˈdəʊlə, əŋˈguːni are fine for us, there is
a further problem with a putative əmbæˈbɑːni or məbæˈbɑːni (and equally
with possible variants having ə or ɑː in what is now the second syllable).
The same problem arises if we attempt an initial syllabic nasal, m̩ bæ-. These
are problematic because the phonetic structure of English words is subject
to what is sometimes called the ‘Teutonic Rule’, namely that the first two
syllables of a word cannot both be lexically unstressed. In words with the main
stress on the third syllable or later, we need a secondary stress on the first
or second syllable.
Attempting to put this secondary on the first syllable, unfortunately, would
violate another rule of English phonetics, namely that schwa can never be
lexically stressed.
Even putting it on the second syllable, thus əmˌbæˈbɑːni, or əmˌbɑːˈbɑːni, is
still rather awkward. Perhaps we should go with AmE and make the first syllable
ˌem-. This at last gives us a form that is truly pronounceable in English: ˌembə
ˈbɑːni. I wonder what non-Nguni-speaking native speakers of English who live
there actually say.
By the way, I have discovered that in the 1880s one Michael Wells was
one of the earliest entrepreneurs to set up an industry in Mbabane, namely
a brewery. Since I know that my grandfather Edward Wells was a brewer
in South Africa in the 1900s, I surmise that this was probably my great-
grandfather.
1.24 brand names

Volkswagen, Weetabix, Whiskas. . . brand names are a familiar part
of every native speaker’s vocabulary, and often thoroughly global. Yet ordinary
dictionaries mostly don’t include them.
Whether we want to tell EFL students how we pronounce them, to reassure
native speakers on what to say, or to document the current state of the language
for the instruction and delectation of future generations, brand names (trade
names, commercial names, trademarks) do in my view deserve a place in a
pronunciation dictionary.
When Jones was editor, there were, I think, no brand names in his English
Pronouncing Dictionary (EPD). So the likes of Mazawattee, Dolly Blue and
Craven A, now only of historical interest, remained undocumented. Even Coca-
Cola was not mentioned. The publishers of EPD’s successor, the Cambridge
English Pronouncing Dictionary (Cambridge University Press 2011), have now
changed its policy about trademarks and include them.
Brand names are not without their problems. Many years ago I noticed
that various pronunciations of Hyundai were in use in British television adver-
tisements, though they now seem to have settled on ˈhaɪʌndaɪ – very different
from the Korean source hjəːndɛ. The American version, ˈhʌndeɪ, is closer.
From the English pronunciation of Kyocera, ˌkaɪəˈsɪərə, you’d never guess
that it is an abbreviation of Kyoto Ceramic (Kyōto Seramikku).
1.25 Sexwale 19
The modern lexicographer has resources unknown to earlier generations.

If I want information on a trade mark, I can go on line and consult Wikipedia
or Google. Sometimes the brand owner’s website even carries an indication of
how the name is to be pronounced.
Recently I was trying to discover how to pronounce Freixenet. On the
company’s British website I found clips of their television advertisements, in
which the voice-over clearly says ˈfreʃənet. In the original Catalan, on the other
hand, this word has final stress: fɾəʃəˈnɛt. The company’s American website
is useless in this respect, offering no video or sound clips, just the ambiguous
specification ‘pronounced fresh-eh-net’.
The German electrical brand Braun is usually brɔːn in English, even
though this word, the German equivalent of Brown, happens to coincide with
the English word in both sound and meaning. Spelling pronunciation rules.
For Nestlé, on the other hand, we have generally abandoned our earlier ˈnesl̩ in
favour of ˈnesleɪ.
As for Audi – which in English, as in German, is ˈaʊdi – the slogan Vorsprung
durch Technik does present problems to English speakers. In German it is
ˈfoːɐʃpʁʊŋ dʊʁç ˈtɛçnɪk. But non-German speakers are not familiar with the
idea that the letter V can stand for f, and they do not understand the German
distribution of ç and x, both spelt ch. Anyhow, the cluster ʁç – a uvular fricative
(or approximant) plus a palatal fricative – is tricky by any standards. (Actually,
Germans often make it dʊɐç, so we could attempt at least dʊəx.) We usually say
something along the lines of ˈvɔːsprʊŋ dɜːx ˈtexnɪk.
1.25 Sexwale
No, that isn’t a misprint for something else. Tokyo Sexwale is a

leading South African politician.
In the Bantu languages of South Africa, the letter x can stand for at least three
sounds, depending on the language: a lateral click, a palatoalveolar fricative or
(as in IPA) a velar fricative. Conveniently for the phonetically unsophisticated
speaker of English, the first and last of these map onto English k. So it would
probably be safe to pronounce his name seˈkwɑːleɪ (or perhaps sɪˈkwɑːli).
In the name of the language Xhosa the X stands for a voiceless alveolar lateral
click, ǁ (or in old money ʖ). The following h shows that the accompanying velar
component is aspirated, so in full we have ˈk͡ ǁʰɔːsa. This spelling convention
applies in Zulu, too.
In Tsonga, the letter x stands for ʃ. This is presumably due to Portuguese
influence, since most speakers of Tsonga live in Mozambique. But in Venda (aka
TshiVenda) the letter x stands for the voiceless velar fricative x.
Mr Sexwale’s background is Venda, although he was born in Soweto.
The name of the company he founded, Mvelaphanda, is certainly a Venda word.
The BBC Pronunciation Unit confirms that the orthographic x in this name is
pronounced as a velar fricative.
So in English we should call him seˈxwɑːleɪ or, failing that, seˈkwɑːleɪ.
1.26 tortoise
‘When we were little,’ the Mock Turtle went on at last, more calmly, though
still sobbing a little now and then, ‘we went to school in the sea. The master
was an old Turtle—we used to call him Tortoise—’
‘Why did you call him Tortoise, if he wasn’t one?’ Alice asked.
‘We called him Tortoise because he taught us.’ From Alice in Wonderland
by Lewis Carroll
The pun works perfectly in non-rhotic accents (¼ most English, Welsh and
southern-hemisphere accents) providing you use the pronunciation of tortoise
that I grew up with, namely ˈtɔːtəs.
But every now and again I hear people say ˈtɔːtɔɪs or ˈtɔːtɔɪz. I think these must
be fairly recent spelling pronunciations. In LPD I gave them a warning triangle.
The OED is not very clear about how the spelling came to be -oise, given an
origin in the Latin tortūca, French tortue. The earliest (fifteenth-century) English
spellings included tortuce, tortuse, tortose, as well as the French tortue. Never-
theless, the spelling tortoise is also quite early (mid-sixteenth century). There
seems to be a possibility that the final sibilant may have arisen from, or been
reinforced by, a genitive form as in tortue’s shell (tortoiseshell).
The 1913 OED entry for this word also reported a pronunciation ˈtɔːtɪs, which
I have never heard.
We have a similar situation with porpoise, though here the etymology and
history are rather different. The form porpoys goes back to Middle English,
a borrowing from Norman French, based on some unattested Latin form of
the type porcus piscis ‘pig fish’.
1.28 my liege, we are besieged 21
As with tortoise, so with porpoise. The pronunciation I recommend in LPD is

ˈpɔːpəs, with a warning triangle for ˈpɔːpɔɪs, ˈpɔːpɔɪz.
For both tortoise and porpoise Merriam-Webster reports only -əs.
I remember a misunderstanding when I was a boy. A dead porpoise had been
washed up on the beach in Cumbria where we were on holiday. When we told
the cleaning lady about it, she thought we were telling her about a dead cat – a
poor puss.
1.27 continental and English values
On BBC World television news 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. 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 in 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
(¼ continental); but for Vigo Street in central London it was traditionally the
vowel of tiger (¼ 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 (¼ continental, but
locally æ) and on the other hand tornado and Barbados (¼ English).
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. And then there’s yet another in France, lagoːs.
1.28 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.
The first dictionary to mention this variant seems to have been Webster’s
Third International (1961). You will still not find it in most dictionaries.
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 other long 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. And most English
people seem to use ʒ rather than dʒ in raj, which is not French at all. (Nor, for
that matter, is Beijing.)
So we seem to favour the fricative rather than the affricate after a long vowel.
Is this an incipient sound change? Well, I’ve not heard anyone pronounce huge
as hjuːʒ. . . yet.
1.29 Rothersthorpe
The service area that I know as Rothersthorpe Services on the M1
motorway in the English midlands has been renamed Northampton Services.
This got me idly thinking: how is it that we know immediately how to interpret
these three th spellings? Obviously the pronunciations are ˈrɒðəzθɔːp and
nɔːˈθæmptən (or minor variants thereof). So ð, θ, θ: but given that the spelling
th is used indifferently for both θ and ð, how do we know which one to use where
in an unfamiliar place name?
The spelling-to-sound rules can be summed up as saying that word-initial
th spells θ except in function words such as articles and determiners. Word-
medial th is ð in Germanic words but θ in Greek or Latin words. In the present
discussion we can ignore word-final position.
At first it looks as if getting these names right is crucially dependent
on recognizing morpheme boundaries. We have to be able to see that
Rothersthorpe – an obviously Germanic name – consists of Rothers plus thorpe.
The initial th in -thorpe therefore counts as being initial and hence voiceless.
Compare Scunthorpe and indeed the surname Thorpe. The medial th in Rothers-
follows the rule by being voiced. This is no doubt reinforced by the well-known
place name Rotherham.
Northampton is slightly trickier. Although we say θ in north, we say ð in
northern, northerly and place names such as Northall, Northam, Northenden,
Northiam, i.e. when th is medial – yet θ in Northumberland. The distinguishing
feature of Northampton and Northumberland seems to be that the vowel after the
th is strong, thereby attracting the dental fricative into (syllable-) initial position
and ensuring that it is voiceless. Perhaps that applies to Rothersthorpe, too,
where the strong vowel attracts the second dental fricative into the last syllable.
So it is not the morphology we depend on, but the syllabification.
Another place name with repeated th is Rotherhithe in London. Its polite
pronunciation is ˈrɒðəhaɪð, but a correspondent wonders about ˈrɒðəraɪð,
1.30 Campbell 23
which he feels sounds ‘more Sarf London’. Indeed, in non-rhotic accents

such as received pronunciation (RP) and London r and h are mutually
exclusive in names such as Leatherhead and Wolverhampton. The standard
pronunciation retains the h, thus ˈrɒðəhaɪð, ˈleðəhed, ˌwʊlvəˈhæmptən.
Local working-class accents tend to drop h, giving ˈrɒðəraɪð, ˈleðəred,
ˌwʊlvərˈæmptən. Dropping the h in these names is comparable to drop-
ping the h in phrases such as over here, a bigger house: if the h goes,
linking r is automatically triggered. But dropping the h is clearly stigma-
tized and regarded as uneducated: it would indeed sound ‘Sarf (¼ south)
London’.
Rotherhithe also has a variant Redriff, which attests not only H dropping but
also a centuries-old case of TH fronting. In 1665 Samuel Pepys wrote of walking
to Redriffe. According to Ekwall, the name comes from Old English (OE)
hrȳther þ hȳth ‘landing place for cattle’. But according to others, the first
element is OE redhra ‘sailor’.
The only times I have passed along Redriff Road, Rotherhithe, were while
running the London Marathon. The marathon course takes the runners right
around the Rotherhithe peninsula.
1.30 Campbell
Campbell is a pretty common surname throughout the English-

speaking world.
The mismatch between its pronunciation and its spelling has always puzzled
me. We pronounce it ˈkæmbl̩ yet the spelling has an irrelevant (‘silent’) p before
the b. Why?
The origin of the name Campbell lies in a Scottish clan name. The Campbells
occupied an extensive area of Kintyre and the western Highlands. According to
the Oxford Names Companion (Oxford University Press 2002), which incorpor-
ates A Dictionary of Surnames (1988) by Patrick Hanks and Flavia Hodges,
its likely etymology is Scottish Gaelic, in which cam means ‘crooked’ and
beul ‘mouth’. So Cambeul or Caimbeul was a nickname, ‘wrymouth’, apparently
borne by the clan founder Gillespie Ó Duibhne, who lived at the beginning of
the thirteenth century.
According to Wikipedia, there is a separate surname Campbell, of Irish origin,
from the Irish Mac Cathmhaoil meaning ‘son of the battle chieftain’ (though
personally I’d have thought that an Irish name of that spelling would yield the
English form McCall).
But what about the spelling? As so often, it turns out that Norman scribes and/
or lawyers writing in Latin were to blame. They came up with an elegant Latin
equivalent of the Gaelic name, turning it into de Campo Bello ‘of the fair field’.
Just as the b in debt and doubt is down to Latin etymologizing, so is the p in
Campbell.
1.31 malapropisms
When people speak of some man having an enlarged ˈprɒstreɪt
or undergoing a ˈprɒstreɪt operation, is this just a non-standard pronunciation
of ˈprɒsteɪt prostate?
Not really. Rather, it reflects a confusion between the words prostate and
prostrate. People who pronounce the gland in question as the ˈprɒstreɪt will
also want to spell it prostrate. For them, the gland and the word meaning ‘lying
flat, face down (in exhaustion or submission)’ are perceived as homonyms.
Actually, since the latter is a relatively learned and unfamiliar word, they may
know only the glandular meaning.
I wonder what they make of the words of the hymn.
All hail the power of Jesu’s name,
Let angels prostrate fall;
Bring forth the royal diadem
And crown him Lord of all.
A similar case is the frequent confusion between silicon (chemical element, Si)
and silicone (polymeric compound, used e.g. for breast implants). Referring in
writing to a ‘silicone chip’ or in speech to a ˌsɪlɪkəʊn ˈtʃɪp is a solecism.
In a pronunciation dictionary, ought ˈprɒstreɪt to be included as a variant
pronunciation of prostate, perhaps with a sign to show that it is not considered
correct? My answer is no. We cannot cater for Mrs Malaprop. But it can be
difficult to know where to draw the line.
1.32 ho, ho, ho

How do we pronounce the name of the letter H?
All dictionaries agree that it is eɪtʃ. Only LPD gives an alternative possibility,
heɪtʃ, which it marks with the symbol § to show that it is not the RP form. I added
a note, ‘The form heɪtʃ is standard in Irish English, but not BrE or AmE’. This
claim is possibly too sweeping, but on visits to Irish universities I had noticed
Irish academics at all levels using this form. I was particularly struck by the
expression piː heɪtʃ diː PhD.
At that time, twenty or thirty years ago, calling the letter heɪtʃ was perceived in
England as simply wrong, a pronunciation likely to be used only by the illiterate and
uneducated – the people who were prone to H dropping and to hypercorrect H insertion.
But things seem to have changed. Nowadays many younger Londoners, at any
rate, call the letter H heɪtʃ. I have taught undergraduates who would defend this
as the correct pronunciation, perhaps on the grounds that the name of a conso-
nant letter should contain an instance of the corresponding consonant sound.
So I included the question of aitch or haitch in the online survey of pronunci-
ation preferences that I conducted in 2007.
1.34 ring in the old ear 25
As reported in the third edition of LPD (2008), the overall voting figures were
84% for the traditional form, 16% for the newcomer. Comparing the preferences
of different age groups shows that heɪtʃ is indeed an innovation that appears to
be spreading rather fast. As many as 24% of those born since 1982 voted for it.
It’s sometimes claimed that we divide on religious lines. I mentioned this point
25 years ago in my Accents of English (Cambridge University Press 1982, vol. 2,
p. 432) in the chapter about Ireland.
The letter H itself [. . .] is called heːtʃ, at least by Catholics. (This is in fact
widely considered a sectarian shibboleth, with the Protestants calling it eːtʃ,
cf. RP eɪtʃ.)
It’s clear that in London, at least, this no longer applies: in no way is the
h- pronunciation restricted to Roman Catholics or those taught at Catholic
schools. Shortly after I wrote this I had a conversation with one of the youngest
members of my partner’s family, who was then aged five. ‘He knows the names
of the letters,’ I reported, ‘and one of them, he declares, is heɪtʃ. His family are
Methodists, educated at state schools.’
1.33 d’oh, doh… duh
What does Homer Simpson say when he realizes that he has said
or done something stupid?
You’ll find the interjection spelt variously as d’oh, doh or duh. It’s a kind
of euphemism for damn.
Of course it’s not really Homer who says it, but the actor Dan Castellaneta,
who plays the part of Homer.
From analysis of TV clips of the cartoon I can report that it’s pretty clear
that he, at least, pronounces this interjection as doʊ (though in some cases
it seems almost to approach daʊ). It is thus a homophone of dough, and is often
pronounced with a special harsh or constricted voice quality. Like all exclam-
ations, it is always said with a falling, or at least a non-rising, nuclear tone.
The interjection spelt duh is different. Whereas Homer says d’oh, and was
perhaps the first to do so, duh was in use in AmE long before the Simpsons
appeared on the scene, going back at least to Laurel and Hardy. It is pronounced
dʌ, or like duck without the final k, and is thus not a homophone of dough.
The meaning of duh is different, too: it is used to comment on someone else’s
stupidity, while d’oh is a comment on one’s own stupidity.
1.34 ring in the old ear
Tinnitus is the perception of sound in the absence of corresponding

external sound: an auditory disorder in which you hear a ringing in your ears.
How do we pronounce it? Is it tɪˈnaɪtəs, with stress on the penultimate, or

ˈtɪnɪtəs, within initial stress? Dictionaries differ.
Here’s what the American Tinnitus Association says:
TINNITUS is pronounced either ti-NIGHT-us or TIN-i-tus. Both pronunci-
ations are correct; the American Tinnitus Association uses ti-NIGHT-us.
I remember my colleague Michael Ashby telling me he thought that in LPD

I ought to have prioritized ˈtɪnɪtəs, for BrE at least. On reflection I rather
suspected he might be right. My problem is that, being a classicist by training,
I am perhaps too strongly influenced by my awareness of the quantity of the
second vowel in Latin, tinnītus, which implies penultimate stress and aɪ in
English.
It was one of the items about which people voted in the BrE survey I carried
out in 2007. The results were interesting. Overall, respondents preferred
ˈtɪnɪtəs by a wide margin. However, the one age group for whom tɪˈnaɪtəs
made a strong showing was those aged below twenty-six years. There is no
sudden recent swing towards studying Latin in our schools, so I am at a loss to
explain this.
1.35 /Who did you say were all liars?

I heard an ad on a commercial radio station enumerating some of
the many attractive features of Greece, including ˈkriːʃn̩ olive groves.
I don’t know if this is phonetics or more properly morphology, but surely the
adjective from Crete is more usually Cretan ˈkriːtn̩ .
At first I assumed that Cretian was simply a mistake. A check in the OED,
though, shows that things are more complicated. Here’s the OED’s etymology
for Cretan, a. and n:
ad. L. Crētānus. The forms used in the various translations of the Bible are,
in Acts ii. 11 Cretes (Middle English (a 1400, ed. A.C. Paues), Geneva and
A.V.), in Titus i. 12 Cretayns (Tindale and Coverdale), Cretyans (Cranmer),
Cretians (Geneva and A.V.); Reims and Douay have Cretensians, and R.V.
Cretans in both places.
My Cassell’s Latin Dictionary (1887, last revision 1928) confirms Crētānus, and
adds a variety of other forms for a dweller in Crēta, Crētē, noun or adjective –
among them Crēs, Crēsius, Crēticus, but no *Crētiānus.
So it’s really just the Geneva Bible and the Authorized Version (aka the King
James version) that speak of Cretians. Here are the A.V. verses in question.
Cretes and Arabians, we do hear them speak in our tongues the wonderful
works of God. – Acts 2:11
One of themselves, even a prophet of their own, said, the Cretians are alway
liars, evil beasts, slow bellies. – Titus 1:12
1.36 anglicized Welsh place names 27
There’s also a metrical foot called a cretic (ˉ˘ˉ). Another fascinating fact: in
Modern Greek the word for ‘Cretan’ (Κρητικός) and the word for ‘critical’
(κριτικός) are homophones.
Anyhow, I think an olive grove ought to be Cretan.
1.36 anglicized Welsh place names
In Gwent, southeast Wales, in areas where the English language

displaced Welsh centuries ago, there are nevertheless plenty of small villages
with names that are obviously Welsh, both in appearance (their spelling follows
Welsh spelling conventions) and in meaning (in Welsh they mean something).
I wonder how they are in practice pronounced by their monoglot English-
speaking inhabitants.
Between Chepstow and Newport, for example, just across the border from
England, we find villages called Mynydd Bach, Gaerllwyd and Pen-y-cae-mawr
(‘small mountain’, ‘grey castle’, ‘top of the big field’). In standard Welsh they
would be ˈmənɪð ˈbaːx, ˈgəɪr ˈɬʊɪd, ˈpen ə kəɪ ˈmaʊr.
Many of these village names are listed in the recently published Dictionary
of the Place-names of Wales by Hywel Wyn Owen and Richard Morgan
(Gomer Press 2007), but without any systematic indication of pronunciation.
None of them are mentioned in the BBC Pronouncing Dictionary of British
Names (Oxford University Press 1990, second edition).
Let’s consider the elements of the place names one by one. As an English
name, though of Welsh origin, mynydd becomes Mynd mɪnd in the name
of the Shropshire hill The Long Mynd. Gaer is the mutated form of caer, which
gives us Car kə in Carmarthen/Caerfyrddin and Carnarvon/Caernarfon. Welsh
Llwyd becomes English Lloyd lɔɪd. Pen anglicizes with no difficulty; the
article -y- becomes English i in for example Pontypool/Pont-y-pŵl; and -cae-
gets reduced to kə in Gwaun-cae-Gurwen ˈgwaɪn kəˈgɜːwən.
Of course what I ought to be doing, instead of speculating, is going out there and
interviewing local people. I don’t have anyone from east Gwent here in London
that I can consult, but I do have someone from further west, a lady from near
Neath. She says she speaks no Welsh, and indeed remembers from her schooldays
that there was only one child in her class who spoke Welsh at home, which made
him something of an outsider. (This was before the days of Welsh–English
bilingual education.) When I asked her about the village of Blaengwrach, she
had no hesitation in pronouncing it as blaɪnˈgraːx. Similarly, she confidently
pronounces Llanelli as ɬəˈneɬi. So she clearly has the non-English consonants x
and ɬ in her system. She has the usual anglicization of Pontypool with -i-, but
pronounces Pontypridd as pɒntəˈpriːð, with Welsh-style -ə-.
I asked her about Mynydd Bach. Straight away: ˈmʌnɪð ˈbaːx. So members of
the non-Welsh-speaking majority in Wales can often nevertheless pronounce
Welsh names in a Welsh, as opposed to an anglicized, way.
Downloaded from http:/www.cambridge.org/core. New York University Libraries, on 20 Dec 2016 at 21:13:21, subject to the Cambridge Core terms of use, available at
http:/www.cambridge.org/core/terms. http://dx.doi.org/10.1017/CBO9781139871327.002
(Jack Windsor Lewis points out that although this lady and other non-Welsh-
speakers like her may ‘have the item ɬ in her system’, many actually pronounce
not exactly ɬ but perhaps rather ɬl or, worse, xl or θl, none of which would pass
muster in a phonetics exam.)
1.37 Jersey h
Jersey, in the Channel Islands, was in the news for a time because
of a child abuse scandal. The former children’s home then under investigation
was called Haut de la Garenne, in the parish of St Martin.
It is striking that radio and television newsreaders pronounced this name with
initial h, despite the fact that they must all be aware that in French the letter h at the
beginning of a word is silent. Indeed, all British newsreaders must certainly know
enough French to be aware that the French for ‘high’, haut, is pronounced o.
(Let’s not get involved in the question of so-called h aspiré.)
The BBC Pronunciation Unit tells me, ‘Our newsreaders are following our
advice by pronouncing the h-. This is the local pronunciation according to
staff we consulted at BBC Radio Jersey. We tend to check Channel Island place
names directly with people there because the level of anglicisation of French-
looking names, though usually high, can be a bit unpredictable.’
Although the 1990 BBC Pronouncing Dictionary of British Names does not
include any entry for Haut de la Garenne, its Channel Islands Appendix does list
fourteen French-looking names spelt with h-. Thirteen of them are pronounced
with h, including Hacquoil, Hautes Capelles and Houiellebecq ˈhuːlbek.
I don’t know anything about Jèrriais (French jersiais), the Norman French
dialect of Jersey, but presumably it has retained historical h despite the disap-
pearance of h from standard French.
Or is it to be explained as spelling pronunciation, like the h in Terre Haute,
Indiana?
1.38 Friern Barnet

In north London, in the London Borough of Barnet, is a place called
Friern Barnet.
The earliest recorded form of this name is Frerenbarnet (1274). The first
part of the name, Freren, is the Middle English (from French, from Latin) for ‘of
the brothers’, and refers to its early possession by the Knights of St John of
Jerusalem.
Not too far away there is another place, Monken Hadley. It is so called after
the monks of the monastery which was once there.
As Monken is to monk, so Friern is to friar. In Friern Barnet there is a
Friary Park.
1.40 Marcel Berlins and Sarkozy 29
Obviously, therefore, you would expect Friern to be pronounced ˈfraɪən, just

like friar. And that is indeed how many people say it.
Others, though, call it frɪən. Presumably this variant arose as a spelling
pronunciation, on the analogy of words like pier pɪə, fierce fɪəs and so on.
(But compare fiery with aɪə.)
Both forms are given in Graham Pointon’s BBC Pronouncing Dictionary
of British Names (Oxford University Press 1990, second edition). Both are
duly recorded in EPD and LPD. But the Oxford Dictionary of Pronunciation
for Current English (Oxford University Press 2001) has only ˈfraɪən, which
I think is OK prescriptively but not descriptively.
1.39 veterinary
My bedside radio comes on at 06:30. It is set to BBC Radio 4. So in
the early morning every Saturday I get a dose of Farming Today.
One day there was a discussion about vets (i.e. animal doctors) and the
problems that arise from the fact that working with companion animals (cats,
dogs, rabbits) is nowadays much more lucrative than working with farm animals.
This causes a shortage of vets prepared to work down on the farm.
Naturally, the word veterinary was used repeatedly. I was struck by the fact
that the only pronunciation I heard, whether from farmers, from vets themselves
or from their teachers at university, was ˈvetn̩ ri or the phonologically equivalent
ˈveʔn̩ ri, ˈvetənri. Only three syllables, only one r.
Dictionaries tend to show only spelling-based forms such as ˈvetərɪnəri, with
more consonants and more syllables.
In LPD I do give both types, though with priority to forms that have r near
the beginning as well as near the end. You have to infer ˈvetn̩ ri by unpacking the
rather complex ˈvet ən ər‿|i. (Perform this operation by ignoring the raised and
italic letters and the cutback line, and compressing the last two syllables into one.)
Actual usage is clear – at least as exemplified in Farming Today. Three
syllables only.
1.40 Marcel Berlins and Sarkozy
The columnist Marcel Berlins pronounces his name ˈbɜːlɪnz, and

would like everyone else to do so too. In an article in the Guardian he com-
plained about the BBC’s ‘widespread mangling of the [then] president
of France’s surname’. ‘The three syllables Sarkozy’, he continued, ‘should be
stressed pretty much equally’.
It is ridiculous to suggest that we should pronounce all three syllables of
Sarkozy with equal stress: we’re speaking English, a language in which no
non-compound three-syllable word has three stresses. (In compounds, perhaps
OK: deep-sea-blue.) I assume Berlins is really calling for final stress, which is
what with our English ears we tend to hear in French words: saʁkɔˈzi.
I expect American announcers pronounce it with final stress. But the BBC
is not American.
The name is originally Hungarian, and Hungarian has initial stress. So that
could be seen as justifying stress not on the last syllable but on the first. All the
same, only an out-of-touch pedant would demand retention of the Hungarian
stress.
In BrE we tend to give French words penultimate stress. That’s how we mostly
say cliché, cachet, idée (fixe), Calais, Orly and many others. It’s how many of us
would stress Chirac, and I think it’s fine to do it in Sarkozy.
Actually, in Hungarian the name is Sárközy, so that the middle vowel ‘ought’
to be ø, as if spelt eu in French. In English this Hungarian vowel would normally
map onto ɜː (the NURSE vowel), not əʊ (the GOAT vowel). So should we be
saying sɑːˈkɜːzi? I think not.
The Hungarian spelling also implies that the initial consonant should be not s
but ʃ, which is something that, like the French, we can happily ignore.
1.41 repertory
In an interview that I found on the internet, my eminent colleague and
old friend David Crystal is quoted as saying ‘I’ve been in an amateur repartee
company for many, many years. . .’
For someone so verbally dexterous and quick-witted, perhaps that explains
a lot. Repertory, repartee.
Actually, repertory is phonetically quite an interesting word. It’s OK for the
Americans, since they maintain a strong ɔː vowel in the -ory ending. But we Brits
weaken it to schwa, which leaves weak vowels in three successive syllables:
ˈrepətəri. As usual, the penultimate schwa is subject to possible disappearance
through compression, giving just ˈrepətri.
So perhaps it’s not surprising that it sometimes gets misheard.
1.42 Penwortham
Penwortham is a village on the outskirts of Preston, Lancashire.

Those who live there call it ˈpenwə(r)ðəm, with initial stress. (Preston is on
the edge of the shrinking rhotic area of Lancashire, so there may still be a few
people who pronounce it with -r-.)
Non-locals, particularly southerners, tend to assume that it’s penˈwɜːðəm
(or -ˈwɜːθəm), with penultimate stress; they’re wrong.
(I don’t know how the Australians pronounce the place with the same spelling
in South Australia.)
1.43 hello? 31
This is one of several such place names in the north of England. In Lancashire
just near where I lived as a boy is the mining village of Winstanley. We called
it ˈwɪnstənli, but those unfamiliar with the place, and indeed bearers of the
corresponding surname, almost all seem to say wɪnˈstænli. The BBC Pronouncing
Dictionary of British Names gives both. Etymologically, the name derives from
the OE proper name Wynnstān, modern Winston, plus lēah ‘wood, clearing’,
which explains its initial stress.
Just over the summit of the Pennines as you go east into Yorkshire lies
Todmorden. I’ve always known this place as ˈtɒdmədən, though according
to the BBC PDBN it can also be ˈtɒdmɔːdən. But not tɒdˈmɔːdən, which is
what parachuted-in TV reporters tend to go for.
Conversely, we all know about Newcastle (upon Tyne), which locals stress on
the -cas- but most other people on the New- (a stressing ‘firmly established in
national usage’, in the words of the BBC PDBN). But Newcastle(-under-Lyme)
in Staffordshire has initial stress, and so do various other Newcastles.
Just to keep us on our toes, however, near Todmorden there is Mytholmroyd.
That’s actually ˌmaɪðəmˈrɔɪd, with final stress.
I console myself with the thought that all of this must be good for the sales
of pronunciation dictionaries.
1.43 hello?
How do we spell and pronounce this greeting? You can find any
of three spellings: hello, hallo and hullo. If I am not mistaken, Americans
always spell it hello. (My Webster’s Collegiate also gives hollo, but perhaps
this is a different word.)
We can agree on its usual pronunciation: həˈləʊ. But instead of a schwa it
can sometimes in BrE have a strong vowel in the first syllable; and this strong
vowel may correspond to any of the spellings hello, hallo, hullo. In the first
syllable, as well as ə I think I’ve heard e, æ, ʌ and I also suspect I’ve heard the
stress on the first syllable, though maybe this was in stress-shift positions.
The way to determine whether or not there is a secondary stress on the first
syllable is to see whether stress shift is possible. And here there is a bit of a problem.
As far as I can tell, stress shift happens only if hello (however spelt) is
followed by the name of the person being greeted, i.e. a vocative. To trigger
stress shift, this vocative has to be accented.
The difficulty is that final vocatives are normally not accented.
Furthermore, as far as I can see the intonation pattern under stress shift has
to be fall plus rise.
\Hello, | /Mary.
I don’t think you can say:

*\Hello, Mary.
though of course you can say the neutral

Hel\lo, Mary.
(or the same thing with a rising or falling-rising nuclear tone).

This stress shift triggered by the anomalous accenting of a vocative is paral-
leled by the cry of despairing British tennis fans,
\Come on | /Tim!
where logic would lead you to expect the unremarkable

(ˈ)Come \on, Tim!
But in Hello, Mary the stress shift itself is anomalous, too, given that the usual
first vowel in hello is ə. The only explanation seems to be that we’re dealing
with intonational idioms.
1.44 Lucida
How do you pronounce the font name Lucida?

Actually, Lucida is not the name of a single font but rather of a family of fonts:
hence the names Lucida Sans Unicode, Lucida Grande, etc. They were designed
by Charles (‘Chuck’) Bigelow and Kris Holmes in 1985.
The question is whether we should say ˈluːsɪdə (or with lj-), just like lucid
plus a final schwa; or whether it should be luːˈsiːdə or even a great-vowel-shifted
luːˈsaɪdə or an Italianate ˈluːtʃɪdə.
As an Italian word, meaning ‘shining, lucid’ (f.), lucida is indeed ˈlutʃida.
Spanish has both lúcida and lucida, again feminine adjectives, with slightly
different meanings. The first is ˈluθiða, ˈlusiða and the second luˈθiða, luˈsiða.
However, we are talking here about English, not Italian or Spanish. And in
English we look to Latin.
English lucid is taken from the Latin adjective lūcĭdŭs. Since the penultimate i
is short, the regular Latin stress rule means that stress goes on the antepenulti-
mate, the lū-. Dropping the masculine ending -ŭs, as is usual, yields English
ˈl(j)uːsɪd. The feminine form of the adjective is lūcĭdă, with the same stress.
Keeping the feminine ending, we would get ˈl(j)uːsɪdə.
There are no other English words ending -ida that might provide an analogy,
except perhaps in a learned singular of such zoological taxonomic terms as
Psittacidae, Vespidae, Ovidae, all with antepenultimate stress.
The penultimate-stressed armada and pagoda do not provide models. The first
comes via Spanish from the Latin feminine past participle armāta, with a long
penultimate vowel and therefore penultimate stress. The second is from Portu-
guese, ‘probably ultimately from Persian butkada idol temple’ (Concise Oxford
Dictionary) and so not Latin-derived.
1.45 phonetic numerals 33
Charles Bigelow tells me that he and Kris Holmes (the designers of the Lucida
typeface) pronounce it ˈluːsɪdə, just like lucid plus a final schwa. ‘Frankly, we
don’t object to any of the pronunciations. We’re just glad to hear that people are
using it.’
He adds, ‘Because Lucida was designed to optimize legibility in early laser
printing and computer screen display, we wanted to give it a name that could
suggest it was made of light and was clear despite the low resolutions of screens
and printers at that time, but we didn’t want the name to be a mere description.
“Lucidus” or “Lucido” might have been fine, but the success of Futura, Optima
and Helvetica led us to consider a name ending in the Latin feminine singular.
Maybe it was just superstition that a name could influence popularity.’
So it’s ˈluːsɪdə.
1.45 phonetic numerals
Many people are familiar with the NATO/ICAO ‘phonetic alphabet’

(alfa, bravo, charlie, delta, echo, foxtrot. . .). In LPD I list these ‘communications
code names’ under each alphabetic letter.
In the NATO/ICAO scheme it is not only the letters that are supposed to
be pronounced in specified ways, but also the numbers. Their pronunciation is
indicated by respellings, hopefully to avoid any confusion. The two best-known
special pronunciations are fife for ‘five’ and niner for ‘nine’ – a sensible way to
avoid confusion between the two over a crackly radio channel.
But one confusing respelling is fower for ‘four’. In discussion of this recom-
mended pronunciation people tend to assume that it’s meant to rhyme with
flower (perhaps because it’s immediately preceded by tree). Probably, though,
this respelling originated in an attempt to train people to say something like ˈfəʊə
rather than the monosyllabic, non-rhotic fɔː. I very much doubt whether it was
intended that radio operators should actually say ˈfaʊə.
The question, then, is whether fower is meant to rhyme with flower, tower
or rather with grower, mower. Fower seems also to be the Scots dialect spelling
for four.
An instructional CD-ROM I have (the excellent ‘VFR RT Communications’,
produced by Oxford Aviation Training) pronounces the word ˈfaʊə when intro-
ducing the pronunciations in isolation – but then goes on to pronounce four in
the usual way [i.e. fɔː] in examples of actual transmissions. Some speakers on
the CD do make more of an effort to say ˈfəʊə during transmissions, it’s true;
but none goes so far as to say ˈfaʊə. All this confusion could have been avoided
if only the NATO people had used IPA to describe what they meant! But then
pilots aren’t trained to read IPA. . .
Or they might even have asked a phonetician to advise on making respellings
unambiguous.
1.46 Llantwit Major

A German teacher of EFL asked me, on behalf of one of his students,
how to pronounce the name of Llantwit Major, a small town in the Vale of
Glamorgan.
This is an English-speaking area, and Llantwit Major is actually its English
name: in Welsh it is called something else. Nevertheless, the name looks Welsh,
with the characteristic initial Llan-.
So I was interested recently to see on TV a feature about the place. It was
striking that the locals all pronounced the name with an ordinary voiced approx-
imant initial l. The visiting reporter, on the other hand, made a great effort to
produce a Welsh ll, ɬ. He shouldn’t have bothered. It’s ˈlæntwɪt ˈmeɪdʒə, as
confirmed by the BBC Pronouncing Dictionary of British Names.
The Welsh name is Llanilltud Fawr ɬaˈnɪɬtɪd ˈvawr. Fawr, the mutated
form of mawr, just means ‘great, big’, and corresponds to the Latin Major
of the English name. According to the Dictionary of the Place-names of
Wales, both the English Llantwit and the Welsh Llanilltud have the same
Welsh origin, llan- ‘church’ plus Illtud, the name of a monk who reputedly
established a monastery there in the sixth century. English -twit, according to
the authors, is traceable to an early variant Illtwyd from an Irish (!) genitive
Illtuaith.
There’s a BBC web page where you can hear audio files of the pronunciation
of various Welsh place names. Trying it out, I was struck by the way the speaker
pronounces the Welsh ll (e.g. in Llandeilo, Llangwm, Llanddewi Brefi, Benllech).
Instead of the standard voiceless alveolar lateral fricative ɬ, he uses a voiceless
palatal fricative ç.
I’m not sure whether this delateralization counts as a speech defect. I have
come across it before, in the speech of a Welsh-speaking undergraduate we
had at UCL a few years ago. But it is certainly not the mainstream pronunciation,
and I do not think it ought to be given to the general public as a model.
1.47 apostasy
The egregious Archbishop Akinola of Nigeria was reported as

wanting to save the Anglican Church from apostate leaders. I have not heard
him personally say apostate aloud (and he now apparently denies that he said it),
but when he was reported on BBC radio the reporter pronounced it as ˈæpəsteɪt.
However, the only stressing of apostate given in dictionaries is əˈpɒsteɪt,
with penultimate stress.
Historians know the word as the appellation of Julian the Apostate, Roman
Emperor 355–63 AD. In the modern world, ‘apostate’ is a favourite word not
only of intolerant evangelical Christians but also of Islamic jihadist extremists
denouncing their political opponents.
1.48 pwn 35
Why, though, is it standardly pronounced aˈpostate, with the stress on the

penultimate? Why doesn’t it follow the antepenultimate pattern of other trisyl-
lables in -ate such as ˈdelegate, ˈvertebrate, ˈreprobate? I am not sure what
the historical answer is, but it must be something to do with the fact that unlike them
apostate does not contain the Latin participial ending -āt(us), in which the vowel is
long. It comes from Greek, and consists of a prefix apo- and a stem -stăt- with a short
vowel. Literally, it means one standing (stat) away (apo) from something. In
classical Greek ἀποστάτης apostatēs meant a runaway slave or a deserter.
Presumably this Greek origin is also the explanation for the unexpected
spelling of the corresponding abstract noun apostasy əˈpɒstəsi (Greek ἀποστᾰσία
apostasia; compare stasis).
The same applies to ecstasy, fantasy and idiosyncrasy, which are also Greek-
derived. Confusingly, it doesn’t apply to Greek-derived democracy, theocracy
and other -cracy words, which have -κρᾰτία -kratia in Greek and follow
the familiar spelling pattern of Latin-derived accuracy, celibacy, delicacy,
candidacy, profligacy, etc.
1.48 pwn
What is to pwn, and what is a pwn? According to Wikipedia, this is

a leet slang term (see http://en.wikipedia.org/wiki/Leet), originating as a mistyp-
ing of own and meaning to own in the sense of being able to dominate or
humiliate a rival. It is used in the internet gaming culture ‘to taunt an opponent
who has just been soundly defeated’ (charming!).
Facebook users who have installed SuperPoke can give someone else a
virtual pwn.
So how do you pronounce it? Wikipedia offers a range of possible pronunci-
ations ranging from poʊn through puːn to more imaginative suggestions such as
pəˈoʊn and pwəʔˈn̩ .
The past participle is apparently sometimes pwn3d.
My first feeling on encountering this word pwn, though, was that it should be
pronounced pʊn, which is how it would be if it were Welsh. (Or if spelt pŵn,
then puːn.) According to my Welsh dictionary there is indeed a Welsh word with
this spelling and pronounced pʊn, though it means ‘pack, burden’. For compli-
cated but predictable reasons of Welsh phonology, its plural is pynnau ˈpənai.
I’ll take that as a virtual pwn.
My friend Tim Owen tells me:
I am a frequent user of message boards and forums. Owned and pwned
(alternatively ownt) have been around fairly commonly since I first started
surfing regularly in 2003. Every time I use it and have heard it (which runs
into hundreds of times), it’s consistently been pronounced as the word own
prefixed with a p.
So for BrE that’s pəʊn, then.

I see the word has now spawned a derivative pwnage.
1.49 Entwistle
I was struck by the fact that two presenters on BBC television
pronounced the surname Entwistle as ˈenʔwɪsl̩ . Others, like me, pronounce it
ˈentwɪsl̩ . Why the difference? Why my surprise? It᾿s not because I wouldn’t
glottal a t between n and w: I am perfectly happy to say slantwise as ˈslɑːnʔwaɪz.
It’s all to do with syllabification. The medial consonant sequence ntw is one of
the few such sequences that allow two phonotactically well-formed possible
places for the syllable boundary, nt.w or n.tw.
Like many English surnames, this one is derived from a place name. Entwistle
is a village between Bolton and Blackburn in Lancashire. There’s another place
in Lancashire called Oswaldtwistle, and a Derbyshire village called Tintwistle.
So it’s natural for anyone who knows of these places to see -twistle as a separate
morpheme.
According to my syllabification rules, dubious consonants go with the more
strongly stressed of the flanking syllables, unless there is a major morpheme
boundary blocking this. In the first case we would have ˈent.wɪs.l, where the t is
in an environment that favours glottalling. In the second case we would have
ˈen#twɪs.l, where t cannot be glottalled because it is syllable-initial.
The announcer’s pronunciation of Entwistle is therefore what would be pre-
dicted in the absence of a morpheme boundary. Mine, on the other hand, is what
is to be expected if there is a morpheme boundary between En- and -twistle.
I looked up the etymology of Entwistle in the Oxford Names Companion. Its
origin is OE henna, ‘water hen’ (or perhaps ened, ‘duck’) plus OE twisla ‘tongue
of land in a river fork’. So the historical morpheme boundary corresponds to my
contemporary perception. That explains my non-use of a glottal stop in this word
(and also my non-use of pre-fortis clipping for the first vowel plus nasal). That’s
why it’s ˈen.twɪs.l̩ .
1.50 Chinese into English

A correspondent wrote to complain about the ‘mangling’ of Chinese
names by English speakers, particularly that of the tennis player Zheng Jie.
‘Why’, he asked, do commentators ‘insist on saying something like ʒʌŋ ʒiː
when something like dʒəŋ dʒeə is quite close to the proper pronunciation and
perfectly easy in English?’
In Chinese her name is written 郑洁, Pinyin Zhèng Jié, pronounced 4tʂəŋ 2tɕi̯ ɛ
(or you may find it easier to think of the distinctively unaspirated obstruents
as devoiced dʐ, dʑ). Zheng is the family name, Jie her personal name.
A reasonably thorough anglicization would be ˈdʒʌŋ ˈdʒɪə, or as a halfway
house perhaps ˈdʒəŋ ˈdʒiːe. (Those who have a monophthongal SQUARE vowel
could pronounce the personal name dʒeə, i.e. dʒɛː.) So why do commentators
not say something like that?
1.51 inter(n)ment 37
I’ve commented from time to time on the problem of dʒ in foreign names.

We are so brainwashed by the influence of French that we find it difficult to
believe that any foreign language has an affricate dʒ rather than a fricative ʒ.
Hence pronunciations such tɑːʒ in Taj Mahal instead of tɑːdʒ. Hence ʒ instead
of dʒ in Zheng.
In the case of Chinese zh- this tendency is exacerbated by the prevalence of
zh as a respelling for ʒ and as a romanization of Cyrillic ж (Zhukov, Zhivago).
And there’s always Beijing, which ought to have dʒ rather than ʒ in the
middle. How difficult is it to use a perfectly ordinary sequence of English
phonemes beɪdʒɪŋ in saying a name? The BBC Pronunciation Unit gives very
clear guidance on this. But it’s one thing to tell announcers and commentators
what they ought to do, another thing to get them to actually do it.
1.51 inter(n)ment
One of my social duties from time to time is to create service

sheets for religious services, including in particular funerals. Many of the
deceased are buried rather than cremated, and in a big city like London that
means that after the funeral service is held in church the actual burial takes
place in a cemetery located possibly several miles away. So the service sheet has
to tell you where this will be.
Rather than ‘burial’, people tend to prefer to use the posh word ‘interment’.
The trouble is that many of them think it is spelt internment and pronounced
correspondingly. In the service sheets I make myself I spell the word correctly,
of course. But at several funerals I have attended the service sheet has been
created by someone else, and they have got it wrong.
As we literate people know, internment ɪnˈtɜːnmənt is imprisonment
without trial. You can see how readily it might be confused with interment
ɪnˈtɜːmənt. Think of the common pronunciation of government with a single
nasal, ˈɡʌvəmənt.
A further stage in the confusion is pronouncing this noun as ˈɪntəmənt. It does
admittedly look confusingly like words with the prefix inter-. If interval is
ˈɪntəvl̩ , you might reasonably expect interment to be ˈɪntəmənt.
The verb to inter, stressed on the last syllable, has been in English since
the fourteenth century, and its derivative noun interment likewise. Just think,
instead, of
The evil that men do lives after them;
the good is oft interred with their bones.
Shakespeare might have written

The evil that men do lives after them;
the good is often buried with their bones.
But he didn’t.
2 English phonetics: theory
and practice
Including discussion of topics such as compression
and weakening, not always well described in
standard textbooks
2.1 assimilation
A Japanese correspondent asked about the pronunciation of the word

classroom, having heard a visiting lecturer from the UK pronounce the word it
with ʃ instead of s before the r. This pronunciation is not recorded in dictionaries.
He asked whether it is a common pronunciation in RP in particular and in BrE
in general.
I can’t find any discussion of this possibility in the literature, although I think
it is by no means unknown. I do mention the possibility in one of my lecture
handouts, where I list various environments in which the assimilation of /s/ to [ʃ]
may occur.
DEALVEOLAR FRICATIVE ASSIMILATION
/s, z/ ! [ʃ, ʒ] before /ʃ, ʒ/, and perhaps before /j, r, tʃ, dʒ/ this shop, this student
You can get this wherever s and r abut: across word boundaries as in this reason,
between the parts of a compound as in the classroom my correspondent heard,
or indeed within a word as in the compressed version of nursery ˈnɜːs(ə)ri,
ˈnɜːʃri or grocery.
There can be assimilation of z to ʒ in the same environments, e.g. miserable
ˈmɪʒrəbl̩ .
I have no statistics about how widespread this assimilation is. I don’t do it
myself, and it’s not a mainstream pronunciation.
I suspect that the assimilation product may in some cases not be identical with
an ordinary ʃ, ʒ, but rather a kind of retroflex ʂ, ʐ, anticipating the place of
articulation of the r.
Rather more widespread, and on the increase, is s ! ʃ before tr, as in strong.
Many of my native-English students have this at least as an optional variant.
The str ! ʃtr assimilation joins the stj ! stʃ ! ʃtʃ assimilation as phonos-
tylistic phenomena we shall have to think of telling our EFL students about. I call
them ‘s-affricate assimilation’.
2.2 internal intrusive r
At the cardiac support group organized by my local hospital

I collected a new example of word-internal intrusive r: meta-analysis.
38
2.3 royal toil 39
The speaker, a general practitioner, used this word several times, and each
time pronounced it as ˌmetərəˈnæləsɪs.
A meta-analysis, in statistics, is a large study that combines the results
of several smaller studies.
Actually, as the classicist in me protests, this word is not well-formed from the
point of view of Greek. In classical Greek, prefixes ending in a vowel, such as
meta- (but also para-, cata-, ana-, epi-, etc.), lose the vowel if the following stem
has an initial vowel (or h plus a vowel). We see this in metempsychosis, category
and metonymy, and also in method and cathode. So the word ought to be
metanalysis.
Indeed, the second edition of the OED does not include meta-analysis, but
only metanalysis. Meta-analysis was added in the third edition (2001), with the
British pronunciation correctly shown as including an optional r. Interestingly,
the third edition records that unhyphenated metanalysis, and with it the verb to
metanalyse, belongs not to statistics but to linguistics. The first citation for the verb
is from Randolph Quirk, in his 1962 book The Use of English (Longman), where
he wrote that ‘when the French word crevice . . . was introduced into Middle
English, its connexion with “sea food” caused people to metanalyse the final syllable
as -fish (“crayfish”)’. The first citation for metanalysis is from Jespersen in 1914.
Somehow it’s not surprising that the linguists get the Greek right, while the
statisticians, medical or otherwise, don’t.
2.3 royal toil

Does mile rhyme perfectly with trial? Does owl rhyme with vowel?
Does oil rhyme with royal? I suspect that for most English people the honest
answer to all three questions is ‘yes’. (I’m prepared to concede that it is not so for
everybody, and in particular not for Geordies, as well as not for the Welsh and
the Scottish.)
This is part of the more general collapse of the contrast between long vowel or
diphthong plus l and long vowel or diphthong plus əl.
You’d never guess, though, from the entries in pronouncing dictionaries
(let alone general dictionaries). Yet Gimson, when I was a student of his in the
early 1960s, was already then pointing out that the historical distinction between
the end of mile maɪl and the end of trial traɪəl was no longer made in practice.
Owl and vowel, oil and royal are now usually pronounced so as to rhyme exactly.
In LPD I apply a complicated convention – perhaps too complicated – to try
and allow for the rhymes to be identical, while not insisting that they must be.
The raised schwa (denoting an optionally inserted sound) in the transcriptions
of mile maɪəl and oil ɔɪəl allows for the ‘breaking’ that leads some speakers
to insert a schwa glide between the vowel and the l. The slur mark (undertie)
in the transcriptions of trial ˈtraɪ‿əl and royal ˈrɔɪ‿əl allows for possible
‘compression’, i.e. the squeezing of the two syllables into one. If mile gets
40 english phonetics: theory and practice
a potential schwa and trial gets compressed, they end up rhyming perfectly.
Similarly for owl and vowel.
LPD also allows for the possibility of ‘smoothing’, i.e. the loss of the ɪ element
in trial and the ʊ element in vowel. That’s how people like me can end up with
trial and vowel as near-rhyming monosyllables traːɫ, va̠ ːɫ.
2.4 wronger
I heard a TV presenter say ‘You’re placing undue reliance on a price
of oil which is already wrong and which may well end up wronger.’ He
pronounced the last word ˈrɒŋə.
It is rare to catch sight of this comparative in the wild. Wrong is the only
monosyllabic adjective ending in ŋ other than long, strong, young. The compara-
tive and superlative forms of these latter three are irregular inasmuch as they are
pronounced with -ɡ-: ˈlɒŋɡə, ˈjʌŋɡɪst, etc. Is this because of some minor rule
concerning the pronunciation of comparatives and superlatives of all adjectives
ending in ŋ, or because long, strong, young are lexical exceptions? Wrong provides
a possible test case to answer this question. If it too has -ɡ-, then there seems to be
some general rule in operation. But if it doesn’t, the other three are exceptions.
The difficulty is that wronger is rarely encountered, because wrong is not a
readily gradable adjective: things are either right or wrong, rather than having
degrees of wrongness. But as the instance I heard shows, this is not always
the case. The presenter’s pronunciation of wronger, which is also my own,
demonstrates that – for him and me at least – wrong follows the ordinary rule
for comparatives, which means that long, strong and young must be exceptions.
2.5 compression
By the term ‘compression’ I refer to variability in the number of

syllables in a word or phrase. A compressed version of a word or phrase has one
syllable fewer than the uncompressed version. In LPD I show sites of possible
compression by using the special mark [‿].
Many words have two pronunciations, differing only in a segment that in
one version is syllabic, e.g. i, u, n̩ , l̩ and in the other non-syllabic, e.g. j, w, n, l.
This variability is exploited by versifiers, who can require a particular sequence
to be pronounced either uncompressed or compressed, to fit the metrical require-
ments of the poem, hymn or song in which it appears.
The crucial point of compression is that the number of syllables is reduced
by one when a syllable beginning with a vowel follows. In English (treating
potential syllabic consonants as phonetic variants of schwa plus the consonant):
cycle ˈsaɪkl̩ – cyclist ˈsaɪklɪst, usually not ˈsaɪkl̩ ɪst;
sparkle ˈspɑːkl̩ – sparkling ˈspɑːklɪŋ, usually not ˈspɑːkəlɪŋ, ˈspɑːkl̩ ɪŋ.
2.6 compression in hymnody 41
In English this process is usually blocked if the following vowel is strong. There
is no possibility of compression in the following (though there is the possibility,
indeed the likelihood, of a syllabic consonant):
cattle ˈkætl̩ – catalogue ˈkætəlɒg or ˈkætl̩ ɒɡ, but not ˈkætlɒg;
summer ˈsʌmə – summarize ˈsʌməraɪz or ˈsʌmr̩ aɪz, but not ˈsʌmraɪz.
There are other cases, too, where it fails. Vitally ˈvaɪtəli or ˈvaɪtl̩ i does not rhyme
with nightly ˈnaɪtli.
Jack Windsor Lewis has commented on ‘a fast increasing minority ten-
dency for General British speakers to favour the use of schwa plus (unsyllabic)
consonant where previously syllabic consonants were the norm, e.g. in cotton,
garden, bottle and struggle, and even increasingly in such items as assembly,
doubly, gambling, cackling, etc., for which it is doubtful that they ever previ-
ously contained a syllabic consonant’. Usually there has to be a related word
with a syllabic consonant to trigger this, so that e.g. duckling, madly, ugly, for
example, are not usually affected but words such as buckler, burglar, butler,
inkling, spindly, stickler, etc. may well (he thinks) soon be increasingly heard
with an anaptyctic schwa by some British speakers.
I think Jack’s observations are correct. I do quite often hear three-syllable
schwa-containing pronunciations of words such as threatening, doubling,
gathering (though I don’t think I’d ever use such a pronunciation myself),
and even of simpler. I’ve never yet heard a trisyllabic version of gently, though
it’s not uncommon in subtly.
2.6 compression in hymnody
I’ve been musing about the phonetics of hymns. One thing I have
noticed is the way hymnodists exploit compression.
Thus the word victorious is pronounced uncompressed, as four syllables, in
the dactylic rhythm of the British national anthem, and glorious likewise as three
syllables:
Send her victorious (ˈtum-ti-ti ˈtum-ti-ti)
Happy and glorious
Long to reign over us
God save the Queen!
But they are to be sung compressed in a well-known hymn by Charles Wesley:

Ye servants of God, (ti-ˈtum-ti ti-ˈtum)
Your master proclaim,
And publish abroad
His wonderful name;
The name all-victorious (ti-ˈtum-ti ti-ˈtum-ti)
Of Jesus extol
His kingdom is glorious

And rules over all.
Another hymn in which glorious is to be sung compressed (¼ as two syllables) is:

Glorious things of thee are spoken, (ˈtum-ti ˈtum-ti. . .)
Zion, city of our God;
He, whose word cannot be broken,
Formed thee for his own abode. . .
Compression can also operate across a word boundary. The relevant vowels,
weak i and u, arise from the weakening of function words such as me, we, she,
the and you, to. Compression means that i, u lose syllabicity, becoming j, w (or if
you prefer i̯ , u̯ ).
Both within a word and across a word boundary, compression can operate
only if the following vowel is weak – one of i, u, ə, ɪ. Thus, within a word we can
have compression in victorious vɪkˈtɔːri‿əs ! vɪkˈtɔːrjəs, where the following
vowel is the weak schwa, but not for example in Victoriana vɪkˌtɔːriˈɑːnə,
where it is the strong ɑː.
Eighteenth-century hymnodists follow this principle pretty faithfully. Look
for example at the last verse of Charles Wesley’s And can it be. The hymn is in
8.8.8.8.8.8 meter. Each line is an iambic tetrameter: it consists of four feet, the
first syllable of each foot being unstressed and the second stressed (except that
the first foot may be trochaic, stressed-unstressed, instead of iambic – producing
an ‘initial choriamb’):
No condemnation now I dread
Jesus, and all in him, is mine!
Alive in him, my living head,
And clothed in righteousness divine,
Bold I approach the eternal throne,
And claim the crown, through Christ, my own.
In the penultimate line the phrase the eternal throne is to be scanned as four
syllables, ðjɪˈtɜːnəl ˈθrəʊn, as the undergoes compression before the weak initial
vowel of eternal.
Elsewhere in Wesley’s hymns we find the incarnate deity as scanned six syl-
lables, with the undergoing compression before the weak initial vowel of incarnate.
He also scans Immanuel as trisyllabic ɪˈmænjwəl, with the u compressed to w
before the schwa (compare genuine optionally pronounced as two syllables,
ˈdʒenjwɪn) and he can make to appear two syllables, with to (prevocalic weak
form tu) undergoing compression to tw before the initial weak schwa of appear.
However, Wesley would never (for example) have scanned the only or to
answer as disyllables. Compression is allowed only before an unstressed
vowel. Normally, this vowel must also be weak, as set out above. Exception-
ally, though, the hymnodists allow compression before an unstressed
but strong vowel, as in the Christmas hymn Hark! The herald angels sing,
2.7 GOAT compression 43
where With the angelic host proclaim must be scanned as ðjænˈdʒelɪk.

Or perhaps we are intended to sing this with elision of the vowel of the
article, th’angelic ðænˈdʒelɪk. But if that were the case I think Wesley would
have spelt it in that way.
2.7 GOAT compression
OK, this isn’t about the maltreatment of smallholding livestock.

No, it’s about the possible weakening and subsequent compression of the vowel
we use in the GOAT lexical set (see my Accents of English, pp. 146–7). In RP
and similar accents this is the vowel whose strong version we transcribe əʊ.
It used to have a weak version that Daniel Jones transcribed o in the first twelve
editions of EPD, as in November noˈvembə. When Gimson took over the
editorship, he replaced this (quite rightly) with plain schwa, thus nəˈvembə,
and nowadays we reckon that in RP əʊ indeed normally weakens to ə: micro-
cosm, biosphere, proˈtest, Yellowstone, possibly also window-sill, tomorrow
morning.
But this doesn’t apply prevocalically. How can we weaken the unstressed first
vowel of, say, oasis? The answer is, we can’t.
However, there are one or two words in which the GOAT vowel appears
to weaken prevocalically to u. A good example is tomorrow evening.
Weakening to u produces a candidate for compression, since this vowel can
be compressed to w before a weak vowel, so losing its syllabicity. One word
in which this can happen is following, where ˈfɒləʊɪŋ, ˈfɒluɪŋ can be com-
pressed to ˈfɒlwɪŋ. This also applies in the archaic (thou) followest ˈfɒlwɪst,
as proved by verse scansion. As usual, hymns show this well. O Love that
wilt not let me go was written as late as 1882, and has the line O light that
followest all my way, in which followest must, if it is to scan, be pronounced
ˈfɒlwɪst.
2.8 analogical decompression
‘Eng-ger-land! Eng-ger-land! Eng-ger-land!’ chant the patriotic

football crowds, cheering on the English team. But why do they feel the need
to add an extra syllable? Why do they change change ˈɪŋ.glənd into ˈɪŋ.gə.lənd?
There’s nothing phonotactically difficult for English speakers in syllable-initial
gl. Why break up the cluster with an epenthetic schwa? Why the svarabhakti?
(Some speakers don’t have a g in England, but say simply ˈɪŋ.lənd. They have
even less excuse.)
There’s one other instance of this phenomenon in English that comes imme-
diately to mind: the widespread pronunciation of athlete, athletics as ˈæθəliːt,
ˌæθəˈletɪks instead of the standard ˈæθliːt, æθˈletɪks.
I have also heard template pronounced as ˈtempəleɪt instead of the usual
ˈtempleɪt. (A hundred years ago it was a ˈtemplɪt. But nowadays the spelling
pronunciation prevails.)
The explanation must be that the phonetic Cþl sequences in these words could
also arise from a process of compression (schwa deletion), as when tingling
(from the verb to tingle) is compressed from ˈtɪŋgəlɪŋ to ˈtɪŋglɪŋ, ˈkæθərɪn
Catherine to ˈkæθrɪn or ˈdʌbəlɪt double it to ˈdʌblɪt. (I ignore here the possibility
of syllabic l̩ from əl, which is irrelevant to the argument.) Apply this process
in reverse, and we get the svarabhakti forms. (But it doesn’t explain why it applies
only to a very few lexical items.)
2.10 oh, oh! 45
This, together with morphological regularization, presumably explains the

occasional ˈwɪntəri for wintry or riˈmembərəns for remembrance.
Eng-ger-land fans, don’t let it spread any further. Please! Or as we say
nowadays, puh-leeze!
2.9 compression anomalies

English seems to have a rule barring the compression of i, u to j, w
when followed by a strong vowel, though it is fine before a weak vowel.
Hence we can compress radiant to two syllables, ˈreɪdiənt ! ˈreɪdjənt but can’t
really do the same with radiate ˈreɪdieɪt. We can reduce mutual but not bivouac.
At least, that’s the rule I’ve been teaching people for many years.
It also explains why sierra, fiesta, Bianca, Kyoto, which are all disyllables in
the original language, get expanded to three syllables in English. Spanish ˈsjerra
becomes English siˈerə, Japanese ˈkjooto becomes kiˈəʊtəʊ.
Given this, you would expect that French borrowings such as moiré, joie
(de vivre) would be subject to the same pressure. But for some reason we’re happy
with ˈmwɑːreɪ and ʒwɑː. We don’t feel any need to expand them to *muˈɑːreɪ
and ʒuˈɑː – even though English has no native words with mw- or ʒw-. Funny.
2.10 oh, oh!
In the ‘quantitative-qualitative’ English transcription system intro-

duced by Gimson in the sixties, which with minor adjustments remains our standard
transcription today, there is just one change I sometimes wish Gim had not made.
It concerns the symbol he introduced for the GOAT vowel, namely əʊ.
You can see why he wanted to change the symbol that had been standard until
then, namely oʊ. In mainstream RP the vowel no longer had a back rounded
starting point. Rather, the starting point had become central and unrounded.
It was this change that he wanted to symbolize, but there are certain drawbacks
to his choice of symbol.
The letter ə had hitherto been used only for the weak vowel and the
less prominent element of centring diphthongs. Using it for the more
prominent part of a stressable diphthong confused matters, and still
misleads beginners.
People tend to mix up the symbols əʊ (GOAT) and aʊ (MOUTH), both
in reading and in writing, because they look too similar to one another.
Both of these problems would have been avoided if he had chosen to write the
GOAT vowel as ɜʊ. This would also have appropriately signalled that the first
element in GOAT is more or less identical in quality with the vowel in NURSE.
Even better, arguably, would have been to stick with oʊ. This symbol would
have required some flexibility in interpretation, since the quality of the starting
point is indeed nowadays some way from o (cardinal 7). (But we go on writing
the STRUT vowel as ʌ, even though its modern quality is some considerable way
from cardinal 14 – see 2.25 below.)
If we had kept oʊ as the RP symbol, we would have been able to use the same
GOAT symbol in both of the accents taught to EFL learners, RP and General
American. As it is, the use of different symbols implies a greater difference
between BrE and AmE than really exists.
In my own teaching and authorship I have loyally continued to use Gimson’s
system, including əʊ, because I appreciate the importance of having a standard
system that all reference books agree in using. Except for works edited by
Clive Upton, that’s what we have, and I’m glad that we do. And even Upton
uses the əʊ symbol.
2.11 stressing schwa
The discussion in 2.10 above raises the further question: can ə on

its own (i.e. not as part of a diphthong) ever be stressed?
In a universal framework, yes of course it can. There are a fair number of languages
around the world in which stressed schwa is attested, starting in Europe with Welsh,
Romanian and Bulgarian. The Welsh for ‘mountain’ is mynydd ˈmənɪð; the
Romanian for ‘apple tree’ is măr mər; the Bulgarian for ‘I am’ is съм səm. (No
prizes for seeing the Indo-European connections here: Latin mont-, mal-, sum.)
But what about English? You do occasionally get a phonetically stressed
schwa in English through the restressing of a weak form, e.g. (be)cause kəz,
the adverb just dʒəst and going to, gonna gənə. But lexically there are no
stressed schwas in English.
Unless. . . like some authors we use ə for the vowel in luck, cup, done, etc.
(¼ my lexical set STRUT). Many American writers do this, notably Trager and
2.12 Latin stress in English 47
Smith and their disciples, who claim that the STRUT vowel and the commA
vowel belong to the same phoneme, being the stressed and unstressed allophones
respectively. If they are co-allophones of the same phoneme, they should be
written with the same symbol.
But that is a different issue.
2.12 Latin stress in English
The stress rules of Latin still have a kind of ghostly presence in

modern English, as Chomsky and Halle showed in their Sound Pattern of English.
In Latin a noun or adjective, if it has three or more syllables, is stressed on the
antepenultimate if the penultimate vowel is short and followed by at most one
consonant (ˈfacĭlĭs); otherwise it is stressed on the penultimate (impeˈrātor,
treˈmendus). We have taken this principle into English and not only apply it to
words derived from Latin but extend it to words from Greek and sometimes other
languages. So we have antepenultimate stress in ˈdeficit, ˈmiracle, ˈstamina,
ˈmemory, ˈJupiter (Latin) and aˈnemone, geˈography, aˈnatomy (Greek). All of these
meet the criterion of having a short penultimate vowel followed by a single conson-
ant sound. But we get penultimate stress in creˈator, Ocˈtober, bronˈchitis, aˈroma,
with their long penultimate vowels, and equally in treˈmendous, Deˈcember, aˈorta,
with more than one (historical) consonant between the penultimate and final vowels.
That is the etymological reason that we stress poˈlygamous (Greek -γαμ- -gam-,
with short a) differently from Polyˈphemus (-φημ- -phēm-, with long e) and poly
ˈmorphous (-μορφ- -morph-, with two consonants after the vowel).
A nice pair is Diˈogĕnes vs. Dioˈmēdes.
Faced with an unfamiliar word ending -VCV(C), we crucially need to know
whether the penultimate V is long or short. If it is long, it will be stressed; if it is
short, it won’t be, and indeed will undergo vowel reduction.
Nearly everybody pronounces diploma with penultimate stress, because the o
is long, dɪˈpləʊmə. (It comes from Latin diplōma, Greek δίπλωμα.) But I knew
one person (now deceased) who called it a ˈdɪpləmə, perhaps under the influence
of diplomat ˈdɪpləmæt (which originated as a French back-formation from
diplomatique, and does not follow the Latin stress rule).
There’s a kind of dinosaur called a diplodocus. I’ve always called it

a dɪˈplɒdəkəs, which may or may not reflect the fact that I know that its Greek
etymon consists of διπλό- dipló- ‘double’ plus δοκός dokós (short o) ‘beam’.
I was initially taken aback to find that one of my colleagues called it a ˌdɪplə
ˈdəʊkəs. But she’s not the only one. Our ordinary spelling, of course, doesn’t
indicate vowel length in -VCV(C) words. It’s one of the biggest shortcomings
of our spelling system.
And then there is the word diaspora. It has an etymologically short penulti-
mate o (Greek διασπορά diasporá, ‘sowing around, scattering’), and a corres-
ponding traditional English pronunciation daɪˈæspərə. But there are also people
who stress it ˌdaɪəˈspɔːrə. The spelling doesn’t tell you whether the o is long
or short: and that’s the factor that determines the stressing. Because of the
Latin stress rule.
The news we heard recently about the human remains identified as those of
Richard III meant that several newscasters and commentators made use of the
word skeletal. Those I heard on BBC Radio 4 all pronounced it with penultimate
stress, as skɪˈliːtl̩ . There is, though, an alternative pronunciation, with initial
stress, ˈskelɪtl̩ .
So this is a word like palatal, in which uncertainty about the quantity of the
penultimate vowel leads to rival pronunciations, one having the antepenulti-
mate stressing and penultimate vowel reduction predicted if the penultimate is
short (ˈpalətal, ˈskelɪtal), the other having the penultimate stressing to be
expected if that vowel is long (paˈleɪtal, skeˈliːtal). In the case of palatal,
phoneticians and linguists have settled for the first, although anatomists prefer
the second.
Since skeleton is derived from the Greek σκελετόν, where the epsilon spelling
of the penultimate vowel indicates a short vowel, the regular pronunciation
according to the Latin stress rule is indeed the usual one, ˈskelɪtn̩ .
Actually, the adjective-forming suffix -al is normally disregarded for purposes
of the stress rule: we keep the stress on the adjective in the same place as it has
for the naked stem, thus from ˈperson we form ˈpersonal (despite the long ō in
Latin persōna). From ˈuniverse we have ˌuniˈversal, because the extra syllable
in the adjective means that Chomsky and Halle’s Alternating Stress Rule
comes into play. So in skeletal, which does not involve an extra syllable
compared with skeleton, it would be expected that the stress would be located
on the same syllable as in skeleton.
And please don’t ask about adjectival, because I don’t understand why it’s
ˌædʒɪkˈtaɪvl̩ , either.
2.13 gemination and degemination
Geminated (double) consonants are quite common in English. They

are never found within a morpheme, but arise across (i) morpheme boundaries
2.13 gemination and degemination 49
and (ii) across word boundaries, wherever one element ends in a given consonant
and the following element begins with the same consonant:
(i) meanness ˈmiːnnəs, guileless ˈɡaɪlləs, nighttime ˈnaɪttaɪm, midday
ˌmɪdˈdeɪ;
(ii) nice sort naɪs sɔːt, big girl bɪɡ ɡɜːl, bad dog bæd dɒɡ.
Phonetically, geminated consonants are pronounced like ordinary ones but with
extra duration. In the case of plosives, there is a single articulatory gesture but
with a longer hold phase.
same man /seɪm mæn/ ¼ [seɪmːæn], rub both /rʌb bəʊθ/ ¼ [rʌbːəʊθ]
This much is covered by our textbooks. But what I don’t remember seeing much
discussion of is degemination in English, the process whereby a geminate is
simplified, i.e. two consonants are reduced to one.
Degemination is the norm in derivational (fossilized) morphology, but the
exception in inflectional (productive) morphology. So inside the lexicon there are
plenty of cases of degemination such as:
dis þ sent ¼ dissent dɪˈsent (cf. consent), in þ numerable ¼ innumerable ɪˈnjuːmərəbl̩ ;
abbreviate, addiction, aggregation, allocate, connotation, immigration, immature
– though even here we sometimes override degemination to emphasize the

meaning of a prefix, as in dissuade dɪ(s)ˈsweɪd (cf. persuade), illegal ɪ(l)ˈliːɡl̩ .
Germanic affixes are not subject to degemination. So alongside the Latinate
innumerable we have unnecessary with geminated nn and cases like meanness,
guileless in (i) above. Entombment is an interesting case: -ment, despite its Latin
origin, ‘came to be treated as an English formative’ (OED), which means that
entombment ɪnˈtuːmmənt has geminate mm.
What got me thinking about all this was hearing a Conservative MP on the
radio pronouncing the phrase a good deal better with just a single, ungeminated
d. This option is something I drew attention to in LPD, and something
I sometimes do myself, though I admit it is a minority choice and probably
becoming less frequent. It is also irregular, in the sense that we don’t degeminate
other dd sequences (good dog, bad deal).
Cannot is a special case. When can and not come together the result is not
can not kæn nɒt but kænɒt (which in spoken styles is then usually further
reduced to can’t).
Jack Windsor Lewis has pointed out the ‘commonplace’ possible degemination
in some more sə ˈmɔː, good `deal, take `care and prime `minister. Degemination
in prime minister (as praɪˈmɪnɪstə) is certainly widespread: I am not so convinced
about take care, which doesn’t feel right to me with just one k.
The possible elision of t in sit down, let go, whaddya want is not degemina-
tion, though it is obviously similar, and similarly lexically restricted.
When the suffix -ly is attached to a stem ending in l, there are several
possibilities, partly lexically determined. Recall that degemination is the loss
of one of two identical consonants; compression is a reduction in the number of

syllables.
A. stems ending in syllabic l (or əl)
1. degemination with compression, so that the adverb has the same
number of syllables as the adjective. Examples: gentle ˈdʒentl̩ !
gently ˈdʒentli, simple ! simply, single ! singly, noble ! nobly,
able ! ably (and similarly for the suffix -able or -ible: possible !
possibly, visible ! visibly, reasonable ! reasonably, understand-
able ! understandably, etc.).
2. degemination with variable compression. Sometimes compressed,
sometimes pronounced with -əli or -l̩ i. Adjectives ending in -ical,
thus historic(al) ! historically, surgical ! surgically usually -ɪkli.
Adjectives ending in -ful, thus peaceful ! peacefully, beautiful !
beautifully -fli. Also double ! doubly, special ! specially.
3. degemination without compression. Other adjectives ending in -al,
thus natural ! naturally, vital ! vitally ˈvaɪtəli or ˈvaɪtl̩ i, racial !
racially.
B. stems ending in non-syllabic l
1. degemination. Example: full ! fully ˈfʊli (perhaps the only categor-
ical example).
2. variable degemination. Examples: whole ! wholly, dull ! dully
ˈdʌli or ˈdʌlli.
3. no degemination. Examples: pale ! palely ˈpeɪlli, vile ! vilely, cool
! coolly, futile ! futilely.
2.14 reductions in casual speech

Someone asked me how a phonetician would transcribe There you
go!, so I wrote down for him ˈðeə ju ˈgəʊ. ‘OK,’ he said, ‘but what if it was said
like this?’ – and he produced something like ejˈgə. ‘Has anyone attempted to
produce systematic transcriptions of this type of speech?’ he asked.
I referred him to various publications that attempt to do just that, or something
like it – Linda Shockey’s Sound Patterns of Spoken English (Blackwell 2002)
and Gillian Brown’s Listening to Spoken English (Longman 1990, second
edition). I also mentioned Jack Windsor Lewis’s EFL-oriented People Speaking
(Oxford University Press 1977). (There is now also Richard Cauldwell’s excel-
lent and self-published Phonology for Listening of 2013, www.speechinaction.
org/phonology-for-listening.)
I also suggested that such reduced versions could (in principle) always be
derived by rule from careful versions. We may not be sure of all the rules, but in
2.15 abject haplologies 51
this particular case there appears to be no difficulty in deriving the reduced

version by rules of quite general applicability:
input ˈðeə ju ˈgəʊ

1. reduce you to its weakest form, jə ˈðeə jə ˈgəʊ
2. drop the second element of diphthongs ˈðe jə ˈgə
3. drop initial ð ˈe jə ˈgə
4. drop all stresses except the nuclear accent e jə ˈgə
5. drop phrase-internal schwa e j ˈgə
¼ output
QED, or, as young people say these days, ta-dah!
2.15 abject haplologies

We’ve all heard library pronounced as ˈlaɪbri, or probably as
ˈprɒbli. Some of us know that these are examples of haplology, defined as the
omission of a repeated sound or syllable.
I remember at school, when I was in the classical sixth, the teacher wrote
φιλόγος on the board instead of φιλόλογος. ‘Please sir,’ I said, ‘you’ve commit-
ted a haplology.’ (I was a real smart-arse as a teenager.) The teacher replied
sarcastically, ‘I have made a simple mistake’.
If I’d been even smarter, I’d have known that the term for a written as opposed
to a spoken example is a haplography.
Anyhow, I heard a new phonetic haplology the other day. It was from a
medical consultant giving a talk on blood pressure. She repeatedly referred to
the instrument used to measure blood pressure, the sphygmomanometer, as a
ˌsfɪɡməˈnɒmɪtə.
In the BBC Radio 4 panel game Just a Minute the panellists have to attempt
to speak for sixty seconds on a given topic without hesitation, deviation or
repetition. In one episode I listened to, one of the topics was what is usually
written as quantitative easing. I would pronounce the first word here as
ˈkwɒntɪtətɪv. However, the chairman (Nicholas Parsons), and everyone else
on the programme as far as I could tell, pronounced it ˈkwɒntətɪv. Formally,
I suppose this too is a haplology. A repeated consonant gets deleted along with
its support vowel.
The OED has a separate entry for quantitive, which it regards as ‘irregularly’
formed. It adduces citations from 1626 onwards. In Just a Minute, given that
the chairman specified the topic as ˈkwɒntətɪv easing, I wonder whether panel-
lists would have been penalized for deviation if they had referred to ˈkwɒntɪtətɪv
easing. You’re allowed to repeat the words on the card (but no others) without
penalty. So perhaps repeated ˈkwɒntətɪv ought to have been allowed, but not
quantitative ˈkwɒntɪtətɪv, since that was not exactly the word the chairman
had specified as part of the topic. Are the ‘words on the card’ those that the
chairman utters, or those that are written?
Americans and some others prefer ˈkwɒntəteɪtɪv, a variant which I imagine
would be resistant to haplology.
2.16 the happY vowel
I still receive a number of queries from students puzzled by ‘the happY

vowel’ (shown in many dictionaries and textbooks as i, as opposed to iː or ɪ).
The point is that the last vowel in happy and all other similar words
can sound like the iː of fleece or like the ɪ of kit or something intermediate
between them. The distinction between iː and ɪ is ‘neutralized’ in certain
phonetic environments. There are many thousands of other examples besides
happy – very, coffee, valley, spaghetti, sanity, suddenly; also the second vowel
in various, radiation.
Current pronunciation dictionaries and other works use the notation i to
show these cases in which the distinction between FLEECE (iː) and KIT (ɪ)
does not apply. Where it does apply, they write iː (as in green, bead, feet, reach,
seating) or ɪ (as in grin, bid, fit, rich, sitting), as appropriate.
There is a corresponding neutralized back vowel u, as in thank you, incongruous,
situation.
More technically, it seems to be right to recognize two distinct vowel systems
in English. The strong system (in RP) covers ɪ e æ ɒ ʌ ʊ iː eɪ aɪ ɔɪ uː əʊ aʊ ɪə eə
ɑː ɔː ʊə ɜː, while the weak system covers not only i u ə but also ɪ ʊ, as in finishes,
executive. Weak vowels occur only in unstressed syllables, although strong
vowels may be either stressed or unstressed.
The vowels ɪ ʊ therefore belong to both systems. As a weak vowel, how-
ever, ʊ is increasingly replaced by ə. Various other accents, e.g. Australian,
lack the distinction between weak ɪ and ə, or at least give it a very low
functional load. Thus English seems to be moving towards a weak system of
three vowels only: i u ə.
Vowel weakening involves a switching between the strong system and the
weak system. We see this:
in words with strong and weak forms, e.g. can kæn – kən;
and in alternations such as preside – president, variety – vary,
anatomy – anatomical.
One of the reasons our spelling system is less than transparent is that it generally
writes the strong and weak alternants identically.
The happY vowel is good news for speakers of languages such as Spanish,
for whom the distinction between iː and ɪ is difficult. Where we write i, they
needn’t bother to make that distinction.
2.17 crazes and crazies 53
2.17 crazes and crazies

The Guardian newspaper carried the following correction:
A mishearing led us to quote the psychoanalyst Adam Phillips as saying that
‘anybody who’s been in the therapy profession for any length of time will
know that there have always been crazies’. In fact, he was referring to crazes.
I am one of those who pronounce crazies and crazes identically, as ˈkreɪzɪz.

They are absolute homophones for me, so I can sympathize with anyone who
‘mishears’ one for the other.
Nevertheless many other native speakers of English – the majority, I should
think – make a distinction between the two. Either their happY vowel (the final
vowel in crazy) is tenser, [i], or they use a schwa in the plural ending after a
sibilant (crazes), or both.
That is why modern pronunciation dictionaries, including LPD, use the special
symbol i for the weak happY vowel. It covers both an iː-like quality, for those
who use it, and an ɪ-like quality, for those like me who use that. So we write
crazies as ˈkreɪziz, and crazes as ˈkreɪzɪz (or ˈkreɪzəz).
Other potential homophones to test include: bandied – banded, studied –
studded, taxis – taxes, Rosie’s – roses.
You find the happY vowel both at the end of words like happy and also
within a word before a vowel, in words such as glorious. Behind this use lie two
main considerations. One is saving space when specifying pronunciation in a
dictionary or vocabulary list. If I pronounce ˈhæpɪ but the majority of English
speakers say ˈhæpiː – i.e. if like Jones and Gimson I identify the final vowel with
that of kit, while they identify it with that of fleece – then we save space by using
a special symbol, i, distinct from both, rather than by transcribing each such
word twice (and there are a very large number of them).
But the other reason is a more sophisticated one. English arguably distin-
guishes two vowel systems, strong and weak. Weakening means switching
from the strong system to the weak, making a strong vowel weak. Thus ə is
the weak counterpart of strong æ, ɒ, ʌ (and various other vowels), as we see in
the strong and weak forms of at, of, us. In exactly the same way, i and u may be
seen as the weak counterpart of strong iː, uː, as seen in me and prevocalic you, to.
Because the weak-vowel system is much smaller than the strong-vowel system,
in these positions we have a neutralization (or, in Trubetzkoy’s terminology,
an Aufhebung, ‘annulment’) of some of the phonemic oppositions present in the
strong-vowel system.
These are not the only neutralizations we find in English. The opposition
between t and d, exemplified in tamp vs. damp, is neutralized after tautosyllabic
s, as in stamp. So logically we might introduce a new symbol, say T, to show
it, and instead of stæmp write sTæmp. What applies to the alveolar plosives
also applies to the labials, as in pin – bin – spin, and to the velars, as in core –
gore – score.
As a further example, within the strong-vowel system, there is the neutraliza-

tion of iː (bee) vs. ɪə (beer) in the environment _rV, as in the first syllable
of serious.
But the practical needs of EFL learners inhibit us from going any further down
this path. Even the happY vowel is puzzling enough for many of them.
2.18 banded or bandied?
A journalist in the Financial Times spoke of a derogatory term being

‘banded about’ by hip-hop artists. What he meant, of course, was bandied about,
from the verb bandy about ‘to mention an idea, name, remark, etc. several times,
especially in order to seem impressive’. There is no verb band about.
Bandy is one of the words that ends with the happY vowel, which we
nowadays write i, representing the neutralization (non-distinction) of iː and ɪ in
weak final position, and ranging phonetically between tense [i] and lax [ɪ].
The conditions for bandied and banded to be homophonous, as they pre-
sumably are for the journalist in question, are twofold. First, the happY vowel
must be realized with a quality and quantity identical to those of /ɪ/, and
secondly the past tense/past participle suffix -ed must be pronounced as ɪd,
not as əd.
Since my own pronunciation satisfies these two criteria, the words are hom-
ophonous for me, ˈbændɪd. I therefore have to be careful to spell each one
appropriately. But Tony Blair, for instance, with his pervasive use of ə in weak
syllables, would presumably make a difference between them. So would anyone
who has an [i]-like quality for the presuffixal happY vowel of bandied. Some
speakers have both these characteristics. The trend is towards distinguishing
the two words as [ˈbændid] bandied vs. [ˈbændəd] banded.
Other relevant pairs are candied – candid, taxis – taxes, studied – studded and
even Billy’s – Bill is. In the first of these it is a matter not of an inflectional
ending, but of the adjectival suffix -id. But this usually seems to agree with -ed.
2.19 the status of schwa
Ought schwa to be included in the phonemic inventory of English?

After all, it only occurs in the weak-vowel system, and all other vowels in that
system are already members of the strong-vowel system. So isn’t schwa equiva-
lent to the T in sTæmp we discussed above (2.17), and wouldn’t it make
sense to write panda as either ˈpændæ, ˈpændɒ or ˈpændʌ rather than ‘pændə?
This is essentially what the first edition of the OED did, using notations such
as ă. It was abandoned in later editions.
In some words (not panda) this kind of analysis is supported by the use of
strong vowels in singing. Transcribing angel as ˈeɪndʒĕl (where ĕ ¼ weakenable
2.20 charted and chartered 55
e) could be justified on these grounds, even though in all kinds of speech other
than singing we say ˈeɪndʒəl.
But it is precisely cases such as the second syllable of panda that present
an interesting case. There is no style of speech in which we use anything other
than ə for the vowel in this word, and no grounds other than orthographic for
choosing one of the alternatives shown rather than another. As long as it remains
a weak vowel, the final vowel of panda contrasts only with what I write i and u,
i.e. the vowels of happy and thank you. Compare panda and handy. As long
as we know we are in the weak vowel system, schwa does indeed represent a
neutralization of e, æ, ɒ, eə, ɑː, ɔː, ɜː, as we see in the strong/weak alternation
of the function words them, at, of, there, are, for and sir.
Actually, there’s a problem with e, in the endings -ed, -es, -est, which for most
speakers of RP and similar accents weakens not to ə but to ɪ. (In closed syllables
there is a slightly larger weak-vowel system, which includes ɪ.) Such speakers
formerly also weakened it to ɪ in -less and -ness, but now mostly weaken it to ə.
So we have the awkward anomaly that e can have two different neutralization
forms, even for the same speaker.
However, the main argument against this line of thought is rather different.
It is that we never know from the general structure of a word whether a given
unstressed syllable will select its vowel from the strong system or from the weak
system. Alongside words like gymnast (with -æst, strong system) we have words
like modest (with -ɪst, weak system). Think about the unstressed final syllables
of phoneme, Kellogg, cuckoo, syntax, torment (n.) – all strong. No neutralization
there! Chomsky and Halle have some pointers to what’s going on, but in many
cases it remains arbitrary whether or not vowel weakening occurs.
In this respect English differs strikingly from, say, Russian, where weakening
is highly predictable once you know where the word stress is located.
2.20 charted and chartered
Every day the Guardian newspaper publishes ‘Corrections and

clarifications’. One day in 2006 the paper included an apology for writing
‘unchartered territory’ (they meant ‘uncharted’), and another for saying that
a minister had ‘tended his resignation’ (they meant ‘tendered’). The next day
they apologized for allowing the writer of a letter to the editor to say ‘. . . eating
initiatives are having an affect on our school children’, with the undesir-
able effect of suggesting that he did not know the difference between ‘affect’
and ‘effect’.
I am showing my age when I reveal that for me not only does chartered ˈtʃɑːtəd
sound different from charted ˈtʃɑːtɪd and tendered ˈtendəd from tended ˈtendɪd
but also that for me affect (verb) əˈfekt is not a homophone of effect ɪˈfekt.
All my twenty-year-old native-English students seem to have affect and effect
as homophones, both with ə-. An awful lot of them pronounce -ered and -ed as
homophones too, both with -əd. Evidently the Guardian reporters who misheard
or misspelt tend(er)ed and chart(er)ed do the same.
In England it seems to be along the east coast that the drift from weak ɪ to ə is
at its most general: in Norfolk, for instance, or in Geordieland. Elsewhere, it is a
well-known characteristic of southern-hemisphere English, as well as of most
rhotic accents. Among public figures, Tony Blair is one who uses a striking
number of schwas where traditional RP would have ɪ.
Although in EFL we traditionally teach that the verb ending -ed is pronounced
-ɪd after alveolar plosives (as in waited, faded, painted, landed), at least one
recent EFL textbook prescribes -əd.
2.21 singing strongly

Being the son of an Anglican clergyman, I have been accustomed
from my earliest years to singing hymns, both at school and in church.
The pronunciation used in hymn-singing is subtly different from ordinary
conversational pronunciation. It is ‘over-articulated’: in particular, there is
less vowel reduction. The ending -ed after alveolar plosives is usually ed
rather than ɪd or əd. The ending -(e)s after sibilants is usually ez rather than
ɪz or əz. The endings -less and -ness are unreduced les, nes. Not all vowels
are strong, however. The preconsonantal indefinite and definite articles retain
schwa: ə, ðə.
So when we sing we pronounce praises as ˈpreɪzez and kindness as
ˈkaɪndnes, as well as singing to invariably as tuː. In the morning and the sky,
however, the is ðə. When singing, we all make these modifications without
conscious thought.
2.22 happY prefixes
In the 2008 revision of LPD I changed the entries for words with the
unstressed prefixes be-, de-, pre-, re-. Instead of entries like these:
becalm bɪ ˈkɑːm bə-, §biː-

predict prɪ ˈdɪkt prə-, §priː-
– I put:
becalm bi ˈkɑːm bə-

predict pri ˈdɪkt prə-
The LPD symbol § denotes ‘BrE non-RP’. For the last twenty years or so
phoneticians have been using the symbol i for the weak close front vowel used
at the end of words such as happy, coffee, donkey (‘the happY vowel’).
2.23 happY endings 57
In conservative Daniel Jones-style RP this vowel is phonetically identical

with the ɪ of KIT. But for many other speakers it is more like (or indeed
identical with) the iː of FLEECE.
The environments in which I use i in the first and second editions of LPD are:
word-finally: happy, coffee, valley;
before a strong vowel: variation, ratio;
and before a weak vowel: glorious, convenient.
I also use it:
at the end of a combining form: multilateral, polytechnic;

and in inflected forms in which d or z is added to a stem ending in i:
carried, carries.
So it was a very natural extension to use it for the prefixes be-, de-, pre-, re-.
It saves a little space, too.
Just as studied was a homophone of studded for Daniel Jones (and is for me),
or bandied and banded, so now descent was/is a homophone of dissent. But
there are other speakers who make a difference in such pairs.
2.23 happY endings
In the discussion of bandied and banded (2.18 above) I failed

to mention a complicating factor: that there are people who have tense [i] for
the word-final happY vowel, but change it to [ɪ] before inflectional d or z. Such
speakers therefore have [i] in bandy, taxi, study, coffee but [ɪ] in bandied, taxis,
studied, coffees. It follows that for them bandied can still be homophonous with
banded, and taxis (cabs) with taxes, despite the [i] in the uninflected form. Other
speakers, though, do not do this, but keep [i] even before an inflectional suffix.
I have never come across anyone who distinguished the plural from the posses-
sive. Probably cities and city’s are homophonous for everyone.
You can get the same change of [i] to [ɪ] before -ful and -ment, giving [i] in
beauty, accompany but [ɪ] (or even [ə]) in beautiful, accompaniment.
An even more interesting case is suffixes beginning with l: -ly and -less. I have
[ɪ] in happy, angry, but often change it to [ə] in happily and virtually always
in angrily. (As editor of LPD, Gimson gave priority to -rəl- in this word as
long ago as 1977.) A more conservative pronunciation keeps ɪ in this position,
a more progressive one ə.
For those who go all the way (like me) there is a further interesting conse-
quence. The derived schwa can then trigger syllabic consonant formation,
producing a syllabic l̩ in readily ˈredl̩ ɪ, with a laterally released d. Mightily then
rhymes with vitally. Similarly, I think many, perhaps most, people say funnily
and penniless with -nl̩ -.
I wouldn’t do that in guiltily or unwieldily, though.
2.24 some
A correspondent asked about the pronunciation of the word some.
From the point of view of EFL there are three issues here:
does this word have distinct strong and weak forms?
if so, what is their distribution?
if the distribution depends partly on whether this word is accented,
when is it accented?
Let’s take them in order.
In RP some has the strong form sʌm and the weak form səm. The weak form
is susceptible to possible syllabic consonant formation, making it sm̩ . I’ll
ignore this possibility in the remainder of the discussion. Despite ending in a
labial rather than in an alveolar, this word is for many speakers also susceptible
to assimilation of place of articulation, e.g. ˈsʌŋ kaɪnd əv. . . some kind of. . . I’ll
ignore this possibility, too.
As far as other accents are concerned, it depends on whether or not there
is a robust contrast between ʌ and ə (as in RP) or not (as in many other accents).
If there’s no reliable contrast between the vowels, it makes little sense to
distinguish strong and weak forms.
The distribution of strong and weak forms in RP is subject to the usual rules.
The word is pronounced strong if stranded or accented, weak otherwise.
‘Stranded’ means followed by a syntactic gap. In the case of some, this would
be because the noun that would otherwise follow has been deleted (ellipted).
In (1) some has its usual following noun, so is weak. In (2) it is stranded,
therefore strong, even though unaccented.
(1) Would you like some coffee?
(2) A: More coffee? Let me pour you some.
B: No, thanks. I’ve still got some left.
The some in (2) stands for some coffee. But the word coffee has been ellipted
(‘is understood’). That leaves some stranded.
In some more, some is not stranded, so has its weak form.
Let me get you some more.
Some is unaccented when it is merely a quantifier (used like a/an, but before
a non-count noun or a plural: some salt, some books). If the noun is present it
must have its weak form, səm.
There’s some milk in the fridge.
Go and get some potatoes.
Here are some more examples of stranded some, with the strong form.
(You’re eating ice cream.) ˈI want some, | ˈtoo!
(Look, apples!) Why don’t you ˈtake some?!
2.24 some 59
You could say that all ‘pronominal’ uses of some – all cases where
some is not followed by its noun or noun phrase – represent stranding. That
would cover cases like the following, where some is however likely to be
accented.
(Milk?) There’s ˈsome in the fridge.

(Potatoes?) There are ˈsome in the ˈshed.
When some has a more specific meaning, it is usually accented and always has
its strong form sʌm.
(i) as opposed to others or to all

ˈSome ˈinsects are ˈgood for the garden,
| ˈsome are ˈbad, | and ˈothers are ˈneutral.
A: All Cretans are liars.
B: That’s not true, | although ˈsome are.
ˈSome cheese | is made from ˈgoat’s milk.
It’s ˈall right for \/some!
(¼ unlike the rest of us, you [or they] have all the luck)
(ii) meaning ‘a considerable quantity of’
It was ˈsome ˈyears | beˈfore she ˈsaw him again
It’s been ˈsome ˈtime | ˈsince we last ˈsaw one another.
(iii) as a determiner, ‘some . . . or other’
ˈSome ˈidiot’s | ˈleft the ˈlight on.
ˈSome day | I’ll ˈwin the ˈlottery.
ˈSome people | are ˈjust ˈuseless.
For ˈsome reason | she ˈwasn’t inˈvited.
(iv) exclamatory, ‘very remarkable’
That was ˈsome | ˈparty!
They were ˈsome | ˈstudents!
Compare
(i) He put səm ˈfruit on the pizza.

(as well as cheese, ham, etc.)
(ii) He put ˈsʌm ˈfruit on the pizza.
(I can’t identify what fruit.) or (it was a remarkable fruit.)
See Churchill’s defiant words to the Canadian parliament in 1941:
When I warned [the French] that Britain would fight on alone, whatever they
did, their Generals told their Prime Minister and his divided cabinet: ‘In three
weeks, England will have her neck wrung like a chicken.’ ˈSome ˈchicken;
| ˈsome ˈneck!
– with the strong form, of course.

2.25 the symbol for STRUT

A student has taken me to task for using what he sees as inaccurate
phonemic symbols for English.
My /ʌ/ [he claims, very reasonably] seems to be far closer to [ɐ] than to
(cardinal) [ʌ]; my /æ/ I believe to be far more like (cardinal) [a] than [æ]; my
/u/ seems to be a kind of central/lax front vowel of the order of [ʉ], quite far
from (cardinal) [u], my aɪ is far more like ɑɪ or ʌɪ, and my e is far more like
ɛ than (cardinal) e. Now, I understand that phonemic symbols are merely
conveniences, and aren’t necessarily meant to convey any underlying
phonetic reality. I also understand that, historically, it was also more difficult
to type, for instance, the IPA symbol for a close central rounded vowel,
ʉ, than u.
But he’s still very unhappy about it.

I replied along the following lines:
When I first started phonetics myself, half a century ago, I had the same
thought as you about ʌ: that it would be better written ɐ. I made a point of
transcribing it like that in the transcription assignments I was set by my
phonetics teacher.
But when I came to be a phonetics teacher myself, and to publish books, it

seemed better to stick with the same transcription that other people use.
There are millions of people around the world who know that ʌ means the
STRUT vowel of English; there are a few thousand, at most, who know or
care anything about the cardinal vowel system.
The same, mutatis mutandis, applies to /æ/ (better written /a/?) and to /uː/
(better /ʉː/?).
Furthermore, if we want the same transcription system to apply to a wide

range of different varieties of English, there are strong arguments in favour
of sticking with æ and uː.
It is possible that in another fifty years the discrepancies will have become
too great: the tectonic plates will shift, and some new influential phonetician
will establish a different notation. But not me.
2.26 the Teutonic Rule
The ‘Teutonic Rule’ of English says that the first two syllables of
a word cannot both be lexically unstressed. In words with the main stress on the
third syllable or later, we need a secondary stress on the first or second syllable.
So what do I mean by ‘lexically stressed’?
In running speech, as we know, stress (in the sense of a rhythmic beat on a
syllable) is very variable. Depending on intonation, speech rate and other factors,
2.26 the Teutonic Rule 61
we readily suppress the stress on some of the syllables which could have been
stressed. This is ‘utterance stress’.
‘Lexical stress’, on the other hand, is a fixed property of each word in the
speaker’s mental lexicon. Whatever happens in running speech, the second
syllable in each of regret, consider, decisively remains lexically stressed, as we
show by their dictionary transcriptions riˈɡret, kənˈsɪdə, diˈsaɪsɪvli.
These are all single-stressed words. What is the justification for considering
some English words to be lexically double-stressed? i.e. to have two (or more)
lexically stressed syllables?
The crucial point with multiple-stressed words is their susceptibility to ‘stress
shift’. Although in isolation we say the words photographic, catastrophic with
the main stress on the penultimate syllable, and afternoon, fifteen with the main
stress on the final syllable, in connected speech – as is well known – the first
syllable of each of these words may have not just a rhythmic beat but also
intonational prominence (¼ an accent), while the other stress may be suppressed.
photoˈgraphic aˈphotographic ˈrecord

cataˈstrophic a ˈcatastrophic ˈfailure
afterˈnoon ˈafternoon ˈtea
fifˈteen ˈfifteen ˈpeople
So far, so good: students of English phonetics know about this (I hope).

And this is where the Teutonic Rule comes in. Without a lexical stress
(‘secondary stress’, ‘pretonic stress’) before the main stress, and consequent
susceptibility to stress shift, words such as photographic, catastrophic, afternoon
would not be phonologically well-formed.
Other pretonic stresses, though, are not dependent on the Teutonic Rule.
The word fifteen has only two syllables, so it would conform to the Teutonic
Rule even without a pretonic stress on the first syllable. How, then, can we
determine its presence or absence? The only way is to examine its behaviour
under stress-shift conditions.
Consider, as a test case, the word bamboo. In its dictionary entry, should the
pronunciation read bæmˈbuː or ˌbæmˈbuː?
First let it bear the intonational nucleus, and then put it in a sentence in which
it is followed by a word bearing the nucleus, and see what happens.
1. It’s ˈmade of bamˈboo
2. a bamˈboo ˈtable
3. a ˈbamboo ˈtable
We say (1) and (3), but not (2). So bamboo must be lexically double-stressed,
ˌbæmˈbuː.
If we test cosmetic, we get:
1. It’s ˈmerely cosˈmetic
2. a cosˈmetic ˈsurgeon
3. a ˈcosmetic ˈsurgeon
We say (2), not (3). At least, that’s what I do. So cosmetic is single-stressed,
kɒzˈmetɪk.
Sometimes usage is divided.
1. ˈbuying anˈtiques
2. ? an anˈtique ˈchair
3. ? an ˈantique ˈchair
Some speakers go for (2), some go for (3). For those who say (2) this word is
lexically single-stressed; for those who say (3) it is lexically double-stressed.
In LPD I want to cater for both. That’s why I use the rather ungainly notation
(ˌ)ænˈtiːk.
2.27 problems with lexical stress
It is sometimes difficult to decide how to mark word stress in

a dictionary. A dictionary should ideally show lexical stress, i.e. the stress
pattern that the word has when spoken in isolation with no special pragmatic
factors in play (no special contrastive stress, for example). You can usually
check this by making the word a one-word utterance, as an answer to a
question such as What’s that? What’s it like? What did he do? How did
he do it?
There’s a particular difficulty with compound adjectives such as God-given,
hard-drinking, long-term, sabre-toothed. Lexically it is probably correct to
show them all with double stress (end-stress), so ˌGod-ˈgiven, ˌhard-ˈdrinking,
ˌlong-ˈterm, ˌsabre-ˈtoothed.
The trouble is that all double-stressed words are candidates for stress shift. And
adjectives like these are usually attributive: they precede a noun, itself likely to be
stressed, so that in practice they undergo stress shift much more often than not.
our ˈGod-given ˈrights
a ˈhard-drinking ˈsailor
ˈlong-term ˈplans
a ˈsabre-toothed ˈtiger
It may feel slightly awkward to force them into predicative position:
What sort of rights are these? – God-given.
What kind of sailor was he? – Oh, hard-drinking.
What plans did you make? – Long-term ones.
What kind of tiger was it? – Sabre-toothed.
If we do this, however, I think it becomes clear that they are indeed double-
stressed.
What sort of rights are these? – ˈGod-ˈgiven.
What kind of sailor was he? – ˈHard-ˈdrinking.
2.28 twenty-twenny 63
What plans did you make? – ˈLong-ˈterm ones.

What kind of tiger was it? – ˈSabre-ˈtoothed.
If we answered the last question by saying
–ˈSabre-toothed
then we would implicitly be contrasting sabre- with some other kind of tooth-
edness. But sometimes this kind of contrastiveness becomes fossilized, as in
pear-shaped, so that initial stress is the only possibility.
It’s ˈall gone ˈpear-shaped.
There is also a problem in deciding whether to show words such as one-size-fits-
all or MRSA with four stresses or (as they are usually pronounced) just with
two; and likewise whether to show three or two lexical stresses in words such
as non-denominational.
2.28 twenty-twenny
Americans, as we all know, tend to pronounce twenty as ˈtweni rather

than ˈtwenti. That’s because they have an nt reduction rule that allows the
deletion of t between n and a weak vowel, as in winter, painting, center, counter,
gentle, in front of and all other cases satisfying this structural description.
Londoners, too, often say twenty in this way. But they don’t lose the t in winter,
painting, etc. (though they might make it a glottal stop). Londoners don’t say ˈwɪnə
rather than ˈwɪntə. In fact the only other cases for Londoners seem to be plenty,
wanted, went (away), want to, ˈpleni, ˈwɒnɪd, ˈwen (əˈweɪ), ˈwɒnə. (More
generally, they can lose the t from want and went whenever they are followed
by a vowel sound. There are also t-less reductions of trying to and going to.)
So we can say that for Londoners this is not a general phonological rule but
a lexically specific one applying to just a few special cases.
And what about RP? In LPD I do mention the ˈtweni form as a BrE possibil-
ity, but with a warning triangle against it. Yet I am aware that some people
who qualify generally as RP speakers do use it, as well as (some of) the other
London reductions mentioned. Ought I to remove the warning triangle?
You may be surprised that I include gentle as one of the examples in the
first paragraph above. Does it have t followed by a weak vowel, as required by the
structural description of the rule? Well, yes, it does. I regard syllabic [l̩ ] as a
realization of the underlying string /əl/. If this is right, then the nt reduction rule
holds as long as it applies before the syllabic consonant formation rule is applied.
underlying form /ˈdʒentəl/

nt reduction [ˈdʒenəl]
syllabic cons. formation [ˈdʒenl̩ ]
¼ surface form
But this is not without its problems. Before another syllabic consonant, namely
[n̩ ], Americans don’t reduce nt in this way. In words such as Clinton, accountant
they tend instead to use a glottal stop: ˈklɪnʔn̩ . (American /t/ tends to be glottal
before a consonant in the next syllable.)
So the AmE nt reduction rule is a bit messy, because we have to allow it to
operate before syllabic l̩ (and syllabic r̩ , as in winter), but not before syllabic n̩ .
But those are the facts.
The word continental /ˌkɑntəˈnentəl/ is a nice example. In the first /nt/ the /t/
becomes glottal, while in the second it disappears. So we get [ˌkɑnʔn̩ ˈenl̩ ].
Here’s one way of doing the phonology, with some awkward rule ordering.
underlying form /ˌkɑntəˈnentəl/

syllabic n formation [ˌkɑntn̩ ˈentəl]
T glottalling [ˌkɑnʔn̩ ˈentəl]
nt reduction [ˌkɑnʔn̩ ˈenəl]
syllabic l formation [ˌkɑnʔn̩ ˈenl̩ ]
¼ surface form
2.29 phantom r
As we all know, the rulers of Burma would prefer that we call their
country Myanmar.
In Burmese, this name is essentially just a variant of the name Burma. It is
transliterated as Myan-ma or Mran-ma, and in the local language pronounced
something like ma(n)ma, as against bama for the traditional name.
Within the Burmese language, Myan-ma is the written, literary name of the
country, while Bama (from which English Burma derives) is the oral, colloquial
name. In spoken Burmese, the distinction is less clear than the English transliter-
ation suggests.
So the situation is comparable to the former right-wing rulers of Greece, the
colonels who insisted on literary (katharevusa) Ellas rather than popular (dhi-
motiki) Elladha. Right-wing politics becomes associated with the more literary
form, left-wing with the more popular.
What interests me now, however, is the question of how Americans and other
rhotic speakers are supposed to pronounce this name. In both Myanmar and
Burma the English spellings assume a non-rhotic variety of English, in which the
letter r before a consonant or finally serves merely to indicate a long vowel:
ˈmjænmɑː, ˈbɜːmə.
So anyone who says the last syllable of Myanmar as mɑːr or pronounces
Burma as bɝːmə is using a spelling pronunciation based on British, non-rhotic,
spelling conventions.
2.30 Nadsat is non-rhotic! 65
In compiling LPD I faced a somewhat similar problem with the words scarper
ˈskɑːpə ‘vamoose, take off’ and sarnie ˈsɑːni ‘sandwich’. Americans don’t generally
know or use these words, but how would they pronounce them if they did? The first
is believed to be derived from Italian scappa! or by rhyming slang from Scapa
(Flow) (¼ go); the second obviously comes from sandwich. In neither case is there
an r in the source; yet rhotic speakers in Britain and Ireland pronounce both with r.
Another example is the breed of dog called shar pei. This name is Chinese, 沙
皮 shā pí (or the Cantonese equivalent sā pèih): again, no etymological r.
People in the rhotic west of England tend to pronounce r in words such
as khaki, even though there is no r in the spelling. The equivalence ‘non-rhotic
ɑː ¼ rhotic ɑːr’ is a deeply buried part of their phonological knowledge.
Under marm, the OED (2000, third edition) comments:
Variant of MA’AM n.1 In U.S. usage, it is not always clear whether the spelling
with -r- is merely a graphic device to indicate lengthening of the vowel (espe-
cially in representations of the non-rhotic dialects of New England), or else
represents a genuine intrusive /r/: rhotic /mɑrm/ can easily develop from /mɑːm/
by analogy with e.g. rhotic /hɑrm/ harm n. corresponding to non-rhotic /hɑːm/.
It asks us to compare the spellings mars and marse for Mas’, i.e. ‘master’.
Think of the inner-city buses called Hoppa. Some British trade names of this
sort, transparent to us non-rhotic speakers (¼ ‘hopper’ – geddit?), must be either
baffling or irritating to rhotic speakers, not to mention non-anglophones). Eric
Armstrong, writing from Canada, tells me he was baffled by the product names
Whiskas (catfood) and Polyfilla.
As a child, I had no idea that these were derived from words with -er
endings. Their advertising makes the words sound just like we Canadians
would read them. It wasn’t until I went to the UK for a year in my early 20’s
that I began to ‘hear’ the r in those words.
He adds:
As a voice trainer, I have to do mental gymnastics every time I use a book
for actors written by a British (non-rhotic) trainer, such as Cecily Berry,
who writes the ɑ vowel as ar and the ɔ vowel as ore, etc. North Americans
and other rhotic speakers are just stopped dead in their tracks trying to
figure these out. (Why can’t they abandon these spelling conventions and
use the IPA?)
As we say (or rather write) in England, you’ve got to larf.
2.30 Nadsat is non-rhotic!
You may have read the novel A Clockwork Orange by Anthony

Burgess (first published 1962) or seen the 1971 Stanley Kubrick film based on
it. It is set in an imagined England of the future, an England under Soviet

domination. If so, you will either have recognized the Russian origin of various
words in the slang put in the mouths of some of the characters, or alternatively
have been thoroughly puzzled by them. For example, the protagonist Alex refers
to the fellow members of his teenage gang as his droogs. This word is based on
the Russian друг drug, ‘friend’. Similarly, klootch means ‘key’, from Russian
ключ kl’uch.
A correspondent from France tells me of his gratification the day he suddenly
realized that:
Burgess’s British-inspired spelling gulliver actually meant голова
golova (head)
with goober he referred to губа guba (lip)
with morder he referred to морда morda (muzzle, or a human’s
beastly face)
with rooker he referred to рука ruka (hand).
Speakers of rhotic English, or non-natives, do indeed face this additional hurdle
when deciphering Burgess’s invented slang.
My own favourite Nadsat word is horrorshow, meaning ‘good’, which is
cleverly antonymous. It comes from the Russian хорошо, transliterated khorosho
and in Russian pronounced xəɾʌˈʃo.
2.31 non-rhotic loanwords

In Japanese there are plenty of examples of borrowings from English
which rhotic speakers might find difficult to interpret. (To be honest, even we
non-rhotic speakers do not always easily connect such borrowings to the English
origin, either.)
Japanese has an r, but it is used only before a following vowel, i.e. in just
the same positions where non-rhotic English allows it. So English door becomes
Japanese doa.
An interesting example is the word elevator. When written in Japanese kata-
kana it reads, in transliteration, erebeetaa, and is correspondingly pronounced.
Yet this word must have been borrowed from AmE, not from BrE, since we
Brits call the apparatus in question a lift. Nevertheless, the American voiced tap t̬
of American elevator ˈeləveɪt̬ ɚ is mapped onto Japanese voiceless t rather than
onto voiced d or r (¼ tapped ɾ in Japanese), while the American final r-coloured
schwa ɚ is mapped onto a Japanese double aa.
It is regular for English v to be mapped onto Japanese b, and for l to be
mapped onto r, i.e. [ɾ]. But that is not relevant to the current issue.
It also seems to be typical for loanwords to be taken from AmE but then
pronounced as RP-style BrE interpreted in Japanese phonetics.
2.32 Gordon Brown 67
Take another example: hamburger becomes Japanese non-rhotic hambaagaa

(though the place name Hamburg is Japanized as hanburuku).
Actually, there are two possible ways of making a non-prevocalic r fit in
with Japanese phonotactics. One is to ignore it, as in the final syllable of
erebeetaa. The other is to treat it like other non-prevocalic consonants, which
for Japanese means adding an epenthetic u or o, as when McDonald’s becomes
makudonarudo.
The examples door, Japanese doa, and beer, Japanese biiru, illustrate these
two possibilities. Apparently Japanese preserves the American r in beer because
the word was originally borrowed from Dutch rather from English. Compare
the loanwords bia hooru (beer-hall) and bia gaaden (beer garden), which show
the expected treatment of English final r.
There are two interesting Japanese loanwords from German in which precon-
sonantal r is preserved in Japanese: arubaito, meaning a job on the side (German
Arbeit), and the name of the composer Moots(u)aruto (Mozart).
2.32 Gordon Brown

In his regular parliamentary sketch in the Guardian, the late
Simon Hoggart several times commented on the former UK prime minister
Gordon Brown’s ‘peculiar ways of pronouncing certain words’, among them
Bournemouth and reconciliation. Most of us use -məθ in Bournemouth, and
we start reconciliation with ˌrek-. So Gordon Brown’s -maʊθ and ˌriːk- do
sound bizarre to many people.
Hoggart comments that these pronunciations suggest a surprising lack of
social awareness on Brown’s part, since instead of following other people he
seems to have been led to invent his own spelling-based pronunciations.
EFL learners, take note: using wrong pronunciations may convey a hidden
unfavourable message about you as a speaker.
Another unusual pronunciation of Brown’s commented on by Hoggart is that
of pronouncing the word iron as spelt, ˈaɪrən. So for him it rhymes with Byron
and involves the same sequence as tyrant. But I think that this is the usual
Scottish form, and that Brown is merely doing what other Scottish people do.
The usual English pronunciation of iron, on the other hand, namely ˈaɪə(r)n,
must result from a historical metathesis by which ‑rən became ‑ərn. The r then
as usual coalesces in rhotic accents with the preceding schwa in rhotic accents
to yield ɚ, ˈaɪɚn, while in non-rhotic accents, being preconsonantal, it undergoes
deletion, giving ˈaɪən.
The Scottish pronunciation of this word does have the advantage of making
it clearly distinct from ion, wheras for non-rhotic speakers like me iron and ion
are homophones.
2.33 an albatross on the balcony

The phonetician David Deterding tells me that for him all words
with the spelling alC, where C is any other consonant, have the STRUT vowel
ʌ – albatross, altruism, algorithm, alcohol, balcony, calculate. . . He points out
that this does not happen before a clear l, since for him allergy, Sally, callous,
fallacy and so on all have the usual front æ (TRAP) vowel. He says he doesn’t
know whether this is a widespread pronunciation tendency, or whether it is a
personal idiosyncrasy of his own.
I have never come across it before. For most of us balcony and albatross have
æ. So my reaction was that it may well be David’s idiolectal idiosyncrasy.
Allophonic lowering of short front vowels before l is widespread, but speakers
do not usually perceive this as a change of phoneme.
I am aware that a similar treatment of e has been reported for New Zealand
English, where words such as shell and shelf are perceived as having æ rather than
e. But David says he doesn’t do this. For shelf, he says, there is the usual allophonic
lowering before dark l, but otherwise no difference from the vowel in shed.
I wonder if anyone else pronounces these words like David.
2.34 changes in English vowels

Here is a formant plot showing changes in certain vowels of ‘Standard
Southern British English’, comparing the speech of elderly people, born
1928–1936, with that of twenty-somethings born in the 1980s. I have taken it
2.35 Christmas puzzles 69
from an article by Gea de Jong, Kirsty McDougall, Toby Hudson and Francis
Nolan entitled ‘The speaker discriminating power of sounds undergoing histor-
ical change: a formant-based study’, which appeared in the Proceedings of
ICPhS Saarbrücken (2007).
<-- F2
2500 2000 1500 1000 500
200
Ii : l Iu : l
Iυl Iɔ:l
400
Iɑ : l 600
<-- F1
I l
800
H&M’05 (1928–1936) 1000

Dyvis’06 (1981–1988)
1200
You will see it confirms the vowel changes that people have been commenting
on for some time:
the opening/backing of æ (TRAP)
the fronting of uː (GOOSE) and ʊ (FOOT)
– and also to some slight extent a fronting of ɔː (THOUGHT).
What a lot can happen in fifty years!
The vowels iː (FLEECE) and ɑː (START), on the other hand, have remained
static.
2.35 Christmas puzzles
How do we account for the fact that ˈChristmas tree and ˈChristmas
cake are single-stressed, while ˈChristmas ˈDay, ˈChristmas ˈEve and ˈChristmas
ˈshopping are double-stressed?
We do our ˈChristmas ˈshopping in order to get our ˈChristmas cards and
buy some ˈChristmas presents, which we give on ˈChristmas ˈDay before our
ˈChristmas ˈdinner, which includes some ˈChristmas ˈpudding. We might
watch or listen to the Queen’s ˈChristmas ˈspeech or sing some ˈChristmas
ˈcarols round the ˈChristmas tree. Later we eat ˈChristmas cake or might pull
ˈChristmas ˈcrackers.
My colleague Petr Rösel writes to add ˈChristmas ˈstockings, ˈChristmas
ˈholidays, ˈChristmas ˈEve and the now almost forgotten ˈChristmas box. For
many families, he says, ˈChristmas time means a lot of stress. . .
Compounds with the second element Day are just as arbitrarily stressed as
those with the first element Christmas. We have single-stressed (early-stressed)
Boxing Day, St David’s Day, May Day, Labo(u)r Day, Derby Day, Independence
Day and Ascension Day, but double-stressed (late-stressed) Christmas Day,
Easter Day and Michaelmas Day.
These patterns, like all lexical patterns, can of course be overridden for
contrast: not Boxing Day but ˈChristmas Day.
As far as I can see, these lexical stress patterns are ultimately just arbitrary.
There is no logic to them.
2.36 data processing under stress
How do you stress the compound data processing?

When asked, I said I thought that the usual stress pattern for this compound
was double stress: ˌdata ˈprocessing. I could be wrong; like many recently
coined compounds, it has perhaps not yet really settled down lexically. Mean-
while we give it a pragmatic stressing.
?(i) We’re going to do some processing. | Today it’ll be ˈdata processing.
(ii) We’ve got some data. | Now we must move on to data ˈprocessing.
Clearly, (ii) is the more usual implicit situation.

So this pragmatic effect seems to override the usual early-stress rule for
compound nouns. But as the compound becomes more established it would be
expected to fall into line and move towards ˈdata processing (and no doubt
already has done for some speakers).
With word processing, on the other hand, we would never think:
*(ii0 ) We’ve got some words. | Now we must move on to word ˈprocessing.
So it is always ˈword processing, in accordance with the compound rule.

Compare also the American hesitation over the stressing of Thanks-
giving, or the cases of other compounds where we don’t all agree: ice cream,
armchair.
This explanation may be a bit ad hoc. But it’s the best I can manage.
2.37 a cab, innit?
A girl hurriedly dialled directory enquiries to book a taxi from

her home in London to Bristol airport, using the cockney rhyming slang Joe Baxi.
2.38 instances and incidences 71
But the operator told her they were unable to find anyone by that name.
Seething, the youngster snapped back: ‘It ain’t a person, it’s a cab, innit.’
The operator duly gave her what she asked for and put her through to the
nearest supplier of cabinets.
The cabinet saleswoman seemed equally confused.
‘Look love, how hard is it?’ the caller fumed. ‘All I want is your cheap-
est cab, innit. I need it for 10am. How much is it?’ The sales adviser
told her £180. The tantrum-throwing teenager left her address details and
rang off.
The next morning, rather than being picked up by a cab, the young woman had
the cabinet dropped off.
Allegedly.
The segments would have been the same in each case: kæbɪnɪʔ. Theoretically
the two possibilities ought to have been disambiguated by prosody.
(1) It’s a \cab, | \innit?
(2) It’s a \cabinet.
(1) would be expected to have a longer æ than (2), and to have a second accent
on the ɪn of innit (¼ isn’t it). The æ in (2) would be expected to have shorter
duration because it is subject to rhythmic clipping, aka foot-level shortening,
due to the following unstressed syllables.
As we know, you can’t always rely on prosody.
2.38 instances and incidences
It is sometimes very difficult indeed to hear the difference between

instances and incidences. That is why some people confuse the two words.
Here are a few cases thrown up by Google. You can easily find others.
New incidences of human rights abuses in run-up to the Beijing Olympics
revealed
What incidences of racism have you ever had to personally deal with?
Could you please supply me with any information/documents concerning
how many incidences of assault between prison inmates . . . have occurred.
In each case the writer must surely have meant instances. (Alternatively, in the
last case they may have meant incidents, which for many people is a homophone
of incidence.)
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 compen-
satory lengthening of the preceding consonant: ˈɪn(t)sːd̥ ən(t)s. In rapid speech,
however, I think this subtlety of timing can be lost, making incidence as good
as homophonous with instance.
This phenomenon has been termed ‘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.
2.39 settee
A correspondent asked why e (the DRESS vowel) is found at the

end of the first syllable in the word settee, given that this vowel is never found
in final open syllables.
This word, seˈtiː, is one of a small number of words that have a short strong
vowel in this position: another is tattoo tæˈtuː. Why? That’s how it is. English is
irregular.
Not surprisingly, some people (including me) usually pronounce it with a schwa,
səˈtiː. Others, though, say it with se-, and that, I think, is the usual pronunciation.
When I was a boy our family had a piece of furniture we called a settee, so it
is a familiar word for me. Nowadays, though, most people have a sofa.
My correspondent followed this up by asking about the word divorce, which is
shown in LPD with ɪ at the end of the first syllable. Should that not, she asked,
be i (happY) instead?
Yes, in LPD I transcribe divorce as dɪ ˈvɔːs, də-. Why do I not write the main
pronunciation as di ˈvɔːs? After all, I now transcribe words such as depend as di
ˈpend, də-, replacing the dɪ ˈpend, də-, §diː- of earlier editions.
In the third edition of LPD I decided to simplify the treatment of words with
the weak prefixes be-, de-, pre-, re-. I exploit the existing notation i, which
covers a range of possibilities from iː to ɪ, extending its use from words such as
happy, glorious, radiation to these prefixes (see 2.22 above).
However, words spelt di- are different. In the first and second editions I did not
give them a variant with diː-, because as far as I know such a pronunciation is
not found. Their prefix vowel can only be ɪ or ə.
A nice potential minimal pair is dissent vs. descent. Probably for most of us
they are homophones; but not for everyone. I transcribe them dɪ ˈsent and di
ˈsent respectively.
Yes, this does mean that ɪ is thereby left at the end of an open syllable. But in any
case we need to recognize ɪ as one of the weak vowels, sometimes distinct from i,
2.40 imma 73
because of its appearance in words such as finishing. And as schwa shows (banana
bə-, commercial kə-), there is no difficulty about syllable-final weak vowels.
Going back to the word settee, my correspondent asks whether it would
be possible to divide it set.iː, rather than my se.ˈtiː. Although putting the t in
the first syllable might seem to improve the phonotactics, I think it is wrong.
Why? Because the t is strongly aspirated, which it would not be if it were
syllable-final. (The main function of syllabification as I present it in LPD is to
predict the correct allophones in those phonemes that are sensitive to syllable
boundaries.) Conversely, lack of aspiration is what leads me to syllabify nostal-
gic as nɒ.ˈstældʒ.ɪk, leaving a syllable-final short strong vowel as awkward
as the one in settee. Sometimes language is not neat and tidy.
The word mistake brings these various points together. We do not say it as mɪs.
ˈteɪk, i.e. with an aspirated t. Despite its etymology, it must be mɪ.ˈsteɪk, since the t
is unaspirated, just as in stake; because in English t is unaspirated after s in the
same syllable. Furthermore, we are left with a first syllable ending in ɪ.
This analysis goes back to A.C. Gimson, who as editor of the fourteenth
edition of EPD corrected the syllabification of mistake (as shown by the location
of the stress mark) given in earlier editions. As evidence he took the study by
Niels Davidsen-Nielsen, ‘Syllabification in English words with medial sp, st, sk’,
published in the Journal of Phonetics 1974, 2: 15–45.
2.40 imma
I was struck by imma in a cartoon caption, presumably American,

‘Imma get myself a snack’. It evidently means ‘I’m going to’, and I think is
pronounced aɪmmə.
No doubt the initial diphthong can be reduced, too, to some kind of a.
One of my little campaigns over the years has been to try to persuade
learners of EFL that there is a respectable weak form of going to that
people ought to teach and learn, namely ɡənə (or before a vowel ɡənu).
We Brits tend to fight shy of the spelling gonna, thinking of it as too
American. Yet it is clear that we too can and do contract the words going
to (but only when they are modal – so not in sentences like I’m going to
Brighton tomorrow).
One stage further is the fusion of I’m with gonna. In my own speech this can
produce aɪŋ(ə)nə. The nasality of the m of I’m coalesces with the velarity of
the ɡ of gonna to yield a velar nasal ŋ, in a kind of two-way assimilation.
The following schwa can disappear.
But for imma the output is not a velar but a bilabial nasal that spreads over the
following underlyingly velar and alveolar consonants. This assimilation is purely
progressive, like ˈrɪbm̩ for ribbon (2.49 below).
I have no idea what proportion of speakers, American or even British, go for
aɪmənə or aɪmmə rather than the aɪŋənə or aɪŋnə that I would use.
Lynne Murphy, an American living in Britain and author of a blog on

BrE–AmE differences, tells me she thinks that this is from AmE dialectal
(and often humorously applied in other dialects) pseudo-aspectual a-prefixing.
I’m a-going to ! I’m a-gonna ! I’m-a.
Phil Thompson reports from California that aɪŋənə has a variant aämŋənə.
He thinks that the form implied by imma seems to indicate a southern or
African-American Vernacular English (AAVE) speaker and hears it in his mind’s
ear as amə or aːmə or āmə. I was probably wrong in my speculation about
possible double mm or syllabic m̩ .
John Cowan writes:
The form ɑmə began as AAVE basilect and has gone upmarket in the last
several decades, like a lot of AAVE basilect (it’s discussed in Dillard’s Black
English). It’s an open question whether this is nasal assimilation as you
explain, or derived from a Plantation Creole future/volitional particle a (‘I’m
a-quit’), or a mixture of both.
He says he has aɪmənə (not aɪŋənə) most of the time, and aɪmə occasionally. He
claims that aɪmənə is now a standard pronunciation, while aɪmgənə strikes him
as only suitable for formal or emphatic contexts.
With that, I’m’a declare this discussion closed.
2.41 terraced attacks
My summer course colleague Jack Windsor Lewis, listening to the

radio news, reports hearing a disturbing account of police discrimination
against less well-to-do homes. Apparently, a house in Forest Gate was raided
by anti-ˈterəst forces. The many British people who live in terraced houses
(for Americans, that’s row houses) rather than in semis or detached houses
might rightly feel aggrieved. (In case you haven’t caught up, this was meant
to be anti-terrorist.)
For some reason it reminded me of the Scotsman who took a week off work
because he had a wee cough. (If you’re teaching students about pre-fortis
clipping, this minimal pair makes a change from plump eye – plum pie and
might I – my tie.)
With aircraft cabin announcements, I often do a double-take when I hear
about ‘our co-chair partners’ (actually, code-share partners). Because the d in
code-share gets devoiced, the distinction between co-chair and code-share
depends on nothing more than (i) fortis versus lenis and (ii) syllabification,
both of which are pretty inaudible in this context: ˈkəʊ.tʃeə, ˈkəʊd.ʃeə.
Oh, and perhaps the difference between an affricate on the one hand and a
plosive plus a fricative on the other. Pre-fortis clipping is blocked by the
morpheme boundary.
2.43 mine and my 75
2.42 synthetic homographs

I’ve been doing a consultancy job for a publisher, listening to
synthesized speech automatically generated for a handheld computerized dic-
tionary currently under development – not isolated headwords, but phrases and
sentences. Speech synthesis is much better these days than it used to be, but it
still has some way to go, particularly in the matter of working out a plausible
intonation pattern.
I’ve listened to a thousand randomly selected samples. So I can tell you both
what the commonest type of error was, and what the worst type was. (How would
the software have coped with the sentence I’ve just written? It ought to have been
able to recognize the need for contrastive tonicity: both what the commonest type
of error was, | and what the worst type was. Would it have succeeded? No.)
OK, then: the worst type of error was the wrong choice between homographs.
English has a number of words where the same spelling is used for two words
of different sound and different meaning. Thus wind (air movement) is wɪnd,
but wind (turn, twist) is waɪnd; entrance (way in) is ˈentrəns but entrance
(bewitch) is ɪnˈtrɑːns; present (noun or adj.) is ˈprezənt, present (verb) is
priˈzent. In the thousand samples I listened to there were five errors of this
kind. Obviously, choice of the wrong pronunciation would be seriously mislead-
ing for the dictionary user. (See further 2.48 below.)
However, the most frequent error was failure on the part of the speech
synthesis software to recognize compound nouns and therefore to apply
compound stress. In the great majority of cases of noun þ noun in English
it is appropriate to put the primary stress on the first element, as for example
adˈventure story, appliˈcation form, ˈmurder charge. (Of course, you need to
be aware of the occasional exceptions, such as ˌstrawberry ˈjam, ˌgang ˈwarfare.)
The software got this wrong in 65 cases, constituting just over a quarter of all
errors identified.
You could see this, alternatively, as a demonstration of a fault in English
spelling conventions. In the other Germanic languages, compound nouns are
written solid, with no space (thus German Abenteuergeschichte, Anmeldeformular,
Mordanklage). So you can recognize them as such. English sometimes does
this (wintertime, railway) but very often doesn’t. If we wrote adventurestory,
applicationform, murdercharge and so on, then the software (and millions of
EFL students) might get the stress right.
2.43 mine and my

We all know the opening lines of The Battle Hymn of the Republic:
Mine eyes have seen the glory of the coming of the Lord; He is trampling out
the vintage where the grapes of wrath are stored. . .
Given that these words were written in 1861, we may well ask why it begins
Mine eyes rather than My eyes, which is what everyone writes and says today,
and indeed wrote and said in the Victorian era.
From the thirteenth to the eighteenth centuries, my maɪ and mine maɪn were in
complementary distribution as possessive adjectives before a noun. People said
and wrote mine if the following word began with a vowel and my otherwise.
(Before h, though, usage was divided.) This is the usage to be found throughout
Shakespeare (1564–1616), the Authorized Version of the Bible (1611), and the
Book of Common Prayer (1662).
Nowadays, as the OED puts it, mine qualifying a following noun is ‘only arch.
or poet. before a vowel or h; otherwise superseded by my’.
So Julia Howe, the writer of this hymn, was indulging in arch[aic] or poet[ic]
language, which I suppose is fair enough. No doubt she was also echoing the
words of Psalm 121 in the Prayer Book, I will lift up mine eyes unto the hills.
Other hymn writers of this period, though, didn’t hesitate to write my anchor and
my heart.
The OED adduces a quotation from Swift’s Gulliver’s Travels, dated 1726:
Till I had gotten a little below the level of mine eyes.
But that’s their latest citation for mine before a noun. Anything more recent is
indeed consciously archaic – either poetic (up to about a hundred years ago) or
religious, or else jocular, as when you call an innkeeper mine host.
Martin Barry of the University of Manchester, who besides being a phonet-
ician is also a cathedral choirmaster, reminds me of the fifteenth-century carol
with the refrain Lullay, mine liking, my dear son . . . mine own dear darling. Why
do we get mine before l and my before d?
2.44 um… er…
In a conference presentation I attended, Richard Cauldwell discussed

the factors making for fluency in the language of trainee air traffic controllers.
His aim was to isolate the factors that led candidates to be judged as relatively
fluent or non-fluent in their use of English. He found that to progress from level 3
(non-passing, ‘pre-operational’) to level 4 (passing, ‘operational’) you would
need to:
increase the average speed of your speech by 30% (from 100 to
130 wpm);
make your speech units longer (words and syllables);
increase the percentages of medium and fast speech units by 20%;
reduce silent pauses to 15% of duration;
use fillers to fill pauses.
2.45 thou hast, he hath 77
Your fillers should not be distracting. (The term ‘fillers’ here is applied not so much
to expressions such as you know or sort of as rather to hesitation noises.)
To progress to level 5 (‘extended’) or 6 (‘expert’) his data suggests that you
would need to further reduce silent pauses (to 10%), and to eliminate or nearly
eliminate ums and uhs.
Phonetically, um is most often əm. (To what extent do people say ʌm as use
rather than mention?) The variant that Cauldwell spells uh is more usually
written er in BrE, since it is phonetically like a long schwa, ɜː. To my way of
thinking, the spelling uh is appropriate only for American ʌ.
In BrE we also have erm ɜːm. Other languages have other habits. In Japanese
it’s anoo, in German äm or hm. We do say hm in English, too, (or hmm), but it is
not exactly a hesitation noise. As a longish m̥ mm, a voiceless then voiced
bilabial nasal with a falling tone, it shows doubt or disagreement, meaning
something like ‘I understand what you say, but I don’t think I agree with it’.
As m̥ m, with shorter duration and an abrupt fall, it shows mild surprise and
means much the same as ‘Oh!’ or ‘Well, well!’
Unfortunately there are many academics (and others) whose fillers are defin-
itely distracting and indeed annoying. I think public speakers, including lecturers
and conference presenters, ought to train themselves not to um and er.
2.45 thou hast, he hath
I was brought up as an Anglican and have therefore been familiar

since childhood with the Book of Common Prayer and the Authorized Version
of the Bible. A consequence of this is that the old present tense second and
third person singular verb forms (thou givest, he giveth) are part of my linguis-
tic competence – not something learned by study, but part of my everyday
knowledge of English. (They are, though, restricted to the liturgical, biblical
and archaic-literary registers of the language.)
I have come to realize that for many other native speakers of English this is
not the case. To begin with I was amazed to hear people stumbling over these
forms and coming up with nonsense such as thou giveth or I comest.
OK, if you’re not sure, the rule is:
-est only in the second person singular (thou), present tense;
-eth only in the third person singular (he, she, it or a singular noun),
present tense.
Here, from the New York Times Digest, 7 January 2008, is a comment by Paul
Krugman:
You see, for 30 years American politics has been dominated by a political
movement practicing Robin-Hood-in-reverse, giving unto those that hath
while taking from those who don’t.
No, it’s he/him that hath and those that have. Look.
For he that hath, to him shall be given: and he that hath not, from him shall
be taken even that which he hath. – Mark 4:25
Gather my saints together unto me; those that have made a covenant with me
by sacrifice. – Ps. 50:5
One consequence of having these forms as a core part of my knowledge of

English is that I also know that some of them have strong and weak forms in
pronunciation. For example, hath has the strong form hæθ and the weak form
(h)əθ. Similarly, hast has the strong form hæst and the weak form (h)əst. I find
that I use these as naturally as I do the strong and weak forms of has or have.
I know that (thou) sayest is ˈseɪ ɪst but (he) saith is seθ.
I know a further interesting fact: that for do there are doublets doest – dost
and doeth – doth. In each pair the first (ˈduːɪst, ˈduːɪθ) is used as a main verb, the
second (strong forms dʌst, dʌθ, weak forms dəst, dəθ) as an auxiliary.
This makes me realize (si parva licet componere magnis) what it must feel like
to be one of the last speakers of an endangered language.
2.46 rhotic
My own personal contribution to the English language is the word rhotic.

In 1968, inspired by reading Labov’s account of his fieldwork in NYC,
I carried out an afternoon’s fieldwork in Southampton asking random passers-
by on the street about possible ice cream flavours, thereby inducing them to
pronounce the words vanilla and ginger and recording the presence or absence of
final r-colouring in their pronunciation of each. It was in reporting my findings
that I classified the respondents as ‘non-rhotic’ (having no non-prevocalic
r-colouring) or ‘rhotic’ (having r-colouring in ginger even when not followed
by a vowel sound). The existing American term, r-ful, obviously sounded rather
awful if pronounced in an accent like mine.
In due course the word was taken up by other phoneticians and sociolinguists.
But now people are using my word as a noun, in a different sense, which
the 2010 edition of the OED has not altogether caught up with. This new sense is
‘a non-lateral liquid consonant’ (informally, then, an r-sound). Ladefoged and
Maddieson’s 1996 book The Sounds of the World’s Languages (Blackwell)
has a chapter entitled ‘Rhotics’. Back in his 1984 book, Patterns of Sounds
(Cambridge University Press), Ian Maddieson was still calling them r-sounds.
The earliest use of rhotic in this new sense that I can find is in R.M.W. Dixon,
The Grammar of Yidiɲ (Cambridge University Press 1977). He reports the
consonants of Yidiny as including:
a trilled apical rhotic – r;
and an apical-postalveolar (retroflex) rhotic continuant – ɽ [sic].
2.47 progressive assimilation 79
The term subsequently became established in descriptions of other Australian

languages (which typically have two rhotics), and from there was taken up more
generally. Bob Dixon was my colleague at UCL in the 1960s, and no doubt picked
up the word from me there, although he then proceeded to use it in a different sense.
There is uncertainty in some quarters as to how to pronounce the word.
I pronounce it with the GOAT vowel, though some use the vowel of LOT.
You can see their reasoning. After all, -otic has the LOT vowel in antibiotic,
asymptotic, demotic, erotic, exotic, hypnotic, narcotic, necrotic, neurotic,
osmotic, patriotic, quixotic, sclerotic and semiotic.
Note, however, that (with the exception of the maverick Cervantes-hero-
derived quixotic) these words all involve the ending -(o)tic attached to what
etymologically was a separate morpheme. Thus demotic includes the same Greek
stem as democracy and demagogue, neurotic the same as neuralgia.
But rhotic is not derived from a stem *rh- plus -otic. Rather, it is related to
rhō, stem rhōt-, the name of the Greek letter Ρ, ρ (¼ r). The name of this letter
also lies behind rhotacism and rhotacize, which certainly have the GOAT vowel.
Hence my pronunciation ˈrəʊtɪk. For what it’s worth, all the pronunciation
dictionaries agree with me, as does the Concise Oxford.
Plus. . . it avoids the awkward potential homophonic clash with erotic.
2.47 progressive assimilation

An enthusiastic but non-native-speaking teacher of English phonetics
offered the following phonetic version of if you can understand what’s written
here:
ɪf jə kŋ ʌndəˈstæm wɒts ˈrɪtn hɪə
What’s wrong with this?

Personally, I normally keep you (weak form) as ju, rather than reducing it to jə. But
that’s not the point. There are plenty of speakers who would reduce you in that way.
There is a fairly drastic reduction of understand: elision of one stress and one
consonant, assimilation of the newly final consonant. But that’s OK.
No, it’s the word can. Although we can say kŋ in contexts such as you can go,
they can be, we can do, we absolutely cannot do so in you can ask, they can
always, we can only.
So why not? What’s the difference?
It’s the phonetic context. This kind of assimilation can take place only phrase-
finally or before a consonant (go, be, do), never before a vowel (ask, always, only).
Progressive assimilation (aka perseverative assimilation or lagging assimila-
tion) is mentioned in most books about English phonetics, with such examples as
ribbon ˈrɪbən ! ˈrɪbm, bacon ˈbeɪkən ! ˈbeɪkŋ. It is distinguished from
ordinary regressive (anticipatory, leading) dealveolar assimilation of the familiar
tem bɔɪz, teŋ gɜːlz type.
There is a constraint that the assimilation does not happen before a vowel. As
far as I can recall, no one else has ever discussed this constraint. But it’s real
enough: you can say ˈəʊpm before a pause, or before a consonant as in opened,
open them – but not before a vowel as in opening or open it. The constraint holds
even if you change the p to ʔ, as we often do.
Syllabic consonant formation, which changes schwa plus a nasal or liquid
into the corresponding syllabic nasal or liquid, is an absolute precondition to
this type of assimilation. The only possible input to the progressive assimilation
rule is a syllabic n̩ . The point is that if a syllabic nasal immediately follows a
plosive, then there may be assimilation of place to that of the plosive.
n̩ ! m / {p, b} _
n̩ ! ŋ / {k, g} _
Thus from ˈəʊpən open, syllabic consonant formation gives us ˈəʊpn̩ . Then
by progressive assimilation we derive ˈəʊpm̩ . Similarly, for ribbon we have
ˈrɪbən ! ˈrɪbn̩ ! ˈrɪbm̩ , for bacon we have ˈbeɪkən ! ˈbeɪkn̩ ! ˈbeɪkŋ̩ ,
while for organ we have ˈɔːgən ! ˈɔːgn̩ ! ˈɔːgŋ̩ .
Naturally, if the schwa remained it would prevent the nasal from directly
abutting the plosive, so no assimilation could occur.
Secondly, if the nasal is also followed by a consonant then it will be a
candidate for regressive assimilation as well. Sometimes these two assimilatory
pressures reinforce one another, as when steak and kidney becomes ˈsteɪkŋ
ˈkɪdni, organ grinder becomes ˈɔːgŋgraɪndə, or carbon paper becomes
ˈkɑːbmpeɪpə. But sometimes they pull in different directions. In steak and
mushroom (pie) the k of steak tries to make the n̩ (¼ and) velar, while the initial
m of mushroom tries to make it bilabial. I think it was Francis Nolan who
showed that often in such cases both things happen simultaneously, so that we
end up with a velar-bilabial, or indeed velar-alveolar-bilabial, nasal.
Thirdly, in AmE people do not make these assimilations nearly as readily as
we Brits do, or perhaps at all. Nor do they assimilate to m before w (ten ways
tem weɪz) as Brits do.
Fourthly, the corresponding assimilation clearly happens in German. For haben
Sie? you will often hear Germans saying ˈhaːbmzi (or indeed ˈhaːmmzi, with
additional regressive assimilation of nasality). But in German the English constraint
of not-before-a-vowel doesn’t apply. Thus wir haben auch. . . can be viɐˈhaːbm
ˈʔaʊx. It may be that the glottal hard attack of word-initial vowels is what makes this
possible (though not all kinds of German apply this automatic glottal stop).
2.48 homographs
We have a number of cases in English of homographs: pairs of words

spelt identically but pronounced differently. Looked at from the point of view
of text-to-speech rules (which is the way non-phoneticians almost always do
2.48 homographs 81
look at them) they must seem very puzzling, since by definition you cannot
predict the pronunciation from the spelling.
So, as we all know, wind is wɪnd if it is moving air, but waɪnd if it is to turn
or twist. Entrance (way in) is ˈentrəns but entrance (bewitch) is ɪnˈtrɑːns
(or ɪnˈtræns, depending on accent). Present (noun or adj.) is ˈprezənt, present
(verb) is priˈzent. (See also 2.42 above.) Tear (eye water) is tɪə, but tear (rip) is
teə; sow (female pig) is saʊ, but sow (seeds) is səʊ. And so on. Text-to-speech
algorithms may need quite sophisticated ploys to achieve the right result in
such cases.
Very awkwardly for text-to-speech and also for the EFL learner, the words
use, excuse, close and several others have s or z in accordance with rules that
are not always easy to formulate.
I have noticed people doing odd things with the endings -ate and -ment. The
normal rule is for these endings to have a strong vowel in the verb, but a weak
vowel in the related noun or adjective. So we have separ[eɪ]ted but separ[ə]tely,
complim[e]nted but my complim[ə]nts. Every now and again, though, I hear
native speakers, including radio and TV announcers, getting these wrong or at
least violating the usual rule. I suppose that in a literate society such as our own
it is difficult to maintain lexicophonetic distinctions that are not supported by
the spelling. But I find it rather disturbing when someone says they want to
advoc[ə]te or implem[ə]nt this or that policy.
3 Teaching and examining
Teaching and examining phonetics, including
general phonetics and EFL
3.1 deceptive strings
In some phonetics oral exams we test the candidates on reading aloud

short strings of phonetic symbols. Some of these are real English words – dɪˈnaɪ,
ˈpɜːpəs and the like. Others look like English but are not, since they are not
spellings but phonetic symbols, e.g. ˈreply, in which the last vowel is to be
interpreted IPA-style as a close front rounded vowel; or thumb, where we need
[t] followed by [h], a close back vowel, [m] and a final [b]. Candidates find these
sequences remarkably difficult to perform correctly.
3.2 practical tests

The practical test we used to give our MA Phonetics students at UCL
covered transcription, dictation and the oral component (performance and
recognition).
The transcription passage was a colloquial dialogue to be phonetically tran-
scribed and marked up for intonation. Here is a sample passage.
A Oh, hi there! I was wondering if I could have a word with you.
B Yes, of course. What is it?
A It’s about that Birmingham job I mentioned. I’ve finally decided to go ahead and
apply for it.
B Well, good luck then. But what d’you want me to do about it?
A I was hoping you might give me a reference, or at least put in a good word for me.
B OK. You’ll have to tell me what you want me to say, though. I’m not very well informed
about your career to date, you know.
A Oh, I’ll cover all that in my c.v. What I need from you is more a character reference,
saying that I’m trustworthy and honest and so on.
B All right, then. Give my name to the recruitment agency and get them to write to me.
I hope you get the job, though of course I’ll be sad to see you go.
Most of our non-native-speaking students did pretty well on this exercise the year
it was set. Nevertheless, it is instructive to analyse the most frequent errors
they made.
Words frequently transcribed wrongly included then as ðən instead of ðen
(this word has no weak form) and trustworthy with θ instead of ð. We wouldn’t
82
3.4 exotic sounds 83
penalize Birmingham with -ŋɡ-, given that that is how the locals pronounce it.
We also allowed of course with əf instead of əv, although the latter is surely the
usual form, voicing assimilation across a word boundary being unusual.
In the intonation markup, the most frequent errors were a failure to identify
compound stress in ˈBirmingham job, ˈcharacter reference, reˈcruitment agency;
failure to accent ahead; failure to deaccent I mentioned and you know. Only three
of the eight candidates managed to locate the nucleus correctly in what is it; only
two got the most plausible nucleus placement in what d’you want me to do about it.
3.3 faulty abbreviations

‘Final alveolar consonants, i.e. /n/, are often subject to assimilation’,
writes an undergraduate in an examination answer.
But she doesn’t mean i.e., she means e.g. The first stands for ‘that is’, the
second stands for ‘for example’. ‘The consonant /n/’ is not the same as ‘(all)
alveolar consonants’; it is an instance of an alveolar consonant.
Younger native speakers often confuse i.e. and e.g. Both these abbreviations
stand for Latin phrases, id est and exempli gratia respectively. You could
blame their confusion on the decline of classical learning: not many people
learn Latin these days. On the other hand there are millions of people who don’t
know Latin but who manage to use these abbreviations correctly.
You might also think that we ought to have English abbreviations for English
phrases. Foreign learners, in fact, sometimes suppose that to be the case. I have
several times come across ‘f.e.’ or ‘f.ex.’ in English essays written by Germans.
After all, the German equivalent of for example is zum Beispiel, abbreviated as
z.B. I have to explain that ‘e.g.’ is the only way of abbreviating this expression
in English.
Another English abbreviation I have seen invented by Germans is ‘a.s.o.’
What does it stand for? Obviously, ‘and so on’ (German und so weiter, u.s.w.).
But the English for that is etc. It’s Latin again: et cetera.
A popular (mis)pronunciation of this latter phrase is ɪkˈsetrə, as if it began ex-.
3.4 exotic sounds
Courses in phonetics often include a practical examination that

requires the candidates to be familiar with ‘all the sounds represented on the
IPA chart’ (or words to that effect). They are expected to be able to recognize
these sound-types in nonsense words or substitutions and to perform them in
isolation or in simple sequences.
However, in practice certain sound-types tend to be excluded, whether expli-
citly or implicitly. For example, with speech therapy students we usually require
only a subset of the cardinal vowels: all the primaries i e ɛ a ɑ ɔ o u, yes, but
84 teaching and examining
among the secondaries only y ø œ ɯ. We don’t bother with ɶ, with cardinal ɒ ʌ

or with ɤ. On the other hand the MA Phonetics students, and candidates for the
IPA exam, do have to cover these sounds.
The same applies to some of the consonants in the chart. Whereas everyone
has to be familiar with those such as ɸ ʂ ɣ q, only advanced students would
be expected to cope with the pharyngeals ħ ʕ. As for the epiglottals ʜ ʢ ʡ, I don’t
think we require anyone to master them for examination purposes.
I remember being surprised, many years ago, to find that students at another
university with a well-regarded phonetics course were not taught the
alveolopalatals ɕ ʑ tɕ dʑ. We certainly cover them in London, not least because
of having to deal with Polish and Mandarin Chinese, in both of which languages
alveolopalatals contrast with (somewhat retroflex) ʃ ʒ tʃ dʒ.
3.5 the Swedish sj-sound
A correspondent expressed dismay at the ‘defeatist attitude’ he found

in an elementary textbook of Swedish when it came to the distinctive Swedish
consonant ɧ (the ‘sj-lud’). It merely described it as ‘similar to sh in she . . . [but]
difficult for foreigners to produce’.
In its section on Swedish, the IPA Handbook (Cambridge University Press
1999, p. 140) calls ɧ a ‘voiceless dorso-palatal/velar fricative’. In the list of
symbols (p. 167), it gives the phonetic value of ‘hooktop heng’ slightly differ-
ently, as ‘simultaneous voiceless postalveolar and velar fricative’.
To this I would add that in my experience this articulation is often accom-
panied by lip protrusion. The overall effect is of a voiceless fricative with no very
determinate place of articulation but rather a long channel constriction.
My advice to non-Swedes would be to start with a velar fricative x, add a
simultaneous darkish ʃ, and protrude and round the lips.
First listen to a recording of sju, which means ‘seven’, and then practise
saying it.
There is an entire book (in Swedish) devoted to this consonant (Lindblad,
P. 1980. Svenskans sje- och tje-ljud i ett allmänfonetiskt perspektiv). The author
describes two different variants of the sound. And that’s not counting the
retroflex ʂ that a few speakers use instead (and which of course most speakers
use in words spelt rs).
Olle Kjellin, author of several books on teaching Swedish pronunciation to
foreigners, sent a helpful comment.
As for the kind of lip rounding, it has the same lip shape as the long ‘o’
sound (uː). I tell my learners to strongly whisper a long ‘o’, as if in a big
theatre with a thousand spectators.
A peculiar but important characteristic of the sj sound is that it does not get
coarticulated with the vowel that follows, but should keep its o-like lip shape
3.6 pausing a problem 85
at all times, even before front unrounded vowels. I have noticed the same
thing about the English sh sound, as in she, although your twitch of the
Orbicularis oris muscle is much slighter than ours.
I choose practice words with o following, e.g., skjorta ‘shirt’, journalist,
lektion lekˈɧuːn and all those -tion/-sion words. Only after extensive prac-
tice do we turn to skina ‘shine’, giraff, själv ‘self’, etc.: ˈɧiːna, ɧiˈrafː, ɧɛlːv.
Then finally we attempt the irritating word sju ɧʉ ‘seven’, which involves
moving the tongue from back to front (i-like) position but without changing
the lip shape even the slightest. That’s the most difficult word, in my opinion
and experience.
This lip shape, for me, is almost the same as for whistling. So, if I whistle the
uː and ɧ sound, it will be the lowest bass whistle I can manage to whistle at
all. Whereas the hus ʉː whistle will be about 800 Hz. Or preferably 1600 Hz,
which is closer to its F2 frequency. The lower octave generally seems to be
sufficient. It’s quite fun and illustrative to perform this whistling glissando
to demonstrate the ‘true’ pronunciation of sju.
Whistling the lowest note one can is also a good way of achieving cardinal 8 u.
3.6 pausing a problem

It is natural that EFL learners who have no such distinction in their
native language can find it tricky to hear the difference between the GOAT
vowel əʊ and the THOUGHT vowel ɔː.
What surprises me, though, is that even advanced learners may sometimes
be unaware of the relationship between spelling and sound. This became clear
as I was doing a guest ear-training session for the UCL summer course.
My phonetic dictation included the word pose. One of the class members
wrote it as pɔːz. She had, however, correctly identified the word and its
meaning, and so knew that it was spelt pose. When I pointed out that a word
with the spelling -oCe (where C is any consonant letter) would surely not be
pronounced with ɔː, she reacted in surprise. No one had ever told her that.
Yet this is a very simple rule: nose, wrote, hope, alone, code, spoke and many
many others all have əʊ. (Words spelt -ore are of course an exception: more,
store, core. But they are the only cases where oCe corresponds to ɔː.)
Other than to or(e), the vowel ɔː typically corresponds to the spellings au,
aw: author, caught, law, drawn. The spellings corresponding to pɔːz are
pause, paws (and in accents like RP pours, pores). No one had ever told
her that, either.
There are several very common words in which the spelling oCe corresponds
to some vowel other than əʊ, as for example come, love, some, gone, move. This
is one of the particularly poor design features of our traditional orthography.
Nevertheless, poorly designed though it is, English spelling does offer some
pretty reliable reading rules. Learners ought to be aware of them.
3.7 volcano
Parts of Montserrat, the island in the Caribbean, are threatened by

an active volcano. (Contrary to what some people would have you believe, life
still goes on in the ‘safe zone’ in the north of the island, where some five
thousand people continue to live happily and safely.)
The Montserrat Volcano Observatory staff make regular reports on the current
volcanic situation, both written and spoken, the latter on the local radio station
ZJB. Not all the volcanologists and other experts are native speakers of English.
One specialist whom I heard giving a spoken radio report recently is a Dr D.,
whose first language is Italian. Listening to him, I was struck by the fact that
each time he used the word volcano – a frequent and crucial word in his report,
as you might imagine – he pronounced it volc[æ]no.
You would think that as a volcanologist working in an English-speakng
environment he would have noticed by now that all his English-speaking col-
leagues, not to mention all the other half-billion or whatever native-English
speakers in the world, pronounce it volc[eɪ]no. He must have made a wrong
inference many years ago about the pronunciation of this word, on the basis
of the spelling, and somehow become deaf to all the spoken evidence around him
showing that he was wrong. (The error must be spelling-based, attributable to
the uncertainty over whether the letter a represents eɪ or æ. It is not a matter
of negative transfer from his native Italian, which would give what we would
hear as volc[ɑː]no.)
Does this sort of thing matter? Well, yes, it does. The effect of his mispronun-
ciation, on me at any rate, was to make me discount the value of what he had
to say. If he doesn’t register the abundant evidence about the pronunciation
3.9 perceptions of /æ/ 87
of this everyday word, why should we suppose that he pays proper attention
to the evidence on which he bases his scientific findings?
‘To seek to change someone’s pronunciation – whether of the L1 or of an
L2 – is to tamper with their self-image, and is thus unethical – morally wrong’
(D. Porter and S. Garvin, ‘Attitudes to pronunciation in EFL’, Speak Out! 1989,
5:8–15). What rubbish! If some kind person were to teach Dr D. the correct
pronunciation of volcano, they would be doing him a positive and useful service
that would enhance his scientific credibility and thus his professional standing.
3.8 suddenly in Sydney

As part of a phonetic workshop I conducted at the University of
Kōchi there was a question-and-answer session. There were seventy or eighty
people in the audience, a mixture of students and schoolteachers, with a sprink-
ling of academics, and we asked them to submit their questions in writing, giving
plenty of advance warning. The result was a surprisingly large number of
questions, which ranged from the narrowly phonetic (‘What is the difference
between ˈsʌdn̩ li and ˈsʌdənli?’) to the general and personal (‘Do you speak
Japanese?’, ‘How many people do you have in your family?’).
The question about the transcription of suddenly gave me an opportunity to
talk briefly about nasal release and also about the abbreviatory convention I use
in the LPD, where the word is transcribed ˈsʌd ən li.
Japanese learners of English find nasal release very difficult. (They find lateral
release even more difficult, for obvious reasons.) The Japanese version of Sydney
is シドニー (shidonii); for them to pronounce it as two syllables, ˈsɪdni, with
nothing between the d and the n, as native speakers of English do, presents a
real problem.
Although I know a few words and phrases, and have plenty of theoretical
and practical knowledge about Japanese pronunciation, the answer to whether
I speak Japanese is a regretful ‘No’.
3.9 perceptions of /æ/
Some weeks ago a German colleague, an excellent linguist but not

a specialist phonetician, was telling me that he and his wife had taken a short
holiday-cum-English language course in the Republic of Ireland. They were
delighted with the holiday aspect, but he was worried about whether Irish English
was a suitable model for learners such as himself to be exposed to. In particular,
he complained, the Irish EFL teachers mispronounced – as he saw it – the English
æ vowel of that bad man, making it [a] instead of the [ɛ] he was expecting.
I had to tell him that (i) on balance I think it is a good thing, rather than a bad
thing, for EFL learners to be exposed to more than one variety of English; and
(ii) if he is concerned about deviation from the RP standard, the things that might
well worry him about Irish English could include their use of plosives for θ, ð
and of a fricative for intervocalic coda t, but not the quality of their æ, which
is well within the range of variability to be found within England, let alone
the remainder of the British Isles.
I think the problem is really the institutionalized perception in central and
eastern Europe that English /æ/ (the TRAP vowel) is to be mapped onto a variety
of the local short /e/ (phonetically [ɛ]) and not onto the local short /a/. In German,
the English-derived loanword Flashback is pronounced ˈflɛʃbɛk. Polish has
borrowed flash as flesz, Russian as флеш, Serbian as fleš. All too often, this
leads to a failure by Germans and east Europeans to make any distinction in
English between the vowels of pan and pen, bat and bet, or indeed flash and
flesh. (Speakers of Spanish and Japanese, on the other hand, map English /æ/
onto their own /a/, and may therefore fail to distinguish pan and pun, bat and
butt, flash and flush.)
The east European treatment of /æ/ rather shocks us native speakers of
English – British ones, at any rate. We think of our TRAP vowel not as a kind
of [e], but as a kind of [a]. After all, we call Málaga ˈmæləgə, Hamburg
ˈhæmbɜːg, Poznań ˈpɒznæn and Novi Sad ˌnɒvi ˈsæd. We expect the equiva-
lence to work the other way round, too.
3.10 misled by spelling
A correspondent tells me of an Italian lady of a certain age who is

educated, cultured, intelligent, quite refined, self-aware and quite ready to
criticize other people’s mistakes. But when she speaks English – although her
syntax is flawless – she has ɒ (the LOT vowel) in London and money. These
are words she must have heard pronounced correctly with the STRUT vowel a
million times! How can it be that she makes this mistake?
How indeed? I blame our crazy English spelling.
3.11 unaspirated /p t k/
Every now and again I get emails from Chinese correspondents

worried about the statement found in textbooks to the effect that English voice-
less plosives, usually aspirated, are nevertheless unaspirated after s, as in spin,
steam, skin.
The statement is correct. There is an audible difference between the bilabial
plosive in pin and that in spin. In the first it is aspirated, in the second it is
unaspirated. This means that in the first there is a delay after the release of
the bilabial closure before the onset of voicing for the vowel (the duration of
this delay being known as ‘voice onset time’): [pʰɪn]. In the second there is no
3.12 Yokohama and the like 89
such delay: [sp˭ɪn]. (The basic IPA offers no diacritic to mean ‘unaspirated’;
I show it by using a raised equals sign, part of the extension ExtIPA.)
It is difficult to hear the difference between unaspirated [p˭ t˭ k˭] and devoiced
[b̥ d̥ ɡ̊ ]. Theoretically it depends on the difference between fortis articulation, for
[p˭ t˭ k˭], and lenis articulation, for [b̥ d̥ ɡ̊ ]. But fortis-lenis is much easier to hear
in fricatives and affricates, where it affects the volume of air flow and therefore
the intensity of the turbulence, than it is in plosives.
Whereas we like to think of the English plosives as being basically voiceless
vs. voiced (despite positional devoicing), it is clear that the Chinese ones are
aspirated vs. unaspirated. Their representation in Hanyu Pinyin may be p, t, k vs.
b, d, g, but their phonetic representation is usually reckoned to be /pʰ tʰ kʰ/ vs.
/p t k/. Mapping these onto English /p t k/ vs. /b d g/ respectively works well,
except after s. In that position Chinese EFL learners should be encouraged to
use the sound that they usually map onto the English voiced plosive.
It may be interesting to note that when English words are borrowed into
Welsh (which, like Chinese, has basically aspirated vs. unaspirated plosives),
spite becomes sbeit, spell becomes sbel, (di)scourse becomes sgwrs, and screw
becomes sgriw. But Welsh is inconsistent: for some reason, studio is stiwdio,
not *sdiwdio, and station is stesion, not *sdesion.
3.12 Yokohama and the like

Languages differ in the syllable structures that they allow. We can
distinguish (i) languages such as English or French, that allow both (a) consonant
clusters and (b) one or more final consonants (a coda) in a syllable; and
(ii) languages such as Hawai‘ian, that do not allow them.
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 ¼
[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 compli-
cated 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 ¼ [N]
or [Q]. (The Japanese affricates romanized as ch, j and ts count as single
consonants.) It’s true, too, that Japanese i and u are devoiced or deleted in
certain phonetic contexts, thus giving rise to phonetic (but not phonological)
consonant clusters and coda consonants. So sukiyaki sounds as if it begins ski-,
ohayō gozaimasu ‘good morning!’ seems to end -mas and hajimemash(i)te
‘how do you do?’ to have a cluster ɕt. But these are superficial details of
allophonic realization.
Anyhow, 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.
Although I started this by talking about syllable structure, 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 Japanese word for ‘bread’, pan pa-N, has two moras, though we interpret
that as one syllable. 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 used for Japanese,
in which there is one symbol per mora.
The devoicing or elision of certain high vowels does not affect the mora
count. So hajimemash(i)te still comprises six moras despite the disappear-
ance 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
3.13 struggling to teach phonetics 91
making a syllable with it. That is why we still need to talk about syllables in
Japanese. Moras alone are not enough.
3.13 struggling to teach phonetics
I hesitated for some time before finally writing this piece (in
December 2007).
A young man wrote to me recently. He is a student at a university which had
better remain nameless. He is enthusiastic about phonetics, and has taught
himself a good deal about the subject. He had already sent me a number of
well-informed comments on various matters.
But he had never studied phonetics formally – till now.
One of the modules in my third level course [he said] is linguistics. Today
we began with phonetics. [. . .] I was eagerly anticipating my first official
lecture on the topic, when I could experience the know-how of an ‘expert’.
I was shocked and dismayed at the misinformation propagated by my
lecturer. I was spoilt for choice, but here are a few examples:
in the list [p, b, t, d, m, n], she said that only [t] was voiceless;
after introducing English r under the heading ‘place of articula-
tion’, she said that the English phonetic realization of /r/ was
one of the ways that the language was determined to be Germanic
rather than Indo-European;
she described forwards slashes // as those used for phonetic
(rather than phonological) transcriptions;
she said that the Spanish phoneme /j/ was glottal (she meant
orthographic j, which would be correctly transcribed /x/) [in
fact of course in standard Castilian it is velar, or at most uvular];
she said that the difference between English accents is abso-
lutely never in consonants, always in vowels;
she said the difference between Australian and New Zealand
English is slight (true), and that the differences concern a and e
(I’ve no idea what that means, TRAP and DRESS, or FACE
and FLEECE? Anyway it doesn’t matter, because needless
to say the main difference is in the KIT vowel);
she transcribed tree as /tRi/, dubbing it ‘phonetic’.
You might wonder why I go to the trouble of listing these, but when one
feels passionately for a subject, one wants to see justice done to it. I just
cannot understand how somebody with such a poor understanding of a
subject could be allowed to teach it.
What can one say? I did a little web research. The university in question states, in
the prospectus for the BA in question, that the programme ‘provides a foundation
in [. . .] Linguistics’. The teacher in question had recently completed a PhD in a
branch of linguistics. Presumably, as a junior post-doc, she was told by the head
of department that she must teach an introductory linguistics course covering
phonetics. No doubt she has been struggling to do her best.
My correspondent says he is dismayed ‘that anyone can get a PhD in any
branch of linguistics without knowing about basic linguistic concepts such as
competence/performance and grammaticality, and without knowing even elem-
entary phonetics’.
All I could hope was that in due course the external examiner would realize
what was going on, and flag it up. Or even the Head of Department.
But possibly the Head of Department did not know much phonetics and
linguistics either.
3.14 accents and actors
Long ago, when I was young and unknown, I spent a lot of time
looking into material I might draw on for the work that eventually became
Accents of English (Cambridge University Press 1982). As well as pursuing
all the written stuff I could lay my hands on, I enrolled in an evening class
in Accents and Dialects, taught by a drama teacher whose name I have
forgotten.
Each of the weekly three-hour sessions began with extensive warming-up
exercises (going bɑː bɑː bɑː, biː biː biː, buː buː buː and the like). We then
had a little bit of theory, and a coffee break. The second half was taken up by
play-reading to practise what we had learnt (or what the teacher could demon-
strate, or what we could do anyhow).
The ‘theory’ consisted mainly in learning the IPA symbols for RP.
People in the class obviously enjoyed it, and had an opportunity to imitate
the speech varieties demonstrated to them. But for my own purposes I concluded
that on this evidence the speech and drama people couldn’t offer me anything
of serious phonetic interest.
A recent article in the Guardian described a new CD set about English
accents, the work of a former tutor of speech and drama at RADA and a well-
known actress. Reading it, I saw that the speech and drama world is still a land
of fantasy, not facts. The author’s ‘environmental theory of accent formation’
attributes ‘the directness of Geordie’ to the way that ‘in the north-east, the wind
whips off the North Sea and smacks you in the face’, while she tells an actor
struggling with RP to ‘imagine walking across a manicured lawn in England’;
and for south Wales, she imagines the intonation ‘rising to the tops of mountains
and then descending to the bottoms of valleys’. Her advice to someone trying to
sound Welsh is ‘When in doubt, sing,’ while remembering that for some vowels,
e.g. in the word here, ‘the tongue forms a curled, daffodil shape’. As for AmE, its
‘colourful r sound’ calls for ‘speedy contraction in the tongue’; the actor should
‘imagine riding a horse at full gallop and pulling back hard and short on the
3.14 accents and actors 93
reins’. The letters p and b, she suggests, ‘are softened by tapping the tongue
lightly on the bottom teeth’.
I wish someone could persuade the drama schools of the value of real phonetic
knowledge. Can’t they see what nonsense it is to talk of tapping the tongue
on the bottom teeth in order to make an American p-sound (or rather, letter)? Or
curling the tongue into a daffodil shape?
We lucky people who know phonetics can simply say that in parts of south
Wales here is pronounced jœː. End of story.
Eric Armstrong, a voice, speech and dialect coach now working in Canada
as Associate Professor of Voice at York University, supplied a thoughtful and
valuable reaction to my rant.
I was pleased to see your post on Actors and Accents. I’m not in the least
surprised by your point of view. Far too many of my colleagues haven’t
enough knowledge to teach basic phonetics effectively, and essentially pass
on what they were taught, continuing a tradition of ignorance that is decades
if not centuries old.
‘The speech and drama world is still a land of fantasy, not facts.’
This is true, and perhaps to be expected. Actors are creative artists, not
scientists. Most voice and speech trainers come to their work from the world
of the theatre, not from the world of science, and their clients are uneasy with
the world of science. In fact, many actors have made it clear that the reason
they were drawn to their art was because of their disdain for science and
math. To them, phonetics seems an odd combination of science and math-
like symbols, things that they have failed at in the past, and notions that
make them uneasy, fearful and angry.
As I’m sure you’re well aware, there are many types of learners in the world.
We aren’t all blessed with ‘good ears’ – so few seem to be auditory learners.
I find that many actors are kinesthetic learners more than auditory ones, but
there are also visual learners. The world of actor training is immersed in
metaphor and imagery, as is the world of the professional actor. This is the
language of art, not science, and it is used in an uninhibited manner –
splashing around in the pool of metaphor – in order to reach a goal that
may be beyond a performer who has a resistance to more linear thinking,
rational explanation and concrete examples. Many high level actors work in
an almost child-like sense of wonder, playing their way into new roles, with
new physicalities and vocalizations. Short attention spans are, unfortunately,
the norm. There are some who clearly have great ears, brilliant minds, agile
tongues, and open hearts. As a colleague of mine once retorted, ‘We used to
call that “talent”’.
Part of the problem the established voice and speech community faces in
trying to embrace greater phonetic detail and more accurate usage of the
tools available to us from the worlds of linguistics arises, I believe, from
deeply held beliefs about the systems and methodologies in which people
have been trained. When new data is presented to these trainers revealing
that their teaching methods are ineffective, cognitive dissonance, uncomfort-

able feelings arising from having two conflicting ideas or beliefs, forces
them to justify what they have been doing in the past, and to devalue the new
information. Often the systems and methods learned by trainers are the life
work of their mentors, and so rejecting the concepts they learned means
rejecting what is, in many ways, a beloved parent figure for them.
I think there is hope for the future of the speech and drama world. Teachers
like Dudley Knight and Phil Thompson have found ways to bring greater
phonetic detail into the conversation while keeping an actorly playfulness in
their teaching. Their greatest innovation is in guiding the people they work
with to explore the world of sounds available. One small example: making
noises through gibberish play called ‘omnish’, an imaginary language of all
the world’s sounds. Actors love the kind of exploration where they do an
oration in Omnish, and another actor provides simultaneous translation. Yes,
to a scientist, it must sound silly. But silliness has great value to an actor! It
frees her up, lets her connect to those new sounds in a joyous, unfettered
manner, stripping away all those value judgments of sounds (and ultimately
symbols) as being ‘mere math’. And once she can hear those sounds in her
own mouth, to feel the physical action required, then visualize the action of
her articulators, she is ready to begin to learn how to write them down
systematically with IPA.
There is no defense of inaccurate metaphor, I think. But actors need metaphor,
and a trainer of actors is sorely lacking if they don’t have it in their arsenal.
3.15 minimal pairs in action
We were having dinner on a cruise ship. For dessert one of our party
had ordered cherry tart (the elegant Frenchified menu called it a clafoutis).
But the waiter, who was Thai, brought us what looked like Black Forest gateau
instead. We protested, and he took it away and brought us the correct order.
He explained that it wasn’t his fault. He had asked the kitchen for tart tɑːt.
But they had given him torte tɔːt.
Very few of the catering staff on the ship were native speakers of English.
This was English as a lingua franca in action. It failed.
(At least, I suppose we have to say it’s English. The words in question are
loanwords.)
Distinguishing between minimal pairs, in perception and in production, is not
just a classroom exercise. And don’t say that bothering about that sort of thing is
not important, and that the only thing that matters is communication, that the
context will sort things out. Sometimes communication breaks down precisely
because the context does not sort things out. Getting it wrong has a practical
consequence.
In this case, the result was inconvenienced customers and an indignant waiter
convinced it was not his fault.
3.17 Taiwan English 95
3.16 a French user of English

I attended a university public lecture on cognition. The speaker’s
native language was French, but he spoke English fluently and fully intelligibly.
He had a number of typical French (mis)pronunciations: uvular r, no contrast
between ɔː and əʊ (law – low), occasional spelling-based oddities such as ˈɒðə
other, infɔːmation, ripresent, together with word stress errors such as chaˈracterize,
satisˈfy. But most word stresses were OK, and he had a good robust contrast in
iː–ɪ with its high functional load.
This was a good demonstration of the point that in EFL you can get away
with quite a few pronunciation ‘errors’ as long as your speech overall is fluent,
with good rhythm and acceptable intonation.
3.17 Taiwan English

Many thanks to Karen Chung for allowing me to reproduce this
transcription of ‘Taiwan English’ – English as it might typically be pronounced
by someone from Taiwan. Bear in mind that schools in Taiwan are all supposed
to teach AmE, not BrE.
However, I think most of this would apply to English in much of mainland
China, too, where BrE is the norm.
The instruction Karen gives her students is ‘First, read the text out loud to
decipher it, then write it out in regular English orthography.’ Readers might like
to try their hand.
ɛs ˈkjuz mi iz ˈðis ˈsɪt ˈtæk ən

ˈnɔ is ˈnɑt ˈsæŋ kju nɑʃ ˈwe ðɚ ˈizən ɪt
ˈjɛs is ˈnɑs æ̃ ˈwɔm ɪs biŋ ˈkoʊ fɔ sɔ ˈmɛni deɪs lɪs ɪz ma ˈfʌst ˈtãĩ tu ˈdʒɔk
lɪs ˈmɑŋθ du ˈdʒu laɪ ˈʒɔk iŋ
ˈwɛl ˈnɑ ˈɹɪ li ˈju ʃu ˈtɹaɪ ˈɪt ˈdʒɔk iŋ is maɪ ˈlaɪ ˈluk æ maɪ ˈdʒɔɡ iŋ ˈʃus
ˈleɪ ɑ ˈmæt ĩ ˈiŋ ɡə lən æŋ aɪ ˈboʊ ˈlɛm iŋ ˈhoŋ ˈkon ˈleɪ ɑˈɹɪ li ɡʊt
ɪf ˈju wɑŋt tu ˈlʌŋ ˈxaʊ tu ˈðʒɔk ju ˈnɪt ə ˈɡu ˈpɛɹ ɑf ˈsus laɪ mãĩ
ˈwɛl aɪ ˈluk æ ˈlos mɛn ˈleɪ ɑ ˈɹɪ li s̩ ˈloʊ ˈleɪ ɑ ˈwok iŋ ˈnɔt ˈdʒɔk iŋ
ˈxu ɑ ˈju ˈtok iŋ əˈbaʊ də mæ̃ wɪf loŋ ˈxɛɹ hiz ˈxæt luks ˈfa ni
o læs ˈmaɪ ˈbɹɑ ðɚ o am ˈso ɹi aɪ ˈmin ðə ˈpʌ sən ˈnæs tu ˈxim lə ˈwʌ̃ ĩ lə
ˈstju pɪ ˈɡɹiŋ ˈswe tɚ æn bəˈlæk ˈpæs̃ ˈwɛl lə ˈmæ̃ ˈju a ˈtok iŋ əˈbaʊ is
ɑˈpi ˈʔi ˈtɪ tʃɚ æ̃ ˈbɑɪ lə ˈweɪ hiz ˈo soʊ ˈmaɪ ˈxɑs bɛnd
Notice how:
final consonants tend to be lost, particularly if the next word begins
with a consonant; if present, they may be wrongly voiceless;
consonant clusters tend to be simplified or broken up with an

intrusive vowel;
the voiceless dental fricative θ tends to be replaced by s or f, and
the voiced ð by l or d;
final nasals often have the wrong place of articulation, or nasalize the
preceding vowel and are then lost;
the vowels FLEECE-KIT, DRESS-TRAP-FACE and STRUT-
START are often confused;
and so on.
You would think the loss of final consonants would be particularly devastating
for intelligibility. However, this seems to be characteristic of some African-
American English, too, and the results are apparently not too catastrophic.
3.18 the IPA Certificate
The year 2008 marked the centenary of the setting up of the Inter-
national Phonetic Association’s Certificate of Proficiency in Phonetics.
Here is the exact text of the IPA Council’s decision establishing it
(m.f. 23:5–6, 1908, p. 67). It is in phonetically transcribed French.
eɡzamɛ̃ d fɔnetik. a l ynanimite, lə kɔ̃ːsɛːj s ɛ prɔnɔ̃ːse ɑ̃ favœːr d œ̃ sɛrtifika
d etyd fɔnetik, a delivre ɔfisjɛlmɑ̃ o nɔ̃ d l af par dez eɡzaminatœːr dymɑ̃
kalifje. kɔm prəmjez eɡzaminatœːr (a defo də Sweet ki avɛ rfyːze) ɔ̃t ete
nɔme :
pur l almɑ,̃ W. Viëtor;
pur l ɑː̃ ɡlɛ, E. Edwards;
pur lə frɑː̃ sɛ, P. Passy.
nu dɔnɔ̃ si-aprɛ lə mɔdɛl adɔpte.
So there were three Certificates: one for German, one for English and one for
French. On the following page the m.f. shows a model of the French certificate.
The various topics were:
Dictée phonétique française
Dictée en langue inconnue
Transcription
Lecture d’un texte phonétique
Lecture d’un texte orthographique
Questions théoriques
Prononciation en parlant.
Most of these topics remain in the examination to this day, though in recent years
only the certificate in English has been examined. We still have dictation of
English and of nonsense words (words in an unknown language), transcription,
3.19 dead letters 97
reading from a phonetic text and questions on theory. Only reading from an
orthographic text and ‘spoken pronunciation’ have been dropped (since many
candidates are now native speakers of English). The standard required has always
been uncompromisingly high.
(The 1908 examiner for English, Ernest Edwards, was appointed to teach
phonetics at UCL in 1903. It was he who recommended Daniel Jones as his
successor at UCL to carry on his work.)
The IPA examination was particularly important at a time when few univer-
sities held examinations in the phonetics of foreign languages or indeed of oral
proficiency in general. As this changed, the number of candidates fell. Latterly,
though, it has been given an entire new lease of life by the IPA Certificate strand
of the UCL Summer Course in English Phonetics, which culminates with sitting
the examination.
You can read about it and download sample questions at www.phon.ucl.ac.uk/
courses/ipaexam/ipa-exam.htm.
3.19 dead letters
From the IPA exam regulations:
In the dictation of nonsense words or an unknown language, candidates

should be prepared to recognise any sounds occurring in the IPA chart.
Let’s leave aside the interesting question of whether any sounds truly occur in
the chart, which on the face of it seems to betray a naïve confusion of speech
(sounds) and writing (symbols).
Instead, here is something candidates sitting this and other examinations in
practical phonetics might like to know.
There are a few symbols in the chart that you can ignore. You will not be
tested on them. They stand for sounds that you will not actually be expected
to recognize or reproduce.
Take, for example, the vowel ɞ. In the current IPA chart it is shown as
standing for an open-mid central vowel, the rounded equivalent of ɜ. But
rightly or wrongly it is simply not among the sounds we drill advanced
students on. (Rightly, I think: when did you last see this symbol in use, as
opposed to sitting in a chart?)
The same applies to ɘ, the symbol for an unrounded close-mid central vowel.
You will indeed be tested on the cardinal vowels i e ɛ a ɑ ɔ o u y ø œ ɒ ʌ ɤ ɯ,
though probably not on ɨ ʉ ɵ and certainly not on ɶ. (Although there are some
quite well-known languages for which one or other of these symbols is needed.)
Among the non-cardinal vowels the only ones you must know are the ɪ ʊ æ ɜ ə
that you need for RP English, plus the ɐ that you may know from the phonetics
of German or Portuguese.
As with vowels, so with consonants. You can ignore the ‘other symbols’
ɺ ɧ ʜ ʢ ʡ. Concentrate on mastering the recognition and production of the
consonants shown in the main table at the top of the chart, plus clicks, implosives
and ejectives.
Even in the main table I think you could probably ignore ʟ, the velar lateral.
3.20 alveolopalatals
The IPA chart includes ɕ and ʑ, the ‘alveolopalatal’ fricatives. Listed

under ‘other symbols’ on the chart, these are nevertheless two sounds that
the examination candidate needs to be familiar with.
There seems to be some confusion over the terms ‘postalveolar’, ‘alveolopa-
latal’ and ‘palatoalveolar’. I came across a claim that most (if not all) Australian
accents of English have alveolopalatals, whereas RP has postalveolars; and some
American textbooks call AmE tʃ and dʒ ‘alveolopalatal’.
I suspect that this may be just a difference of terminology, without any
significant difference in the sounds referred to. I can’t say I have ever noticed
much of a difference between Australian or American tʃ dʒ and my own.
Like many other phoneticians, I do not follow the post-Kiel IPA chart in
classifying tʃ dʒ as ‘postalveolar’. For me, the English affricates that are post-
alveolar are the ones in trace and drab, while those in chase and jab are
palatoalveolar.
How do alveolopalatals and palatoalveolars differ?
One important difference between ɕ ʑ tɕ dʑ and ʃ ʒ tʃ dʒ is that the fricatives
and affricates in the latter of these two sets of four sound ‘darker’ than those in
the former.
One of the exercises I used to do with students was to make them slide back
and forth along the continuum ɕ – ʃ – ʂ.
I think the important thing is the auditory/acoustic effect rather than the precise
configuration of the articulators (since speakers differ so much in the anatomy of
the mouth). For alveolopalatals you can also imitate those of Japanese, Chinese
or Polish.
We must also bear in mind that the symbols ɕ ʑ are only needed in languages
that have a phonemic contrast between them and a darker type of ʃ ʒ (e.g. Polish,
Mandarin Chinese). For languages where this is not the case (e.g. Japanese)
people often use just the symbols ʃ ʒ whatever the exact phonetic quality. This is
in line with the principle of simplicity in transcription.
Equally, you could write the postalveolars as ʃ ʒ tʃ dʒ in Polish and Chinese.
I usually do this for Polish sz rz (ż) cz dż, but for Chinese sh r ch zh it seems
better to write ʂ ʐ ʈʂʰ ʈʂ.
Indeed, the ʃ ʒ symbols can be used very flexibly. If you compare the Japanese
s(i), the English sh, the French ch and the Russian ш, you notice a gradual
change in the friction noise from clear (very palatal) to dark (not at all palatal).
3.21 BrE or AmE for TEFL? 99
Yet they can all reasonably be written ʃ in simple transcription of those lan-
guages. For comparative work you can choose as appropriate from ɕ ʃ ʂ or even,
following Ladefoged, ṣ.
I could add that in my experience Poles are not really happy about accepting
Chinese x q j as exact equivalents of Polish ɕ tɕ dʑ, while the Chinese in turn
are not sure about accepting the ś ć dź of Polish as the exact equivalents of
their own sounds. But the difference (perhaps mainly one of degree of voicing
and aspiration) is below the radar of what IPA oral examiners, for example, are
on the lookout for.
A Polish word including both types is cześć tʃeɕtɕ ‘hello’, for which you
can easily find audio clips on the web. For comparison, listen to Chinese xi ɕi
and shi ʂɨʵ.
3.21 BrE or AmE for TEFL?
A Dutch correspondent writes:

The Dutch have always been at the forefront as foreign learners of English
and RP has been the standard accent for EFL in the Netherlands for many
years. This is now coming under pressure from GA [General American]
(among others) and we are slowly being faced with the fact that we shall
have to let RP go, that is if we want any of our students to graduate.
This is a difficult issue, and one people often ask me about. Is RP (‘BrE’) still
the right pronunciation model for EFL, or ought it to be replaced by some form
of American or Lingua Franca English?
The first thing I usually say is that it is much more important to concentrate
on the common phonetic core of all kinds of English than to worry about the
minute details of this or that model. First any learner has to get a good grasp of
such matters as:
syllable structure – final consonants, consonant clusters;
word stress placement;
tonicity (nucleus placement) in intonation;
vowel weakening;
important phonemic oppositions that all native speakers make –
particularly those with a high functional load, such as p – f, r – l,
iː – ɪ, ɔː – əʊ/oʊ (pin – fin, write – light, seat – sit, law – low);
spelling-to-sound rules and the most important exceptions to them.
Only when those are all thoroughly under control should you start worrying
about whether to take British or American or some other kind of English as your
model. And let’s face it, for most learners of EFL they are not all thoroughly
under control.
Second, the ideal circumstances for learning to pronounce English are when
the learner has a close relationship with a single native speaker or a group
of native speakers. If you are lucky enough to be in that position, imitate what
you hear. Your native speaker friend or colleague can operate in any kind of
native-English-speaking environment, so if you are like them you will be able to
too. (In my time I have met a German who sounded as if he came from Essex, a
Japanese who sounded as if he was Texan and a Korean who sounded like an
upper-class Indian. I consider all three to be highly successful learners of English
pronunciation, even though none of them pronounced in exactly the way any
textbook prescribed.)
However, that does not altogether address the question put.
From Germany, Petr Rösel reports:
My experience with German students of English who come to my phonetics
classes at university is that they have [typically] been taught English at
secondary school for seven to nine years by different teachers; for example,
during the first two or three years their teacher spoke British English (with or
without a German accent), for the next three years their instructor favoured
American English and the rest of the time at secondary school they heard
British English again.
When they come to my diagnostic sessions I’m confronted with a hodge-
podge of accents: British English with patches of Americanisms and a
strong tinge of German spoken by one and the same student. And I am
not speaking of errors that consist in pronouncing some isolated word in
the accent they are not aiming at. On the contrary! They pronounce the
non-prevocalic r in some words and leave it out in others. They pronounce
stop with an open ɑ, but dog with an ɒ. They flap the t in city, but not in
letter etc., etc. If I were sarcastic I would call it International English. No,
those poor souls are to be pitied. They should not have been confronted
with this diversity during their first steps into the world of English and
they should have been warned and encouraged by their teachers not to
mix varieties. This is not to say that they should not be able to understand
different varieties!
Mainz University, where I teach English phonetics (among other things),
accepts both standard varieties, which is a double-edged thing (see above).
This is the main reason why I split my phonetics classes when it comes to
practising pronunciation. I hired a native speaker of American English as a
tutor. She takes care of those students aspiring to speak AmE and I look
after those who want to speak BrE. But, alas, this is just a drop in
the ocean.
. . .while another Dutch correspondent wrote:

Although BrE is still the norm taught at school, I think the average Dutch
student of English is exposed far more to AmE (in all varieties) than BrE, as
a result of American movies, TV series, pop music, etc. being predominant.
EFL learners should be exposed (passively) to a wide variety of types of English

even if their teachers, dictionaries and textbooks offer a single standardized model
3.23 excuse my excuse 101
for imitation. As native speakers we do admittedly sometimes have difficulty in

understanding an accent we haven’t encountered before, or even one we have.
We need time to ‘tune in’. And it’s interesting that New Zealanders – with less than
two centuries to create their own variety – should be right up there with Ulstermen
and American southerners in presenting a challenge to the rest of us.
3.22 spot the mistake

I’d been doing some exam marking. It was a phonetics written exam
in which candidates, among other tasks, had to convert a short English passage in
orthography into phonetic transcription.
Some candidates were native speakers, some were not (and none of them were
from north Wales). The examiners are extremely liberal in allowing all sorts of,
shall we say, less usual pronunciations. Native speakers, in particular, are
encouraged to transcribe their own accent (but must say what it is). Non-native
speakers normally attempt RP, but there too the examiners are very flexible.
Here, however, are some transcriptions offered by candidates but penalized by
the examiners. These must be wrong, and candidates ought to know that they are
wrong. Readers might like to check that they can see where and why they’re wrong.
Each is a single word, extracted from the passage in which it was embedded.
James dʒeɪms
laws lɔːwz
wrong wrɒŋ
talking ˈtaʊkɪŋ
magazine mægˈzɪŋ, mægˈzɪn, məgəˈziːn
managed ˈmænɪʒd, ˈmænɪdʒɪd
dangerous ˈdængərəs, ˈdeɪŋgrəz, ˈdeɪŋdʒərəs
think ðɪŋk
news njuːs
ignoring ˈɪgnɔːrɪŋ
conservative cənˈsəvətɪv
3.23 excuse my excuse
For an examiner, marking phonetic transcriptions of English from

dictation can be rather depressing.
In one test passage the first two words were excuse me, which I dictated in
a perfectly ordinary way, as ɪkˈskjuːz mi. Among the versions I was offered
by the examinees were ɪˈkjus me, ɪˈkjʊz ˈmiː, ˈıkskjuz mi and several cases of
ɪkˈskjuːs mi. Now these were proficient speakers of English, often professional
teachers of the language, students of phonetics. And this was from my very
clear, slow, repetitious dictation, so all they had to do was listen and transcribe.
Those who were not native speakers didn’t have to work out that excuse is a verb
here, not a noun, and therefore has final z, not s.
That’s assuming they actually knew about the s ~ z noun~verb alternation in
use, abuse, excuse, refuse, house. And that it doesn’t apply in ease, tease, phase,
pause, bruise, cruise, surprise (always z) or in base, case, promise (always s).
And that in advice~advise we have the same alternation but change the spelling.
And that in practice~practise both noun and verb have s, but we Brits change the
spelling. In choice~choose there is a vowel change in addition to the consonantal
alternation, reflected in spelling. And in close it’s mainly an adjective~verb
alternation kləʊs ~ kləʊz, as it is in diffuse and loose~lose luːs ~ luːz, this latter
pair with a spelling change. And used is juːzd when it means ‘employed’ but
juːst when it means ‘accustomed’. Ah, English!
In these dictation exams candidates tend to regard the nonsense words as a
big challenge, while assuming the English will be easy to do. In practice, almost
everyone does better on the nonsense than on the English.
Are you ready to be a phonetics examiner?
3.24 miscellanea
Here are some jokes for your EFL class (or your seven-year-old).
How do you weigh a whale?
– You go to a whale-weigh station!
What do you call a deer with no eyes?
– No idea!
And a deer with no eyes and no legs?
– Still no idea!
***
And here’s a nice mispronunciation recently heard from a BBC Radio 4 news-
reader. He wanted to refer to the newspaper City AM, given away free every
weekday morning to London commuters. Its name derives from City (¼ financial
centre) and AM (a.m. ¼ morning). But the newsreader didn’t realize this.
He called it ˌsɪti ˈæm.
4 Intonation
Assorted comments on English intonation
4.1 introduction
Intonation is best seen, in my opinion, as reflecting three types of

decision that speakers repeatedly make as they speak. They are how to break up
the material (chunking, the signalling of syntactic boundaries); what tones to use
(e.g. fall vs. rise, signalling certain grammatical functions as well as such things
as the speaker’s attitude to what they are saying); and where to place sentence
accents (particularly nuclear accents, mainly used to signal focus). We call these
systems tonality, tone and tonicity respectively. Informally, they are the three Ts.
In tonality the speaker has to decide how to break the stream of speech into
successive intonation phrases (IPs). By default, each clause will be made a
separate IP; but the speaker is free to make the IPs longer or shorter than this.
I think that the principles of chunking are probably pretty universal. No EFL
learner needs to be taught to make or hear the difference between (1) and (2) in
speech:
(1) I don’t know. (one IP)

(2) I don’t, | no. (two IPs)
Where it is relevant, I show IP boundaries with the mark ‘|’, as in example (2).
Clearly, tone varies wildly across languages and dialects, at least at the
superficial level. The question is what underlying regularities there may be.
We can dismiss obviously untrue claims such as that statements always have a
fall, while questions always have a rise. (Just think of Belfast and the west of
Scotland.) But there are other candidates for the status of universal that are worth
considering. How general is it that, as in English, wh-questions tend to have the
same pattern as statements, while yes-no questions are different? Pretty wide-
spread, I think.
When discussing tone in English, it is enough at the basic level to distinguish
only three types of nuclear tone: fall (‘\’), rise (’/’) and fall-rise (’\/’). I write the
tone mark immediately before the syllable that bears it, and underline that
syllable.
(3) Do you want /coffee | or /tea | or \cola?

(4) \Marvellous!
103
104 intonation
(5) What lovely \flowers!

(6) A: I’ve just \spifflicated them.
– B: You’ve just /what? | /Spifflicated them?
Note that the syllable with the tone mark is where the falling, rising or falling-
rising pitch movement begins; particularly in the case of non-falls, it may not be
completed until the end of the IP. In (6), the fall in A’s utterance is manifested
as a high pitch on spiff followed by an immediate drop to a low pitch for the
rest of the utterance: typically, -licated them will be on a low level pitch. In B’s
response, the rise in the second IP starts on the syllable spiff and continues
steadily to the end of the utterance: the unaccented them, paradoxically as it may
seem, has the highest pitch in the IP.
(7) I \/think it’s true | (. . .but I could be wrong.)
In (7) the pitch starts high on think, drops abruptly down, and then rises on true.
If tone and/or tonicity are not relevant to the discussion, I leave them unshown,
as in examples (1) and (2) above.
4.2 intonation notation
I hope you like the notation system used here. It is the system I use
in my book English Intonation (Cambridge University Press 2008). My aim is
a notation system that is intuitive for the reader to interpret, while being easy
to type on a computer.
Given that I want to be able to combine it with ordinary phonetic transcription,
it has to be based on adding to the orthographic or segmental text (text-
decoration) rather than changing it (text-alteration). This means that although
I approve of Brazil’s principle of symbolizing the place of the nucleus (the tonic)
separately from the choice of nuclear tone (see Discourse Intonation and Language
Teaching, Longman 1980), I have rejected his device of using capitaliZAtion for
this purpose, and have instead adopted Halliday’s old idea of underlining the
nucleus (see On Language and Linguistics, Continuum 2003).
In the notation of tone, I want to use marks that are iconic, i.e. that suggest by
their shape the pitch characteristics involved, as do those of the O’Connor
and Arnold system (see Intonation of Colloquial English, Prentice Hall 1973).
But keeping the notation easily typable means rejecting some of their special
marks, since some of them are not even Unicode symbols. In any case, I think
that for most EFL purposes (and even for native speakers) the distinction
between different kinds of fall (high fall, low fall, rise-fall) is relatively unim-
portant, and that for most purposes it is sufficient to symbolize them all with the
backslash symbol [\]. Likewise with the various types of rise [/] and fall-rise [\/].
So I operate with a basic three-tone system.
Where the choice of nuclear tone is irrelevant, I just put [ˈ] as a place holder.
4.3 idiomatic intonation 105
For non-nuclear accents I think it is generally adequate to use a simple straight

apostrophe [’], or if available its phonetic equivalent the stress mark [ˈ]. In the
basic markup I do not mark simple rhythmic stress at all, though the symbol [ ]
is available for those who want to show it.
\/ðɪs | ʃəʊz ˈhaʊ tə kəm\/baɪn | fəˈnetɪk sɪmbl̩z ən ɪntə\neɪʃn̩ mɑːks ||
In pedagogical applications, I think that ToBI-style notation (H*L-L% and

the like) is a non-starter (see www.ling.ohio-state.edu/~tobi/). But I could
be wrong.
4.3 idiomatic intonation
English has various tone idioms – words or phrases for which the
choice of tone (rise, fall or fall-rise) is fixed rather than free. For example, the
interjection oops or whoops, used when you’ve fallen, dropped something or
made a mistake, can only have a rise. You can’t say it with a fall.
/Oops!
\Oops!
On the other hand the phrase by the way, used in spoken English to introduce a
side issue not connected with the main subject you were talking about before,
seems (at least for me) always to have some kind of fall (high fall, low fall, rise-
fall), never a rise or fall-rise.
By the \way, | have you ˈseen my um/brella anywhere?
By the /way, | have you ˈseen my um/brella anywhere?
This is despite the fact that (as Alan Cruttenden has shown us – see Intonation,
Cambridge University Press 1997, second edition) most limiting adverbials tend
to have a rise; it is reinforcing ones that usually take a fall.
You can say hello! with any tone. But its newer equivalent hi! seems to
demand a fall.
Hel\lo!
Hel/lo!
Hel\/lo!
\Hi!
/Hi!
(I suppose there is also a cutesy way of saying hi with a fall-rise, but I think that
only a few speakers would use it. When I was young, people in my circle didn’t
say hi at all. We thought of it as an Americanism. But now you hear it
everywhere. I think I started using it myself in the mid-1990s, on the principle
that if you can’t beat them join them.)
106 intonation
You can use any tone for goodbye, though probably the most usual one is a
rise. But its informal equivalent see you sounds odd with a simple rise: it seems
to need a fall-rise.
Good/bye! (also with a high head or prehead, ˈGood/bye! or ¯Good/bye!)
\/See you!
?? /See you!
A rise seems OK if we add an adverbial. So is a fall.
ˈSee you to/morrow!
ˈSee you to\morrow!
There are also tonicity idioms – in which the place of the nucleus is fixed, though
not necessarily the tone. Here are some examples involving personal pronouns
which bear the nucleus despite seeming from a pragmatic point of view not to be
in contrastive focus.
ˈGood for ˈyou! (¼ congratulations!)
ˈBully for ˈyou! (¼ sarcastic congratulation)
ˈBlow ˈme! (¼ I am surprised or annoyed)
ˈGet ˈher! (¼ look at her putting on airs)
ˈSearch ˈme (¼ I don’t know, I’ve no idea)
The fixed tonicity is necessary for the idiomatic meaning. If, instead, you say:
ˈSearch me!
– that’s an invitation (perhaps to a policeman or immigration official) to do just
that, rather than a confession of ignorance. And if you say:
ˈBlow me!
– then that’s an invitation to perform a sexual act. The same tonicity-dependent
interpretations apply if we replace blow by the f-word.
4.4 systems
A student from Argentina wanted to know how to account for the word
‘system’ in the expression ‘the English intonation system’. ‘I know Brazil talks
about the system of English intonation,’ she wrote, ‘but why is it called a system?’
Intonation is systematic, or a system, in the technical sense of having a finite
set of either/or options (e.g. a syllable either is, or is not, prominent). In linguis-
tics, a system is a fixed set of choices. Nouns in many (but not all) languages
exhibit a NUMBER system – every time we utter a noun we have to make it
either singular or plural. Verbs usually display a TENSE system (present vs. past,
etc.). Most people analyse English verbs as also displaying an ASPECT system
(progressive vs. non-progressive, perfective vs. non-perfective).
4.5 politeness 107
Similarly, ever since Halliday’s work in the 1960s, people analysing English
intonation have identified three systems in operation (see 4.1 above):
a tonality system, or ‘chunking’, whereby we divide the spoken
material exhaustively into successive IPs (aka word groups, tone
units, etc.);
a tonicity system, or ‘nucleus placement’, whereby we select one
lexical item to highlight by placing on its stressed syllable the nucleus
(aka tonic) of the IP – in English this has the pragmatic function of
marking the end of the focus domain;
and a tone system, whereby we select fall, or rise, or fall-rise, each
with a characteristic tone-meaning, as the tone for the IP (located
physically on and after the nucleus but pragmatically characterizing
the whole IP).
The claim here is that in these three respects we make a succession of choices all
the time as we speak, as we decide how to structure our material, what to focus
on and what attitude (etc.) to express towards what we are saying. Some would
claim that this is all also part of a complicated battle with our interlocutor(s) for
control of the conversation.
4.5 politeness
As I was jogging in a local park, a man coming the other way asked me:
You ˈhaven’t got the /time on you, have you mate?
Since I was indeed wearing a watch, I duly answered him:

Eˈleven fifty-\seven.
The wording and intonation of his question was appropriate for the situation of
an English person addressing a stranger. But an EFL student would probably
have cast the question differently:
ˈWhat’s the \time?
– which in this context would have sounded rude.

Why is this? We use the simple wording and the fall when we wish to be
businesslike with someone we are already talking to. We would also use this
pattern if repeating the question to someone who seemed not to have heard it,
or who was refusing to answer, or if we were otherwise annoyed at their non-
cooperation. But when starting a conversation, particularly with a stranger,
we prefer to be indirect. The negative statement followed by a reverse-polarity
tag question in the same IP does the job nicely.
You ˈhaven’t got the /time on you, have you mate?
108 intonation
Alternative polite formulae in BrE would be:

ˈHave you got the /time on you by any chance?
Ex\/cuse me, | d’you ˈhappen to know what /time it is?
\/Sorry, | d’you ˈhave the /time?
ˈGot the /time, mate?
Ex\/cuse me, | I was \wondering, | ˈwhat /time d’you make it?
Some speakers (particularly younger ones?) might alternatively use a highish

fall-rise instead of the rise.
Americans would probably not be so indirect. And they would not use the
BrE working-class non-deferential male-to-male vocative mate, but probably
bud(dy).
The formulae you sometimes hear from speakers of Russian, involving a
negative question:
ˈHaven’t you got the /time on you?
ˈDon’t you know the /time?
sound like accusations and are therefore very inappropriate ways to ask the
question. They would be heard as reproaches to someone who is late.
4.6 fall-rise or fall plus rise?
As I breasted the brow of a rather steep hill during my running club’s

Sunday morning social run, a fellow runner sweetly enquired, ‘Did you run up
that hill?’
My answer was:
\No, | I’m aˈfraid I \didn’t. k I ˈhave to admit I \walked up.
However, what interests me here is the intonation used by my interlocutor.

She had a wide falling pitch on run and then a wide rising pitch on hill.
The problem is to decide whether she was using (a) a fall tone followed by a
rise tone, or (b) a fall-rise tone:
(a) Did you \run up | that /hill?
(b) Did you \/run up that hill?
I think the answer is (b), for three reasons.
First, we both knew we’d just got to the top of a hill. So from the pragmatics
of the situation, that hill was old information (given). Accordingly, we would
not expect focus on those words: we would not expect them to be accented.
Analysis (a) places not just an accent but a nuclear accent on hill; analysis
(b) treats it as unaccented.
If, introspecting, I change the tone on run into a rise, but otherwise keep the
pragmatics the same, I find myself saying not (a0 ) but (b0 ):
4.7 empty things 109
(a0 ) Did you /run up | that /hill?

(b0 ) Did you /run up that hill?
And if I had not heard clearly and had asked her to repeat the question, I think
she would have then said not (a00 ) but (b00 ):
(a00 ) Did you \run up | that \hill?
(b00 ) Did you \run up that hill?
Thirdly, if – introspecting again – I ask myself how she would have said the same
thing without the last three words, but keeping the pragmatics the same, I feel
pretty sure she would have said not (a000 ) but (b000 ):
(a000 ) Did you \run?
(b000 ) Did you \/run?
Conclusion: she used a single IP, with a fall-rise tone on run. I have noticed that
a fall-rise tone on a yes-no question is heard increasingly frequently.
I think that phoneticians, or at least intonologists, have to be ready to perform
mental speculations like this to elucidate structures, just as syntacticians do.
4.7 empty things
In English intonation there are certain types of word and phrase

that tend not to receive the nucleus, even though they may be the last lexical
item in an IP. They include various nouns that have very little meaning of their
own, particularly vague general nouns such as things, people – often known as
‘empty words’.
I ˈkeep ˈseeing things.
ˈWhat are you going to ˈtell people?
A correspondent, having read the section in my book English Intonation that
deals with this point, mentions the greeting expression How are things? ‘Where
would the nucleus occur?’ he asks. ‘Probably on the question word how? Yet
don’t some native speakers put the nucleus on the empty word things?’
The problem is that the nucleus has to go \/somewhere, and if all the other
words in the IP are function words then it will even go on a weak (‘empty’)
content word. In fact, this is the usual tonicity for this question.
ˈHow are \things?
Actually, for reasons I cannot explain, people like me usually violate number
concord here, and say:
ˈHow’s \things?
– though that is not relevant to the intonation.
110 intonation
A nucleus on how would constitute marked tonicity, and might be, for
example, a disbelieving echo question.
A: ˈHow’s \things?
B: \Oh, | \otiose.
A: /What did you say?! | /How are things?
My correspondent then offered the following counter-examples, in which things
receives the nucleus.
(1) a: Mary, I’m sorry to hear about your father.
b: Thank you, John. | It was ˈone of those \things.
a: When did he pass away?
b: He was buried on December 20th.
(2) a: What does your father do for a living?

b: He’s into ˈmany \things. (¼ jobs)
a: Does he have a major line of work or. . .?
b: Yes, he’s a waste material dealer.
(3) ˈGet your \things. (¼ belongings) You’ll be leaving with the police!
(4) We’ll drive out right after dinner | and ˈget your \things.
Numbers (1), (3) and (4) are clear: we have to accent things. You could consider
just one of those things an idiom, and exempt things ‘belongings’ from the
category of empty words. It seems to me that (2), though, could go either way:
the nucleus can go on things or, alternatively, on many, given that the speaker
can imply that jobs is given (¼ predictable from the context).
There are similar difficulties with people. Although we say:
(5) I ˈwant to \meet people.
(6) I ˈwant to com\municate with people.
(because a nucleus on people might suggest a contrast with, say, machines or

even animals), we nevertheless say:
(7) There were a ˈlot of \people in the room.
4.8 counterpresuppositional insists
One area of English intonation that is really difficult to explain is

found in exchanges like this:
A: Have you /eaten?
B: \No, | there was ˈnothing \to eat.
Why does B put the nuclear accent on to, a function word that appears to have
virtually no meaning of its own?
4.8 counterpresuppositional insists 111
This is an example of a class of utterances known as counterpresuppositionals.

Here, A says something that presupposes that there was something available for
B to eat. That is, A treats the availability of food as something that is shared
ground (mutual knowledge) for the two participants in the conversation. B’s
reply involves a rejection of this presupposition: it counters it. The supposed fact
of the availability of food was not in fact shared ground: B denies that it was
the case.
We know that in English it is usual to avoid putting the nucleus on a repeated
item. So we don’t expect a nucleus on eat. Logically, we might expect a nucleus
on nothing. But that’s not where native speakers put it. They usually put it on to
(above), or sometimes, less commonly, on was, or indeed on eat:
B: There \was nothing to eat.
B: There was ˈnothing to \eat.
In my book English Intonation (p. 183), I give various other examples of this,
including:
A: They ˈwant you to a\pologize.
B: But I’ve ˈno reason \to apologize.
A: ˈNever /mind. | ˈWorse things happen at /sea.
B: We’re ˈnot \at sea.
Alan Cruttenden, in his book Intonation (Cambridge University Press 1997,
second edition, p. 85), calls this kind of answer an ‘insist’. He says that in
these ‘the nucleus is only rarely on a final noun, may fall on the main verb,
but may also fall on classes of word not usually accented at all, in particular
on auxiliary verbs, on a negator, and on a preposition immediately following
copula “be”’.
Among his examples are the following.
(1) A: Did you meet the Ripper when you were in prison?
B: But I haven’t been in prison.
(2) A: I put my bag in your study because there was nowhere else to
put it.
A correspondent asks, ‘Is that always the case? I mean, cannot the verb been
be stressed instead of the preposition? Would that imply a change of meaning
of the sentence?’ My answer is that it \/is possible to put the nuclear accent
on been. But it’s also possible to put it on in, and that’s the mysterious fact
that Cruttenden is discussing at this point of his chapter. As to whether there is
any difference in (pragmatic) meaning, I confess I find it very difficult to put my
finger on what, if anything, it is.
With regard to (2), my correspondent says, ‘I myself would have placed
the nucleus on else instead’. I think that it is indeed possible to put it on else,
but that I myself might well, like Cruttenden, have put it on to. I can’t really
explain why.
112 intonation
In reply to the above, Alan Cruttenden writes as follows.

I used the term ‘insists’ in my book because I deliberately wanted a fairly
vague term which covered many types of preposition and auxiliary verb
accenting. These include presuppositionals but many other uses which are
only presuppositionals by a considerable extension of the meaning of ‘pre-
suppositional’. David Faber and I published an article some years ago in the
Journal of Pragmatics (1991, 15: 265–86) called ‘The accentuation of
prepositions’ in which we discussed examples like:
(a) This poem describes London and the journey to London

(b) The rules and the interpretation of those rules
(c) A woman shouldn’t be a politician; she should be the wife of
a politician
(d) I like children; what I don’t like is looking after children
(e) You keep on going on about Guyana but have you ever met
anyone from Guyana?
(f) They’re not particularly good but there are masses of the brutes.
In all these cases there are alternative nucleus placements which more easily
fit the general rules and non-stressing of function and ‘given’ words, e.g on
journey, interpretation . . . etc. (though this is not always true). The prepositional
placement does a variety of things like contrasting two noun phrases ‘London’
and ‘to London’, suggesting that the preposition should be considered as
a lexical item rather than just a function word, and suggesting the newness
of the preposition as opposed to the givenness of the following noun. But a
characteristic of them all is that the impression is given that the user is some-
how showing off his sophisticated use of language; as such they are used
by articulate middle-class speakers (frequently by journalists and reporters).
I cannot imagine a road sweeper saying ‘I hate litter and the leavers of that litter’.
The last point, linking this usage to social class, is a very acute one and I think
probably correct. Unlike the road sweeper, I think that Alan and I could well say:
I ˈhate \/litter | and the ˈleavers \of that litter.
Touché.
4.9 that’s funny
Quite a common idiomatic intonation pattern in English consists of a

fall on the word That with the verb to be, followed by a rise on the complement.
\That’s | /funny.
\That was | /good.
\That’s | a re/lief.
\That’s | a /pity.
\That’ll be | the /day!
4.10 where? wohin? 113
A: Your ˈkeys aren’t \here.

B: Well \that’s | rather /odd. k
(I could have \sworn | I ˈleft them here \/somewhere.)
A: I’m ˈgoing to start going to the \gym.
B: \That’ll be | the /day!
This pattern is rather difficult to explain. The word that is merely anaphoric – it
refers to some situation or thing already mentioned, rather as a pronoun does. But
if we replace that by it, we cannot use the same tone pattern.
* \It’s | /funny.
* \It was | /good.
Obviously, this is connected to the fact that the default for demonstratives is to
be accented, whereas the default for pronouns is to be unaccented. But why is
that so important that it is not merely accented, but carries a nuclear tone?
We might wonder whether this pattern is not a fall plus rise but a fall-rise.
? \/That’s funny.
? \/That was good.
But in this analysis, if the single nucleus is on the demonstrative, the words funny
and good are deaccented, postnuclear – which would normally imply that they
must be given (old), or repetitious, or predictable in the context, none of which is
the case here. So that can’t be right. And in any case we would still be left with a
difficult-to-explain nucleus on the demonstrative.
Compare contrastive focus on that, with a genuine fall-rise:
A: They’re ˈboth pretty \good.
B: Well \/that one is | but \/this one isn’t.
Unfortunately, I don’t think I discussed this intonational idiom in my English
Intonation book. Your appropriate response, reader, is
\That’s | a /pity.
4.10 where? wohin?
I had been invited to Berlin to give a lecture on intonation.

I mentioned an example I have used several times before: an exchange that is
ambiguous in writing but probably not in speech:
A: I’m off to England tomorrow.
B: Where?
My claim was that with a fall on Where B’s response means ‘What part of
England? Where in England?’ But if the intended meaning is ‘I didn’t quite catch
what you said. Please repeat’ then Where? must have a rise (or a high fall-rise).
114 intonation
Carefully worded like that, I think the claim is correct. However, it doesn’t
necessarily follow that a rise on Where? must always signal a pardon-question.
It could just be a friendly way of asking for more precise details (as with the
fall). Although the default/neutral/unmarked tone for a wh-question is a fall,
we can also use a rise or (low) fall-rise. Hence a rise on Where? might be
ambiguous as between the two possible readings.
One of the audience pointed out that the German equivalent is not open
to ambiguity in this way. In German you can’t use Wo? (the word for ‘where?’)
in this context. You have to say Wohin? ‘where to? whither? where hence?’.
And the word will be stressed differently depending on the intended meaning.
(1) a: Morgen fahre ich nach England.
b: Wohin? (Also nach London, oder. . .?)
(2) a. Morgen fahre ich nach England.

b: Wohin? (Das kann ich kaum glauben!)
No matter whether the nucleus on -hin in (1) has a fall or a rise, the meaning must
be a request for further details. With a (rising) nucleus on wo-, as in (2), it must
be a pardon-question, ‘I didn’t quite catch what you said. Please repeat’, or ‘I
can’t really believe that’.
I then realized that the same applies to the alternative way of framing the
question in English, Where to?
(1) a: I’m off to England tomorrow.

b: Where to? (Liverpool, Norwich, Southampton?)
(2) a: I’m off to England tomorrow.

b: Where to? (Where did you say you were going?)
(1) asks for amplification; (2) is a pardon-question.

It’s fun giving lectures. Sometimes even the lecturer learns something.
4.11 examining intonation
When we ran an undergraduate course on English intonation at UCL,

mainly for native speakers, how did we conduct the practical part of the exam?
The candidates were set three tasks: a prepared reading, an unseen reading and
the description of their own and the examiner’s intonation of a given sentence.
For the prepared reading, candidates were given in advance a short piece of
conversational English marked up with intonation marks. The candidate had to
read the passage out loud, using the intonation patterns indicated.
Here’s part of a sample passage. The further intonation marks are those I use
in English Intonation: an introduction. \ ¼ falling head; / ¼ rising head; ¯ ¼ high
4.12 the importance of being accented 115
level prehead; ¼ rhythmical stress. For nuclear tones, we distinguish \high fall
from \low fall, /high rise from /low rise.
He /seems to be en joying the job im\mensely
though I must \/say | it ˈwouldn’t suit \me very well.
\
All that \travelling a\/bout | sounds ex/\hausting.
\
/Still | ¯I sup/pose some /people thrive on that sort of thing.
This is a much more complicated pattern than one would usually recommend
to EFL students. But as a performance exercise it certainly tests the abilities of
native speakers and foreign learners alike. (Candidates lost one mark per symbol
incorrectly realized.)
If I were setting this as a test for EFL teachers and learners, I would treat as
unimportant everything except the three Ts: tone (candidates should make rises,
falls and fall-rises as required, but without worrying about such distinctions as
low fall vs. high fall), tonicity (put the nuclear tones in the right places as
required: preferably get the other accents in the right places too) and tonality
(break the material into chunks as required).
So I might mark it up in this simpler way.
He ˈseems to be enˈjoying the ˈjob im\mensely
though I must \/say | it ˈwouldn’t suit \me very well.
ˈAll that ˈtravelling a\/bout | sounds ex\hausting.
/Still | I supˈpose some ˈpeople \thrive on that sort of thing.
This is still a good test of performance ability, since not all the tones and
tonicities are those that would first come to mind in the given situation. Accord-
ingly, even those with a good feeling for English intonation might be misled into
producing a different intonation pattern if it seemed to make more sense. For
example, I think that in real life I would be more likely to say the following rather
than the patterns asked for in the exam:
. . .it ˈwouldn’t suit \/me very well

(with a fall-rise to show the implicit contrast between me and others) and
. . .I suppose ˈsome people \thrive on that sort of thing

(some people do, some don’t).
4.12 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:
116 intonation
Bill threatened Jim, | and then he ˈhit him. (¼Bill hit Jim.)
Bill threatened Jim | and then ˈhe ˈhit him. (¼Jim hit Bill.)
Bill threatened Jim | and then ˈhe hit ˈhim. (¼Jim hit Bill.) English
Intonation, p. 239
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
taken from a newspaper report (American, as you can tell from the spelling), the
subject of an item in Language Log.
If leadership never takes time off, people will be skeptical whether they can.
Does the pronoun they to refer to the company’s leadership? In that case, they
will be unaccented when you say it aloud. Or is the intended referent of this
pronoun actually people? If so, there will have to be a contrastive accent on they.
The difficulty is that the distinction is not normally shown in writing. Hence
the newspaper report was ambiguous. If the writer of the report had realized this,
and had wished to avoid the ambiguity, he might have chosen one of various
ways of resolving it: for example, he could have emboldened they, or changed it
to they themselves.
Take the written sentence A third of the women surveyed thought that their
husbands were better cooks than they were. As it stands, it is ambiguous. But in
speech it would be disambiguated by intonation.
1. . . . thought that their husbands were better cooks than they were
¼ than the women
2. . . . thought that their husbands were better cooks than they were
¼ than in reality
In (1) the accenting of they signals the change of grammatical subject. In (2)
I suppose we have to call the accented modal a ‘marked positive’.
4.13 accented be
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 (¼ positive vs. negative) or contrastive tense.
4.14 O lift: your voice 117
You ˈthought he wasn’t ˈready, | but ˈactually he ˈwas.

(Why didn’t they ˈmake you ˈmanager?) – Well I ˈam manager, | ˈactually.
Well I ˈwas happy, | but ˈnow I’m ˈnot.
I ˈused to be at the top, | and I ˈstill ˈam at the top.
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 plus be plus
pronoun, or syntactic manipulations thereof. I devote a separate section (3.18) to
this in my intonation book.
ˈWhat ˈis it?
ˈWhere ˈare you?
ˈHow would it ˈbe | if we ˈmet for ˈlunch?
ˈWho do you think you ˈare?
However, none of these explanations fit two sentences I have heard recently.
(ˈWelcome ˈback!) – We’re ˈhappy to ˈbe back! (at the end of a TV programme
about the history of East Anglia)
. . .and ˈso we ˈsee | what a ˈhoneypot East Anglia ˈwas.
(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’ (pp. 182–4 in
my English Intonation). 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.
4.14 O lift: your voice
At the dental surgery where I have my teeth done there are five floors,
or six counting the ground floor. There is a lift with an automated recorded voice
telling you what the lift is doing and where it has got to.
I was struck by the intonation patterns of the recorded voice as I went up to
the top.
ˈDoors \closing.
ˈGoing /up. . .
ˈFloor \/four.
ˈDoors \/opening. . .
ˈGoing /up. . .
ˈFloor \five.
These patterns are just what a human operator might use, which is perhaps only
to be expected given that they were recorded by a living human being. We have
a fall on a declarative statement, followed by a non-final rise (because there is
118 intonation
more to come), followed by another non-final tone, this time a fall-rise

(because there’s still another floor to go), then another fall-rise (a warning), a
repeat of the second line, and finally a fall proclaiming that we’ve reached
our destination.
This was followed of course by a repeat of the warning:
ˈDoors \/opening.
What is interesting is the difference of tone between four and five. Although this
fitted in excellently with my progress upwards, presumably it arose by accident,
since the voice-over actress probably recorded:
ground floor || floor one || floor two || floor three || floor four || floor five
– with the usual listing sequence comprising a rise or fall-rise on the non-final
list items and a fall on the final one. Then the speech engineer chopped them
up into separate sound files for the onboard computer to use.
Coming down again I had the lift to myself.
ˈDoors \closing.
ˈGoing \down. . .
ˈGround /floor.
With a fall instead of a rise on floor here it would have been perfect.
I have to congratulate the anonymous speaker on saying going up with an
iconic rising tone and going down with an iconic falling tone. Not all would-be
students of intonation, let alone members of the general public, can identify and
reproduce the difference between rises and falls.
4.15 stating the bleeding obvious

I saw what is described as a ‘terrible Essex joke’ in the Guardian.
nurse, in casualty, to injured patient: Where are you bleeding from?
patient: I’m from bleedin’ Romford!
(Explanatory note for non-Brits: the patient wrongly interprets bleeding as

an expletive. See the Concise Oxford Dictionary: bleeding adj. & adv. Brit.
coarse slang expressing annoyance or antipathy. Casualty ¼ the E.R. The
inhabitants of Romford, a suburb in Essex out beyond London’s East End,
are stereotypically white working-class Londoners and thus supposedly likely
to swear a lot.)
The trouble is that as a spoken joke this doesn’t actually work. The reason is
that the two readings of the ambiguous question differ in tonicity (¼ nucleus
placement).
The nurse’s question asks what part of his body is the source of the blood:
(1) ˈWhere are you ˈbleeding from?
4.17 des accents enfantins 119
But the patient’s answer is an answer to a different question, one about the
patient’s origins:
(2) ˈWhere are you bleeding ˈfrom?
Version (1) is parallel to other examples such as:
ˈWhat are you ˈlooking at?
ˈWho were you ˈthinking of?
ˈWhere do you ˈcome from?
Version (2) is just an expansion of the expletive-free
ˈWhere are you ˈfrom?
The way I explain this point in English Intonation (section 3.17) is that the
nucleus goes on the last lexical item (¼ the last content word) if there is one.
That accounts for (1). If there is no lexical material – no content words – it goes
on the preposition. That accounts for (2).
And what the tonicity difference seems to demonstrate is that bleeding and its
more forceful variants do not count as content words. So a nuclear accent on
bleeding, as in (1), is a signal that the word is to be interpreted literally, lexically.
4.16 chunking
Here are two nice examples I have recently come across which are
ambiguous in writing but normally distinguished by tonality (or ‘chunking’) in
speech.
People who do this often get bored.
Subscribing now means you will receive all three issues promptly.
In speech we could have either:
People who do this | often get bored, or
People who do this often | get bored.
And either
Subscribing now | means. . . or
Subscribing | now means. . .
The adverb of time can go either way. So there is no ambiguity in speech.
4.17 des accents enfantins

The other day I overheard a snatch of conversation.
A father has just got onto a bus with his two children, a girl aged about
five and a boy aged about seven. They sit right at the front of the upper deck of
120 intonation
the bus, so they all have a good panoramic window view. The boy takes the
seat next to the side window. The girl protests.
Girl: \/Daddy, | ˈI want to sit on the \window side.
Boy: But you’re \on the window side!
My first reaction was to think that if it were me I would probably have said:
But you \are on the window side!
General principles lead us to say that accenting on would be avoided, since it

is a repeated item (it has just featured in the girl’s utterance). Hence, I assume,
my immediate reaction that the boy’s command of English intonation was not
yet quite adult-like, and that the proper place for the nucleus is somewhere else,
namely on are.
This is the word that carries the polarity (positive as against negative). We
have what Halliday calls a ‘marked positive’ (see my English Intonation, p. 135).
But on reflection I think this is not the whole story, and that adult native
speakers could after all put the nucleus on on, just as the boy did. (/Couldn’t they?)
I find it very difficult to put into words what difference, if any, there is
between a nucleus on the preposition on and a nucleus on the verb are.
We are dealing here with a counterpresuppositional. The girl’s utterance
implies that she is not sitting by the window. The boy (deviously, perhaps)
challenges this presupposition. So it’s comparable to:
A: Never ˈmind. | Worse things happen at ˈsea.
B: But we’re not ˈat sea.
(p. 183 in my English Intonation). And the intonation patterns of counter-
presuppositionals are notoriously difficult to explain (4.8 above).
Perhaps someone who knows more pragmatics than I do can shed some light.
Christopher Bergmann points out there that in German you cannot put the
nucleus on the preposition in such cases.
A: Ich möchte auf der ˈFensterseite sitzen.
B: Aber du ˈsitzt doch auf der Fensterseite!
and not
B: *Aber du sitzt doch ˈauf der Fensterseite!
The only reason you would put the nucleus on the preposition would be
for contrast. That would be pretty impossible to contextualize here: not on the
window side but ??? below/beyond/outside it.
4.18 intonation idioms in the Germanic languages
There are some arbitrary or illogical intonation patterns in English

that are also found in German and Swedish.
4.18 intonation idioms in the Germanic languages 121
One concerns the grammatical structure Wh – be – pronoun. My Japanese

colleagues tell me that their advanced students of English are surprised at the usual
English tonicity in questions like What ˈis it? Where ˈam I? Logically, they feel,
the focus ought to be not on the verb to be but on the interrogative, the wh-word.
But in English that would happen only in a pardon-question: (What ˈis it? –
It’s a ˈslurge.) – /What is it? or, more likely, – It’s a /what?
My only explanation of the pattern What ˈis it? is that it is idiomatic, and
something we can’t really explain logically.
It is, however, found also in German: Was ˈist es? Wo ˈbin ich? (but as pardon-
questions /Was ist das? Das ist /was?).
And in Swedish: Vad ˈär det? Var ˈär jag?
In each language, however, if this kind of sentence ends in a demonstrative
rather than a personal pronoun, the nucleus may go in the expected place, namely
on the demonstrative: What’s ˈthat? Was ist ˈdas? Or it can stay on the verb:
What ˈis that? Was ˈist das?
The other pattern concerns apparently predictable words. The nucleus, we
know, goes on the last item that adds new information. That’s where the focus
domain ends.
If we’re in England and you ask me how much, say, a computer costs, you
know that the answer will involve pounds: the name of the currency is part of our
shared knowledge rather than truly new information. So it is rather illogical
that by default we nevertheless accent pounds. You would expect that the
name of the currency, being already known, would be out of focus. Yet in a
broad-focus answer we use the pattern It costs four hundred ˈpounds, rather
than the apparently more logical It costs four ˈhundred pounds. (The latter is
of course also available if we want to narrow-focus on the exact number).
It’s the same in German: Es kostet sechshundert ˈEuro. And it’s the same
in Swedish: Den kostar tjugo ˈkronor.
Other examples of the same sort of thing include It’s a lovely ˈday ¼ Es ist ein
wunderschöner ˈTag ¼ Vilken härlig ˈdag! (but we both know that it’s daytime);
She’s a beautiful ˈgirl ¼ Sie ist ein schönes ˈMädchen (but we know she’s a girl);
It’s seven o’ˈclock ¼ Es ist sieben ˈUhr.
I suspect that these patterns are common to all the Germanic languages, and
would accordingly expect that they are also found in Dutch, Danish, Norwegian,
Icelandic and Faroese. They may well also extend further afield.
One difficulty about doing comparative intonation studies is that we are
interested (or at least I am) not so much in the surface differences in tune
shape, but in the deeper differences, if any. Do the two languages differ in such
matters as:
the principles of chunking;

the nuclear tones available;
and, most importantly, the principles governing the location of the
nucleus (focus)?
122 intonation
That is, are there differences in the three Ts: tone, tonality and tonicity? And to
discover whether such deeper differences exist, you need to have access to native
speakers who have insight into what is going on. You won’t get very far by just
sitting in a phonetics lab analysing the pitch patterns of random utterances in a
spoken corpus.
I think it’s clear that English and German have a good deal in common, not
just in vocabulary, syntax and morphology, but also in intonation, if seen in this
light. The same is probably true for all the Germanic languages.
There are differences, of course. We have already discussed one. Another
concerns sentences consisting of a major part with a minor part, for instance
some adverbial or other subordinate material. English has two favourite
intonation patterns for these. If the minor part comes first, it tends to have
a fall-rise, followed by a fall on the major part. If the major part comes
first, it has a fall, usually followed by a rise on the minor part. Thus,
depending on the order we choose for the two parts, we might have either
of the following.
This \/evening | I’m reˈturning to \London.
I’m reˈturning to \London | this /evening.
In German there is a strong preference for a rise on the minor part and then a
fall on the major part. (The other option, with the adverbial at the end, so-called
Ausklammerung, is apparently now becoming commoner than it once was
but is still very restricted.) There are two straightforward possibilities for the
word order.
Ich ˈkomme heute /Abend | nach \London zurück.
Heute /Abend | ˈkomme ich nach \London zurück.
It would be awkward or impossible to say, as in the second English pattern:
?
* Ich ˈkomme nach \London zurück | heute /Abend.
Though of course it would be OK with a change of speaker and with heute Abend
as a question.
A. Ich ˈkomme nach \London zurück.
B: Heute /Abend?
(Thanks to Petr Rösel and Christopher Bergmann for the German and to Gunnel
Melchers for the Swedish.)
4.19 compound stress in English and German

Although German and English share the general principle of putting
primary lexical stress on the first element of a compound, English has a rather
large number of exceptions where German does not. These include cases where
4.20 accenting phrasal verbs 123
the first element names a material, ingredient, time or place. English usually
stresses the second element, German the first.
ham ˈsandwichˈ Schinkenbrötchen

evening ˈpaperˈ Abendblatt
summer ˈholidaysˈ Sommerferien
kitchen ˈtableˈ Küchentisch
leather ˈjacketˈ Lederjacke
(A ˈleatherjacket, on the other hand, is a kind of grub.)

Assorted other cases:
town ˈhall ˈRathaus

Leicester ˈSquare Aleˈxanderplatz
capital ˈcity ˈHauptstadt
There are exceptions in English where the first element names an ingredient (etc.)
but we nevertheless have regular compound stress on the first element.
ˈfruit cake
ˈsoda water
ˈorange juice
ˈbreadcrumbs
And there are certain place names where German nevertheless has stress on the
second element.
Kurfürstenˈdamm
Friedrichsˈhagen
Heiligenˈhafen
Just as English is hopelessly inconsistent in ˈChristmas cake/card/tree/present
but Christmas ˈDay/ˈEve/ˈpudding/ˈcrackers, so German is inconsistent in
ˈHeiligenberg/-haus/-stadt/-wald but Heiligenˈbeil/-ˈblut/-ˈdamm/-ˈhafen/-
ˈkreuz. (German data from the Duden Aussprachewörterbuch.)
My impression is that the other Germanic languages (Dutch, Danish, Swedish,
Norwegian, Icelandic, Faroese) mostly have fairly regular compound stress.
In Norwegian and Swedish it involves a special word tone, in Danish their stød
(glottalization).
4.20 accenting phrasal verbs

If you manoeuvre an English phrasal verb into a position where it
bears the intonation nucleus, the nucleus will be placed either on the verb word
or on the particle. Generally speaking, the nucleus goes (i) on the verb word if the
124 intonation
particle is prepositional, but (ii) on the particle if it is adverbial. Usually,

prepositional (¼ pseudo) phrasal verbs are lexically single-stressed, but adverbial
(¼ true) phrasal verbs are lexically double-stressed.
(i), prepositional ˈWhat are you ˈlooking at?

ˈTell me what you’re ˈlooking for.
(ii), adverbial ˈDon’t (ˈ)look ˈback!
ˈLook aˈway!
They ˈall (ˈ)look ˈup to him.
Accents shown in the examples as (ˈ) are likely to be downgraded for rhythmical
reasons (‘Rule of Three’, English Intonation, p. 229; and 4.22 below).
There are a few phrasal verbs which rather rarely bear the nucleus. With a
transitive verb, where there is an object following, the nucleus naturally tends to
go on the object. In the resultant string Verb þ Particle þ Object, it can be
difficult to tell whether the lack of an accent on the particle is because the particle
is lexically unstressed (type (i)), or because it is rhythmically downgraded by the
rule of three (type (ii)).
ˈLook up the ˈanswer.
You can usually resolve the uncertainty:
by replacing the object noun phrase with a pronoun (and extraposing
the particle if appropriate);
by switching to a wh-question;
or by switching to a passive construction.
This forces the nucleus onto the phrasal verb.
ˈLook it ˈup.
ˈWhat did you (ˈ)look ˈup?
The answer | was ˈduly (ˈ)looked ˈup.
Hence look up ‘find in a list’ is type (ii), adverbial.
For most phrasal verbs I have no hesitation with this classification. My
native-speaker intuition, supported by observation, readily tells me whether
they are type (i) or type (ii). However, there are cases where I find myself
hesitating. One such is to look after. Applying the techniques mentioned
resolves the uncertainty. (This is a verb in which the particle cannot be
extraposed.)
What about Mary? | We must ˈlook ˈafter her.
ˈWho do you want me to (ˈ)look ˈafter?
She’s ˈbeing (ˈ)looked ˈafter.
Hence look after, despite seeming to be prepositional, is actually double-
stressed.
4.21 train times 125
Michael Ashby thinks that these are fixed focus patterns rather than matters
of lexical stress. For EFL purposes it comes to the same thing.
4.21 train times
‘The ˈtrain at platform \/three | . . .’, says the automated train

announcement at the railway station, ‘is the four/teen | forty-/two | to \Woking.’
(In Britain we use the 24-hour clock for travel announcements.)
What’s wrong with that?
Well, it’s not how a live native speaker of English would actually say those
words. The announcement was obviously based on recordings of individual words
or phrases spoken by a human being but concatenated on the fly by a computer.
The announcer or actor being recorded must have gone through all the
numbers that were needed, probably using a rising tone until he got to the end
of the list.
/One, | /two, | /three, | . . . e/leven, | /twelve . . .
Being a trained speaker, he must have drilled himself to say each numeral as if
it were in isolation. Because he seems to have continued:
thir/teen, | four/teen, | fif/teen . . .
where a person counting would ordinarily use contrastive stress among the teen
numerals:
/thirteen, | /fourteen, | /fifteen . . .
The result is that the computer putting the message together could draw on
recorded instances of four/teen and forty /two but was not programmed to apply
‘stress shift’ (perhaps better termed ‘accent shift’), in the way that native
speakers do when a teen numeral (or any other double-stressed item) is followed
by another accented word:
ˈfourteen forty-ˈtwo
The irony is that if he’d let himself follow his instinct and had recorded:
/thirteen, | /fourteen, | /fifteen . . .
then everything would have come out fine.
The ˈtrain at platform \/three | is the /fourteen | forty-/two | to \Woking.
But there’d still have been a problem, because the version of each number above
eleven suitable for the hours would not have been suitable for the minutes:
(hours) /thirteen, | /fourteen, | . . . /twenty-one . . .
(minutes) thir/teen, | four/teen, | . . . twenty-/one . . .
126 intonation
The only satisfactory solution would be to have made two recordings of each
item, and to have built in to the software the equivalent of the stress-shift rule.
But unfortunately speech engineers tend not to consult phoneticians.
4.22 the rule of three
There are plenty of words in English that seem to change their stress
depending on the phonetic context. Typical examples are afternoon, unknown,
sixteen. We say the ˈlate afterˈnoon but an ˈafternoon ˈnap, ˈquite unˈknown
but an ˈunknown asˈsailant, ˈjust sixˈteen but ˈsixteen ˈpeople.
The usual explanation of this is that the words in question are lexically double-
stressed. Dictionaries show them with a secondary stress on the early syllable,
a primary stress on the later one, thus for example ˌɑːftəˈnuːn or àfternóon.
I think that really the two stresses are of equal lexical status. The supposed
difference between secondary and primary merely reflects the fact that when we
say one of these words aloud, in isolation, the intonation nucleus necessarily
goes on the last lexical stress, making it more prominent than the first.
The alternation goes by various names, including ‘stress shift’ and ‘iambic
reversal’, but I call the general principle involved the ‘rule of three’. This means
that when there are three successive potential accents (¼ syllables that could
be realized with pitch prominence plus a rhythmic beat), the middle one can be,
and often is, downgraded, losing its pitch prominence and possibly its rhythmic
beat too.
Thus a ˈnice ˈold ˈdog becomes a ˈnice old ˈdog, and ˈvery ˈwell deˈsigned
becomes ˈvery well deˈsigned. The ˈBˈBˈC becomes the ˈBBˈC, and our ˈafter
ˈnoon ˈnap becomes an ˈafternoon ˈnap. Likewise an ˈun(ˈ)known asˈsailant,
ˈsix(ˈ)teen ˈpeople.
Anyhow, the point of all this is that in Italy some years ago I got caught
out through applying the same principle, wrongly, to Italian. My room number
in my hotel was 202, which in English is ˈtwo ˈhundred and ˈtwo, which by the
rule of three becomes ˈtwo hundred and ˈtwo. In Italian it’s duecento due, which
I discovered does not become *ˈduecento ˈdue. It has to be dueˈcento ˈdue.
That’s because (most) Italian words can have only a single lexical stress. So ‘two
hundred’ is dueˈcento, not ˈdueˈcento.
The consequence is that English people sometimes put accents in the wrong
place in Italian, as I did; and conversely that Italians find it difficult to apply
the rule of three to English double-stressed words.
4.23 Are you asking me or telling me?
English intonation is not very good at making clear whether an

utterance is a statement or a question.
4.24 accent on a VP subordinator 127
(1) So he’s ˈright.

(2) So he’s ˈright?
The statement, (1), is typically said with a fall.
(10 ) So he’s \right.
But it can also have a rise, to show truculence or mere non-finality (not to
mention the new fashion of uptalk).
(100 ) So he’s /right, | what\ever you say.
The yes-no question, (2), ‘so do you mean he’s right?’ is said by default with
a rise.
(20 ) So he’s /right? | Is /that what you mean?
But it can also have a fall, if presented as a kind of challenge to the
interlocutor.
(200 ) So he’s \right!? | Is \that what you mean!?
You could also use a fall-rise, in either meaning. . .
A correspondent writes:
I heard an interesting example of contrastive intonations at the pub tonight.
Someone asked ‘What wines do you serve by the glass?’ and the barman
asked ‘White or red?’, but the customer misheard it as an uninformative
answer ‘White or red’.
At first I thought it was obvious: the answer has falling intonation but the
question has rising. But it doesn’t. The clarifying question ‘White or red?’ had
a fall-rise on both ‘white’ and ‘red’. So it’s rather more subtle than sentence-
question intonation, and I’m less surprised that the customer misunderstood.
4.24 accent on a VP subordinator

I continue to fret about anomalous tonicity. Consider the accenting of
the word to in exchanges like this, heard recently.
A: ˈWhat are you going to ˈdo?
B: There’s ˈnothing ˈto do.
Logically, you would think, B’s intonation nucleus ought to go on nothing. But
it doesn’t. It goes on the function word to. Why?
(We ought not to call this to a preposition. In its role as part of an English
infinitive, as here, Huddleston and Pullum call it a ‘VP subordinator’ in their
Cambridge Grammar of the English Language.)
Other ways for B to say the same thing in his answer include these.
128 intonation
B: There’s ˈnothing I ˈcan do.

B: There’s ˈnothing I ˈcould do.
Or the idea can be lexicalized.
B: There’s ˈnothing ˈleft.
B: There’s ˈnothing ˈpossible.
Or, going back to the first version, we can change nothing to not anything.
B: There ˈisn’t anything ˈto do.
This is another of those counterpresuppositionals (4.8 above), and one explan-
ation that has been offered is that the accented to signals marked polarity.
The speaker emphasizes the absence of anything that is doable, as against what
is not. So we have a marked positive.
However, there is a difficulty here. Taking the sentence as a whole, the speaker
is saying that nothing is possible. The polarity is negative, so it seems perverse
to talk of a marked positive.
A marked positive would be expected to be the dubiously acceptable:
B: ?There ˈis nothing to do.
Compare what happens if B gives an answer with negative polarity for do. The
nuclear accent goes on the negator.
B: There’s ˈone thing ˈnot to do. (And ˈthat’s reˈsign.)
B: There’s ˈone thing I will ˈnot do.
B: There’s ˈone thing I ˈcan’t do.
Notice that it is impossible with negated do to accent to.
B: *There’s ˈone thing not ˈto do.
B: *There’s ˈone thing ˈto not do.
I feel confused.
I gather that in German things are somewhat more straightforward. You don’t
accent zu where you would English to.
B: *Da ˈgibt’s nichts ˈzu tun.
Instead, you accent the word corresponding to English is.
A: ˈWas wirst du ˈtun?
B: Es ˈgibt (aber) nichts zu tun.
or
B: Da ˈgibt’s nichts zu tun.
Apparently, though, while you cannot have accented zu to show marked polarity,
you can have it to show marked tense. So the following might be OK.
4.25 on the train 129
B: Es ˈgibt nichts ˈzu tun, | es hätte schon ˈlängst geˈtan werden müssen.
‘There’s nothing to do (now), it ought to have been done long ago’.
How difficult it all is!
What happens with a slightly different wording?
A: What are you going to do?
B: There’s nothing to be done.
I think I could then accent be or to.
B: There’s ˈnothing to ˈbe done.
B: There’s ˈnothing ˈto be done.
4.25 on the train

Masaki Taniguchi has sent me a recording made aboard a train
from Canterbury to London. Here is a fragment of the announcement made by
the train manager to the customers (or, as we used to say, by the guard to the
passengers).
Good afternoon, ladies and gentlemen, and welcome.
We are interested not so much in the segmental details – the extreme reduction of
and gentlemen and – but in the intonation.
Clearly, the fragment ends with a falling nuclear tone on \welcome. But what
about the earlier part, the first IP up to gentlemen?
Hearing a high level pitch on good and a low, gradually rising pitch extending
from aft- to -men, Masaki wondered whether to analyse this as a high head accent
on good followed by a low-rise nuclear accent on af-.
ˈGood /afternoon ladies and gentlemen
The difficulty with this analysis, as he realized, is the unexplained stress (accent)
shift on afternoon. We expect to lose the accent on -noon before a following
accented syllable, but not otherwise.
We’ve come for ˈafternoon ˈtea.
It’s a ˈnice afterˈnoon.
*It’s a ˈnice ˈafternoon.
So what’s going on? After all, we don’t expect any accents at all in a routine trailing
vocative such as ladies and gentlemen. What causes the stress shift in this case?
In my opinion, there isn’t one. The correct analysis is:
¯Good ˌafter/noon, ladies and gentlemen
– the high pitch on good is a prehead, not a head (the word is not accented);
the low pitch on after- is a low level head; and the rising nucleus starts at its
130 intonation
expected place, -noon. That seems to be the only analysis that corresponds (i)
to the perceived pitches, and (ii) to what we know linguistically about
English intonation. Here’s a spectrogram of the utterance with a tracing of the
fundamental frequency.
Time (s) 0.0 0.2 0.4 0.6 0.8 1.0 1.2

8944
–9952
Hz 20k
15k
10k
5k
Hz
200
150
100
An
gUd A: ft 8nu:n 1eI =z en nwel k0 n
Time (s) 0.0 0.2 0.4 0.6 0.8 1.0 1.2
As is so often the case, a physical analysis of the speech signal doesn’t help us
resolve the issue. The high pitch we perceive on good doesn’t show up at all on
the frequency tracing, perhaps because I didn’t adjust the levels properly. Af- is
indeed physically lower than -ternoon, as Masaki perceived, and the whole
of -ternoon la- is pretty level.
Not only is the articulatory detail very reduced as compared to the underlying
linguistic facts, so is the fundamental frequency detail too. How marvellous
it is that the human perceptual mechanism can nevertheless reconstruct it all.
And how difficult for non-native speakers.
4.26 idiomatic nucleus on a pronoun
A few English intonation idioms involve a falling tone on a nuclear-

accented pronoun (English Intonation, p. 126).
4.27 not a drop 131
ˈGood for \you!

ˈGet \her!
ˈSearch \me!
There are more.
They’re ˈnot going to fool \/me.
You ˈcertainly told \/him where to get off!
I ˈdon’t think we’ll be hearing from \/them again!
I ˈdon’t expect any more trouble from \/you lot.
You ˈcertainly left \/her in no doubt.
I’ll ˈsoon show \/him what’s what.
I have shown these all with a fall-rise tone, though I think they can also, less
usually, have a simple fall.
The unspoken corollary of the fall-rise is something like the following.
They’re ˈnot going to fool \/me again (| though they ˈmay still fool \/you).
You ˈcertainly told \/him where to get off (| ˈeven if you didn’t say anything
to the \others)!
and similarly with the remaining examples.
Perhaps these are not exactly idioms. But they’re certainly cases in which the
location of the nucleus is unexpected | ˈeven for the ad\vanced EFL learner.
4.27 not a drop
‘Petrol is expensive’, warns the newspaper, ‘don’t waste a drop’.

If you say this aloud, the intonation nucleus goes on drop.
ˈDon’t waste a ˈdrop.
You would think the most important word would be waste, or even don’t. Yet we
place the main accent on neither of those items, but on the apparently rather
unimportant drop.
There’s no kind of contrast involved. We’re not contrasting drop with bucket-
ful or litre. So what’s going on?
Not . . . a drop is a more complicated way of saying nothing. Don’t waste a
drop is an idiomatic way of saying Waste nothing. It parallels other similar
expressions.
I ˈcan’t see a ˈthing. (¼ I can see nothing.)

I ˈwon’t tell a ˈsoul. (¼ I’ll tell no one.)
(Have you ˈheard from them?) – ˈNo, | ˈnot a ˈdickybird. (¼ nothing)
(Have you ˈheard from ˈTom?) – ˈNo, | ˈnot a ˈsausage.
I ˈhaven’t got a ˈdime. (¼ I have no money.)
132 intonation
The nucleus that would otherwise be located on nothing or no one is

retained on the dummy word that is left behind after the negative has been
fronted.
This is different from what happens with the dummy words I call ‘empty
words’ (4.7 above). They do not get an accent.
I keep ˈseeing things.

I ˈpromise I won’t ˈtell people.
Hurry up!! | ˈDon’t waste a ˈminute.
4.28 a little while
My Japanese colleague Masaki Taniguchi has been reading the Bible,

and asked me about the likely intonation of John 16:16.
In a little while you will see me no more, and then after a little while you will
see me.
His suggestion, which he asked me to comment on, was:

In a ˈlittle \/while | you will ˈsee me no /more, | and then \/after a little while |
you \will see me.
I told him that this was fine, except that I think I would use a fall rather than a rise
on more, since that is the end of Christ’s first declaration.
In a ˈlittle \/while | you will ˈsee me no \more, | and then \/after a little while |
you \will see me.
The reason for the choice of nucleus placement in the second half is that a little
while and see me are repeated (old, given) material, which in English means they
are likely to get deaccented.
I noted that the wording he quoted from was the New International Version. In
the Authorized Version (known to some as the King James version) that I was
brought up on it reads:
A little while, and ye shall not see me: and again, a little while, and ye shall
see me.
I then found an American audio clip of the chapter from which this verse is taken.
The reader chose the intonation pattern:
A ˈlittle \/while | and you will ˈnot \see me. || A\/gain a little while | and you
\will see me.
– which nicely vindicates Masaki’s suggestion, as amended by me.

Interestingly, in the Greek original there are two different words translated into
English as ‘see’.
μικρὸν καὶ οὐκέτι θεωρεῖτέ με, καὶ πάλιν μικρὸν καὶ ὄψεσθέ με.
mikron kai ouketi theōreite me, kai palin mikron kai opsesthe me.
‘little and no-longer you(pl.)-look-at me, and again little and
you(pl.)-will-see me.’
The first word, θεωρεῖτε theōreite, means ‘look (at)’ rather than ‘see’, and
is in the present tense. The second, ὄψεσθε opsesthe, is the ordinary verb ‘see’,
and is in the future tense. The first verb is from the stem that underlies our
modern word theory, the second from the stem that underlies (syn)opsis. These
subtleties are lost in the English versions. They are also lost in the Latin vulgate:
Modicum, et jam non videbitis me; et iterum modicum, et videbitis me.
– with videbitis (you will see) in both places.

A literal English translation would indeed be awkward.
A little, and you no longer look at me; and again a little, and you will see me.
I think it might affect the intonation, too.
4.29 international intonation
Most learners of English as an additional language (the latest expres-

sion for EFL/ESL) are not taught intonation and do not study intonation. Yet they
do not speak English on a monotone. A few may be gifted mimics who succeed
in imitating intonation along with everything else in the phonetics of the target
language. For most, though, their intonation patterns are presumably those of
their first language, transferred to English.
The same applies to English learners of foreign languages.
On the whole, even though this may make the speaker sound strange, typical
of their origin, boring or annoying, it seems not to cause much of an actual
breakdown in communication. How can this be?
It must be because the principles of intonation in language are sufficiently
universal for us to be able to rely on them even in a foreign-language situation.
However, it may also be the case that many of the nuances of meaning carried
by intonation in the speech of native speakers are overlooked or misinterpreted
by some or most non-native speakers. But this may also apply between different
native varieties of the same language: think only of Ulstermen interacting with
Londoners.
This leads me to ask which parts of intonation are universal and which are
language-specific. I am referring only to the linguistic (systemic) use of the pitch
of the voice, not the paralinguistic (presumably non-systemic) factors such
as pitch range, speech rate and voice quality.
Analysing intonation in terms of the three systems discussed above (4.1, 4.4),
we can say that the principles of chunking are probably pretty universal. Those of
tonicity (accentuation) clearly are not. Is choice of tone universal in any way? In
134 intonation
English, when we pronounce lists we signal that the list is incomplete by using a
rise, and that it is complete by using a fall.
(3) /One, | /two, | \three.
(4) Do you want /coffee | or /tea | or \cola?
Is it a universal that the signal for list completion (if there is one) is always a
falling tone?
Is it a universal that exclamations have a fall, pardon-questions a rise?
(5) \Marvellous!
(6) /What was that again?
How general is it that, as in English, wh-questions tend to have the same tone as
statements, while yes-no questions are different? Lastly, there is the matter of
tonicity (accentuation). By this I mean the location of sentence accents, and
particularly their use to signal focus. Two principles that apply in English and
seemingly in many languages are (i) that the last sentence accent within a given
stretch (the IP) is particularly important – which is why we call it the nucleus or
tonic – and (ii) that it signals the material on which the speaker chooses to focus,
i.e. mark as foreground as against background, new as against given, comment as
against topic, rheme as against theme.
What is not clear is that either of these principles is actually universal. Unless
contrastive focus is involved, it is not clear that all languages impose a focus
at all (it’s been claimed that Danish is like that, not to mention French). And
intonation is not the only way to signal focus, you can also do it syntactically.
This presumably is why nucleus placement is arguably the most complicated
part of English intonation, and the hardest to learn.
It is not difficult, in simple cases, to understand the different meanings
associated with switching narrow focus around.
A: What did you see? A red car?
B: (i) No, | a red bus.
B: (ii) No, | a green car.
But here the focus is obvious from the wording.
In core English we readily exploit tonicity to mark focus in examples like this:
A: I want to buy a comb.
B: But you’ve already got a comb!
Yet this doesn’t seem to happen, or not as much, in African English.
In English we are so anxious to avoid accenting repeated items that we say:
Would you like your coffee with milk | or without milk?
But in Spanish, I gather, this would be:
. . .con leche | o sin leche?
In languages with a freer word order than English you can often use a change
of word order to show focus instead of, or as well as, using intonation.
In an interesting unpublished experiment an MA student at UCL tested
English people and speakers of Cantonese on whether they could correctly match
up question and answer in pairs such as this (quoted from memory, to the best of
my recall):
Q1. Have you any special dietary requirements?
Q2. Would you prefer beef or pork?
A1. I don’t eat meat.
A2. I don’t eat meat.
(A1 goes with Q1, A2 with Q2.)
The English score was near perfect, the Chinese no better than random. So no
universal here.
Daniel Hirst and Albert di Cristo contributed an introductory article ‘A survey
of intonation systems’ to the book they edited, Intonation Systems (Cambridge
University Press 1998). Their first sentence ends: ‘. . . intonation is paradoxically
one of the most universal and one of the most language specific features of
human language’. We can all agree with that.
These points about intonation in EFL apply equally to intonation in Esperanto:
somehow speakers manage to understand one another in the language very well
despite the lack of any agreed, taught or described intonation system.
5 Symbol shapes, fonts and spelling
Topics in the design of the International Phonetic
Alphabet and in English orthography
5.1 labiodental flap
In 2005 the International Phonetic Association approved a new

phonetic symbol, to represent a voiced labiodental flap. It is like a letter v, but
with a hook on the right-hand side, thus ⱱ. It is at Uþ2C71 LATIN SMALL
LETTER V WITH RIGHT HOOK.
It was adopted by the IPA too late to make it to the most recent printed
Unicode standard, 5.0. However, it was added for the Unicode 5.1 revision, and
you’ll find it in the online Unicode charts at www.unicode.org/charts. It may not
yet be found in the Microsoft fonts routinely supplied with new PCs, nor in
the Lucida Grande that comes with Mac computers. So to use it you may need
to download and install a specialist font. Try the SIL fonts Charis SIL and
Doulos SIL at www.sil.org/computing, which you can download free. (You will
then need to unzip the files and install them.)
Tiresomely, Unicode did not add it to one of the blocks where phonetic
symbols are usually found (IPA Extensions, Uþ0250–Uþ02AF; Phonetic
Extensions, Uþ1D00–Uþ1DBF). Instead, it is in a small block called Latin
Extended-C (Uþ2C60–Uþ2C7F), nestling unobtrusively between Glagolitic
and Coptic. Within that block it lies among the ‘miscellaneous additions’ located
between ‘Additions for Uighur’ and ‘Claudian letters’ (Latin letters invented by
the emperor Claudius but soon dropped).
Early versions of Doulos SIL did not contain the symbol. Nor does Lucida
Sans Unicode. Nor, of course, did any of the old single-byte fonts such as SIL
Doulos or the ipa-sam fonts that I developed many years ago.
SIL also offers Mac users a keyboard facility called Ukulele that will enable
them to insert all the Unicode symbols they need. Users of Word/Windows
don’t really need any such extra software, since they can immediately ‘see’ all
installed Unicode characters with Insert Symbol, and quickly create macros using
AutoCorrect. However, for those who like a keyboard facility, SIL offers one
called Keyman. My colleague Mark Huckvale has made a rather simpler one
called simply Unicode Phonetic Keyboard, which you can download from www.
phon.ucl.ac.uk/resource/phonetics.
136
5.2 similar symbols 137
5.2 similar symbols

The IPA symbol for the vowel of English foot, put, good, ʊ, some-
times gives trouble to authors, printers and publishers. It tends to get confused
with the symbol for the labiodental approximant, ʋ.
So it is important that font designers (and the rest of us) differentiate clearly
between the two symbols. The foot symbol, known to Unicode as Uþ028A
LATIN SMALL LETTER UPSILON, is symmetrical about the vertical axis.
Despite the name, it is not so much a Greek upsilon (υ) as an inverted small
capital omega (Ω).
The labiodental approximant symbol, on the other hand, is asymmetrical, with
the right side having a leftward-pointing lip or hook. In Unicode it is Uþ028B
LATIN SMALL LETTER V WITH HOOK.
There are other IPA symbols which authors and printers often confuse with
other letters or symbols:
θ and ɵ voiceless dental fricative (Uþ03B8); rounded schwa (Uþ0275);

ɣ and ɤ voiced velar fricative (Uþ0263); close-mid back unrounded
vowel, cardinal 15 (Uþ0264);
ɑ and a back open vowel, cardinal 5 (Uþ0251); front open vowel,
cardinal 4;
ə and ɘ schwa (Uþ0259); close-mid central unrounded vowel
(Uþ0258);
ɫ and ɬ voiced velarized alveolar lateral, dark l (Uþ026B); voiceless
alveolar lateral fricative (Uþ026C);
ŋ and ɧ voiced velar nasal (Uþ014B); voiceless postalveolar-velar
fricative (Uþ0267).
It is also important not to confuse IPA symbols, which are all lower-case, with
the upper-case versions used in the orthographies of certain languages:
ɪ not I lax close front unrounded vowel (Uþ026A); and upper-case I;
ʊ not Ʊ lax close back rounded vowel (Uþ028A); and Uþ01B1;
ɛ not Ɛ open-mid front unrounded vowel, cardinal 3 (Uþ025B); and
Uþ0190;
ɔ not Ɔ open-mid back rounded vowel, cardinal 6 (Uþ0254); and
Uþ0186;
ə not Ə schwa (Uþ0259); and Uþ018F;
ʒ not Ʒ voiced postalveolar fricative (Uþ0292); and Uþ01B7;
θ not Θ or Ɵ voiceless dental fricative (Uþ03B8); and Uþ0398 or
Uþ019F.
Lastly, be careful to write:
ɑ not α unrounded back open vowel (Uþ0251); Greek alpha
(Uþ03B1);
138 symbol shapes, fonts and spelling
ɣ not γ voiced velar fricative (Uþ0263); Greek gamma (Uþ03B3);

ð not δ voiced dental fricative (Uþ00F0); Greek delta (Uþ03B4);
ɛ not ε unrounded open-mid front vowel (Uþ025B); Greek epsilon
(Uþ03B5);
ŋ not η voiced velar nasal (Uþ014B); Greek eta (Uþ03B7);
ʃ not ʅ voiceless postalveolar fricative (Uþ0283); retroflex apical
vowel (Uþ0285).
5.3 clicks
Clicks are sounds articulated with an ingressive velaric (oral) air-

stream mechanism. As such, they necessarily have a simultaneous velar closure.
When we write ǁ, say, to symbolize a voiceless alveolar lateral click (former IPA
symbol ʖ), we imply a simultaneous k. The click articulation is produced during
the hold stage of a voiceless velar plosive.
In the case of a voiced click, e.g. ɡǁ, or a nasalized click, e.g. ŋǁ, we symbolize
the velar component explicitly. (Many people show the simultaneity explicitly,
too, by linking the two symbols with a tie: ɡ͡ ǁ, ŋ͡ ǁ.)
Treating the voiceless clicks as symbolically unmarked and the voiced or
nasalized ones as marked has its parallel in the orthographies of the click
languages. Zulu ǁ is written as x, ɡ͡ ǁ is written as gx, and ŋ͡ ǁ is written as nx.
There are also the voiceless aspirated ǁʰ, written xh, and the breathy-voiced
(depressor) nasalized ŋ̤͡ǁ, written ngx.
Some phoneticians symbolize all clicks by a double symbol, always making
the nature of the velar component explicit. They write the voiceless alveolar
lateral click as k͡ ǁ, and likewise for the other places of click articulation. This
makes the treatment of all classes of click parallel, at the expense of necessitating
an extra letter for each voiceless click. So is it better to transcribe the Zulu for
‘no’, cha, as kʰ͡ ǀa or just as ǀʰa? Do we transcribe Zulu xoxa ‘narrate!’ as k͡ ǁɔk͡ ǁa,
or is ǁɔǁa sufficient?
In click languages there are no click consonant clusters. So it doesn’t
matter that the symbol for an alveolar lateral click, [ǁ], is barely distin-
guishable from two successive instances of the dental/alveolar click, [ǀ].
Unfortunately the English non-speech interjection sometimes written
‘tut-tut’ (to show disapproval) may consist of just that: two successive central
coronal clicks. To show it you must be careful to use good spacing between
the two click symbols . . . or write each with its non-click component,
thus k͡ ǀk͡ ǀ.
Or, like some reference books, you may think that the tut click is properly
symbolized [!s]. But it sounds to me not like Nama !, which always has an abrupt
release and is apical, but more like Nama |, which is laminal and somewhat
affricated.
5.5 dashes 139
5.4 IPA capitals

Kera is a language spoken by some 50,000 people in Chad. It is a
tonal language, with three tones (high, mid, low). It has no voicing contrast
in consonants, but VOT (voice onset time) varies depending on the tone of the
following vowel: so a velar plosive before a high or mid tone sounds like k,
but before a low tone like ɡ.
The SIL organization (sil.org) is involved in both Bible translation and
a literacy project, and has produced both the New Testament in Kera (JAA
TƏMARWAŊ) and a reading primer (Aŋ aŋkə ku keera la). As you can see, Kera
orthography uses the Latin alphabet supplemented by IPA symbols: not only
ə and ŋ but also ɓ and ɗ. Following the conventions of Latin-based ortho-
graphies, each letter requires an upper-case as well as a lower-case form: hence
Ə Ŋ Ɓ Ɗ. So although phonetic symbols are normally lower-case, and not subject
to the usual capitalization rules for proper names or the start of a sentence, we do
need capitals when we use IPA symbols outside a phonetic context.
Some Kera proper names have clearly passed through French. The names
of the four evangelists are Matiye, Markə, Liki and Zaŋ.
5.5 dashes
What do you call the punctuation mark ‘-’, as in re-enter?

I have always called it a hyphen. So I was surprised when I heard native-
English-speaking undergraduates refer to it as a ‘dash’. My reaction was that this
was a simple error, and if they did it in writing I corrected them. A dash, for me,
is the similar but longer sign ‘–’.
This was the terminology I was taught at school—and this is the usage
of professional typographers and indeed phoneticians. After all, Daniel Jones’s
1955 article (Zeitschrift für Phonetik 9) was entitled ‘The hyphen as a phonetic
sign’, not ‘The dash as a phonetic sign’. MS Word’s AutoCorrect options include
‘Replace hyphens (-) with dash (–)’. This and other word-processing software
usually includes a hyphenation function that can be toggled on or off. The
distinction between ‘hyphen’ and ‘dash’ is a useful one.
Recently I have come across the dash-for-hyphen solecism even in serious
printed works.
I see that the Unicode reference materials use ‘dash’ as a general term for
horizontal line punctuation of various widths: ‘Unicode Dash Characters’
include:
hyphen (-);
en dash (–), the same width as the letter n;
em dash (—), the same width as the letter m, and others.
Another problematic question is whether the dash ought to be surrounded

by spaces – like this – or not (as a few lines above), when used to make a break in
the flow of a sentence. I’m happy to go either way on that one.
5.6 linguolabial trill
(I posted this note online on the first day in April 2006.)

I hear from my colleague Olaf Lipor that the International Phonetic Asssocia-
tion is considering recognizing a further new symbol, in order to cater for the
voiced linguolabial trill, a sound-type recently discovered to be used contras-
tively in Caslon and Ki-Flong, languages spoken on the island of San Serriffe.
Linguolabials, articulated by the tongue tip against the upper lip, are very rare
in the languages of the world. Nevertheless linguolabial plosives, fricatives and a
nasal are known to occur in a cluster of languages in the island state of Vanuatu.
Among these languages are Tangoa and Vao. But until now there had been no
report of a linguolabial trill.
The way in which the IPA is expected to symbolize the new sound is with the
‘combining seagull below’ diacritic, Uþ033C, thus r̼ .
We are hoping to have the Serriffean phonetician Dr Charis Doulos, a native
speaker of Caslon and the person who first described the linguolabial trill,
come to UCL Phonetics & Linguistics as an academic visitor at this time next
year. She will no doubt be willing to act as a language consultant for our
practical phonetics class, so that the students can have the opportunity of
observing the sound first-hand and of learning to perform it to the native
speaker’s satisfaction.
San Serriffe sprang to world fame as a consequence of a feature article in
the Guardian newspaper, published on 1 April 1977, the tenth anniversary of its
independence. But at that time its native languages had not been thoroughly
investigated.
(Subsequently there appeared a report of an Amerindian language, Santo
Domingo Coatlán Zapotec, in which this sound is now, excitingly, further
attested, though disappointingly not as part of the phonemic inventor, but only
in an ‘onomatopoeic word’ denoting the ‘sound of a child’s fart’.)
5.7 the symbol ɮ
In 1938 the Council of the IPA took a decision concerning the

symbol for the voiced alveolar lateral fricative. It read as follows (m.f. 61, p. 15):
5.8 scenes from IPA history 141
Bear in mind that in those days the official language of the Association was
French, and that its journal Le Maître Phonétique was printed entirely in phonetic
transcription. So this decision is expressed in phonetically transcribed French.
Translated, it reads as follows:
The proposal to replace ɮ by [the second symbol shown] is therefore
approved. We think nevertheless that the form proposed by Chatterji can
be employed without disadvantage by those who prefer a symbol less remote
from the old ɮ.
Chatterji’s suggestion was a compromise between the ʒ-shaped right-hand side

of the ‘old’ symbol and the straight side of the ‘new’. The compromise shape
is what Daniel Jones adopted for the symbol in the 1949 Principles booklet
(p. 11, also p. 50):
However, this decision by the IPA seems subsequently to have been overturned
as we moved from hot metal typography to software fonts. After the 1989 Kiel
Convention of the IPA both printed and on-screen usage reverted in practice to
the older shape. The IPA Handbook presents the symbol as an ‘L-Ezh ligature’
and our current chart shows the form ɮ. Unicode calls it LATIN SMALL
LETTER LEZH (Uþ026E). All available computer phonetic fonts show this
same ɮ shape.
By the way, the Unicode documentation (www.unicode.org) also repeats the
Principles booklet’s assertion that ɮ is dhl in Zulu orthography. But Zulu has had
a spelling reform, and (as rightly recognized in the IPA Handbook) the current
spelling is dl. The site of the famous battle that used to be spelt Isandhlwana is
now Isandlwana. It’s still pronounced ísanˈdɮwáːna. (After n you get a laterally
released affricate rather than a lateral fricative.)
5.8 scenes from IPA history
Here, extracted from back issues of Le Maître Phonétique, is a list of

the changes to the IPA Alphabet agreed during the 1940s.
1943: the symbols ɩ and ɷ were introduced as alternatives to ɪ and ʊ
respectively, which nevertheless remained ‘authorized alternatives’
and ‘could be used instead of the new symbols, except in the official
publications of the IPA’. This ban was not maintained in practice after
Jones’s retirement. Recognition of ɩ and ɷ was withdrawn in 1989.
1945: the symbol ɼ (voiced dental/alveolar fricative trill) was
approved as the replacement of ř. Recognition of the new symbol
was withdrawn in 1989.
1945–7: vowel symbols with a retroflex hook were introduced, to

represent r-coloured vowels: ᶏ ᶐ ᶒ ᶓ ᶔ ᶕ ᶖ ᶗ. Recognition was
withdrawn in 1989.
1947: the alveolar affricate symbols ƾ and ƻ were approved, ‘but only
by a small majority’. Recognition was withdrawn in 1976.
1947: the symbols ʆ and ʓ were introduced for palatalized ver-
sions of ʃ, ʒ. They were considered necessary for Russian, which
Daniel Jones was then working on. Recognition was withdrawn
in 1989.
1947: the diacritic ˍ , or a serifed variant, was approved to denote
retraction, including use with t, d, n (thus ṯ ḏ ṉ ) to denote
alveolars in languages where they were distinct from, and rarer
than, dentals.
1947: the symbol ƞ was approved for the Japanese ‘syllabic’ (i.e.
moraic) nasal. ‘It will nevertheless be desirable to consult Japanese
phoneticians [. . .] and if necessary to revise our decision when we
know what they think of it.’ Withdrawn 1976.
1947: the symbol ɧ was approved for the ‘combination of x and ʃ
(one variety of Swedish tj, kj, etc.)’. (See 3.5 above.)
1948: the symbol shape g (spectacles) was approved as an alternative
to ɡ (nine). ‘They will be considered as typographic variants of the
same letter’, except in Russian, where the former could symbolize the
non-palatalized voiced velar plosive, the latter the palatalized. Where
separate letters were needed for ordinary and advanced g, as in the
narrow transcription of Russian, g (spectacles) was recommended for
the ordinary and ɡ (nine) for the advanced, i.e. palatalized. I don’t
think anyone ever applied this distinction in practice. Certainly Jones
himself didn’t: see below what the symbols look like in his The
Phonetics of Russian (co-authored with Dennis Ward, Cambridge
University Press 1969).
Judging by the comments I received on the publication of LPD, where I use g

(spectacles) for the voiced velar plosive, some phoneticians are unaware that the
decision to treat the two shapes of that letter as glyph variants, equally valid, was
ever taken.
In fact, we see that most of the decisions taken during that period of fiddling
around with the Alphabet were subsequently either overturned or ignored.
5.10 orthographic schwa 143
5.9 affricates
A correspondent was worried that the IPA allows for the representa-
tion of the palatoalveolar affricates as c, ɟ instead of what s/he considers to be the
correct way of writing them, namely ͡ tʃ, d͡ ʒ.
Why does this provision exist? It is because there are some languages in which it
doesn’t seem very satisfactory to write affricates with the plosive-plus-fricative
notation. A speaker of Italian, for example, reports that he is very conscious of the
difference in tongue configuration between the ordinary Italian plosive t and the
first element of the Italian affricate spelt c(i), c(e), usually represented in IPA as tʃ.
He would be happier with a notation that does not imply their equivalence. That is
what the use of c, ɟ for the affricates provides, as was implicitly recommended in the
1949 Principles of the IPA booklet, similarly to that of the then official ƾ and ƻ. You
can only do this, of course, in a language in which you do not need to symbolize
a voiceless palatal plosive, which is the default general-phonetic meaning of c.
Another such language is Hindi, in which the affricates very obviously pattern
as single units, not as sequences.
Many linguists use the symbols č, ǰ for these affricates, although they do not
have the approval of the IPA.
In ordinary orthography, although the Latin alphabet offers no way to write affri-
cates without using diacritics or digraphs, other alphabets do: for example, in Cyrillic
the voiceless palatoalveolar affricate is written Ч ч and the voiced one (in Serbian) as
Џ џ. There is clearly a perceived need for a unitary way of writing these affricates.
Returning to the two-symbol notation, my correspondent also assumed that the
correct way to write them is with a tie bar: ͡ tʃ, d͡ ʒ. Personally, I normally omit
the tie bar, and write just tʃ, dʒ. That is what you find in most pronunciation
dictionaries and textbooks, too. It does involve the convention that a sequence
of plosive plus fricative that does NOT form an affricate must be written some
other way. Daniel Jones does this with a hyphen (see his article ‘The hyphen as
a phonetic sign’, Zeitschrift für Phonetik 1955, 9), thus t-ʃ, d-ʒ. This enables
us to show the difference in the Polish minimal pair trzy t-ʃɨ vs. czy tʃɨ (though
the former is perhaps better analysed as tʃʃɨ, i.e. affricate plus fricative).
In English any such sequence must straddle a syllable boundary, so you can
show it using a full stop, as in Wiltshire ˈwɪlt.ʃə vs. vulture ˈvʌltʃ.ə (if you think
I am right about English syllabification) or ˈvʌl.tʃə (if you don’t).
Another way is to use the ligatured symbols ʧ, ʤ for the affricates, leaving the
separated tʃ, dʒ for the non-affricate plosive-plus-fricative sequences.
5.10 orthographic schwa
Which language makes use of the schwa symbol, ə, in its everyday

orthography?
There may be various such languages only relatively recently reduced to
writing by missionaries or linguists. It is difficult to find information about
how many they might be, or what they are called. I do know that it is used in the
Latin Chechen alphabet, though this language is more usually written in Cyrillic.
However, there is one well-established language, spoken by tens of millions of
people, that uses the schwa symbol: namely the Turkic language of Azerbaijan,
Azeri, also known as (North) Azerbaijani. You can see plenty of examples on the
web if you go to the Azərbaycan (sic) version of Wikipedia.
The schwa letter comes in upper-case as well as lower-case. However, the
sound denoted by Azeri Ə, ə is not the IPA’s mid central vowel but a front open
vowel, æ.
Strangely, Unicode includes three identical-looking versions of the schwa
symbol. One is, as expected, in the IPA Extensions block at Uþ0259. Another
is in the Cyrillic block, at Uþ04D9, with an upper-case version at Uþ04D8 –
this is fair enough, given that schwa is also a Cyrillic letter, used when writing
Azeri (and some other languages) in Cyrillic. The third, however, is at Uþ01DD,
and belongs to the Pan-Nigerian alphabet; it is called ‘turned e’ and is distin-
guished from the ordinary schwa inasmuch as it has a different upper-case
form (Ǝ). This is also used in the transliteration of Avestan. ‘All other usages
of schwa are 0259 ə’, says the Unicode standard; and this basic schwa has a
similar but larger upper-case form, Uþ018F Ə.
Although the current official orthography in Azerbaijan is Latin-based, during
Soviet times it was Cyrillic, and before that Arabic. There are also millions of Azeris
living in Iran, but their Southern Azerbaijani has its own distinctive phonology,
lexicon, morphology, syntax and loanwords. They mostly use Arabic script.
5.11 placing the suprasegmental symbols
Where should phonetic symbols for suprasegmental features (tone,

stress, etc.) go?
One of the things in LPD that people knowing Chinese have commented
on is the fact that in transcriptions of Chinese words I decided to place the tone
mark at the beginning of the syllable in question, rather than over the vowel or
at the end.
The sinologist tradition is to use a trailing tone numeral, thus ma2 (‘hemp’,
with tone 2, a rise). This is also how you input tones in the Chinese version of
Word (the word processor). In printed Hanyu Pinyin, however, the tone mark, if
used, goes over the vowel, thus má. If I had wanted to include this word in LPD,
I would have transcribed it as 2ma, with the tone number superscript and leading.
So: is the best place for the tone mark trailing (at the end), central (in the
middle, superscript), or leading (at the beginning)?
My motivation in choosing to write tone marks as leading was for consistency with
the IPA placement of stress marks in stress languages. For the same reason, I place the
pitch accent marks for Japanese at the beginning, not the end, of the mora concerned.
The IPA, too, used to place tone marks at the beginning, thus ˊma. The tone
symbols were iconic, i.e. their shape suggested the relevant pitch movement.
5.12 why Hawai’i? 145
This was changed at the 1989 Kiel Convention to the current IPA rule, which is
to place a non-iconic mark over the vowel, thus mǎ, or to use an iconic Chao
tone letter, presumably trailing, thus ma /ǀ.
I am convinced that the best place for all such prosodic marks is at the
beginning of the syllable, before the symbols for the segments. In order to
pronounce a syllable with the appropriate tone, you have to know what the tone
is before you start. You can’t wait until you’ve finished making the vowels
and consonants and then add the tone as an afterthought. Logically, the tone is
the first thing you need to know about a syllable. Articulatorily, for the Chinese
word for ‘hemp’, ma with a rising tone, you have to get low pitch in place right
at the beginning, so that the m is low-pitched, ready to start the rise. Acoustic-
ally, the F0 in this word is lowest at the beginning of the initial nasal. You’re
too late if you start to think about the tone only when you get to the vowel
(Hanyu Pinyin) or when you reach the end of the syllable (Sinologist practice).
The same applies to stress marking. You need to know whether a syllable is
stressed before uttering it, not afterwards. That’s the justification for the IPA
practice, revolutionary in its day, of placing the stress mark at the beginning of
the stressed syllable, as against the traditional English practice, still found in
some dictionaries, of placing it at the end.
In my view the IPA’s Kiel change in tone notation was a misguided concession
to the Africanist and Sinologist traditions. (They disagree with one another,
anyhow. In the Africanist tradition, for a rising tone – Low, High – we have
to write mǎ.)
5.12 why Hawai’i?
The name of the American state Hawaii is often written

Hawai’i. Why?
Phoneticians will not be surprised to learn that the apostrophe reflects a glottal
stop in the Hawaiian (or Hawai’ian) pronunciation of the name: haˈwaiʔi,
haˈvaiʔi.
However, they may be surprised – I certainly was – to be told that the preferred
way of writing this is as a LEFT apostrophe (an OPENING quote), i.e. Hawai‘i,
not Hawai’i. The Hawaiians call this letter the ‘okina.
Its shape is surprising, given that the IPA symbol for the glottal stop
was chosen for its resemblance to a RIGHT apostrophe (a CLOSING quote):
the top of <ʔ> has the same shape as <’>. Correspondingly, the top of
the symbol for the so-called voiced pharyngeal fricative, the Arabic ‘ayn
<>‫ﻉ‬, namely <ʕ>, has the same shape as <‘>. This reflects scholarly
usage of <’> and <‘> in the romanization of Arabic and other Semitic
languages.
The spelling <‘> for the glottal stop is part of the official orthography not
only of Hawaiian but also of the related Polynesian language Tongan.
Hawaiian is well known for having very few consonant phonemes and no
consonant clusters. There are just eight consonants: p, k, ʔ, h, m, n, l, w. Notice
that the glottal stop is one of them, and the difference between it and zero can
distinguish words. Here are some minimal pairs taken from Pukui and Elbert’s
New Pocket Hawaiian Dictionary, which I bought when I visited the islands
some years ago.
pau ‘finished’ pa‘u ‘soot’

kau ‘to place’ ka‘u ‘mine’
koi ‘to urge’ ko‘i ‘adze’
On the other hand, t and k are in free variation.

The paucity of consonants and absence of consonant clusters makes for some
interesting adaptations of European names. My name, John, can become Keoni,
Ioane (in the Bible), or. . . Keanu (which latter is also a Hawaiian word meaning
‘coolness’). Now you know where Keanu Reeves gets his name from.
Towards the end of December, Hawaiians say Meli Kalikimaka.
5.13 funny letters
In the village outside Birmingham where my brother lives there is a

dress shop. The name board on the fascia reads MAΨA.
My schooling was unusual in that I started to learn (Classical) Greek at the age
of twelve. I went on to read Classics at university. So I have been very familiar
with the Greek alphabet ever since childhood. Naturally, therefore, I read this
name as ˈmæpsə. After all, the letter between the two As is a Greek psi.
But no, this is just a fancy way of writing MAYA. The name of the dressmaker
is Maya, ˈmaɪə, like the Central American Indians.
Maya is following the same path as people who invite you to visit GRΣΣCΣ,
where the Greek capital sigma does duty for E. Indeed, there is a well-known
5.14 anomalous g 147
chain of toyshops called TOYS ‘Я’ US, whose logo exploits the similarity
between the Cyrillic capital Ya and the Latin R.
And phoneticians who ought to know better sometimes use a Greek eta, η,
instead of the velar nasal symbol ŋ.
5.14 anomalous g
We all know about the spelling-to-sound oddity of:

gaol dʒeɪl, alternatively (and more logically) spelt jail;
and margarine ˌmɑːdʒəˈriːn.
Their oddity consists in the fact that the letter g normally corresponds to dʒ
(rather than to g) only when followed by e, i or y. Even that applies mostly
to words of Romance rather than Germanic origin: compare gem and get.
In gaol and margarine, though, we have g ¼ dʒ even when followed by
something else.
Ted Carney, in his comprehensive Survey of English Spelling (Routledge
1994, p. 327) also mentions:
veg vedʒ, the colloquial abbreviation of vegetable – compare the

abbreviation of refrigerator, usually now spelt fridge;
and (rather strangely) syringa, which nevertheless for me and all
other reference books I can lay my hands on has g, not dʒ.
But now I have a number of further candidates for this category – words that
I have noticed native speakers pronouncing with dʒ even though the spelling
would suggest g.
Meningococcal: here it is clearly the influence of meningitis with
regular dʒ that leads people to say meˌnɪndʒəˈkɒkl̩. The word
frequently occurs in the phrase meningococcal meningitis.
Digoxin: similarly, it must be the influence of digitalis and digit that
leads people to call this heart drug dɪˈdʒɒksɪn. (That’s what everyone
at my cardiac support group calls it.)
Purgative, which I heard on the radio the other day as ˈpɜːdʒətɪv
rather than the usual ˈpɜːgətɪv: it must be the influence of purge.
Analogous: I have quite often heard it pronounced with dʒ, no doubt
because of analogy with analogy.
What we have in each case here is morphological regularization. The mor-
pheme in question (mening-, dig-, purg-, analog-) keeps the same phonetic
shape (with dʒ) despite suffixation. Chomsky and Halle’s rule of Velar
Softening clearly does not work quite as regularly as some would have us
believe.
5.15 the spelling wor
We have a number of words spelt worC, where C is a consonant

letter. Some of them are of quite high frequency. In almost all, the vowel is ɜː
(NURSE) rather than the vowel that usually corresponds to or, namely ɔː
(FORCE). They include word, work, worth, worm, worse, worst and worship.
We find ɔː, on the other hand, in the verb forms wore, worn, swore, sworn.
The noun sword sɔːd is clearly orthographically irregular, as is Worcester
ˈwʊstə. These are exceptions to the general rule above.
Overall, you’d think that the equivalence worC ¼ wɜːC was pretty strong, and
that speakers would apply it to new words of unknown pronunciation. Strangely,
though, that does not happen.
In Wimbledon where I live there is a road called Worple Road. Everyone calls
it ˈwɔːpl̩.
And now there is the newly popular herbal remedy St John’s wort. I have known
this plant since boyhood, when my mother taught me the names of the wildflowers
around us. She called it sn̩ ˈdʒɒnz wɜːt, as I do in turn. But the herbal remedy
enthusiasts all seem to call it seɪnt ˌdʒɒnz ˈwɔːt. I wonder if they would do the
same for the many other compounds of -wort that I remember from my child-
hood, all pronounced -wɜːt: glasswort, woundwort, sneezewort, ragwort, mugwort,
bladderwort, liverwort and so on. They’ve probably never heard of them.
This new form wɔːt is presumably a spelling pronunciation. Yet its rise is
mysterious: you would think that the large number of common words spelt with
wor- and pronounced wɜː – word, work, world, worm, worse, worth – would
rather reinforce the traditional pronunciation, the one my mother used.
5.16 practicing (sic) wot I preach
At a question-and-answer session on a UCL Summer Course in

English Phonetics one of the students asked about the possibility of reforming
5.17 why ph? 149
the chaotic and inconsistent spelling of English. I expressed the view that this
would be a Good Thing, though for social and political reasons difficult
to achieve. Jack Windsor Lewis then challenged me – given that I, like Daniel
Jones before me, was President of the English Spelling Society (formerly the
Simplified Spelling Society) – to publish some of my writing in reformed
spelling instead of in traditional orthography. He claimed that he himself always
writes re’d for the past tense of the verb to read.
So heer goes. In my view, English spelling reform shoud be gradual rather
than raddical. As Patricia Ashby suggested, we aut to be like the Dutch and hav
a minor reform evry few yeers rather than a big bang awl at wunce.
The Society has a House Stile, wich probbably goes further than we woud
wont to go in the first instance, tho not as far as we mite wish to go in the end.
The most important principals in this House Stile ar these.
No chanje is compulsory. U can go on using the traditional orthog-
grafy if u prefer.
Redundant letters ar cut: giv, hav, det, thum, thaut. We retain the use
of e to sho that a preseeding vowl is long: drive, shave.
Consonants are dubbled after stressed short vowels consistently
rather than sporaddicly: rappid, coppy, nevver, sitty, reddy (just as
in happy, coffee, etc.). Dubbling is also made consistent in -ing
forms: givving, driving.
Few!
These proposals are probbably at leest too hundred yeers old.
5.17 why ph?

Why do we write ph, not f, in words of Greek origin?
As usual with the oddities of English spelling, the reason is historical.
Although the letter phi (ϕ) of Modern Greek is pronounced f, and usually
transliterated into English as f, the phi of Classical (Ancient) Greek is transliter-
ated 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 same applies to th for theta (θ), as in Theocritus, theory and athletic,
classically an aspirated th. In English, as in Modern Greek, this has become a
fricative, θ.
However, ch for chi (χ), as in Achilles, echo and Christ, classically an
aspirated kh, has reverted to being a plosive in English, since we don’t have
the fricative x in our modern English consonant system.
(The evidence in support of this claim is summarized in W. Sidney Allen’s
Vox Graeca, Cambridge University Press 1987, third edition, pp. 16–27.)
The voiced plosives of Ancient Greek have become fricatives, too, in Modern
Greek, so that classical Euboea is now called Evvia, and as a modern placename
Delphi (the home of the Delphic oracle) is sometimes transliterated Dhelfi.
Just occasionally, a classical education is really useful.
This does not cover all cases of the English spelling ph. It doesn’t explain
the ph in caliph, nephew or various other words.
Caliph comes from Arabic khalīfah via medieval Latin calīpha. Probably
the Romans 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 erudite 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.
Then there is 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 was Mac an Phearsain, or in English Macpherson.
As for the 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’.
5.18 yogh and ezh

My colleague John Maidment says he’s ‘slightly annoyed’ at the
pronunciation of Beijing with ʒ instead of dʒ, and similarly in raj, adagio and
several other borrowed words that have an affricate in their original language.
‘This creeping yoghism should be stamped out’, he says.
His neologism refers to yogh, not to yoga. However, yogh and ezh are different
letters. And the Unicode name for ʒ is ezh. So perhaps he ought to be protesting
against ‘creeping ezhism’.
In Unicode, the phonetic symbol ʒ has the number Uþ0292 and the full
name LATIN SMALL LETTER EZH. Yogh, on the other hand, i.e. the letter
ȝ, has the Unicode number Uþ021D. This letter was used in Middle English and
Middle Scots to stand for a velar fricative or palatal approximant. 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, in a few Scottish words and names,
it came to look just like z, which is why we have a spelling-to-sound mismatch
in such names as Dalziel diˈ(j)el and Menzies ˈmɪŋɪs.
5.20 low and long 151
Ezh has an angled top, while yogh has a curly top.

The confusion between the two names is found, I am sorry to say, in the
Phonetic Symbol Guide by Geoff Pullum and Bill Ladusaw (University of
Chicago Press 1996, second edition), where the phonetic symbol is unfortunately
labelled YOGH. But the IPA Handbook calls it Ezh or Tailed Z.
5.19 bevy, bevvy
One of the conventions of English spelling concerns consonant

doubling. In derived and inflected forms we double a consonant letter between
two vowel letters to show that the first vowel is short. So we have shop, bat,
swim but shopping, shopped, shopper; batting, batted; swimming, swimmer. In
morphologically simple words, however, consonant doubling is extremely
erratic: rabbit but robin, eddy but edit, stagger but dragon.
The advantage of having (havving?) the rule is that in inflected forms it
enables us to drop the silent e that (thanks to another convention) shows the
vowel as long. So a single consonant letter signals a preceding long vowel in
hope, hoping, hoped; rate, rating, rated; time, timing, timed. Given an under-
standing of our spelling-to-sound conventions, these spellings are unambiguous:
shopping ˈʃɒpɪŋ, but hoping ˈhəʊpɪŋ.
With v, however, this breaks down. Alongside give, giving, giver gɪv, ˈgɪvɪŋ,
ˈgɪvə and have, having hæv, ˈhævɪŋ consider drive, driving, driver draɪv,
ˈdraɪvɪŋ, ˈdraɪvə and rave, raving reɪv, ˈreɪvɪŋ. Because v does not follow
the two rules, these spellings are ambiguous.
Correspondingly, we write clever, devil, cover rather than *clevver, *devvil,
*covver. The spelling is again ambiguous, as we see by comparing these
examples with fever, evil, over.
The only cases that do have medial vv or final v are slang, humorous forms or
abbreviations, e.g. navvy, flivver, civvy, skivvy, revving, bovver ; gov or guv, lav, rev,
spiv. There is also bevvy, meaning a drink, especially an alcoholic one. This noun is
slang, originating as an abbreviation of beverage. It is an interesting case, because
there is a homophone bevy, meaning a flock (of certain birds) or a group (of women).
In the advertising world, new product launches may involve both bevies
and bevvies.
5.20 low and long

A correspondent wrote to me saying that for work on Hausa
he needed to use a combination of macron and grave on all five (Hausa) long
vowels, but on his computer could only find this combination on e (ḕ) and o (ṑ).
What can he do about the other vowels?
It was not my correspondent who was missing something. It’s Unicode that
is missing something, namely precomposed characters for a, i, u with both a
macron (ˉ) and a grave accent (`).
Unicode 5.0 does indeed include provision for e and o with the desired
combination of diacritics. But not for the other vowel letters.
This must mean that no one has yet made a case for these combinations to
the Unicode Consortium committee that recommends the inclusion of new
symbols. To succeed in such a case I think the applicant would have not only
to demonstrate substantial existing printed usage of the symbol but also to show
that a combination of base letter plus diacritic would not be workable.
The experts’ advice in such cases is that we should use just such a combin-
ation. We should combine a-macron (ā, Uþ0101), i-macron (ī, Uþ012B) and
u-macron (ū, Uþ016B) with the combining grave diacritic (`, Uþ0300). If the
results you get from doing this are not very good, then the problem lies with
the fonts and/or the rendering software that you are using (word processor,
browser), rather than with Unicode per se.
Or we could fall back on IPA and write length with an IPA length mark
(ː, Uþ02D0) following the grave-bearing vowel (thus àː, ìː, ùː). Historically, it
was because of the difficulties of multiple diacritics and multistorey symbols
that the IPA adopted that mark for length rather than the macron.
5.21 palatoalveolars and the like
With reference to alveolopalatals, palatoalveolars and postalveolars

(3.20 above), we might note some other phonetic symbols for this general type of
sound – symbols at one time approved by the IPA but from which recognition
has since been withdrawn.
In the early days of the IPA, the symbols σ ƍ ƪ ƺ were approved for labialized
sibilants which we would nowadays probably write sʷ zʷ ʃʷ ʒʷ. The title of
Daniel Jones’s Sechuana Reader (1916) reads ɺipɑlɔ tsɑ̄ -sɪ̶cƪɑ̀ːnɑ̱ . Apart from
this, I don’t think they were ever used very much. By 1949, when the Principles
of the IPA booklet was published, Jones recommended sf, sv or sɥ, zɥ instead of
σ ƍ for the ‘whistling fricatives’ of Shona, which are adequately reflected in the
orthography as sv, zv. (Coleman, though, regards the voiceless one as a simul-
taneous ɸ and s.) In the same booklet Jones wrote the Tswana sound he had
formerly written ƪ (‘a single sound having a peculiar type of lip-rounding
involving an articulation by the inside of the lower lip against the upper teeth’)
as ʃw. Recognition of all these earlier symbols was finally withdrawn in 1976.
In 1947 the IPA approved two symbols to stand for palatalized ʃ, ʒ (‘for use
when these occur as phonemes separate from non-palatalised ʃ, ʒ’). They were ʆ
and ʓ. Daniel Jones (posthumously) and Dennis Ward used these symbols in
their book The Phonetics of Russian (Cambridge University Press 1969, p. 139).
5.22 IPA reforms, 1976 153
The sound ʆ occurs in pronunciation where the letter щ occurs in spelling.

Where this letter occurs some speakers pronounce ʆ ʆ, i.e. a long, palatalized
(‘soft’) ‘ʃ’.
As far as I can see the symbols ʆ and ʓ are more or less synonymous with ɕ, ʑ.
In any case for Russian they could be written ʃʲ, ʒʲ. Recognition was withdrawn
at the IPA Kiel Convention in 1989.
5.22 IPA reforms, 1976
In 1976 the IPA Council voted to add a number of symbols to the

chart, notably ʘ, ʝ and ɰ, which we use to this day. At the same time it withdrew
recognition from various symbols that had never been widely taken up, including
the σ ƍ ƪ ƺ ʆ and ʓ discussed above.
The IPA has recommended various palatalization diacritics over the years.
A dot above, as in ṫ , was used in the early days. Although impractical and hard
to see, it was retained in 1928 as an optional alternative when the attached
j-shaped hook, as in ƫ, was adopted. From 1976 to 1989 the hook was the only
approved way of showing palatalization, other than possibly by a digraph (tj);
but it in turn was superseded at the 1989 Kiel Convention by the raised j that
we use now, as in tʲ.
Until 1976 the symbols ƾ and ƻ appeared in the chart, standing for the
voiceless and voiced alveolar affricates. I have never seen them actually in
use except in the 1949 booklet The Principles of the IPA, in the transcription
of Georgian, where the first two lines have the words ƾʼamosxmulma and
ƻlieri. Nowadays we write ͡ ts and d͡ z, with a combining tie bar, or just ts and
dz. (You may occasionally also see ʦ and ʣ, though these ligatures are not
actually IPA-approved.)
Recognition was also withdrawn from the letter ƞ, which stood for the
Japanese moraic nasal, a phonological unit with various possible nasal and
nasalized realizations. In accordance with the principle of simplicity, it is usually
sufficient to write this entity as ɴ or indeed n (provided you distinguish it from
syllable-initial n in some way). If relevant we can also use various narrower
symbols as appropriate, e.g. m, n, ŋ, ɯ̃. (There is a Unicode number for this
disused symbol, Uþ019E LATIN SMALL LETTER N WITH LONG RIGHT
LEG, but not all fonts distinguish it properly from the Greek eta η.)
Lastly, the way of writing r-coloured vowels was changed. The symbols
ᶏ ᶓ ᶕ etc. were withdrawn; as the r-colouring modifier the IPA now recog-
nizes a retroflex hook higher up, thus ɛ˞ a˞ ə˞. The symbols ɚ ɝ, which include
this diacritic, were first introduced by Kenyon and Knott in their Pronounc-
ing Dictionary of American English, 1944, and have been in widespread use
since then.
5.23 Ortuguese
There is one column in the Guardian newspaper that is an unfailing
source of delight: ‘Corrections and clarifications’. Here’s an apology I saw there.
We wrongly reported that a proposed reform of the Portuguese language
could mean that the letters c, p and h are removed from the alphabet.
After the original report appeared a Dublin reader wrote in to say:

I fear on my next visit the atmospheric second city of Oorto, the ancient
university centre of Oimbra and many other places I have loved in Ortugal
will be, well, just not quite the same.
The Guardian’s correction continued,

The Orthographic Accord signed by Portuguese-speaking countries seeks
only to remove these letters from words where they are never pronounced,
such as húmido (humid). The letters will remain in the alphabet for words
where they are pronounced (World’s Portuguese speakers in new attempt
to unify language, page 16, March 26). In the same article we said one of the
signatories to the accord is Western Timor when we meant Timor-Leste
(formerly East Timor).
If the Guardian journalist were a student of mine I would also have pointed
out to him that a minor change in spelling conventions is not a ‘reform of the
Portuguese language’. A language remains the same no matter how it is repre-
sented by marks on paper.
German remains German, no matter whether you write daß or dass. English
would still be English if we were to reform its chaotic (cayottic) spelling.
Portuguese remains Portuguese with the spelling úmido just as much as with
the spelling húmido. On the other hand getting rid of gender, for instance, or
regularizing irregular verbs like fazer – that would be language reform.
5.24 Panglish, or unintelligibility?
Oh, Jennifer Jenkins, what have you started? The Daily Mail reported
‘How English as we know it is disappearing . . . to be replaced by “Panglish”’.
According to experts, it claimed, a new global tongue called ‘Panglish’ was
expected to take over in the decades ahead.
Er, up to a point, Lord Copper.
‘As English becomes more common, it will increasingly fragment into
regional dialects, experts believe.’ Braj Kachru, of Ohio State University, is
credited with saying that non-native-English dialects were already becoming
unintelligible to each other.
The Mail had lifted its story from the New Scientist.
5.25 definitely 155
. . . the ‘great vowel shift’ . . . shortened some vowels – like ee to aye, as in

mice – and pushed others to the front of the mouth, for example the Middle
English vowel pronounced oh, which became oo as in boot.
Readers of this book hardly need to be told that the great vowel shift affected quality
but not length, while uː is a back vowel, not a front one. I wonder if the New Scientist
would allow crass errors of this kind in an article on, say, physics or biology.
The article goes on to mention various well-known linguists: as well as
namechecks for Kachru, Al Marckwardt and Geoff Pullum, we also find the
usual suspects Jennifer Jenkins and Barbara Seidlhofer.
Jennifer Jenkins is credited with finding that the th sounds of thus and thin
are often dropped and replaced with either s and z, or t and d.
No, the other way round. These foreigners’ mispronunciations typically do not
affect voicing, only place or manner. So the ð of thus might become z or d, but
not s or t. The θ of thin, on the other hand, could indeed become s or t.
Another consonant that causes problems is the l of hotel and rail, which is
often replaced with a vowel or a longer l sound as in lady.
What? Replaced by a vowel, yes. But by a longer l-sound? (And anyhow the l is
surely not longer in lady than in hotel and rail.) Perhaps originally, before
garbling, this was a reference to the loss of l-allophony, i.e. to the neglect in
foreigners’ English of the clear-dark distinction.
The source of the Mail’s headline is the American author Suzette Elgin, who is
reported as musing:
I don’t see any way we can know whether the ultimate result of what’s going
on now will be Panglish – a single English that would have dialects but
would display at least a rough consensus about its grammar – or scores of
wildly varying Englishes all around the globe, many or most of them
heading towards mutual unintelligibility.
People think I’m eccentric for knowing and using Esperanto. One of the regular
arguments deployed against the idea of Esperanto is that it would be at risk of
disdialektiĝo, splitting into mutually incomprehensible dialects. But it seems that
this may be precisely the danger threatening English, its more successful rival for
the role of everyone’s second language. And I don’t find any unintelligibility Japlish-
style on Esperanto websites based in Japan or Chinglish-style on those from China.
5.25 definitely
It has always surprised me how many people misspell definitely as
definately.
On the web, authors get it wrong about one time in seven.
The difficulty, of course, stems from the fact that adjectival weak -ate and -ite
are pronounced identically.
It’s particularly surprising that students of phonetics and linguistics would
get this wrong (which some of them do). You’d think they would be aware of
orthographic and etymological relationships such as those between definite(ly)
and definition, where the stressed ɪ vowel of -ition in the latter shows clearly how
-ite in the former is to be spelt. Not to mention definitive, finish, infinite and the
foreign but widely known finis and finito.
As (in)considerate(ly) is to consideration, so (in)definite(ly) is to definition.
Why don’t people get it?
Anyone who has learnt Latin should have no difficulty, since verbs belonging
to the first conjugation give us -ate -ation -ative, while those belonging to
the fourth conjugation give us -ite -ition -itive. (OK, since you ask, the second
conjugation is exemplified by complete completion expletive and the third
conjugation, in which the stem is consonantal, by correct correction corrective.)
I try not to be annoyed by spelling mistakes. After all, I do think it would
be a good idea to reform English spelling so that we don’t obsess so much about
such ultimately trivial matters.
Nevertheless, as things stand educated people are supposed to get it right.
Until we have a spelling reform. Deffo.
5.26 lexicographers and spelling
English lexicographers over the past four centuries have not in

general been innovators in spelling, but rather have felt constrained to follow
contemporary practice.
Samuel Johnson declared that it was necessary to ‘sacrifice uniformity to
custom’ and thus for example to accept the difficult inveigh alongside the
etymologically parallel and easier convey, receipt alongside deceit, and phantom
alongside fancy.
Furthermore, by no means all of the spellings in Johnson’s great dictionary
(1755) met general acceptance. We do not nowadays write persue, raindeer,
spunge, villany or musick.
By 1806 we were ready to agree with Noah Webster and drop the k from
musick. Webster persuaded the Americans (but not the British) to write favor,
honor, color, but failed to convince anyone that we should drop the final e from
determine and examine and the final b from thumb. By 1828 he had retreated
on these latter points.
Even some of the OED’s preferred spellings have not succeeded in displacing
deprecated alternatives: ax and tire (tyre) are the norm in the States but not in
Britain, while the OED’s preference for rime and connexion has not won out over
rhyme and connection. And its preference for -z- in organize and organization,
5.27 respelling 157
etc., has not persuaded the British to abandon the -s- spellings in these words
as the Americans have.
5.27 respelling
Would-be dictionary users who are native speakers of English

are often reluctant to get to grips with proper IPA transcription. The phonetic
symbols are unfamiliar; you have to learn them. Anyhow, most dictionary users
never read the front matter where such things are explained.
So, despite the widespread adoption of IPA in British-published monolingual
dictionaries, others hold out for something perceived as more user-friendly.
That means a respelling system.
In the US the IPA has made less headway than in Britain among dictionary
publishers. In US-published dictionaries (as far as I am aware) there is no use
of IPA, only of respelling.
In a respelling system we ideally use the familiar letters of English in their
familiar meanings, as far as possible without diacritic marks.
This poses some difficult problems. How, for example, can we represent
the ‘long i’ aɪ diphthong of price? We need to distinguish it clearly from the
short ɪ of kit. Otherwise, how could we discuss the two competing pronunci-
ations of dissect? Or distinguish the two words spelt wind? One traditional
solution is to use a macron diacritic, as in ī, so that aisle aɪl is shown as īl.
But people don’t like diacritics. Another is to write y, thus yl. But this is in
danger of being misinterpreted, since traditional spelling uses y in this meaning
only when word-final (cry, shy). A third possible solution is igh, thus ighl,
which is unambiguous but clumsy. Ordinary iCe would do for ile but not for
winde (wind around, not the wind that blows). There is no ideal solution.
There is also no really unambiguous way to respell the diphthong aʊ of cow:
both ou and ow are open to misinterpretation (soul, mouth, soup; know, how).
And what about schwa? The best solution, ə, requires a special letter. The
alternative, uh, involves an arbitrary meaning for a digraph not used at all
in traditional orthography. Non-rhotic speakers might be happy with er: but
spellings such as bernáhner (banana) will shock rhotic speakers.
There is also sometimes a problem distinguishing clearly between s and z,
since in non-initial position the letter s is commonly used for both: alongside
base beɪs we have rise raɪz. In the respelling systems I designed first for
the Reader’s Digest Great Illustrated Dictionary (1984) and then later for the
Encarta World English Dictionary (1999), with their spin-offs, I adopted the idea
of making use of doubled consonant letters. The BBC has used this device in
its respelling system since 1928; it is not only a familiar way to indicate that
a preceding vowel is short (rívvər), but also means that ss will automatically
be read as voiceless (bayss as against rīz).
So I respelt rubber as /rúbbər/, petrol as /péttrəl/, travel as /trávv’l/, inward as

/ínnwərd/, deputy as /déppyəti/, supposition as /súppə zísh’n/, and teakettle as
/tée kett’l/. To show that s was required, not z, I doubled /s/ in face /fayss/,
miscue /miss kyóo/, and mincer /mínssər/; though there is no need to do that in
wasp /wosp/, first /fúrst/, and tax /taks/.
5.28 pea’s, plea’s
From time to time I get invited to take part in a radio discussion

on the topic of the apostrophe. Its use or misuse is something of an obsession
of the kind of listener who likes to bemoan the deteriorating standards of the
younger generation, in this case their failure to master the principles of English
spelling and punctuation.
One question that arose was the correct spelling of expressions like sports day /
sport’s day / sports’ day. On that occasion there were three of us ‘experts’ being
interviewed. We all agreed that we need no apostrophe in sports day, since it is
a day for sports, not a day of a sport or a day of sports. Going further, somewhat to
my surprise, we all agreed that possessive apostrophes could well be entirely
banished from English spelling. Since we do not distinguish boys, boy’s and boys’
in speech (all are bɔɪz), there should be no need to distinguish them in writing.
I pointed out various inconsistencies in current usage such as the adjacent
stations in the London Underground officially spelt Barons Court and Earl’s
Court respectively.
When I was a boy people quite often still wrote ’bus and ’phone, to show that these
were shortenings of longer words. No one does that any more. Well, hardly anyone.
It is clear that the rules for using the apostrophe are quite difficult for people
to learn. Although they may seem quite straightforward to the linguistically
sophisticated, you only have to observe public signs or internet postings to see
that the rules are nevertheless problematic for many people. Other than confusion
of its and it’s or your and you’re the commonest error is actually the overuse of
the apostrophe – the ‘greengrocer’s apostrophe’ as when banana’s and avocado’s
are offered for sale.
The only apostrophes we really do need, I suggested, are those in the contrac-
tions he’ll, she’ll, I’ll, we’ll, since otherwise they could be misread as hell, shell,
ill, well. Other than that, we can do without.
5.29 sibilant genitives
A correspondent has a query about apostrophes. She wants to know

whether it is correct to write Liz’ diary. Is Liz’s diary better? In her dictionary,
she finds the examples St James’ Palace and King Charles’ crown.
5.29 sibilant genitives 159
This topic is covered in LPD, under -’s. There are longer and more detailed
discussions in the big grammars: Pullum and Huddleston Cambridge Grammar
of the English Language (Cambridge University Press 2002, pp. 1595–6) or
Quirk et al. Comprehensive Grammar of the English Language (Longman 1985
pp. 320–1).
Stems ending in sibilants (i.e. s z ʃ ʒ tʃ dʒ) form their plural by adding an extra
syllable, namely ɪz (or for some speakers əz). They form the genitive (possessive)
singular in the same way in speech, although we write this ending as ’s, using an
apostrophe.
So we have
spoken: ˈlɪzɪz ˈdaɪəri
written: Liz’s diary
No one speaking standard BrE or AmE would say ˈlɪz ˈdaɪəri. Nor do we write
Liz’ diary. Similarly, it’s Liz Jones’s diary.
This is the general pattern applying to words and names ending in a
sibilant.
maɪ ˈniːsɪz ˈwedɪŋ my niece’s wedding
ˈrɒsɪz əˈdres Ross’s address
sən ˈdʒɔːdʒɪz ˈhɒspɪtl St George’s Hospital
ˈdʒəʊnzɪz səkˈsesəz Jones’s successors
ˈdʒɒn ˈwelzɪz ˈblɒɡ John Wells’s blog
But there are certain exceptions, most of them being classical, literary or reli-
gious names ending in the sound z (not necessarily the letter z). With these you
can optionally pronounce zero and write just the apostrophe.
ˈdɪkɪnzɪz ˈraɪtɪŋ Dickens’s writing OR
ˈdɪkɪnz ˈraɪtɪŋ Dickens’ writing
ˈbɜːnzɪz ˈpəʊɪtri Burns’s poetry OR
ˈbɜːnz ˈpəʊɪtri Burns’ poetry
This explains St James’ Palace and King Charles’ crown above.

Personally I usually follow the regular rule with these names, i.e. I would
say ˈbɜːnzɪz ˈpəʊɪtri and write Burns’s poetry; I would say kɪŋ ˈtʃɑːlzɪz ˈkraʊn
and write King Charles’s crown. The only cases where I might go for the
irregular version would be with Greek or Hebrew names.
juˈrɪpɪdiːz ˈpleɪz Euripides’ plays
ˈsɒkrətiːz fɪˈlɒsəfi Socrates’ philosophy
ˈməʊzɪz ˈlɔː Moses’ law
There are also two special cases, which end in s rather than z. One is the fixed
expression for goodness’ sake, which never has a spoken extra syllable or a
written extra s. The other is the name Jesus. For this, we can write either Jesus’s
love or Jesus’ love, and pronounce correspondingly. Poets and hymnodists
can also write Jesu’s and pronounce ˈdʒiːz(j)uːz or even ˈjeɪzuːz.
5.30 greengrocers’ spellings
I saw signs on a greengrocer’s stall in a London street market offering

for sale obo-jeans and monge-two. Brits ought to be able to interpret them.
Non-Brits may find it more difficult.
Obo-jeans is straightforward. This is clearly a pronunciation spelling of
ˈəʊbədʒiːnz, or more elegantly ˈəʊbəʒiːnz, which is properly spelt aubergines.
Americans know these vegetables as eggplants.
Monge-two reads ˌmɒndʒ ˈtuː, or more elegantly ˌmɒ̃ ʒ ˈtuː, properly spelt
mange-touts. These are peas of which you eat the whole pod with its contents.
Hence the name mange-tout, French for ‘eat (it) all’. I think Americans call them
snow peas or sugar snap peas.
5.32 the cost of spelling 161
If nothing else, the mis-spellings prove the lively existence of these vege-
table names in the spoken English of people who are not very literate.
(Or, I suppose, it just might be a literate greengrocer having a laugh.) Despite
their French origin, they are well known to the general public – they are
not restricted to educated speech. They may not even be perceived as French
by the ordinary user.
Educated people, though, recognize that they are French and therefore may
well pronounce them in a more sophisticated, semi-French way.
In native-speaker French they’d be obɛʁʒin, and mɑ̃ ʒ(ə)tu or mɑ̃ ʃtu.
5.31 O tempora! O mores!

The previous section (5.30) was about what you might call pronun-
ciation spellings. This one is about the more familiar territory of spelling
pronunciations.
The word mores, meaning ‘the customs, social behaviour, and moral values
of a particular group’ (LDOCE) is a word that only educated people know.
And most of them know that it is disyllabic, pronounced ˈmɔːreɪz (or in a more
old-fashioned way ˈmɔːriːz).
So it was quite a surprise to hear someone on BBC Radio 4 pronounce it
monosyllabically as mɔːz, as if it were the plural of more.
It isn’t, and it isn’t etymologically related to more. It is the Latin plural of
mōs ‘custom’, of which the stem is mōr-. The nominative plural ending is -ēs.
The word moral is derived from the corresponding Latin adjective.
I expect that one day soon we shall hear lares et penates (household gods)
pronounced ˈleəz et peˈneɪts. But the mayor of London, Boris Johnson, wouldn’t
commit such a solecism. He is a classicist. I have to admit, though, that I think
he’s being overoptimistic in suggesting (as reported in the press) that teaching
unruly youths Latin would stop them knifing one another.
(Hey, it worked for me. I was taught Latin when young and I’ve never knifed
anyone.)
5.32 the cost of spelling
In 2008 I attended a conference with the topic ‘The cost of English

spelling’. What is the cost to the British economy of having a difficult-to-learn
spelling system?
The only presentation that really attempted to address this issue seriously
was a paper by Zuzana Kotercová of the University of Coventry, ‘The eco-
nomic cost of current English spelling’. She started from the claim (how
well based I do not know) that English schoolchildren take three years to reach
the same level of literacy skills as Finnish schoolchildren achieve in three
months.
In Finnish, learning to read and write requires no more than learning
the letters of the alphabet and the sounds they correspond to: then you
just sound out every word as it is written, and write down every word in
the way it sounds. In English, this approach gives you a good start (and is
now known by the buzzword ‘synthetic phonics’). Children can also take
in their stride conventions such as the effect of silent e (hop – hope, rat –
rate). But all too soon the learner is then faced by hundeds of words
whose pronunciation is not what you get by sounding out the spelling
(money, two, find) or in which a rule that they’ve just learnt (such as the
silent e) doesn’t apply (have, give) – and words whose spelling you would
not get right if you based it just on how you say them aloud (friend, head,
knife, climb).
One of the other conference presenters claimed that approximately 25% of
learners cope well with this extra learning burden, 50% manage but with a lot
of extra time and effort, while 25% do not cope and end up disheartened and
ultimately functionally illiterate.
To return to Kotercová’s work: she started by estimating the cost
of the extra teaching time required to achieve literacy in English. She
investigated four Coventry primary schools and calculated the number of
hours of additional teaching expended on spelling instruction as opposed to
general literacy. The average annual cost per teacher came out at £556.
Multiplying by the number of primary teachers in England, we arrive at a
lowest estimate of just over a hundred million pounds per annum: the
additional expense we incur in England through the inadequacies of our
spelling system.
This figure does not include the additional cost of remedial literacy teaching in
secondary schools, which must be substantial. Nor does it take into account the
lower lifetime earnings expectations of the functionally illiterate when compared
with the literate. Nor does it address the similar additional costs incurred in other
English-speaking countries, still less the additional hours of teaching and learn-
ing required in the case of those millions who attempt to learn English as a
foreign language.
It would be particularly difficult to quantify the adverse effects on the
intelligibility of EFL learners – who naturally assume (unless taught otherwise)
that broad must have the same vowel sound as road and that son must rhyme
with on.
So her figures are not only based on a very small sample, they must
also be pretty conservative. But I hope she will inspire some team of
economists and educational researchers to attempt to get a robust answer to
the question.
5.33 capital eszett 163
My sister-in-law, who has extensive experience in educational policy, com-

mented as follows.
Whilst I cannot categorically state that the spelling system in English is not a
factor, what I can say is that the fundamental difference between the English
educational system and that of Finland is definitely a factor. Finland, in
common with most, if not all, Scandanavian countries, begins formal educa-
tion at the age of 7 years, i.e. when children are developmentally ready to
learn to read, write and spell (I’m thinking especially in terms of the fine
motor control needed to control a pencil successfully), whilst here in England
the Government persists in the notion that ‘earlier is better’ and insists on
starting formal education at age 4/5 years.
5.33 capital eszett

Standard German spelling includes the special double-s letter ß, as in
Fuß ‘foot’. The spelling reform has removed it from some words, so that what
used to be daß is now dass; but it remains in plenty of others. (That’s in Germany
proper and Austria. Not in Switzerland, where they write ss instead.)
This letter is known as eszett or scharfes ss (‘sharp s’). It is now used only
after long vowels and diphthongs, which means among other things that it is
never used at the beginning of a word, only in the middle or at the end.
If it’s never initial in a word, it will never be found at the beginning of a
proper name or at the beginning of a sentence. Therefore, you might conclude, it
is never capitalized. And traditional German printing in blackletter never used
all-caps.
Thus ß is the only letter of the Latin alphabet that has no upper-case form.
Or is it? In 2008, Unicode added a capital ß (ẞ) as Uþ1E9E LATIN CAPITAL
LETTER SHARP S. Why? Because modern signage and product design does
sometimes call for it.
The Unicode Standard (5.1) comments:
In particular, capital sharp s is intended for typographical representations of

signage and uppercase titles, and other environments where users require the
sharp s to be preserved in uppercase. Overall, such usage is rare. In contrast,
standard German orthography uses the string ‘SS’ as uppercase mapping
for small sharp s. Thus, with the default Unicode casing operations, capital
sharp s will lowercase to small sharp s, but not the reverse: small sharp s
uppercases to ‘SS’. In those instances where the reverse casing operation is
needed, a tailored operation would be required.
Nevertheless, upper-case ß ‘ist nicht Bestandteil der offiziellen deutschen

Rechtschreibung’ (‘is not a component of official German orthography’).
5.34 Jumieka
There has been a vigorous debate in Jamaica, sparked by the publi-
cation of a translation of part of the Bible into Jamaican Creole.
A web page devoted to ‘Jumieka Langwij’ explains:
Di hiem a dis sait a fi bring tigeda haxpek ahn suos a Jumiekan langwij fi
chrai prizaabi fi paasteriti ahn fi di huoliip we lib a farin wid Jumiekan
kanekshan uu maita hinchres iina dehn linguistik eritij. hUoliip a dem kuda
gat pierans ar grampierans uu kiahn kot di patwa, bot dem siem wan kiaahn
piik di mada-tong, ar els piik wahn luokalaiz verjan, laik Landan patwa.
‘The aim of this site is to bring together aspects and sources of Jamaican
language in an effort to preserve it for posterity and for the many living
abroad with Jamaican connections who may have an interest in their linguis-
tic heritage. Many of these may have parents or grandparents who are or
were patois speakers, but are themselves not fluent in the mother-tongue, or
else speak a localized variant, such as London patois.’
This spelling you see here is more or less the one used in Cassidy and Le Page’s
Dictionary of Jamaican English (Cambridge University Press 1980, second
edition). It differs by adding one or two more symbols: an optional italic h for
the h that comes and goes before word-initial vowels, particularly after a final
vowel in the preceding word, and hn to stand for what the web page author
5.35 fame at last! 165
(not Cassidy and Le Page) calls a ‘soft, breathy n’ (actually nasalization of the
preceding vowel), as duohn duõ ‘don’t’.
This spelling is phonemic (‘ebri leta fi soun, aalwiez soun siem wie’) and uses
the ordinary alphabet with no diacritics. Its main disadvantage for those unfami-
liar with it and with IPA is the use of ai for the vowel of PRICE and of ie, uo
(phonetically spot on) for FACE-NEAR and GOAT-FORCE respectively.
Fred Cassidy (1907–2000), the deviser of this spelling, was an enthusiast
for the spelling reform of Standard English, too. When I was much younger,
and not then involved in the Spelling Society, he invited me to dinner and urged
me to campaign for the cause. His co-author, Bob Le Page (1920–2006), was the
external examiner for my PhD.
You can read about the serious academic side of language in Jamaica – the
Jamaican Language Unit of the UWI – at www.mona.uwi.edu/dllp/jlu/about/.
5.35 fame at last!

Gosh, I’ve been denounced by David Cameron!
Here’s what he said in his speech to the Conservative party conference yesterday:
The election of a Conservative government will bring – and I mean this
almost literally – a declaration of war against those parts of the educational
establishment who still cling to the cruelty of the ‘all must win prizes’
philosophy. . .
. . .and the dangerous practice of dumbing down.
Listen to this.
It’s the President of the Spelling Society.
He said, and I quote, ‘people should be able to use whichever spelling they
prefer.’
He’s the President of the Spelling Society.
I say that he’s wrong. And by the way, that’s spelt with a ‘W’.
He is quoting me out of context. Here’s what I actually said at the Spelling

Society centenary dinner, as quoted in my press release:
. . . Let’s return to a time when English spelling allowed greater variation . . .
Nowadays we often see light written lite and through as thru. Let’s not hold
up our hands in horror – people should be able to use whichever spelling
they prefer.
I didn’t mean that we should abandon all standards in spelling. I was referring to
pairs such as those I mentioned. I added a few more, such as the text messaging u
alongside you. And I suggested we get rid of the apostrophe.
I’m not calling for any dumbing down. I’m not calling for a free-for-all. I’m
not suggesting we abandon all rules, just that we might relax some of them in a
controlled way.
I’m in favour of consistency, logic and obeying the rules, in spelling as in
other things. I just think the rules need to be modified.
6 English accents
Regional and social accents of English
6.1 po(r)ker
On the quayside at St John’s, Antigua, where the cruise ships dock,

there used to be a painted hoarding inviting the cruise passengers to a casino
offering various forms of gambling, including stud porker (sic).
In the English of Antigua, pork is a homophone of poke poːk, which explains
this spelling mistake. It is not a matter of ‘r-fulness on the march’ or ‘a sneaky
little creeping /r/’, as one linguist characterized it. It is not a matter of sounds
at all, just of spelling. Neither the Antiguans nor most English people have an r
in pork. The problem is not the sound r but the letter r. Just as with larva and
lava, we speakers of non-rhotic English may have difficulty in knowing which
words to spell with r and which not.
When we look at different accents of English, there are two sets of words
spelt with -or-. In terms of the Standard Lexical Sets of chapter 2 of my
Accents of English, some belong in the FORCE set and some in the NORTH
set. It is the FORCE words that in Leeward Islands English fall in with the
GOAT words, producing such homophones as court-coat, porch-poach and
indeed pork-poke. The NORTH set, on the other hand, fall in with START
in the local creole, producing homophones such as form-farm, born-barn, lord-
lard. North naːt and bath baːt rhyme. But pairs such as hoarse-horse are
distinct, hoːs vs. haːs.
Most Scots and Irish people preserve the distinction between the FORCE and
NORTH sets. They, the Jamaicans and the Leeward Islanders are conservative
in this aspect of their pronunciation. They retain a historical distinction that most
of us have lost.
6.2 holey ground
In roll, cold, bolt and other words where the vowel is followed by a
morpheme-final or preconsonantal l, some English people use a special vowel
quality ɒʊ, a quality so different from their usual əʊ (as in grow, code, boat) that
they are reluctant to view it as an allophone of /əʊ/. You can hear the difference
if you ask such speakers to pronounce, for example, cold and code, or bolt and
167
168 english accents
boat. They may also not rhyme goalie and slowly and have different vowel
qualities in rolling and Roland.
Unlike such speakers, I personally use pretty much the same əʊ quality in
all these words, no matter what the following consonant might be. In support
of this contention (which surprises some people), let me tell you how, when I was
a small boy and couldn’t sleep one night, my father told me the Bible story
of Moses and the burning bush (Exodus 3). In the words of the Authorized
Version, God spake unto Moses from out of the midst of the bush and said,
‘Draw not nigh hither: put off thy shoes from off thy feet, for the place whereon
thou standest is holy ground’.
But I heard this as hole-y ground, ground with holes in it. (If Moses kept
his shoes on, I thought, perhaps he would get them caught in the holes.)
This implies that my late father pronounced ˈhəʊli holy ‘sacred’ and ˈhəʊli
hole-y ‘containing holes’ identically – like me, and unlike the speakers
mentioned above.
6.3 hair lair!

Rory Bremner’s Foreword to Andrew Taylor’s A Plum in your
Mouth (HarperCollins 2006) starts with a little dialogue which English people
will understand, but probably not many non-English people.
man a: Try pronouncing the word ‘a-i-r’.
man b: Air.
man a: Good. Now try pronouncing the word ‘h-a-i-r’.
man b: Hair.
man a: Excellent. Now try the word ‘l-a-i-r’.
man b: Lair.
man a: Splendid. Now put them together and what have you got?
man b: Air, hair, lair.
man a: Welcome to Sandhurst.
Explanation: this pokes fun at upper-class speech (U-RP). Sandhurst is the Royal
Military Academy where army officers are trained. One of the stereotypical
features of U-RP is a front starting quality to the GOAT vowel, roughly eʊ,
but without much backing or rounding in the second element. Thus U-RP oh
can be caricatured as air, and what Man B ends up saying is Oh, hello.
6.4 minor royalty
Like her mother, whom journalists once caricatured as Princess

Unne, Zara Phillips (born 1981, daughter of Princess Anne and thus grand-
daughter of the Queen) uses a very open, retracted-cardinal-four kind of æ.
You should be able to find a video clip of her on YouTube, and observe
6.6 fricative t 169
several other striking characteristics of the pronunciation used by her and

others of her age and general background.
The replacement of word-final t by sound-types other than a voiceless alveolar
plosive proceeds apace. For post-sonorant word-final t Zara often uses a glottal
stop, as in i[ʔ], bu[ʔ], wen[ʔ] ou[ʔ]. In this she is indeed probably representative
of people of her age and class (though not of people of my age and class, and
sharply different from her grandmother).
When she says No her lips are unrounded throughout.
It is fascinating to be able to observe the really rather substantial change
of accent within three generations of one family as we proceed from mother
to daughter to granddaughter.
6.5 Prince Harry
There are various TV and radio clips available of Prince Harry

being interviewed. He provides excellent exemplifications of the RP (?) spoken
by upper-class young men of his age (born 1984). On one such clip I noted
extensive T glottalling (quite a lot kwaɪʔ ə lɒʔ) and l-vocalization (well,
people, April weo, ˈpiːpo, ˈeɪpro), including remarkable pronunciations of
multi-tasking and pulled, in which where I would have ʌl, ʊl he has something
like oː for both.
6.6 fricative t
In Irish English when /t/ is intervocalic (¼ between vowels) and in

coda position (¼ at the end of a syllable) it is often fricative rather than plosive.
I am thinking of the /t/ in examples such as atom, bottom, jetty, ditto; get off, shut
up, quite easy, wait a second; or, in utterance-final position a lot, who’s that, very
late, all right.
In all of these cases the usual southern Irish sound is a kind of alveolar slit
fricative. It is an example of lenition (weakening), a characteristic feature of Irish
consonants in certain positions.
In Accents of English (p. 429) I used the symbol [t̨ ] for this sound,
employing what was then the official IPA diacritic for ‘opener articulation’, an
ogonek (reversed subscript comma). At the 1989 Kiel Convention the IPA
changed this to a down tack, which means that the proper symbol would now
be [t̞ ].
The contrast between /t/, /s/ and /ts/ is maintained: hit 6¼ hiss 6¼ hits. (The first
two end in fricatives, the last in an affricate.)
You get this kind of /t/ in Scouse (the accent of Liverpool) as well: it’s one of
the typical Irish characteristics of Scouse. But there in certain environments you
get further lenition, all the way to [h].
170 english accents
6.7 hypercorrection
We non-rhotic speakers find it very difficult to avoid intrusive r
while retaining linking r (should we wish to do so) – there just isn’t time, in
the flow of speech, to stop and think whether a given word is spelt with or
without an r. To stop ourselves pronouncing an r at the end of Malta in Tell me
how far away Malta is we usually also have to stop ourselves pronouncing an r
at the end of Gibraltar in Tell me how far away Gibraltar is.
We have a corresponding difficulty in putting on a rhotic accent if we
want to imitate American or Scottish speech. The rule is simple: if there’s an
r in the spelling, pronounce an r. But it can be quite difficult to apply this rule
in real time. Hugh Laurie in House is unusually talented in this respect:
he gets it right. Personally, if I try to talk American it is only with conscious
self-monitoring that I can stop myself from saying things like ˈfɑːrðɚ
instead of ˈfɑːðɚ father. In my own accent, of course, father is a homophone
of farther, both ˈfɑːðə. My unconscious default stratagem to make my speech
rhotic clearly does not refer to the spelling, but says ‘add r-colouring to
every ə or ɜː and insert r after every ɑː or ɔː’. That’s why it gives quite a
few wrong results.
6.8 Montserrat Creole

My partner and I were invited to sing a duet at the funeral in London
of a young man from Montserrat in the West Indies. After the service, a
Montserratian I have known for many years, who like the deceased was forced
to evacuate to England by the volcanic eruption of 1997, congratulated me by
saying mi no mi no ju kudu siŋ so.
This was in the local dialect of English, Montserrat Creole. I wonder how far
people unfamiliar with the dialect would have understood.
mi no mi no ju kudu siŋ so
I NEG PAST know you could sing so
Me no me know you couldoo sing so!
‘I didn’t know you could sing like that!’
Unlike in Jamaican Creole, in Montserrat the past tense marker is mi, homoph-
onous with the first person singular pronoun mi ‘I, me, my’. And the mark of
negation, no, is, like standard no, homophonous with the verb know.
Montserratians speak something much closer to Standard English than this
when talking to outsiders, so I felt complimented to be addressed in dialect: it
meant I was being treated as a belonger, an insider.
6.9 television newsreaders’ RP 171
On a different note, here’s something I overheard in the playground of a

primary school in Montserrat.
a no \/miː
is not me
‘It wasn’t me!’
Although the grammar in this utterance is the local creole, the intonation is
English. It is the same pattern as I would use myself, with the same pragmatic
meaning: a fall-rise on a negative statement (English Intonation, p. 31). This is a
distinctively English pattern, which does not seem to be used with this meaning
in (most) other languages. It reinforces my view that Caribbean English creoles
are best considered dialects of English, not separate languages.
6.9 television newsreaders’ RP

Bente Hannisdal’s PhD thesis (University of Bergen 2007) is entitled
Variability and Change in Received Pronunciation.
For this brilliant piece of work she had collected a large number of recordings
of British television newsreaders (thirty people; half of them men, half women;
from the BBC, ITV and Sky channels; an hour of speech from each individual).
This constitutes an extensive corpus of contemporary RP, and has enabled her
to discover some interesting facts, not previously established, about this variety
of English. Among them were the following.
GOAT allophony
In words such as sold, roll, where the GOAT vowel is followed by
dark l, 24 of the 30 newsreaders use the ɒʊ allophone categorically, and a further
four of them do so variably. Only two do not use it at all. We need to update our
descriptions of RP to cover this.
CURE lowering
Despite reports of its imminent demise, it turns out that ʊə is alive and
well, persisting particularly after j and before r, as in Euro, Europe, European,
security, secure, during. The switch to ɔː is found principally in the complementary
phonetic environments, e.g. in poor, sure, tour. The only important exceptions
to this generalization are your and you’re, which – despite the j preceding and
the possible linking r following – the newsreaders pronounced 100% with ɔː.
T voicing
A remarkable 35% of final /t/s, prevocalic across a word boundary,
were voiced, i.e. pronounced American-style as ɾ. Unlike in American English,
172 english accents
this is not found in content words: Brits don’t make atom sound like Adam.
(There were one or two exceptions, including British and pre-adjectival pretty.)
In Hannisdal’s corpus it is mainly found only in function words, across a word
boundary. It was particularly prevalent in such cases as it is, that it’s, but if, at a,
what about, a lot of, a bit of, not exactly, get elected. Yes, we do need to update
our descriptions.
Yod coalescence
Words with traditional tj, dj before a stressed vowel (tube, Tuesday,
student, due, reduce, duty, during) were pronounced instead with tʃ, dʒ in a
massive 46% of cases. We can no longer consider these variants ‘non-RP’ (even
though that’s what they were for people of my own advanced age).
Smoothing
Pronunciation of aɪə (fire) and aʊə (power) as aə, aː proved to be ‘very
frequent in broadcast RP’. My native-English-speaking undergraduate students
often used to refuse to believe that any such thing was possible: how wrong they
were! Among the newsreaders, there was a highly significant statistical difference
between the sexes, with men smoothing much more than women. As you might
expect, smoothing was less common across a morpheme boundary (e.g. high#er
as compared with its possible homophone hire), and less common before a pause.
R sandhi
There was a tendency for men to use more linking/intrusive r than
women, but the difference was not significant. The /r/ was realized as [ɾ] only in
foreign names (Igor Ivanov); otherwise it was an ordinary [ɹ]. (Compare Daniel
Jones’s pronunciation, where the tap was the norm for all linking r.) The factors
favouring r sandhi are exemplified, in descending order, by for a, four o’clock,
former economic. It was least likely in proper names (Doctor Austen, Sir Alex). The
main factor disfavouring it was another r adjacent (e.g. terror alert). Overall, the link
was used in 60% of possible cases. Interestingly, intrusive r was more prevalent after
ɑː, ɔː than after ə. That is, it was more likely that there would be an r in Jack Straw_
is than in Nelson Mandela_ is. However, the raw number of cases of intrusive r after
ə was higher because of the greater frequency of schwa in word-final position.
6.10 accents change over time shock horror
The Times newspaper carried a story about the supposed disappear-

ance of middle-class speech in England, ‘Where are the gels who can talk proper
and pirouette?’
6.11 bad sociophonetics 173
Casting directors are lost for words because the next generation of British
actors just cannot speak proper. The rise of ‘Estuary English’ has left
children with the intonation patterns of Lily Allen and Jonathan Ross,
regardless of their background.
The decline in Received Pronunciation has not just transformed the presentation
of BBC News. Film and drama producers are struggling to fill period roles that
require unrepentantly middle-class vowels. BBC One is holding an open casting
session tomorrow to try to find two girls to star in a film-length adaptation of the
classic children’s novel Ballet Shoes. Victoria Wood and Marc Warren have
signed up to star in the story, by Noel Streatfeild, set in 1930s London. But the
challenge of finding two ballet-dancing leads who can act, twirl and – most
importantly – speak in middle-class accents has defeated the producers.
Notice anything odd here? First, there is the headline. The spelling gels alludes to
an obsolete upper-class pronunciation of girls (gelz or perhaps geəlz for gɜːlz),
which is surely not relevant to the task of casting child actors to play parts
requiring the middle-class accents of the 1930s.
In my view, middle-class children today speak with middle-class accents.
If the British social class system persists (and it does, though much less sharply
stratified than was the case a hundred years ago), then by definition middle-class
accents are what the middle classes continue to speak with. Middle-class and
working-class accents have certainly converged over the last half-century, but
accent differences haven’t disappeared.
That middle-class accents have changed over time need not surprise anyone:
language does change. Rare indeed are children who speak like their grand-
parents or great-grandparents.
What the producers are really looking for is child actors who can speak with
a particular historical accent. It is like casting a play in which the actors must
speak with Victorian, or indeed Shakespearian, accents. People usually can’t do
it without training. Unless you are particularly talented in that direction and can
do it on your own, you have to study. You must go to drama tuition, elocution
lessons or even, dare one say it, phonetics classes.
6.11 bad sociophonetics
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 [. . .].
I am interested in doing a report on the British language since I read a piece in the
newspapers recently about how the Australian accent is well and truly creeping
into the everyday [sic] of Britons. Apparently Neighbours, Home and Away,
Kylie etc have a lot to do with it as well as lower socio [sic] not wanting to sound
as such and the upper class wanting to sound a little more middle class.
174 english accents
– to which I replied:
You really mustn’t believe everything you read in the papers. The Australian
accent is not in fact ‘well and truly creeping into the everyday [speech] of
Britons’.
There is one, just one, survey – dating from seven years ago! – that uncovered
one apparently Australian-type variant pronunciation of one vowel in one
place and speculated that it might be of Australian origin, only to reject this
explanation in favour of a more plausible one. The co-author of that survey
was Paul Kerswill, whom you could contact for further information.
I feel like Ben Goldacre with his ‘Bad Science’ column in the Guardian, who
says his archives at www.badscience.net:
are overflowing with just a small sample of the media’s crimes: preposterous
cherry-picking, outrageous overextrapolation, startling ignorance or white-
washing 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.
6.12 The Book of Dave
I have recently been reading Will Self’s novel The Book of Dave.
I found it brilliant but not exactly enjoyable. Do read it.
The action takes place partly at the present day and partly in the distant
future, perhaps a thousand years hence, by which time England has been
inundated by rising sea levels, leaving only scattered patches of former high
ground to support human life.
This future is a despotic feudal-theocratic dystopia. Its ideology is a reli-
gion derived from the ravings of Dave, a taxi-driver of the present day, driven
mad by his ex-wife’s refusal to allow him access to their son. He wrote down
6.12 The Book of Dave 175
his disturbed misogyny in a book which he then buried. Most of his book is
a recitation of the Knowledge required of London cabbies, a detailed famil-
iarity with all the streets and routes within six miles of Charing Cross.
The rest is Dave’s misogynistic proposals to reform society, with ‘dads’ and
‘mummies’ strictly segregated from one another, the children moving
from one to the other every WED and SUN. Dug up centuries later, this
book is the foundation of a new religion which becomes the established
orthodoxy. In this dystopic future, prayer (addressed to Dave) takes the form
of reciting the Knowledge, even though the places referred to have long been
submerged.
As a middle-class dropout who has moved down-market to become a cab
driver, Dave speaks to himself ‘with Received Pronunciation’, but to the world in
general in ‘Mockney’ (which is the name we use for the variety in which people
consciously and intentionally make their speech less standard, more demotic).
For the latter, writes the author, ‘he hadn’t dropped his Hs – he’d flung them
away from himself, ninja stars that stuck quivering in the smoky bacon Victorian
woodwork’.
What kind of language does the author puts into the mouths of the follow-
ers of Dave? They have two varieties of speech. One is a formal variety
known as Arpee, which is ordinary Standard English except for various
cabbie-derived vocabulary innovations (screenwash for rain, the headlight
for the moon, and so on). The other, informal, register is known as Mokni,
and bears a remarkably close resemblance (given the passage of centuries) to
contemporary cockney. As well as having vocabulary innovations and typical
cockney slang and phraseology, this is represented in what people would no
doubt call phonetic spelling, with some elements of txt conventions. Here is
an example:
Eyem gonna go onna bí! C if vairs anyfing bettah ovah vair! – Yeah, orlrì,
but nó 2 fa.
– which being interpreted is ‘I’m going to go on a bit! (I’ll) see if there’s anything
better over there! Yes, all right, but not too far’.
So the convention is that an acute accent denotes a short vowel
followed (today, at least) by a glottal stop, and a grave accent denotes a
long vowel.
Much of this is simply eye dialect. That is to say, the spelling is changed in a
way that doesn’t imply any difference of pronunciation, as eyem for I’m or onna
for on a.
Other examples of this include slorta for slaughter and carn for can’t – the
sort of thing that is transparent to non-rhotic readers but that may prove trickier
for the rhotic.
This book will surely spawn a thousand PhD dissertations in English depart-
ments around the world as non-Brits struggle to make sense of it.
176 english accents
6.13 character, calendar

In Caribbean English certain words have what to me is an unusual
word stress. The words character and calendar are generally pronounced there
with penultimate stress, kæˈræktə, kæˈlendə.
As far as I know, this stressing is not found in other kinds of English. Our
usual British/American initial stressing, ˈkærəktə(r), ˈkæləndə(r), is at least
as old as Shakespeare.
The Caribbean stress pattern is actually what Chomsky and Halle’s stress rules
in The Sound Pattern of English predict for words of this phonological structure.
My paraphrase of part of their Main Stress Rule would be:
In a morphologically simple noun, disregard the last vowel if it is short.

If what now remains as the final syllable is ‘heavy’ (¼ if it has a long vowel
or ends in two or more consonants), stress it (yielding penultimate word
stress). If it is ‘light’, stress the preceding syllable (yielding antepenulti-
mate word stress).
We see the operation of this double-consonant effect in such items as philander,

November, momentum, sequester, meniscus, colostrum, conundrum, all stressed
on the penultimate.
The ordinary British/American stress pattern of character and calendar violates
the SPE rules – though, as usual, you can get round this by arbitrarily positing
a particular underlying representation, in this case one with final vowelless -ktr,
-ndr. The same applies to other everyday items, apparently exceptional, such as
cucumber, cylinder, canister, barrister.
The mystery remains: why do West Indians stress character and calendar
differently from the rest of us? Further research shows that they are not alone: the
Irish, too, often have penultimate stress at least in cylinder and cucumber. And there
are even English people who stress cucumber on the penultimate. And Singapor-
eans put the stress on the second syllable in various words in which it has first-
syllable stress in other varieties: abacus, faculty, penalty, abdomen, industry,
precedent, Agatha, lavender, predator, cucumber, maintenance, prevalent.
6.14 Grand Turk

These are notes based on informal observations of the speech
of our taxi-driver-cum-tour-guide on Grand Turk island, part of the British
Overseas Territory of Turks and Caicos, lying to the north of Hispaniola and
to the southeast of the Bahamas. He told us he was a native of Grand Turk.
Our driver consistently pronounced town as taːn. I think this was probably
the same vowel as he had in pass paːs, but I was unable to test this by eliciting
6.15 popular phonetics 177
for example mouth and bath. This monophthongal MOUTH vowel is not uncom-
mon in the north of England, but I have never before heard it from a West Indian
(though it has been reported by others writing on Bahamian).
He said coffee with a definitely back, mid, rounded vowel similar to RP
thought: ˈkɔːfi. (Compare Jamaican speech, where coffee typically has the same
vowel as pass.)
His PRICE vowel, at least before a voiceless consonant, was very narrow: right rəit,
like ləik. Trudgill reports this for various remote islands in New-dialect Formation:
The Inevitability of Colonial Englishes (Edinburgh University Press 2004, p. 52).
His accent was consistently non-rhotic. His NEAR vowel was wide, thus here
hiɐ. His NURSE vowel was open, back and lightly rounded, thus first fɒːs(t),
Turk tɒːk.
When he pointed out the Anglican church, it sounded like ˈẽlɪkən.
But casual notes from a forty-five minute tour are no substitute for a proper
analysis.
There are some general comments on Turks speech in Trudgill’s ‘The history
of the lesser-known varieties of English’, in R. Watts and P. Trudgill (eds.),
Alternative Histories of English (Routledge 2002, p. 37).
The speech of the islands is often described as being very close to Bahamian
English.
. . . [John Holm] reports (personal communication) that certain students at
the College of the Bahamas in Nassau where he taught in 1978–80 were said
by other students to have ‘Turks accents’.
Turks Islanders claim that people in the Caicos Islands speak differently
[from them] . . .
I’m kicking myself that I did not notice or elicit the pronunciation of the only
town, Cockburn Town. I bet it contains only one k and rhymes with Oban.
6.15 popular phonetics

The Guardian’s television critic was discussing a programme in
which two people had the job of scouting the streets looking for new [fashion]
models. He commented:
‘Model’ needs to be pronounced as it is in the industry, with no ‘l’ sound at
the end, just a glottal stop instead.
I put a comment on the Guardian blog saying:

No one pronounces this word with a glottal stop. What your critic means is
‘with a vocalized l’.
I will try and get a life (though my livelihood is phonetics).
Other readers chimed in:

178 english accents
Yes, it would be strange to hear ‘model’ end with a glottal stop. . .

And I was just practising saying it that way. . .
I’ll have to start all over again now.
What is interesting about the original comment is that glottal stop – perhaps the
only one of our phonetic technical terms to have become truly popular in the
wider world – seems to be being used as a catch-all term for any and every
pronunciation feature the writer disapproves of.
6.16 changing London speech

A correspondent wrote to say he had read about the development of a
‘multi-ethnic dialect’ in London, seemingly independent of social origin and
including patterns ‘from Cockney, Jamaican creole, Bengali and other lan-
guages’. He wanted to know whether this was true – or was it ‘more like a hype
such as about Estuary English?’
My reply was that this research is highly respectable, being the work of well-
known sociolinguists.
The research findings are described here.
Jenny Cheshire, Sue Fox, Paul Kerswill and Eivind Torgersen, ‘Ethnicity,
friendship network and social practices as the motor of dialect change:
Linguistic innovation in London’, Sociolinguistica 2008, 22: 1–23.
Susan Fox, ‘The demise of Cockneys? Language change among adolescents in
the ‘traditional’ East End of London’, unpublished PhD dissertation, Univer-
sity of Essex 2007.
Paul Kerswill and Jenny Cheshire, ‘Linguistics innovators: the English of ado-
lescents in London. Project overview’, 2007 www.lancs.ac.uk/fss/projects/
linguistics/innovators/overview.htm.
Paul Kerswill, Eivind Torgersen and Susan Fox, ‘Reversing “drift”: Innovation
and diffusion in the London diphthong system’, Language Variation and
Change 2008, 20: 451–91.
Prof Kerswill believes this new form of speech, which he and his associates refer
to as Multicultural London English (MLE), arose as more recent immigrants
learnt English with few English-speaking role models. He says:
I think there has been a change in the last 30 years. The first generation of
Afro-Caribbean children in London would have spoken Cockney, but they
would have kept their own Caribbean patois and switched between the two.
Now their descendants – and young people of other ethnicities – are
inventing a new form of English.
The teenagers also used words like blud (blood, meaning mate), creps (trainers),
yard (house) and endz (area or estate).
Here is a chart taken from the Cheshire et al. article. What particularly interests
me is the change in vowel qualities among young inner-city speakers.
6.17 American T voicing and sentence 179
200
u:
u:
II
350
F1 (Hz)
e D
500 e D
L ɑɑ
æ
650 L
æ
800
2500 2000 1500 1000 500
F2 (Hz)
The southeastern short vowel shift in Hackney among elderly speakers

(circles); young speakers (diamonds)
You can see in this formant plot the striking anticlockwise movement of TRAP,
STRUT, FOOT and GOOSE (æ, ʌ, ʊ, uː). This makes cat, for example, into [kat],
so pretty similar to older cockney cut, and leaves cut as [kɑt], with a very similar
quality to that of cart [kɑːt]. Meanwhile FOOT and GOOSE lose their rounding
and are no longer back but central or even front: /ʊ/ becomes more schwa-like and
/uː/ turns into a sort of [ɨː] that can sound alarmingly similar to RP /iː/.
The authors say this confirms what they had already observed in other south-
eastern localities. In London, however, they find that the changes in STRUT and
GOOSE are led by young non-Anglos (of African, West Indian, etc. heritage)
and by Anglos (white English) with non-Anglo friends rather than by Anglos
who socialize mainly with other Anglos. It makes the new London STRUT
similar to Jamaican STRUT but makes GOOSE very different from Jamaican
GOOSE, which is firmly back and rounded.
The authors go on to describe the ‘reversal’ of the London diphthong shift,
leading to the restoration of RP-like qualities for FACE, GOAT, PRICE and
MOUTH. Furthermore, they report, the young speakers drop h very much less
than the older people; and they have an innovation called K-backing, involving [q]
or something similar before non-high back vowels, as in cousin, car, cot, caught.
6.17 American T voicing and sentence
A Japanese correspondent asked for an ‘expert opinion’ on American

voiced t, shown in some dictionaries as t̬ . He thought it would be used:
1. After the stressed vowel, and followed by another vowel, e.g. city
180 english accents
. . .providing the following vowel, if within the same word, is weak (which
it usually is). So although we get T voicing (aka ‘flapping’) in city, better, we
don’t get it in latex ˈleɪteks. Strangely enough, this restriction doesn’t apply
across word boundaries, so we do get T voicing despite a following strong vowel
in put up, get over, quite obviously.
2. After the stressed vowel plus n or r, and then followed by another
vowel, e.g. party, twenty
Yes, but the same rule about the next vowel being weak applies: so there is
no T voicing in syntax ˈsɪntæks. Again, across a word boundary a strong vowel
is as good as a weak one, so there is T voicing in start off, count up.
3. After the stressed vowel, and followed by syllabic l, e.g. little
Yes, and even after an unstressed vowel: capital.
He asked why we don’t have any T voicing or elision for the pronunciation
of sentence. This falls under the same heading as words such as button. Before
n Americans typically pronounce t as a glottal stop, ʔ, which is not susceptible
to voicing.
Although we British say sentence as ˈsentəns, Americans have different
rules about syllabic consonant formation (ən ! n̩ ). Americans typically use a
syllabic n after the t in sentence. So the t is immediately followed by a nasal,
which triggers t ! ʔ (the t becomes glottal). You can get ˈsentəns in very careful
speech, but mostly Americans say ˈsenʔns. Since the t is pronounced as a
glottal stop, it is not voiced. There can also be an epenthetic t (i.e. ʔ) between
the last two segments, giving ˈsenʔnʔs. And the vowel can coalesce with the
nasal, so that we end up with something like ˈsẽʔnʔs.
We get the same thing in the American pronunciation of words such as
accountant əˈkaʊnʔnt, mountain ˈmaʊnʔn, Clinton ˈklɪnʔn. (Using a glottal
stop is not the same thing as eliding t. People who claim they elide the t in these
words are mistaken.)
So the crucial thing is the set of environments in which the speaker applies
syllabic consonant formation. In the word sentence, Americans do form a
syllabic nasal, the English don’t (though people from northern Ireland do).
Tricky, isn’t it?
6.18 sexy accents
‘He speaks in a seductively French accent’. Just why a French

accent in English is considered sexy I’m not sure. Like other stereotyped reac-
tions to accents it presumably cannot be taken literally: there can hardly be any
objective phonetic correlates of sexiness that are found in French voices and
absent from English ones.
Presumably it is a relic of the late-Victorian and early-twentieth-century
British attitude to France as the home of uninhibited sexuality. Dirty weekends
6.19 Maori names 181
took place in Deauville or Nice. Aristocrats or other public figures implicated in

sex scandals would seek refuge in Paris. Condoms, condemned for encouraging
immorality, were known as ‘French letters’. We borrowed the word risqué (first
use in English 1867).
Which country has now replaced France in this stereotypical role? Thailand?
Perhaps. But a Thai accent is not considered particularly sexy.
6.19 Maori names

Some years ago I went on a cruise around New Zealand. Our ship
sailed from Auckland, and the first port of call was Tauranga. I heard several
different pronunciations of this name.
The chap on the ship’s public address system called it təˈræŋgə – but
he was an Australian, so not reliable as a source of information about the
pronunciation of NZ place names. The Oxford BBC Guide to Pronunciation
gives the pronunciation taʊˈræŋə, and this is indeed one of the versions
I heard locally.
However, the guide on our shore excursion bus, who was a Maori, stressed the
name on the first syllable. On the whole he seemed to be saying ˈtaʊræŋə. But
sometimes the first syllable seemed to be ˈtɔː- and/or the last part -rəŋə.
His MOUTH vowel in this word (if that was what he was using) sounded rather
un-NZ. Most Kiwis have a quality in the æʊ region, but our guide’s diphthong
started further back. Then I realized that I had been a bit slow on the uptake.
Although most of the time he spoke with what sounded like an ordinary Kiwi
accent, when he came to a Maori name he used not Kiwi English phonetics but
Maori phonetics. Very probably what I heard was (in terms of the Maori phonetic
system) ˈtauraŋa.
He also seemed to pronounce the word Maori itself as ˈmɑːɾi, with a tapped
r perhaps preceded by some sort of central onglide. I find that the Maori form
is actually Māori, i.e. with a long first vowel.
In Rotorua I bought a booklet entitled Pronounce Māori with Confidence,
by Hoati Miwa (Auckland: Reed, 2003). Aimed at the general public, it isn’t
really too bad at all. The material in it is clearly set out, though I wish it wasn’t
presented entirely in the form of reading rules. It tells us that the letters are
such and such, and that they are pronounced (¼ read aloud) like so and so. When
serious phonetic description is called for, the author gets a bit lost. Fortunately
there is a CD attached, so where the written word falters we can just listen
to what it sounds like.
The phonemic system of Māori seems to be very straightforward. Like
other Polynesian languages, it has a five-vowel system i u e o a (short and long),
a very small consonant system and strictly (C)V(V) syllable structure. The Māori
consonants are p t k f m n ŋ ɾ w h.
For the consonant written wh, Niwa tells us that:
182 english accents
Wh is most commonly pronounced f, though the pronunciation varies from

region to region. In the north, for example, you will hear h or hw. In
Taranaki and Whanganui, they say w.
He recommends f. According to Wikipedia, generally it is more precisely a

bilabial fricative ɸ.
When we come to the letter r, Niwa struggles.
[It] is never rolled or trilled. Its sound is very close to d or l. When you make
these two sounds, notice how you flick your tongue against your palate. To say
r correctly on Māori, you flick just the tip of your tongue in the same way.
The examples on the CD show very clearly that the sound is a voiced alveolar
tap ɾ (or occasionally even a trill). So Niwa’s first sentence is way off-beam.
He struggles again when we come to t.
Listen closely and you’ll notice that it sometimes sounds like the t in
English. This is before an i or a u. But when it comes before an a, e or o,
the sound is softer – almost like a d in English.
What he is trying to say is that before high vowels the consonant is aspirated (and
to my ear sometimes a bit affricated). But before non-high vowels it is unaspi-
rated (though still voiceless).
As with the letter r, it’s to do with how you place your tongue against the palate.
(Oh no it isn’t. It’s to do with glottal timing, aka VOT.)

Trying to describe pronunciation without a knowledge of phonetics is like
fighting with one arm tied behind your back. You just don’t have the terminology
or the theoretical framework that you need.
A correspondent comments:
It was interesting reading your take on how Maori words are pronounced in
New Zealand English. It’s been a mess, really, as long as I can remember
(I’m in my mid-thirties). Everyone would like to pronounce things right, but
hardly anyone speaks the language, so hardly anyone gets it right. When
I was a kid, au was usually pronounced to rhyme with cow. Now it seems to
be supposed to rhyme with flow.
That’s probably what I was commenting on for the first syllable of Tauranga,
when I said the guide’s MOUTH vowel sounded rather un-NZ. He was using the
GOAT vowel.
I gather that since about 1980 there has been a drive to pronounce Māori
names in an authentically Māori way, rather than in anglicized form. As a result
there is a lot of variability in what people say. Another complication is that only a
minority of Maoris can actually speak Māori, and fewer still are native speakers.
We passed through the townships of Te Puke and Mount Maunganui, which
our guide pronounced tə ˈpʊki and ˌmɔːŋəˈnuːi (note anglicizing treatment of au
in the latter).
6.20 rich man, poor man 183
When I saw a road sign saying we had reached Ngongotaha I listened out
eagerly for an initial velar nasal. Disappointingly, our guide called it ˌnɒŋɒ
ˈtaːhaː with an ordinary alveolar nasal at the beginning. (But then I read in
Wikipedia that ‘in Tu-hoe and the Eastern Bay of Plenty some speakers merge
ng into n’.)
6.20 rich man, poor man
One of the upmarket areas in pre-volcano Montserrat, just outside

the capital Plymouth, was called Richmond Hill. Now, after the 1997 destruction
of Plymouth by pyroclastic flows, it is uninhabitable. So the locals wittily now
call it Poorman Hill. Why? Because in the local creole and creole-flavoured
English two characteristic processes apply: (1) cluster reduction means that d is
lost from final nd, and (2) the distinction between schwa and the TRAP vowel
is lost.
Hence the antonym of ˈrɪtʃman is obviously ˈpuoman.
7 Phonetics around the world
Languages other than English
7.1 Scottish Gaelic
Scottish Gaelic is a language with remarkable phonetics and a remark-

able orthography. The name of the language, in Gaelic, is Gàidhlig, pronounced
g̊ aːlɪg̊ ʲ. The consonant system includes dental and palatal plosives, palatal
and velar fricatives, three laterals and three r-sounds. There are no voiced plosives,
but fortis consonants are distinguished from lenis by being aspirated or (in non-
initial position) preaspirated. Consonants are divided into ‘slender’ (palatalized)
and ‘broad’ (non-palatalized), and come in slender-broad pairs.
The vowel system includes back unrounded vowels ɯː and ɤː, as in caol
‘narrow’ and ladhran ‘sandpiper’ respectively.
Like the other Celtic languages, Gaelic has initial mutation (lenition) of
consonants in various grammatical or positional environments. In the vocative
case, masculine names undergo both lenition of the initial consonant and ‘slen-
derization’ of the final consonant. Hence Seumas ʃeːməs, the Gaelic for ‘James’,
becomes a Sheumais ə heːmɪʃ. Now I understand why it has come into English
as Hamish.
7.2 tangnefedd a thragwyddoldeb

The thing that first attracted me to the Welsh language was an LP of
community hymn-singing (cymanfa ganu) that I borrowed from a colleague. As a
serious linguist, I am of course disposed to set aside all aesthetic considerations
when discussing language – but I would be being less than honest if I did not
admit that words such as tangnefedd taŋˈneveð ‘peace’ and tragwyddoldeb
tragwɨˈðoldeb ‘eternity’ seemed to have a wonderful ring to them. And I have
always enjoyed singing hymns.
When I was on one of my visits to Montserrat the local church choir had
chosen to do a hymn set to the tune of All through the night. Since this is actually
a Welsh melody, Ar hyd y nos, they asked me to sing a solo verse in Welsh
between the two verses in English.
Holl amrantau’r sêr ddywedant
Ar hyd y nos,
Dyma’r ffordd i fro gogoniant
184
7.2 tangnefedd a thragwyddoldeb 185
Ar hyd y nos.
Golau arall yw tywyllwch,
I arddangos gwir brydferthwch,
Teulu’r nefoedd mewn tawelwch
Ar hyd y nos.
ˈhoːɬ amˈrantajr ˈseːr ðəˈwedant

ar ˈhɨːd ə ˈnoːs
ˈdəmar ˈfɔrð i ˈvroː goˈgonjant
ˈgolaj ˈaraɬ ɨw təˈwəɬux
i arˈðaŋgos ˈgwiːr brədˈvɛrθux
ˈtəjlɨr ˈnevojð mewn taˈwelux
‘All the twinkling stars proclaim

Throughout the night,
This is the way to the vale of glory
Throughout the night.
Other light is darkness
To reveal true beauty;
The family of the heavens in quietness
All through the night.’
Note on the phonetic transcription: Some speakers say i rather than ɨ, ej rather
than əj. In the first syllable of the words brydferthwch and tywyllwch, some sing i
rather than ə. I have analysed the falling diphthongs as vowel plus j or w.
Look for the recording of this by Bryn Terfel (on YouTube).
In the Welsh text you may have noticed the rhymes dywedant – gogoniant,
tywyllwch – brydferthwch – tawelwch.
These are acceptable rhymes within the Welsh literary tradition. But they
wouldn’t do in English, for two separate reasons.
1. The stressed syllables don’t rhyme (stress in Welsh is always on the
penultimate), only the unstressed syllables (-ant, -wch). This would
be like rhyming lemon and bacon in English, or window and yellow.
2. In the second set of words, the rhymed element is actually a suffix,
a morpheme like English -ness. Tywyll-wch ¼ dark-ness; prydferth-
wch ¼ beautiful-ness; tawel-wch ¼ quiet-ness. So this would be like
making rhymes in English such as gladly, brightly, surely, or indeed
darkness, fairness, quietness. You might get near-rhymes like these
in non-traditional verse forms, but not in a hymn. (In Esperanto
poetry, suffix rhyming is a well-known no-no, so much so that it
has its own name, adasismo.)
Just as in Old English and Old Norse poets used alliteration rather than rhyme, so
in the Welsh literary tradition the important thing is not rhyme but cynghanedd
186 phonetics around the world
kəŋˈhaneð, in which patterns of consonants are repeated. (But there’s no

cynghanedd in Ar hyd y nos.)
Here’s an example of cynghanedd from the poem composed by R. Williams
Parry in memory of Hedd Wynn, killed in the First World War when about to
be crowned bard at the eisteddfod.
Y bardd trwm dan bridd tramor, – y dwylo
Na ddidolir rhagor:
Y llygaid dwys dan ddwys ddôr,
Y llygaid na all agor.
‘The poet heavy under foreign earth, – the hands

that will never part again:
the eyes thick beneath a thick door,
the eyes that cannot open.’
The first line up to the dash repeats the pattern b – r – ð – tr – m twice, then
across the first two lines we have d – l – d – l, then in the third line d – d – ð – ð
and in the fourth ɬ – g – ɬ – g. There is a straightforward rhyme, too, between the
second and fourth lines.
Parry continues this linguistic virtuosity for seven more verses. As with the
sonnet, the ghazal or the haiku, truly beautiful poetry somehow combines
memorable words with severe formal constraints.
7.3 Welsh letter names
The letters of the alphabet have their own names in Welsh. In the case of
vowel letters, it is the corresponding vowel sound; in the case of consonant letters,
some have the same name as in English, others have a distinctively Welsh name.
The learner of Welsh needs to know these names in order to talk about Welsh
spelling and to play word games.
A problem arises with the name of the letters I and U. In northern Welsh the
letter I is called iː and the letter U is called ɨː. But in the south there is no ɨ in the
vowel system, and speakers use i in words where northerners would have ɨ. This
makes the names of the two letters homophonous unless we can find a way to
distinguish them.
How can we do that? Rather as the French call the letter Y i grec i gʁɛk to
distinguish it from I i, so the solution for the southern Welsh is to call I i dot and U u
bedol. The reason for the first is obvious: dot means ‘dot’, seen in the lower-case i.
The reason for the second is that pedol means ‘horseshoe’. (It gets soft-mutated to
bedol.) And with a little imagination you will agree that U is shaped like a horseshoe.
An alternative name is u gwpan (cwpan ‘cup’).
I haven’t seen these names in any textbook. I learnt them by listening to
the teacher when I was going to Welsh evening classes and by watching
7.4 Welsh ll 187
Welsh-language game shows on television – on S4C, the name of which is

spoken aloud as es pedwar ec.
Like the newly popular English haitch, for H, the name ec for C includes
an instance of the sound associated with the letter.
7.4 Welsh ll
The BBC has a web page where you can hear audio files of the
pronunciation of various Welsh place names. Trying it out, I was immediately
struck by the way the speaker pronounces the Welsh ll (e.g. in Llandeilo,
Llangwm, Llanddewi Brefi, Benllech). Instead of the standard voiceless alveolar
lateral fricative ɬ, he uses a voiceless palatal fricative ç. (It was Jack Windsor
Lewis who first pointed this out, and Amy Stoller who drew it to my attention.)
I wrote to the BBC about this, and was delighted to receive a reply from BBC
Cymru Wales, whose Ceri Davies wrote:
Thank you for your comments about the Welsh pronunciation guides on bbc.
co.uk/wales. I am not a specialist in the Welsh language, but happen to be a
native Welsh speaker. To my ear, and to the ears of my Welsh-speaking
colleagues here at BBC Wales Education & Learning, we cannot discern a
problem with the pronunciation of the letter ‘ll’ on BBC Wales’ pronunci-
ation guide. We have listened carefully, concentrating on ‘Machynlleth’ and
‘Benllech’ in particular. The audio . . . sounds good to us.
It’s interesting to consider the different ways in which ‘ll’ might be pro-
nounced, but (unless someone is actually saying a soft ‘ch’ in error) all
techniques sound acceptable to us. I wonder however whether in future we
should bear in mind a particular nuance of pronouncing ‘ll’ when recording
audio intended for Welsh learners?
Martha Figueroa-Clark of the BBC Pronunciation Unit (who holds an MA in

Phonetics from UCL) comments:
Following my conversation with Ceri . . . I have listened to many of the
soundfiles repeatedly and, to my ear, there are still clear instances of
realisations with the unexpected palatal fricative as well as the alveolar
lateral fricative (Llanwrtyd Wells, Llanelli and Machynlleth in particular
sound palatal to me). I find it very intriguing that these differences do not
seem to be discernible to native Welsh speakers’ ears.
So there we have it. The use of a palatal fricative rather than an alveolar lateral
fricative, which is immediately obvious to us four phoneticians (Amy, Jack,
Martha and me), is beneath the radar as far as (some) native Welsh speakers
are concerned.
I wonder if there is a sound change in progress, and that ɬ is indeed giving
way to ç. If so, and the palatal succeeds in driving out the alveolar lateral,
speakers of Welsh will no longer be able to boast of having a really exotic sound
in their consonant system. They’ll be no more able to lay claim to exclusivity

than the Germans, and the use of the true alveolar lateral fricative will be left
to Zulu and Xhosa.
In the words of the Zulu farewell, ˈhámb̤ a ˈg̊ aːɬé hamba kahle! Go well!
7.5 c what I mean

I recently watched some ski-jumping on the digital television channel
British Eurosport. The anonymous commentator coped pretty well on the whole
with the many non-English names that inevitably came up. In particular, he knew
the spelling-to-sound value of the letter j – the same value as in IPA – in the
whole swathe of central European languages from Norwegian through Finnish
and Polish to Croatian. He didn’t, for example, make the mistake of calling the
Czech competitor Janda ˈdʒændə or ˈʒændə instead of ˈjændə.
Foreign sounds were mostly mapped onto English sounds, which is fair
enough. But the commentator was also sufficiently familiar with German
phonetics to supplement the English consonant system with not only x but also
ç, and in the correct German distribution to boot.
He even produced a very creditable ˈmaʊɪʃ for the Polish ski-jumper Małysz.
Where he really fell down – and he is not the only one – was over the letters c
and z. Given his apparent familiarity with German, it was disappointing that
he referred to Schlierenzauer as ˈʃlɪərənzaʊə, treating the z as if it were English.
(In German the last part is pronounced -tsaʊɐ.)
Then, oh dear oh dear, we heard that the next meeting would be in ˈlɪbərek.
But the Czech city of Liberec is pronounced with final ts, not k.
Why this reluctance to use an alveolar affricate? Heavens to Be[ts]y, it’s as
bad as when we used to hear about Vá[k]lav Havel. English people seem to have
a complete blind spot about the letter c having the value ts in eastern European
languages.
As the 2008 Wimbledon fortnight was in full swing, I watched a tennis match
on the television in which one of the players bore the name Troicki. The Euro-
sport commentator called him ˈtrɔɪki.
But he’s Serbian, and in Cyrillic his name is written Троицки. It ought to be
pronounced, in anglicized form, as ˈtrɔɪtski. (I think it would be unreasonable for
him to insist on our retaining the Serbian stress, troˈitski.)
On the other hand, the final -ić in the names of various Serbian, Croatian and
Bosnian players does not offer a problem to the commentators. The diacritic may
be omitted in newspaper reports, on the scoreboards and in television captions,
but there is no hesitation in giving Ančić, Bogdanović, Djoković, Ivanović,
Jugić-Salkić, Karlović, Ljubičić and Tipsarević a final tʃ. (Ljubičić gets his
medial tʃ in English, too, although Ivanišević, like the late Milošević, usually
misses out on his ʃ.)
‘Djokovic’ is properly Djoković or Đoković, or in Cyrillic Ђоковић.
7.6 Leewards, ABCs and Virgins 189
In one tournament there was a match between Aleksandra Wozniak and

Caroline Wozniacki. The commentators had an amusing time trying to make
sure we knew the difference between ˈwɒzniæk and ˌwɒzniˈæki. In Polish these
surnames are respectively Woźniak ˈvoʑɲak and Woźniacki voʑˈɲatski. So they
are quite a lot more different from one another in Polish than in English.
But Aleksandra is Canadian and Caroline Danish. Although both are native
speakers of Polish, in their adopted countries they have to be content with the
foreigner’s version of their names, which doesn’t include (in Caroline’s case)
any awareness that c might denote ts.
With Turkish names there’s a different problem with the letter c, namely that
in Turkish it stands for the voiced affricate dʒ. Most foreigners don’t know this.
So Incirlik air base (in Turkish spelling, İncirlik) gets mispronounced as ˈɪntʃ- and
former prime minister Ecevit as ˈetʃ-. In Turkish they are inˈdʒirlik and ˈedʒevit.
There’s a different problem for the commentators with the Slovak player
Hrbatý. In Slovak he’s stressed on his syllabic r, thus ˈɦr̩batiː. In English they
call him hɜːˈbɑːti or hɜːˈbæti.
Turning from sport to food, at one time a television advertisement enticed us
to eat a particular brand of pizza by telling us that it was topped with tʃəˈrɪtsəʊ
(a kind of sausage). That’s chorizo.
So our confusion about the letter Z continues. OK, the ad is indeed about
pizza, in which – as everyone knows – the letter z, or rather double zz, is
pronounced ts.
But pizza is Italian, chorizo is Spanish.
If there were an Italian word pronounced tʃoˈɾitso it would probably be spelt
ciorizzo. The sequence cho- would not be found in an Italian word.
However, as I say, chorizo is actually a Spanish word. And in Spanish
the letter z corresponds to θ or s, depending on the kind of Spanish involved.
Hence the Spanish pronunciation of chorizo is tʃoˈɾiθo (or in Andalucia and the
Americas tʃoˈɾiso).
In English, though, z always means z. We don’t generally know about Spanish
phonetics. Anglicized, chorizo becomes tʃəˈriːzəʊ. Usually, at least.
7.6 Leewards, ABCs and Virgins
In the Caribbean, one of the Leeward Islands, St Martin (Sint Maarten,

Saint-Martin), is divided into a Dutch part and a French part. The official lan-
guages are Dutch, English and French. (By the way, the islands we call the
Leeward Islands are called the Windward Islands by the Dutch. It depends on
where you are standing.) However, virtually the only Dutch to be seen in public
signage is in street names.
Otherwise, the language in the Dutch part is English, and what you mostly
hear on the street from the locals is a form of Caribbean English creole. In the
French part of the island, as you might expect, everything is in French.
A cash machine (ATM) I looked at in the Dutch part is trilingual, but French
is not one of the languages. Papiamento, on the other hand, is – no doubt for
the benefit of Netherlands Antilles citizens from the ABC islands off the
coast of Venezuela, whose language it is. (The other two languages were English
and Dutch.)
APRESIABEL CLIENTE, DESPENSA; E MASHIN AKI TA TEMPOR-
ALMENTE FOR DI SERVISIO POR FABOR HUSA UN DI NOS OTRO
BANKOMATIKONAN
Papiamento or Papiamentu is the everyday spoken language in Aruba, Bonaire

and Curaçao. It is a Spanish/Portuguese-based creole. (The spelling of the name
of the language depends on which island is concerned.)
The plural ending in Papiamentu is -nan, pronounced naŋ (with the final nasal
being velar, as in Venezuelan Spanish). So in the bank ATM screen the last part
of the word bankomatikonan is the plural ending. Nan is also the third person plural
pronoun, ‘they, them’. What its etymology is, I don’t know, though it looks African.
The word for ‘street’ is kaya, from Spanish calle ˈkaʎe or ˈkaje.
In St Thomas (American Virgin Islands) the only official language is naturally
enough English. However, the local dialect is a Caribbean English creole com-
parable to that of the other Leewards, which means it sounds more British than
American. Other matters unusual for an American territory are that they drive on
the left and that many of the street names are Danish. This is because the US
Virgin Islands were Danish until purchased by the Americans in 1917.
7.7 clicks demonstrated
YouTube is a marvellous resource for everyone interested in sound

and voices. One of the things I found there is a very clear demonstration of
clicks. They are among the speech sounds of the Namibian language that the
speakers in the video clip call KhoeKhoegowab. If you do a search on that term
in YouTube you will find a set of language lessons. The clicks are demonstrated
in ‘Lesson No: 1’.
KhoeKhoegowab is an alternative name for the language otherwise known
as Nama or Nàmá, and formerly known as Hottentot. It is a national language
in Namibia, with a quarter of a million speakers.
The video brings out the differences between the four click types very clearly.
First there is the dental affricated click, Nama spelling /. This is the same as the
click spelt c in Zulu and Xhosa, and in IPA written ǀ (formerly ʇ). Then there is
the retroflex or (post)alveolar click, with an abrupt release, Nama !, Zulu-Xhosa
q, IPA ǃ (ʗ). It has a lower-pitched resonance than the alveolar (so-called
‘palatoalveolar’) click, also with abrupt release, Nama #, IPA ǂ, which is not
found in Zulu-Xhosa. Lastly we have the alveolar lateral affricated click, Nama //,
Zulu-Xhosa x, IPA ǁ (ʖ).
7.8 an impossible sound 191
For #, printed books often write 6¼ (the doesn’t-equal sign).

From the example sentence in the clip of Lesson 1 –
/khim !nu #hab //ga
– it seems that the language has not only plain clicks (voiceless, unaspirated,
probably with an accompanying glottal stop) but also aspirated (/kh), nasalized
(!n) and voiced (//g) – though the last-named doesn’t always sound very voiced
here. The click in the third word, #hab, sounds to me breathy-voiced and nasal,
as if it were Zulu ng# (i.e. the pulmonic-air part sounds like that of the Zulu
breathy-voiced nasalized clicks ngc, ngq, ngx – Zulu doesn’t have #); however
it is also described as involving a ‘voiceless nasal with delayed aspiration’.
7.8 an impossible sound

In the 1949 edition of The Principles of the International Phonetic
Association one of the IPA symbols listed is ʞ, defined as standing for a velar click.
This definition is repeated in the current Unicode standard, where ʞ, code
point Uþ029E, LATIN SMALL LETTER TURNED K, is glossed as ‘proposed
for velar click’. The Unicode standard continues, ‘withdrawn by IPA in 1970’.
Correct, though I suspect the date is wrong. In fact ʞ had disappeared from
the IPA chart by 1951, though not from the Principles booklet.
And why was this symbol dropped? One answer is that no language had
been found to have the velar click as one of its sounds. But the real reason is that
a velar click is anatomically impossible.
By the 1960s we understood non-pulmonic airstreams better than our prede-
cessors did. We knew that clicks are made with an ingressive velaric airstream
mechanism, involving a closure between the back of the tongue and the velum
(as in k). Since this closure is used to separate off the mouth cavity from the
pharynx ready for initiation of the air flow, it cannot simultaneously be used for
a velar velaric articulation. As well as the velar closure, clicks must have some
closure forward of the velar place, involving therefore the front/blade/tip/side
of the tongue or the lips. Muscular expansion of the cavity between the two
closures provides the suction needed to make a click on release.
What you can have simultaneously with a click is a velar articulation involving
a non-velaric airstream mechanism, typically a pulmonic one. That is why clicks
can be nasal (simultaneous ŋ rather than k), voiced (simultaneous ɡ), etc. Indeed,
one way of checking that a sound is a click is to see whether you can do ŋ
simultaneously with it: if you can, it’s a click.
As Pullum and Ladusaw put it in their discussion of ʞ (Phonetic Symbol
Guide, University of Chicago Press 1996, second edition, p. 101):
The IPA’s recommendation was simply a mistake, and the symbol could never
be validly used. It was dropped from the IPA chart with the 1979 revision.
7.9 fricative trill, trill fricative

It’s one of the rarest sounds in the world’s languages. There may well
be no language other than Czech that has it. I’m referring, of course, to the
fricative trill, orthographically ř, which we all know through the name of the
composer Dvořák (go to Wikipedia to listen to a sound file with the pronunci-
ation of this name.)
In 1945 the IPA adopted the symbol ɼ to represent this sound, a voiced
alveolar fricative trill. The composer’s name could then be transcribed ˈdvoɼaːk.
One of the decisions made at the IPA’s Kiel Convention in 1989 was to
withdraw recognition from this symbol. The IPA members voting at the Convention
took the view that it was adequate to use the ordinary symbol r (voiced alveolar trill)
together with the ‘raised’ diacritic to show the closer, hence fricative, articulation.
So in the Handbook of the IPA (Cambridge University Press 1999) you will
find the Czech consonant written as r̝ . Dvořák is now ˈdvor̝ aːk.
Some people say they regret this decision, and I am half inclined to agree with
them, though it does seem an unnecessary luxury to have a special symbol
needed only for one language. (But compare Swedish ɧ.)
The fricative trill also occurs voiceless, though only as a positional allophone
after voiceless consonants (tři ‘three’) and word-finally (pekař ‘baker’). Both
with the old symbol and with the new one there is a problem about transcribing
this variant (if you feel you need to). There’s no room for a voiceless diacritic
underneath, so it has to go on top: r̝̊ .
Bill Lee’s Teach Yourself Czech (English Universities Press 1964) offers
excellent pronunciation advice to the learner. See also the Handbook of the IPA.
7.10 all the speech sounds in the world
The emails I receive about phonetics range from the very naïve to
the very sophisticated. Sometimes it is difficult to know which is which. One
correspondent writes:
I am particularly keen to know where I could find a single list that would
include all the sounds found in all the languages. Does such a list exist?
Is my correspondent a beginner who should be advised to consult a textbook

of general phonetics, e.g. Peter Ladefoged’s Vowels and Consonants (Blackwell
2001), or, even better, to enrol for a course in general phonetics?
Or is this a serious question for which a serious answer might be sought in
Maddieson and Ladefoged, Sounds of the World’s Languages (Blackwell 1996)?
I replied, I hope helpfully, that an obvious starting point would be the
International Phonetic Alphabet.
But I think a qualified phonetician would not have asked the question in this
way unless putting on an act of being naïve. There is a difficulty not always
7.11 WALS 193
appreciated by non-specialists: you can’t enumerate ‘sounds’. This is because

there is no real way to say whether sounds are ‘the same’ or ‘different’ except
with reference to a specific language or dialect. For example, English /t/ is by
default a voiceless aspirated alveolar plosive, Dutch /t/ is ditto but unaspirated,
Swedish /t/ is aspirated but dental and French /t/ is both unaspirated and dental.
So depending on language /t/ may by default be aspirated or unaspirated,
alveolar or dental.
Once we start to add in allophonic variation within a language, the difficulty
is multiplied. Sometimes English /t/ is not voiceless, sometimes it is not aspir-
ated, sometimes it is not alveolar (in certain positions it can be dental or glottal),
and sometimes it is not even plosive.
So how many ‘sounds’ are we dealing with here? For everyday purposes,
they are all transcribed with the same t symbol, because we apply a principle
of transcriptional simplicity. But in a sense they are all different.
The English l phoneme comprises two sounds (two allophones), as we know.
The clear l is clear (¼ has a palatal resonance, sort of), but not as clear as the
German l and nothing like as clear as the Russian ‘soft’ lʲ; the place is alveolar,
i.e. retracted from what we might think of as the cardinal dental position, but not
as retracted as that of the Korean l. Factor in the various l-sounds of thousands
of other languages: how many l-sounds are there altogether?
Even if you start with a list of the 109 alphabetic IPA symbols plus 33 stress,
length and tone marks or thereabouts, there are still 32 diacritics that can be
applied to the alphabetic characters to symbolize secondary articulations and
various other attributes. Sometimes more than one diacritic can be applied to
the same base. The combinatorial explosion makes it unrealistic to attempt to
count the resulting ‘sounds’. And we haven’t even started on voiced or nasalized
clicks, double articulations, affricates and diphthongs, all of which can be
represented in IPA by combining two alphabetic symbols with a tie bar, but
are arguably single ‘sounds’.
7.11 WALS
If you want to follow up the question of the range of vowels and

consonants to be found in the world’s languages, you can now access WALS
(the World Atlas of Language Structures) online at wals.info. You can read the
entire text of each chapter and look at all the maps. The size of the consonant
inventory in a language is revealed as ranging from 6 to 122, the record
holder being !Xóõ, a Southern Khoisan language of Botswana, which has 122
consonants, mainly because it has a very large number of different click sounds
with which a word may begin. Most languages, though, have 20–25. Vowel
inventories range from 2 to 14, with an average of 5 or 6. The record holder here
is German. That’s among the languages included in the WALS sample, and using
its criteria for analysis. (There are difficulties in deciding how to treat long
vowels and diphthongs when determining how many ‘vowels’ a language

has; after all, for EFL purposes we usually consider ‘British English’ to have
20 or so.)
7.12 IPA in a logo
The city of Szczecin in Poland got itself a snazzy new logo incorpor-
ating an IPA transcription of the name as ʃ t̺͡ ʃɛˈt͡ ʃʲin.
This provoked the local paper to ask Czy nasze miasto nazywamy SzczeCIN?
(Do we call our city SzczeCIN?)
Answer: no, you don’t. You call it SZCZEcin. Like almost all other Polish
words this name is stressed on the penultimate.
This could be a delicate matter, since before the Second World War the city
was in Germany and called by the German name Stettin ʃtɛˈtiːn, which is indeed
stressed on the last syllable. But now it’s part of Poland, and is called by its
Polish name, which is stressed on the penultimate.
It was a nice idea to use eye-catching IPA symbols as the main content of the
logo, with the usual orthographic form written very small underneath. But what
a funny transcription!
I have touched elsewhere (3.20 above) on the question of the Polish sz, cz, ś, ć
and how to transcribe them. (The letter c before i, as in Szczecin, is pronounced
like ć.) Pretty well everybody agrees on writing the latter two, the alveolopala-
tals, as ɕ, tɕ. The former two, postalveolars, are usually written simply as ʃ, tʃ,
though you could argue for ʂ, tʂ.
What I have never seen before is the use of diacritics for apical and
palatalized applied to the basic palatoalveolar symbols. I don’t think they are
needed here.
We can argue about whether the first vowel is best written with the simple
symbol e or the comparative symbol ɛ. We can argue about whether or not we
ought to use tie bars with the affricates to show that they’re not plosive-fricative
sequences. We can agree that the stress mark ought to be in the right place,
showing initial stress.
Me? I would write the simple ˈʃtʃetɕin. That’s if I wanted to transcribe the
Polish pronunciation. The nearest we can get in English is ˈʃtʃetʃiːn.
7.13 frozen in Italian

Pronouncing Italian words is fairly straightforward, as long as you
know where the stress goes.
All too often, the classical music commercial radio announcers don’t.
More than once I have heard them say Che gelida manina with the stress
in gelida on the -li- instead of on the ge-.
7.14 ugh! 195
If you watch or listen to this aria being sung (try YouTube), you can hear the
correct stress. It’s ke ˈdʒelida maˈnina. It has the same rhythm as its English
version your ˈtiny hand is ˈfrozen. Duh!
In Latin we have adjectives ˈacidus, ˈgelidus, ˈrapidus, ˈsolidus. The suffix
-id- has a short vowel, therefore stress goes on the antepenultimate. Correspond-
ingly, in Italian (and also in Spanish) it’s ácido, gélido, rápido, sólido; in English
acid, gelid, rapid, solid ˈæsɪd, ˈdʒelɪd, ˈræpɪd, ˈsɒlɪd. I could go on: it’s just
the same for the Latin and Italian words corresponding to English avid, horrid,
liquid, rigid, splendid, valid and many others. Where’s the problem? Why would
anyone suppose that the stress fell on a different vowel in Italian?
Perhaps the French that we British learn at school is to blame. Although
French acide, rapide, solide, etc. in theory have no word stress, in practice
we treat them as having final stress: aˈsid, raˈpid, sɔˈlid. Taken over into Italian,
this could be the source of the announcer’s mistake. (The actual item *gelide
seems to have been lost in French: they say gelé.)
And by the way it’s Plácido Domingo.
7.14 ugh!
All good students of phonetics learn to recognize and perform a close

back unrounded vowel, ɯ.
But because it is perceived by most of them as a very strange and exotic sound
they tend to overarticulate it: tense jaw muscles, tensely spread lips, extra length,
grimace.
Those in whose language it occurs, on the other hand, naturally treat it as a
perfectly ordinary and everyday sound. They pronounce it in quite a relaxed way.
As far as the languages I have come into contact with are concerned, that
means speakers of Korean or of Scottish Gaelic. In both of these languages it
contrasts with a close back rounded u.
The Korean word for ‘writing, a piece of writing’ exemplifies this vowel: it
is 글 kɯl (or you might prefer to show the initial consonant as a devoiced g̊ ). Of
the two competing romanizations of Korean, one represents the word as geul, the
other as kŭl.
You usually get something similar for the close back vowel of Japanese.
But that’s not so good for demonstration purposes, since Japanese has no contrast
of rounding.
In Scottish Gaelic both long and short ɯ occur, in contrast (7.1 above). Thus
for the long vowel we have, for example:
caol kɯːɫ̪ (narrow)
saor sɯːɾ (free)
laogh ɫ̪ɯːɣ (calf)
aois ɯːʃ (age)
faoin fɯːɲ (silly)
and for the short:

uisge ɯʃg̊ ʲɪ (water – the etymon of whisky)
uiseag ɯʃag̊ (lark).
7.15 Icelandic
Iceland’s financial difficulties brought the name Kaupthing to our

television screens and newspapers. Its Icelandic spelling is however Kaupþing,
which includes the special letter þ (thorn). None of our newspapers attempts to
reproduce this character. Even the bank itself writes its name with th outside
Iceland: KAUPÞING at home, KAUPTHING abroad.
The sound corresponding to þ is a voiceless dental fricative, θ. Icelandic
orthography also has ð, with the same sound as in IPA, ð. I believe that these
are in complementary distribution, with the letter þ used morpheme-initially and
ð elsewhere, and the two sounds likewise (except that phrase-medial function
words can have initial ð, spelt þ).
It is remarkably difficult to find a good account of Icelandic phonetics in
English. I have a book-length description in Icelandic, Hljóðfræði, by Árni
Böðvarsson (Reykjavík: Ísafoldarprentsmiðja 1979), which the author, now
dead, gave me on the occasion of my visit to his country in 1988, but I can’t
say that I can actually read it. Fortunately he also gave me his Esperanto booklet
Islanda lingvo (Reykjavík: Islanda Esperanto-Asocio 1977), which furnishes the
most comprehensive account that I can understand.
When we did Icelandic in my informant class at UCL the sounds that people
found most interesting were the nasals and liquids devoiced (i) before voiceless
consonants, as in verk ‘work’ with r̥ , and (ii) in final position after a consonant,
as in vatn ‘water’ with n̥ and fugl ‘bird’ with l̥ . Another tricky consonant is
the laterally released voiceless alveolar affricate tɬ spelt ll, as in fjall ‘mountain’.
7.16 wie war das?

As I finished checking in to my Berlin hotel, the receptionist asked
me Möchten Sie ˈveːlan? ‘Would you like [something not understood]?’
The nearest word that I could think of here was wählen ˈvɛːlən ‘to choose’,
but that didn’t seem to make much sense.
It turns out what she had said was W-LAN, the German for Wi-Fi, the facility
for connecting a laptop to the internet wirelessly. So having sorted that out I said
yes, please.
Wi-Fi is one of those computery words that I expected, evidently wrongly,
to be international, borrowed by every language you come across; but it isn’t.
I gather that it is actually a proprietary trade name. According to the German
7.17 simplicity 197
Wikipedia, it doesn’t stand for ‘wireless fidelity’, as you might think. It is a name
(a Kunstbegriff) invented merely for marketing purposes.
I also discovered that the Germans believe their terms Handy and USB stick
to be (i) English and (ii) international. But the first is what we Brits call a mobile
and the Americans a cellphone. The second is normally known, in Britain at
any rate, as a memory stick or a memory key or a flash drive or a thumb drive:
we can’t agree on just what to call it, but ‘USB stick’ is not one of the options.
7.17 simplicity
We normally transcribe the English word red as red.
But there are two things here that worry some people. First, they argue, the
initial consonant is a voiced postalveolar approximant and should therefore be
written ɹ. Second, the vowel is lax (short) and should therefore be written ɛ.
Let us take first the question of how to transcribe the initial consonant.
It can indeed be narrowly represented as ɹ. That would be an appropriate
‘impressionistic’ transcription.
However, in the old IPA Principles booklet (1949) we read:
27. (a) . . . it is desirable to substitute more familiar consonant letters for less
familiar ones . . .
(e) The letter r may, when convenient, replace ɹ, ʀ or ʁ in the transcription
of a language containing one of these sounds but not a rolled lingual r . . .
[Note to the young: until quite recently, to substitute A for B always meant to use
A instead of B, not the other way round.]
This idea was put most clearly by David Abercrombie in his English Phonetic
Texts (Faber 1964, pp. 17–18).
A simple phonemic transcription uses letters of the simplest possible shapes.

This is a typographical principle: the simplest shapes are the most familiar,
the most typographically satisfactory.
He goes on to say that ŋ ʒ ɯ ʎ are less simple, because less ‘romanic’, more
exotic, than the traditional letters of the alphabet.
The conventions and traditions of the IPA often allow, for the representation
of a particular sound, a choice between two or more different letters. In these
cases where such a choice exists, one of the letters will usually be found to
be more romanic . . . than the alternative.
Somehow this principle of simplicity seems to have got dropped from the current
IPA Handbook, but I believe it is still valid. It immediately justifies the use of
r rather than ɹ and e rather than ɛ for English.
In the case of the vowel, there is the further point that, as the IPA Handbook
itself mentions (p. 30):
the vowel phoneme of get in Standard Southern British English has allo-
phones, according to phonetic environment, which mostly lie between the
cardinal vowels [e] and [ɛ], some realizations being close to one and some
to the other. It is therefore permissible to choose either symbol as the one to
represent the phoneme.
Since the DRESS vowel can be unambiguously written e, an English transcription

with the symbol e is simpler than one with ɛ. The same applies in those languages
that have just five vowels (or just five peripheral vowels), such as Modern Greek,
Serbian, Polish, Czech and Japanese. In these languages the mid front vowel is
best, and most simply, written e. Whether its quality is closer to that of cardinal
2 [e] or cardinal 3 [ɛ] is irrelevant. On the other hand, in languages that have a
close-mid front vowel in contrast to an open-mid one (e.g. French, German, a
Scottish accent of English), the two symbols e and ɛ must both be used; and
that means ɛ for French même, German Bett and Scottish dress.
There is not some great phoneme system in the sky from which particular
languages select their phonemes, with one IPA symbol always standing for the
same thing. Languages differ. We use the symbol t for the aspirated alveolar
plosive of English and the unaspirated dental plosive of Russian because it is
simpler to do so. We use the same symbol l for the clear alveolar lateral of
German, the variably coloured alveolar lateral of English, and the retracted
postalveolar lateral of Korean, because it is simpler to do that than to festoon
the transcription with diacritics. It is better to state such information in the
conventions that accompany a phonetic transcription rather than in the transcrip-
tion itself. In a language that distinguishes dental and alveolar (or postalveolar)
plosives, we obviously need to symbolize their place of articulation explicitly,
by distinguishing plain t from t̪ and/or ʈ. In a language that doesn’t, we don’t.
Nevertheless, people do quite often query the convention whereby I (and
various other phoneticians) write the DRESS vowel as e.
7.18 Calon lân
Some of the emails that come to me out of the blue are difficult to
know how to reply to. One correspondent asked simply ‘Have you any idea
where I can obtain a phonetic version of the Welsh song calon lan please?’
I should explain that Calon lân (‘A pure heart’) is a well-known Welsh
hymn, often sung for example at international rugby matches. Unlike, say,
Cwm Rhondda, which is often sung in English under similar circumstances
(‘Guide me, O thou great Jehovah’), Calon lân is always sung in Welsh, even
in the non-Welsh-speaking parts of Wales, perhaps because it has no established
English translation.
Welsh spelling is extremely regular. If you know the spelling-to-sound corres-
pondences, you can just read a written Welsh text out, sound by sound, with only
very occasional uncertainties or exceptions.
7.18 Calon lân 199
My correspondent probably didn’t know this, or didn’t know what the

spelling-to-sound correspondences are. Otherwise he wouldn’t have been asking
for a transcription of a particular text.
It’s also possible that he was not familiar with phonetic symbols, either. If he
were he would probably not have asked for a phonetic version but a phonetic
transcription. (People sometimes ask for a phonetic translation, which is
even more confused.)
It may be, then, that what he wants is a guide to the pronunciation using
English spelling conventions. Since Welsh contains both consonants and vowels
unknown in English, this is not a satisfactory way of proceeding. I suppose
it might begin need oo-in govin buh-wid moythis, but already we are up against
the problem of how to represent the Welsh stressed schwa in an open syllable.
The second line, perhaps ire uh beed nye behr-lye mahn, brings us up against
the problem of how to write aj (ai) unambiguously, and how to show er in a way
that won’t be read as ɜː.
So here is the first verse in orthography and in IPA.
Nid wy’n gofyn bywyd moethus,
Aur y byd na’i berlau mân;
Gofyn rwyf am galon hapus,
Calon onest, calon lân.
Calon lân yn llawn daioni,
Tecach yw na’r lili dlos;
Dim ond calon lân all ganu,
Canu’r dydd a chanu’r nos.
nid ujn govin bəwid mojθis

air ə bid naj berlaj maːn
govin rujv am galon hapis
kalon onest kalon laːn
kalon laːn ən ɬawn dajoni
tekax iw nar lili dlos
dim ond kalon laːn aɬ gaːni
kaːnir dið a xaːnir nos
There are several excellent recorded versions of this hymn available. Take
your pick of Kathryn Jenkins, the Welsh rugby team, Bryn Terfel and others.
There are occasional minor differences in the wording (e.g. Bryn Terfel sings
Does ond rather than Dim ond) and various phonetic differences reflecting local
pronunciation (e.g. Kathryn Jenkins pronounces rwyf as if spelt royf, cf. llwyd
becoming the English name Lloyd).
In the refrain, it is interesting to note the successive mutations of the word for
‘sing’, which appears not only as canu (its basic form), but also as ganu (soft
mutation) and as chanu (so-called aspirate mutation, i.e. with its initial plosive
changed to a fricative).
7.19 Ystrad Mynach

There’s a village near Caerphilly in south Wales called Ystrad
Mynach. In a national news bulletin, I heard BBC announcers call it ˌʌstrəd
ˈmʌnəx, as recommended by the BBC Pronunciation Unit.
Phonetically, the name is an interesting mixture of English and Welsh.
In standard Welsh it would be ˌəstrad ˈmənax, with strong vowels in the final
syllable of each word. In Welsh, vowels do not get weakened as in English,
and in particular you never get schwa in the final syllable of a content word. You
do, however, frequently get schwa in stressed syllables.
So the anglicization involves weakening of the final vowels, as well as
the usual mapping of Welsh stressed schwa onto the STRUT vowel (which in
south Wales English is pretty schwa-like anyhow). But it does not extend to the
replacement of x by an English consonant.
Outsiders, though, would surely change x to k. And outsiders wouldn’t know
how to treat the Welsh spelling y (which in proclitics and non-final syllables
stands for schwa). They would probably say something like ˌɪstrəd ˈmɪnək.
So due credit to the BBC announcers for getting it right.
Ystrad, by the way, is the Welsh equivalent of the Gaelic srath that we know
from Scottish place names such as Strathclyde (Srath Chluaidh, Welsh Ystrad
Clud). Mynach means ‘monk’ (< Latin monachus): so on the face of it the whole
name means Monk’s Vale. However, there appears to be no former monastery
nearby, and experts tell us that the name may actually be derived from some
lost personal name.
Postscript
Where did my interest in phonetics come from?

I have always been interested in languages, and particularly in pronunciation.
I started French at school at about eight years of age, then Latin from age nine
and Ancient Greek from age twelve. For my twelfth birthday I asked my parents
to give me an Italian-English dictionary, since I had started to teach myself a little
Italian. I passed my O level exams in French, Latin and Greek at age fourteen,
and my A levels in Latin and Greek at age sixteen. That same year, once my
exams were out of the way, I decided to teach myself Esperanto, which soon
became my best language after English.
Seeing my interest in languages, my parents made the far-sighted decision
when I left school at eighteen to arrange an exchange with a German family.
I stayed in Kiel with the family for six weeks, followed by six weeks for their son
with us in England. I was determined to make good use of this, my first trip
abroad, learning German ab initio by total immersion. My exchange partner,
Klaus, took very seriously the task of ensuring that by the time I came back to
England I could speak fluent German. I can see in hindsight that I also picked
up excellent German pronunciation habits.
So by the time I became an undergraduate at Trinity College Cambridge I had
a thorough knowledge of Latin and Greek, reasonable ability in French, and good
though non-literary fluency in German – oh, and fluency in Esperanto.
For the last year of the Classical Tripos at Cambridge I chose to specialize in
language. This meant comparative philology, plus some general linguistics
and (optionally) an introduction to phonetics. The phonetics teacher was John
Trim. He taught us to recognize and produce most of the sounds represented on
the IPA chart, together with their articulatory classification. We learnt to tran-
scribe English and to make ejectives, implosives, clicks and cardinal vowels.
I was hooked.
In summer 1960, aged twenty-one, I was fortunate enough to win a travel
grant, which enabled me to visit Greece and see the antiquities in Athens, Delphi,
Mycenae, Delos, Knossos and elsewhere. The grant covered my return ticket
to Piraeus by train and boat, but little else. I toured around in Greece, living very
simply and of course doing my best to speak and understand Modern Greek.
I stayed in youth hostels and in private rooms, hitch-hiking and island-hopping
by ferry. (This was before the days of mass tourism in Greece. At the time very
few Greeks spoke English.)
201
202 postscript
The next academic year, following Trim’s advice, I moved to University

College London to pursue a master’s degree in general linguistics and phonetics.
Among my teachers were A.C. Gimson, J.D. O’Connor, Dennis Fry and Adrian
Fourcin. I did a course in French phonetics, taught in French by Hélène Cous-
tenoble and Marguérite Chapallaz, and visited SOAS for linguistics with
R.H. Robins. I could not have asked for better teachers. At the end of the two-
year master’s course UCL offered me the post of Assistant Lecturer in Phonetics.
I jumped at the offer, and was accordingly able to spend my entire career
teaching phonetics at UCL, doing the subject I loved.
I took my PhD in 1971, exploring the phonetic adjustments Jamaican immi-
grants in London made to their English pronunciation in response to their new
circumstances. At UCL I also frequented the English department, benefiting
from the teaching of Randolph Quirk, and the linguistics department, under
Michael Halliday.
Later on I enrolled for an Introduction to Zulu course at SOAS. (But it was
not until 2013 that I actually visited South Africa for the first time and could try
out as much as I remembered of what I had learnt of that phonetically very
interesting language.) I also enrolled for courses in Modern Greek and Welsh
at City Lit in central London. Meanwhile, Esperanto had taken me to many
different countries in Europe, where I regularly tried to find out about the
pronunciation of the local language – Polish, Dutch, Swedish or whatever.
One of the courses I most enjoyed teaching at UCL was a practical class
working on ten or twelve different languages each year, using a native speaker
to demonstrate words and phrases which the students and I would then analyse
phonetically, transcribe and reproduce. This gave me a good knowledge of the
phonetics (if not much else) of languages as varied as Modern Standard Chinese,
Japanese, Korean, Arabic, Polish, Spanish, Portuguese, Danish and Icelandic.
The only language I learnt thoroughly in adulthood was Welsh, which
I continued to study in evening classes in London up to A level standard.
It was marvellous being in a university department along with half a dozen
or more other phoneticians, with whom one could always argue about phonetics
and discuss this or that problem. Teaching students phonetics was always a
pleasure, too – particularly postgraduate (research) students whose dissertations
I had to supervise. We also had many excellent academic visitors over the years,
a stimulus for new directions in my own research.
You can perhaps see why when I retired in 2006 I felt the lack of regular
phonetic discussions with students and colleagues. That is the reason I started
my phonetic blog, in which I could discuss phonetics with the world and conduct
a dialogue with the many followers who commented on what I had written. This
volume, as I mentioned in the Preface, is based on some of my postings there.
Index of words
affect, 55 diplodocus, 48
albatross, 68 diploma, 47
Alcester, 8 divorce, 72
amiodarone, 13 doh, d’oh, duh, 25
analogous, 147 Dominica, 8
angrily, 57 Dvořák, 192
Antigua, 7, 167
apostate, 34 -ed, 56
artisanal, 3 elevator, 66
athlete, 44 England, 44
aubergine, 160 entombment, 49
Audi, 19 Entwistle, 36
escalator, 2
balcony, 68 etc., 83
bandied, 54
Beijing, 37, 150 feng shui, 16
Berlins, 29 fiesta, 45
bevy, bevvy, 151
following, 44
Birmingham, 83 Freixenet, 19
Blaengwrach, 27 Friern Barnet, 28
bleeding, 118
Bournemouth, 67
Braun, 19 gaol, 147
Gauguin, 13
Burgh, 8
genuine, 42
C. diff, 4 glorious, 42
calendar, 176 gonna, 73
good deal, 49
Campbell, 23
cannot, 49 goodbye, 106
cervical, 7 Gwaun-cae-Gurwen, 27
Chagos, 11
character, 176 Hamburg, 88
Charon, 10 Hamish, 184
chorizo, 189 Haut de la Garenne, 28
Christmas, 69 Hawai’i, 145
code-share, 74 hello, 31, 105
crazies, 53 Hokkaidō, 90
Cretan, Cretian, 26 holey, 167
Cymru, 16 Honshū, 90
horrorshow, 66
data processing, 70 hypernymy, 12
definitely, 155 Hyundai, 18
deity, 11
diaspora, 48 i.e., 83
digoxin, 147 imma, 73
diocese, 6 incidence, 71
203
204 Index of words
jaguar, 7 quantitative, 51
Judea, 11
remembrance, 45
Kyocera, 18 repertory, 30
Kyoto, 45 Richter, 15
Rotherhithe, 22
Lagos, 21 Rothersthorpe, 22
Leatherhead, 23 royal, 39
liege, 21
linguolabial, 140 Salida, 8
Llantwit Major, 34 Sarkozy, 29
Lloyd, 27 sarnie, 65
Llwynywormwood, 14 scarper, 65
London, 88 schnitzel, 2
Lucida, 32 sentence, 180
settee, 72
Machynlleth, 187 Sexwale, 19
Málaga, 88 shar pei, 65
mange-tout, 160 shih tzu, 4
margarine, 147 silicon, silicone, 24
marm, 65 sloth, 1
Mbabane, 17 some, 58
meningococcal, 147 sphygmomanometer, 51
meta-analysis, 39 St John’s wort, 148
mine, 76 Sydney, 87
money, 88 syringa, 147
mores, 161 Szczecin, 194
Myanmar, 64
Mynd, 27 tart, 94
Mynydd Bach, 27 tattoo, 72
Mytholmroyd, 31 terrorist, 74
thrombolysis, 13
Nestlé, 19 tinnitus, 25
Newcastle, 31 torte, 94
Nicaragua, 7 tortoise, 20
Novi Sad, 88 trial, 39
nuclear, 2 twenty, 63
um, 77
omega, 5
unchartered, 55
Penwortham, 30 veg, 147
percolator, 2 veterinary, 29
pizza, 189 volcano, 86
Placerville, 9
plenty, 63 wanted, 63
plethora, 5 Whiskas, 65
Polyfilla, 65 wind, 75
Pontypool, 27 Winstanley, 31
porpoise, 20 wintry, 45
Poznań, 88 Wolverhampton, 23
precede, 12 wort, 148
proceed, 12
prostate, 24 Xhosa, 19, 190
prostrate, 24
pukka, 15 ylang-ylang, 16
Punjab, 15 your, you’re, 171
purgative, 147
pwn, 35 Zheng Jie, 36
General index
abbreviation, 83 counterpresuppositionals, 111

accentuation, 115–16, 127 Cowan, John, 74
affricate, 143 Cruttenden, Alan, 111
alveolopalatals, 98, 152 CURE, 171
apostrophe, 158 cynghanedd, 186
Arabic, 150, 202
Armstrong, Eric, 65, 93 Danish, 202
Ashby, Patricia, 149 Davidsen-Nielsen, Niels, 73
assimilation, 38 decompression, 44
Australian English, 174 degemination, 49
Azeri, 144 Deterding, David, 68
devoicing, 90
Bantu, 19 dictation, 82
Barry, Martin, 76 Dutch, 149, 193, 202
Bengali, 178
Bergman, Christopher, 120 Edwards, Ernest, 97
Berry, Cecily, 65 elision, 90
Bible, 26, 77, 132 empty words, 109
Bigelow, Charles, 33 English as a lingua franca, 94
brand names, 18 epiglottals, 84
Brazil, David, 104 Esperanto, 155, 201
Brown, Gillian, 50 Estuary English, 173, 178
Brown, Gordon, 67 eszett, 163
Bulgarian, 46 ezh, 150
Burgess, Anthony, 65
Burma, 64 fall-rise, 108, 171
Figueroa-Clark, Martha, 187
Cantonese, 135 fillers, 77
capital letters, 139 French, 13, 95, 98, 201
cardinal vowels, 83 fronting, 69
Caribbean English, 176
Cassidy, Fred, 165 gemination, 48
Cauldwell, Richard, 76 German, 13, 19, 75, 88, 114, 120, 122, 128, 188,
Chatterji, Suniti Kumar, 141 193, 201
Chinese, 36, 65, 88, 202 Gimson, A.C., 39, 73, 202
Chomsky, Noam, 12, 47, 147 glottal stop, 146, 177
Chung, Karen, 95 GOAT, 44–5, 85, 171
chunking, 103 Grand Turk, 176
clicks, 138, 190 Greek, 35, 39, 47, 201
cockney, 70, 178
compound stress, 75, 83 Halliday, Michael, 104, 202
compounds, 70 Hannisdal, Bente, 171
compression, 39–40 haplology, 51
consonant doubling, 151 happY, 52–3, 56
contrastive stress, 125 Hausa, 151
205
206 General index
Hawai’ian, 90 O’Connor, J.D., 104, 202

hesitation, 77 ogonek, 169
Hindi, 15 Omnish, 94
Hirst, Daniel, 135
homographs, 75 palatalization, 153
Hungarian, 30 Panglish, 154
Hymns, 56 Panjabi, 15
hyphen, 139 Papiamento, 190
pharyngeals, 84
Icelandic, 202 Phillips, Zara, 168
idioms, 120 phrasal verbs, 123
insists, 112 Polish, 84, 88, 98, 188, 194, 202
intonation, 31, 103 politeness, 107
intonation phrase (IP), 103, 107, 134 Portuguese, 154, 202
intrusive r, 38 postalveolars, 98
IPA Chart, 83, 201 prefixes, 56
Irish, 23 Prince Harry, 169
Irish English, 169 pronoun, 130
Italian, 194, 201 prosodic marks, 145
pseudo-elision, 72
Jamaican Creole, 178, 202
Japanese, 45, 66, 202 Quirk, Randolph, 39, 202
Jenkins, Jennifer, 154
Jersey, 28 R sandhi, 172
Johnson, Samuel, 156 reading rules, 85
Jones, Daniel, 44, 143 reduction, 50
respelling, 157
k-backing, 179 retraction, 142
Kera, 139 retroflex sounds, 142
Kerswill, Paul, 178 Romanian, 46
Kjellin, Olle, 84 Rösel, Petr, 70, 100
Korean, 193, 202 RP, 171
Kotercová, Zuzana, 161 rule of three, 126
Russian, 55, 66, 88
labiodental flap, 136
Latin, 10, 35, 47, 133, 150, 156, 161, 201 San Serriffe, 140
lenition, 169 schwa, 46, 54, 143
lexical stress, 62 Scots, 150
lexicographers, 156 Scottish Gaelic, 23, 150
LPD, 29, 159 Seidlhofer, Barbara, 155
Self, Will, 174
Murphy, Lynne, 74 Serbian, 88, 188
macron, 152 Shockey, Linda, 50
malapropisms, 24 sibilants, 159
Mandarin, 16, 84 siSwati, 17
Montserrat, 86, 170, 183–4 Slovak, 189
mora, 90 smoothing, 40
Southern British English, 68
Nadsat, 65 Spanish, 21, 45, 202
Nama, 138, 190 spelling, 161
neutralization, 53, 55 Spelling Society, 149, 165
New Zealand, 68, 101 St Martin, 189
Nguni, 17 St Thomas, 190
non-rhotic, 170 Stoller, Amy, 187
nt reduction, 64 stranding, 58
nucleus, 103–4 stress, 60
numerals, 33 stress shift, 32, 125
General index 207
STRUT, 60 Unicode, 104, 136, 139, 141, 150,

Swedish, 84, 120, 202 152–3, 163
syllabic consonants, 41, 180 Upton, Clive, 46
syllabification, 36, 143 Urdu, 15
syllable structure, 89
system, 106 velar softening, 147
Venda, 19
T voicing, 171
Taiwan, 95 Wales, 93
Teutonic Rule, 18, 60 Ward, Dennis, 152
TH fronting, 23 weakening, 52–3
THOUGHT, 85 Welsh, 5, 13–14, 16, 27, 184, 186–7,
tonality, 103, 107, 119 202
tone, 103–4 Wesley, Charles, 41
Tongan, 146 Williams Parry, R., 186
tonicity, 106–7, 121, 131 Windsor Lewis, Jack, 41, 50, 149,
transcription, 82 187
Trim, John, 201 word boundaries, 42
Tswana, 152
Turkish, 189 yod coalescence, 172
yogh, 150
Ulster, 101
unaspirated consonants, 88 Zulu, 138, 188, 190, 202

Sounds Interesting Observations On English and General Phonetics

Uploaded by

Copyright:

Available Formats

Sounds Interesting Observations On English and General Phonetics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sounds Interesting Observations On English and General Phonetics

Uploaded by

Copyright:

Available Formats

Sounds Interesting

How do you pronounce omega, tortoise and sloth? – and why? Do

j.c. wells is Emeritus Professor of Phonetics at University College

Illustrations by Lhinton Davidson

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of

This publication is in copyright. Subject to statutory exception

First published 2014

Printed in the United Kingdom by Clays, St Ives plc

Library of Congress Cataloguing in Publication data

ISBN 978-1-107-07470-5 Hardback

Cambridge University Press has no responsibility for the persistence or accuracy of

1 How do you say…? 1

2 English phonetics: theory and practice 38

2.33 an albatross on the balcony 68

3 Teaching and examining 82

4.6 fall-rise or fall plus rise? 108

5 Symbol shapes, fonts and spelling 136

5.24 Panglish, or unintelligibility? 154

6 English accents 167

7 Phonetics around the world 184

After my retirement in 2006 I started to write a phonetic blog discussing

I saw an American TV programme recently based on the theme of the

1.2 nucular snitchel

The bacterium Clostridium difficile has been in the news recently.

1.5 shih tzu

How do you pronounce plethora?

I pronounce it ˈpleθərə. That’s also what’s recommended by every dictionary

The area that an Anglican or Roman Catholic bishop controls is a

recognized by Dr Johnson and the other eighteenth-century lexicographers, and

1.9 West Indian islands

Thoughts on returning from abroad: if you fly with Virgin Atlantic

movement in Britain away from the anglicized ˌnɪkəˈrægjuə in the direction

1.10 place names

Driving to Gatwick Airport a few days ago to meet an arriving

Even English-derived names can be surprising. I remember driving through

Any visitor to Reading, on the way from London to Bristol, will be

A few years ago astronomers were discussing just what a planet is

1.15 proceed and precede

1.16 hypernymy and thrombolysis

One of our students chose hypernymy as a dissertation topic. How do

1.18 French, German or Welsh?

1.19 Loony what?

Whenever there’s an earthquake, the newsreaders make refer-

1.21 Punjab and feng shui

One word that I keep hearing (mis)pronounced on television is not

I’m sure we all know that Mbabane is the capital of Swaziland in

1.24 brand names

The modern lexicographer has resources unknown to earlier generations.

No, that isn’t a misprint for something else. Tokyo Sexwale is a

As with tortoise, so with porpoise. The pronunciation I recommend in LPD is

1.27 continental and English values

On BBC World television news I was struck by how one of the

1.28 my liege, we are besieged

which he feels sounds ‘more Sarf London’. Indeed, in non-rhotic accents

Campbell is a pretty common surname throughout the English-

1.32 ho, ho, ho

1.33 d’oh, doh… duh

1.34 ring in the old ear