English Phonetics, Phonology and

Spelling for the English Language

Teacher

This resource supports TESOL preservice and in-­service teachers and cur-
riculum designers in teaching pronunciation more effectively. Laurie Bauer
examines the patterns of pronunciation found in English, comments on
common errors made by learners, provides advice on what must be taught
and what can be allowed to pass, and offers commentary on which parts of
the curriculum are necessary for beginners and which are of value only to
advanced students. Part I introduces the phonetic background; Part II cov-
ers phonetics in more detail (consonants, vowels, prosody, phonotactics and
syllables); Part III covers phonology (sound changes influenced by adjacent
sounds, morphophonemics, stress rules and free variation); and Part IV cov-
ers spelling (English spelling, spelling consonants and vowels, and spelling
particularly difficult words). The helpful content can be tailored to one’s
teaching needs and will support an educator’s efforts to teach pronunciation
seriously, whether it is a matter of pronouncing particular vowels accurately
or knowing how to interpret the spelling system to get at the appropriate
pronunciation.

Laurie Bauer, FRSNZ, is Emeritus Professor of Linguistics at Victoria Uni-

versity of Wellington, New Zealand. He is the author of over twenty books
including being one of the authors of The Oxford Reference Guide to English
Morphology (2013), which won the Linguistic Society of America’s Leonard
Bloomfield Prize. In 2017, he was awarded the Royal Society of New
Zealand’s Humanities Medal.
ESL & Applied Linguistics Professional Series
Eli Hinkel, Series Editor

Doing Reflective Practice in English Language Teaching

120 Activities for Effective Classroom Management, Lesson Planning, and
Professional Development
Thomas S. C. Farrell

Creating Classrooms of Peace in English Language Teaching

Edited by Barbara M. Birch

Shaping Learners’ Pronunciation

Teaching the Connected Speech of North American English
James Dean Brown, Dustin Crowther

Handbook of Practical Second Language Teaching and Learning

Edited by Eli Hinkel

English L2 Vocabulary Learning and Teaching

Concepts, Principles, and Pedagogy
Lawrence J. Zwier, Frank Boers

Praxis-oriented Pedagogy for Novice L2 Teachers

Developing Teacher Reasoning
Karen E. Johnson, Deryn P. Verity and Sharon S. Childs

English Phonetics, Phonology and Spelling for the English Language Teacher
Laurie Bauer

For more information about this series, please visit: www.routledge.com/

ESL-Applied-Linguistics-Professional-Series/book-series/LEAESLALP
ENGLISH PHONETICS,
PHONOLOGY AND
SPELLING FOR THE
ENGLISH LANGUAGE
TEACHER

Laurie Bauer
Designed cover image: © Getty Images
First published 2024
by Routledge
605 Third Avenue, New York, NY 10158
and by Routledge
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2024 Taylor & Francis
The right of Laurie Bauer to be identified as author of this work has been
asserted in accordance with sections 77 and 78 of the Copyright, Designs and
Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in any
information storage or retrieval system, without permission in writing from the
publishers.
Trademark notice: Product or corporate names may be trademarks or registered
trademarks, and are used only for identification and explanation without intent
to infringe.
ISBN: 978-1-032-63701-3 (hbk)
ISBN: 978-1-032-60794-8 (pbk)
ISBN: 978-1-032-63702-0 (ebk)
DOI: 10.4324/9781032637020
Typeset in Times New Roman
by SPi Technologies India Pvt Ltd (Straive)
CONTENTS

List of illustrations vii

Preface viii

PART I
Introductory material 1

1 Introduction 3

2 The phonetic background 12

PART II
Mostly phonetics 25

3 Consonants 27

4 Vowels 43

5 Prosody 55

6 Phonotactics and syllables 63

7 Extra advice for teachers 71

vi Contents

PART III
Phonology 87

8 Sound changes influenced by adjacent sounds 89

9 Morphophonemics 103

10 Stress rules 122

11 Free variation 136

PART IV
Spelling 139

12 English spelling 141

13 Spelling consonants 145

14 Spelling vowels 150

15 Names and other difficult words 166

Reading and references 172

Index 174
LIST OF ILLUSTRATIONS

Figures
2.1 Places of consonant articulation 15
2.2 A vowel chart 18
6.1 Syllables as peaks of sonority 69

Tables
2.1 Places of articulation 14
2.2 Descriptions of the English consonants 16
2.3 Some other consonant sounds 16
2.4 Description of some English vowels 18
2.5 Some non-IPA transcriptions 21
2.6 Alternative IPA transcription systems for English vowels 22
3.1 English plosives 28
3.2 Uses of the glottal stop 30
3.3 English fricatives 31
3.4 English affricates 34
3.5 English nasals 35
4.1 The short vowels 44
4.2 The long monophthongs 47
4.3 The diphthongs 49
4.4 Unstressed vowels 51
6.1 Relative sonority of English sounds 64
6.2 Initial two-consonant clusters 65
6.3 Initial three-consonant clusters 66
6.4 Final two-consonant clusters (no affixes) 67
PREFACE

Every English teacher has to juggle multiple tasks in order to make their
students competent users of the language being taught. Pronunciation is just
one of those tasks, alongside grammar (morphology and syntax), vocabu-
lary (including collocational material and the connotations of important
words), idioms, pragmatics (how to apologise, how to greet someone, how to
make polite requests, and how to engage in small talk) and more general
skills, which may include guessing in context, reading a new writing system,
carrying on a conversation and things which do not even feel like ‘skills’ such
as improving self-­confidence. In every single one of these areas, there is more
to teach than there is room for in the curriculum, and the teacher (to some ex-
tent, the curriculum developer) must prioritise – prioritise within each of these
areas and prioritise these areas against each other.
In order to prioritise properly, teachers must know what they are pri-
oritising: not only do they need to know which bits of pronunciation (for
example) need to be prioritised against other bits, they need to be aware of
what prioritising one thing has the effect of downplaying, since giving one
thing precedence inevitably leaves something else in the background. Part of
this is a matter of what information is vital to making progress, but part of
it is a matter of what the students already know, by virtue of already being
speakers of (at least) one language. Some of that information can be gleaned
from books; some of it must derive from experience.
The aim of this book is to provide information on matters of English
pronunciation (including how the spelling reflects, or fails to reflect, the way
in which words are said) that can provide input to the process of prioritising.
This includes commenting on errors that some learners are likely to make
 Preface ix

and commenting on how to overcome some of these errors. The book

attempts to do this without assuming that the reader is already an expert in
phonetics. At the same time, the book attempts to cover matters of pronun-
ciation which will be relevant only at advanced levels, as well as those which
will be vital for novice learners of English. In the end, though, any book has
to rely on the competence of the teacher – competence to recognise good
pronunciation, competence to know when poor pronunciation is likely to
lead to misunderstanding or lack of comprehension, and competence to
pick out the messages that will help the students the most.
Teachers will need to read this book (or the parts of it they are interested
in) in the light of the experience of the students they are teaching – taking
account of the students’ first-­language (L1) and the level at which the stu-
dents are working. Not only will students who have Greek as an L1 have
different requirements from those who have Japanese as an L1, beginning
students will need a different focus in learning to deal with sounds and with
spellings than advanced students will.
Although this book is aimed primarily at those who want to teach Eng-
lish as an additional language, much of what is said here can also be of value
to those who want to learn English for themselves or for L1 speakers of
English who want improve their overt knowledge of the intricacies of Eng-
lish pronunciation and spelling. In short, if all you want is a book about how
to pronounce the sounds of English as a beginner, this book is not for you.
I should like to thank Olly Ballance, Sasha Calhoun, Tim Edwards, Paul
Nation and the anonymous referees for the publisher for their helpful feed-
back on an earlier version of this work. They should not be blamed for how
I have interpreted their comments.
Laurie Bauer
Wellington
PART I

Introductory material
1
INTRODUCTION

Let us begin with what this book is not. It is not a book to teach you about
phonetics. There are many such books, aimed at people with different levels
of background knowledge or aimed at people with or without a teacher to
guide them. This book provides a recap of the fundamentals of phonetics as
they apply to the concerns here but does not dive into the mechanics of ini-
tiator systems or acoustic measures of vowel quality.
Neither is this a book about phonological theory. It is informed by vari-
ous approaches to phonological theory but does not deal with rule notation,
feature systems, articulatory gestures, or the multiple phonological processes
(e.g. apocope, lenition) which are so crucial for the understanding of lan-
guage change. Rather, where it deals with phonology, it is concerned with
describing the facts of English which a phonologist might wish to explain or
which might provide data for the construction of phonological theories.
It is not a book about spelling systems. It does not look at alternative
approaches to writing languages down or ways in which an alphabetic sys-
tem can be deployed and can become modified over time. The approach to
spelling taken here is based on the current system of writing English, with
only brief excursions in how this spelling system came to be, where such
information might be useful.
Finally, it is not a course book for learners: it does not provide exercises
but it does provide information that might be needed to create good exer-
cises. Since teachers of English are operating in such a wide range of situa-
tions (in terms of the nature and level of their learners, in terms of the
amount of equipment they have available, in terms of the political and
social environment), it would be impossible to provide suitable exercises for

DOI: 10.4324/9781032637020-2
4 Introductory material

all – and more difficult at more advanced levels. While there are many sug-
gestions to help teachers, the teachers themselves need to consider how to
make exercises that will work for them and their students.
In fact, it might be argued that this is not one book at all but three short
ones, linked by their focus on the sounds of English and aimed at the needs
of teachers of English, particularly at the needs of teachers who are dealing
with English as an additional language, but including teachers who are
teaching first-language (L1) users of English. Included in these teachers are
those who are writing textbooks for students of English or those who are
writing curricula for English language teaching. Matters such as the priori-
tising of various skills that students need to achieve are regularly a matter of
focus, for example. This principle is taken to the point where abstruse facts
about English pronunciation are simply ignored, not because they are not of
potential interest but because they are unlikely (in the author’s opinion) to
be of value in the classroom. This is inevitable. Even if the author knew
everything (which nobody does), if every detail were included, the important
points would get lost in the trivia, and the resultant work would be many
times longer than this book.
Any work that deals with the pronunciation of English has to make
choices. Most importantly, a choice has to be made as to the kind of English
that is to be described. Here, the variety of English that is chosen is Standard
Southern British English (SSBE). This is not a matter of perceived value, it
is a matter of practicality. The author speaks SSBE, albeit as an acquired
kind of English, and there are many works on English pronunciation which
use this as their basis. This kind of English is sometimes called the King’s
English, sometimes called RP (which stands for ‘received pronunciation’, the
pronunciation found acceptable in polite society) or Oxford English. All of
these labels are, to some extent, misleading, but all are used to label the kind
of English that is a national standard (i.e. not just one restricted to a particu-
lar area) within England and the wider United Kingdom. This kind of Eng-
lish, like all kinds of all languages, is also changing, and young speakers
sound different from older speakers. Socially, this kind of pronunciation can
have benefits and disadvantages. Learners who use this variety will fit in with
those whose families have professions in the law, the church, business and
academia, but they may be thought stand-offish by people who work in fac-
tories, in shops, in the transport industries, on the land, and so on. One of
the great benefits of this kind of English is that it is understood everywhere
in Britain, even if it is actually spoken by only a small minority of people.
From the point of view of the consumer of this book, however, this choice
may create problems. Many students, wherever they live, are exposed to huge
amounts of American English (of various kinds) in music, in movies, in the
broadcast media, even in print. Standard American English sounds very
 Introduction 5

different from SSBE, and not everything that is said of one will be true of
the other. This becomes particularly important in two very wide areas: the
vowels and the use of the R-sound. Speakers of SSBE do not pronounce an
R-sound in words like far and farm, speakers of American English do, and
this has an important effect on the systems involved in the pronunciation.
Many American speakers find this loss of R incomprehensible in places. It
also means that American English has fewer vowels that SSBE has and that
they are distributed in a slightly different way as a result. Points involved
with the R-sound will be dealt with here, but only sporadic comments will be
made on differences between the vowels of American English and SSBE.
The good news is that most of the rest is transferable, and occasional inci-
dental differences (such as the different stressed vowels in the word tomato)
are commented on. Occasional comments are also made on other kinds of
English: Australian English, Canadian English, Irish English, New Zealand
English, Scottish English, and South African English. Such comments,
though, are sporadic, not systematic. These other varieties also differ from
SSBE mostly in the vowel sounds.
While the discussion is on kinds of English that can act as models for
learners, the question of whether some artificial norm could be used should
be discussed. The idea of an artificial norm – meaning some variety of Eng-
lish which is not natively spoken by anyone but is made up as a compromise
between comprehensibility and ease of production – arises particularly when
it is pointed out that, at least in business encounters round the world, Eng-
lish is used as much between speakers neither/none of whom has English as
an L1 as it is in encounters where one or (even rarer!) both or all speakers
have English as an L1. Since, it is argued (and it is hard to disagree), English
is a language which is phonetically and phonologically difficult to acquire,
having features which are unusual cross-linguistically, it must be possible to
construct a type of English which is easier to pronounce but which remains
comprehensible. If we consider what happens in any language classroom
anywhere in the world, we find that students who come from the same lan-
guage background create a version of English (or whatever their target lan-
guage is) which they clearly find easier than attempting to get really close to
a native version, so that this general principle seems to have merit; after all,
these people remain comprehensible within the confines of their classroom,
and that includes being comprehensible to the teacher. We can deal with the
position of the teacher first. The teachers, in effect, have to train themselves
to ignore certain features of the pronunciation they hear in the classroom in
order to make progress with other aspects of language teaching. This is eas-
ier if they are members of the same language community as the learners –
they already know what to expect. But if they are not, they become
accustomed to the classroom variety and become expert at interpreting it.
6 Introductory material

In the process, they inevitably become inured to a certain amount of devia-

tion from the native model as long as it is in a particular direction. The stu-
dents come to agree on their classroom version because it is based on
pronunciations they are already familiar with. English L1 learners from
almost anywhere, and whatever language they are trying to learn, will replace
a final [e]-type of sound with [eɪ] and a final [ɒ]-type of sound with [əʊ].
Teachers of French, Italian or Spanish to English-language speakers will be
familiar with the phenomenon; they may not consider it particularly prob-
lematic. This pronunciation change is due not to something natural and
inevitable but to the students’ experience of English, where words do not
finish with [e] or [ɒ]. Students do not notice what they are doing unless it is
pointed out to them.
The important thing to notice about such a local version of English,
though, is precisely that it is local. The version of English produced by
French students in Paris will be different from that used by French stu-
dents in Marseilles; the version produced in Rome will be different from
that produced in Helsinki or Tokyo or Hanoi. In fact, these individual
local versions of English may be so different from each other that learners
from one of these cities may be unable to understand learners from another
who are speaking their own local version. Having one of the standard ver-
sions of English (usually SSBE or a standard North American version, but
Australian is being seen in something of the same light) keeps teachers
honest. It gives them a yardstick and keeps learners from all round the
world, all of whom have learnt English because they want to speak what
they see as a World language, from drifting so far apart that the benefits of
a World language are completely lost. If this seems to be an exaggeration,
consider what has already happened for L1 speakers of English, where a
speaker from Gateshead, England may not understand a person from Aus-
tin, Texas and neither may understand a speaker from Aberdeen, Scotland
or Aberdeen, Hong Kong. The reason why English remains a World lan-
guage is that the educated are taught to understand a version of English
which we might call Standard International English, which, although it
shows some variation (especially in pronunciation), allows for very limited
variation away from the standard. What all this means is that there is an
argument from logic for sticking as closely to possible to a standard ver-
sion of English.
There is also an argument from politeness and respect. If you want to
learn a language like English in order to communicate with other people
(some of whom may be L1 speakers of that language), then it is only polite
to do it as well as possible, so as not to put your listeners to too much trouble
when they try to interpret what you say. In general terms, listeners will allow
foreign learners a large amount of leeway and do their best to listen
 Introduction 7

cooperatively. But as a learner, you should be making that job as easy for
them as possible. By going beyond your own comfort zone of what you can
already pronounce, you are being polite to your new listeners and also show-
ing respect to them and to their language. You should be trying to make it
easier for them to speak to you in their language than it would be for you to
listen to them attempting to speak yours. And you should think about how
great an insult to you and your language you would find it if foreign learners
(as is, no doubt, often the case) made no particular effort to get your lan-
guage right.
And if even these weighty arguments do not convince you or are not suf-
ficient to make you change your practice and try to do better, think of the
argument from good will. If you are a learner, you will no doubt make mis-
takes in your grammar and word choice as well as in your pronunciation.
But your pronunciation is what your listeners will judge first. A good pro-
nunciation will create immediate good will and make you audience more
likely to be forgiving of the other mistakes you will make and more likely to
feel positively towards you.
What has been said here might sound as though it is directed in particular
at pronouncing the vowels and consonants of the target language in a par-
ticular way. This is not entirely wrong, but neither is it completely right.
Speakers do not have to get absolutely every aspect of their pronunciation in
line with the values provided by L1 speakers from the same area in order to
speak perfectly acceptable and comprehensible English. Part of the teacher’s
job is to make sure the focus is on the most important of these areas. More-
over, some problems of pronunciation arise because learners fail to apply
appropriate phonological processes – either not using normal English pro-
cesses or, for L2 learners, applying phonological processes from another lan-
guage, even though they do not work in English. Similarly, many
pronunciation problems (even for L1 speakers) arise because the spelling
system is misinterpreted. Knowing what the spelling system tells us (and
what it does not) are important skills in helping avoid such problems.
This book looks at matters of pronunciation from these three perspec-
tives: phonetic, phonological and orthographic. To a certain degree, this
means covering the same material multiple times, but since phonetic aspects
are usually dealt with earlier in the teaching process than phonological ones,
the separation also has pedagogical justification. In some instances, it is not
very clear whether a particular topic falls under phonetics or phonology.
Such matters are treated earlier rather than later.
Since the difference between American English and SSBE has been noted,
a useful bit of terminology is the notion of a variety. A variety of English
can be a national variety (like American versus British versus Australian), it
can be a regional dialect within one of these national varieties (such as the
8 Introductory material

English of Birmingham, England or Birmingham, Alabama), it can be based

on social factors (such as gender or social class or formality), it can be based
on time (such as the English of the 1950s versus the English of the 2020s),
and it can be based on medium (written versus spoken English). SSBE is just
one variety among many. It is localised to Britain but is not regional within
Britain, it is a formal variety but a spoken variety, it develops over time, so
that the SSBE of 1950 is different from the variety heard today. There are
social differences within SSBE, so that the English of King Charles is differ-
ent from the English of most of the professors at the University of Oxford
(even if we exclude those who speak not SSBE but another national or
regional variety).
We also need here to look at some matters of notation and the way the
book is set out. The book is divided into four parts, but the chapters are
numbered consecutively across all four. Tables and figures are numbered
with the chapter number first and then the running number of the table or
figure within the chapter. Things that are presented as rules are numbered
inside the relevant chapter with an identifying one- or two-letter identifier.
For instance, (V2) refers to the second rule dealing with vowels. Examples
are numbered consecutively within in each chapter with an X (for eXample)
preceding the number. Thus, (X10) refers to example 10 in a particular chap-
ter (the current chapter unless otherwise stated). In general terms, rules are
processes which may be obligatory or may be optional but which are expected
to be able to apply. Examples provide words which illustrate a particular
phenomenon.
In the text, words in italics are examples, as in Cat is a three-letter word.
Transcriptions are enclosed in square brackets: [kat]. Spellings, when it is
important to distinguish them from transcriptions, are enclosed in angle
brackets: <cat>. The symbols used in the transcriptions are given in various
places in this book, including in Chapter 2. Standard symbols from the
International Phonetic Alphabet (IPA) are used, but the transcription sys-
tem used here is not one from any particular source. Note that these IPA
symbols include [.] as a way to divide syllables. Words in small capitals are
the names of lexical sets (see Section 2.2), as in kit. Single quotation marks
enclose glosses or meanings, as in cat ‘feline quadruped’, although they are
also used for scare quotes and for marking quotations. Very few references
to other works are provided in the text, but there is a list of recommended
reading at the end of the book, and many examples have been found in such
works. Technical terms of phonetics and phonology have sometimes been
avoided for ease or presentation to a non-technical audience, but some of
the basic terminology is retained, and much of that is explained in Chapter
2 or where it is needed. The asterisk is used, as is common in linguistics, to
indicate that what follows is not English. We might thus have *cqt.
 Introduction 9

Part II deals, as the title says, mostly with phonetics (i.e. with the details
of pronunciation of English). Comparisons are made with other languages
and hints are given to aid the teacher and learner achieve an appropriate
standard of pronunciation. The section is called ‘Mostly phonetics’ because
some of the material on syllables and phonotactics could be seen as being
phonological.
Part III deals with phonology, viewed as dealing with those patterns
which are not learnt by knowing how to pronounce individual words. That
is, this section is not concerned with how to pronounce the [θ] sound at the
beginning of the word thing, or how many syllables there are in spasm, but
will be concerned with generalisations of the ways sounds operate over word
boundaries and across large numbers of different words. Most of this is not
necessary knowledge for early learners of English: in the first instance, learn-
ing how to pronounce individual sounds and individual words is challenge
enough. At later stages of learning, though, being able to see the patterns
can be helpful in allowing the student to understand how the language func-
tions (and thus to be able to improve their own pronunciation) as well as to
understand individual utterances produced by others and to see links
between vocabulary items.
Part IV deals with English spelling and particularly with the way the
spelling has to be interpreted if the words are to be spoken. That is, it goes
largely from spelling to sound. Some of the sound-to-spelling rules can be
deduced from what is said, but they are less predictable.
Although this book is about pronunciation (in various aspects), the top-
ics that are covered intersect on a number of occasions with word structure
(or morphology). We do not need much morphological terminology, but
some is unavoidable. The term ‘word’ is one such fundamental piece of jar-
gon. For our purposes here, a word can be considered an element specified
by the spelling system and marked off with spaces and punctuation marks.
Within the words, we find prefixes and suffixes and what those prefixes and
suffixes are added to, here called a ‘base’. A base may be made up of an
indivisible element, or it may contain several meaningful elements. Not only
the prefixes and the suffixes but also these meaningful elements inside bases
are termed ‘morphemes’. In the word disjointedly, there is a prefix (dis-) and
two suffixes (-ed and -ly). The base for the prefix is joint (with a single mor-
pheme, so we can call it monomorphemic), the base for the addition of -ed
is disjoint (with two morphemes) and the base for the -ly suffix is disjointed.
Affixes which create new words are said to be derivational affixes, and the
process of forming new words by them is called derivation; affixes which cre-
ate new forms of the base word, like the -s on cats or the -ed on walked, are
called inflectional affixes, and the new form is created by a process of
inflection.
10 Introductory material

Teachers might assume that students have an interest in learning to pro-

nounce well the language they are learning. This is not necessarily the case.
For some students, though they may not know it, the way they speak –
even the language they speak – may be an important part of their own vision
of themselves, a part of their identity. If this is the case, speaking another
language is a direct challenge to their perceived identity.
Since anyone’s pronunciation gives clues as to their regional origins, their
social background, their age, their gender and their personality, learning a
new pronunciation which denies their own truth about themselves can be
deeply disturbing. They may want to retain pronunciations which disclose
their language background or their regional origins within their L1. They
may reject a pronunciation which they think makes it seem as though they
come from a different social background, which falsifies their age or sexual
identity or which implies that their personality is other than what they think
they wish to project. They may wish to retain aspects of their L1 accent
which speakers of their new L2 perceive as making them sound sexually
attractive or macho.
At the same time, there will be students who see a chance in learning a
new pronunciation to change their image, perhaps projecting a more confi-
dent or socially superior version of themselves. Any of these things can
cause problems for the language teacher, and the only way to challenge such
views is to present the learners with a new in-group to which they can aspire
to belong. Such a procedure is extremely difficult, even in the country in
which the L2 is spoken. This is a language-teaching problem which is rarely
discussed, but teachers should at least be aware of it as a problem.
The other elephant in the room is that it is extremely difficult for teachers
to get students to perform better than the teachers do themselves (this
applies to grammar and vocabulary as well as to pronunciation, of course).
It is not impossible, however, if students are presented with models other
than the teacher and, where pronunciation is involved, are taught what to
listen for and how to change their habits of speaking. The old language labo-
ratory model often did not work, because students could not hear the differ-
ences between what the model was saying and what they were saying – and
the same is true with a teacher instead of a recording. Overt teaching is not
a panacea, but it is a necessary part of pronunciation teaching.

Highlights
English encompasses a huge amount of variation, and teachers need to
choose a model of pronunciation for their students. Because the situation of
teachers is so different, a single solution cannot be given for all, but the
model used in this book is SSBE, and references are frequently made to
 Introduction 11

other varieties. Teachers need to develop overt strategies to deal with teach-
ing pronunciations that the students quite literally cannot hear.

What should the teacher do?

Teachers must have a model in mind for their students. It need not be the
model described here, but if it is not, teachers need to be able to translate
between the model they are using and the model described here (or in other
textbooks) and reference works (e.g. dictionaries, pronouncing dictionaries).
Some hints for creating a model are given later in the book. Even in places
where English is used in the community, there is some value in choosing a
model which is closer to an international standard than the local usage is.
Teachers who are not used to reading a transcription must practice this skill,
to help to distinguish between spelling and sound. Overt strategies for teach-
ing pronunciation will be provided in later chapters.
2
THE PHONETIC BACKGROUND

The point of this chapter is to provide sufficient background to understand

the technical parts of the book. The idea is not to provide an introduction to
phonetics (as has already been stated, this is not a phonetics textbook) but
to provide a summary of the phonetic classification used here and intro-
duce the technical terminology that will be used throughout the book.
Readers who have no background in phonetics might prefer to read a text-
book introduction to such matters; people with some background in pho-
netics can view this is a rapid overview for purposes of revision; people with
a solid background in phonetics can probably skip this chapter entirely. It
is worth saying that the focus here is the phonetics of English, sounds of
other languages that might need to be contrasted with English sounds are
only briefly acknowledged. Click sounds (like the sound that is usually writ-
ten tut tut in English or the sound made to encourage horses), for instance,
are not discussed.

2.1 Describing consonants
When the airflow from the lungs gets to the larynx (or the voice box, seen
externally as the Adam’s apple), it encounters the first of several possible
obstructions, the larynx, and more specifically the vocal folds (sometimes
called the vocal cords – note the spelling), which are bands of muscle which
can be drawn across the opening to the trachea or windpipe within the lar-
ynx to various degrees. If the vocal folds are as open as possible so as to let
air through easily, we talk about the sound that is produced as being

DOI: 10.4324/9781032637020-3
 The phonetic background 13

voiceless. If the vocal folds narrow the opening without blocking it, we get a
whisper, though this is not linguistically important for our study of English.
If the vocal folds are brought together, but loosely enough for the force of
the air to blow through them, the vocal folds are said to vibrate, in the same
way as our lips vibrate if we try to imitate a horse or the sound of a motor
car engine. This vibration produces a tone, known as voice, and the sound
that is produced is called a voiced sound. If you rest your fingertips lightly
on the outside of the larynx, or if you block up your ears and say a long ah,
you will be able to feel the vibration, the voice. If, instead, you produce a
long sigh, you will find that the vibration is lacking and that a sigh is voice-
less. Similarly, if you make a hissing sound, a long S-sound, you will find it
is voiceless, while a long Z-sound, a buzz, is voiced. Although the distinc-
tions here have been described as having three steps, there are degrees of
voicing and voicelessness, some of which are linguistically relevant in lan-
guages other than English.
When the air that flows out of the lungs is blocked or restricted in some
way, consonant sounds are produced (if the airflow is not restricted, we get
vowel sounds). The consonant sounds are classified according to the degree
of obstruction in the vocal tract – that is, the mouth, the nose and the phar-
ynx, as far down as the larynx.
If the vocal tract is completely blocked at some point, so that the airflow
is interrupted, we talk about a plosive. The words pear, bear, tear, dare, care
and garish in English all begin with plosives. Some speakers of English,
though not usually speakers of Standard Southern British English (SSBE),
have a plosive called a glottal stop in a word like butter, where the T-sound
is missing: in ordinary spelling, we often see this written as bu’er.
If the air stream is not fully blocked but has to squeeze through such a
narrow gap that it creates audible friction, we talk about a fricative conso-
nant. The words fine, vine, sign and shine in English all begin with fricatives,
and the words sooth, soothe and sues all end with one. The sound [h] at the
beginning of high, home and who is also a fricative.
If a sound starts with a plosive bit and then has a fricative bit in the same
sound, we talk about an affricate. The words cheap and jeep both start with
affricates in English.
Plosives, fricative and affricates together form a class called obstruents.
Consonants that are not obstruents are sonorants. Some obstruents are
voiceless, some are voiced. All sonorants are inherently voiced.
If the airflow is blocked in the mouth but still flows through the nose, we
talk about a nasal sound. The English words ram, ran and rang all end with
a nasal sound. Nasals are sometimes called, more fully, nasal stops because
the air in the mouth is stopped, as it is for plosives, plosives being another
kind of stop consonant.
14 Introductory material

Non-nasal sonorants are called approximants. In the lateral consonant,

[l], the air passes over one or both sides of the tongue, as you can feel if you
try saying a long [l] on in-drawn breath. There are various kinds of R-sound,
in English, but they are all count as the same sound. The other two, [j] and
[w], can be found at the start of words like yell and well.
So far, we have discussed two features of consonants which can distin-
guish them: voicing and manner of articulation. The third and last feature
necessary for a description is where the sound is produced. The places where
the sounds are produced are named after the place on the upper jaw towards
which some organ attached to the lower jaw moves. Where plosives are con-
cerned, you can feel the point of contact; where fricatives are concerned, you
can feel the point of near contact if you breathe in while trying to say the
sound. The place of articulation for the approximants is harder to feel. The
places of articulation and some exemplary consonants are given in Table 2.1
and in Figure 2.1.
If we put together the three features, we can define the major consonant
sounds that we will be interested in for English, as is shown in Table 2.2.
Occasionally, we will need to make reference to other sounds, either as
types that occur within English or as sounds of foreign languages. These are
listed in Table 2.3.

TABLE 2.1 Places of articulation

bilabial [b] in butter; [m] in The lower lip approaches the upper
mare lip.
labiodental [f] in fair The lower lip approaches the upper
incisors.
dental [θ] in thing The tip of the tongue approaches the
upper incisors.
alveolar [d] in day; [s] in say The very front of the tongue
approaches the alveolum or teeth
ridge, just behind the teeth.
post-alveolar [ʃ] in shoe; [ɹ] in roe The front part of the tongue
approaches the area behind the
alveolum.
palatal [j] in year The body of the tongue approaches
the hard palate.
velar [ɡ] in goat; [ŋ] at the The body of the tongue approaches
end of sing the velum or soft palate.
glottal [h] in hew The vocal folds approach each other.
labial-velar [w] in way The lips round and the body of the
tongue approaches the velum.
 The phonetic background 15

Nasal Cavity
Alveolar Ridge or
Teeth Ridge

Velum

Top Lip

Top Incisors
Hard Palate

Botom Lip

Uvula
Tongue

Position of Larynx, Epiglottis

Containing Vocal Folds

FIGURE 2.1 Places of consonant articulation.

2.2 Describing vowels
Just as we need three pieces of information to describe a consonant, we need
three to describe a vowel, but a different three. There are two difficulties with
describing vowels: there are no clear places of articulation, and the three
features are not absolutes but matters of degree. This also means that it is
hard to be precise about the qualities of vowels and thus to give precise in-
structions on how to pronounce the vowels of a given language.
The first piece of information we need is how close to the roof of the
mouth the tongue is when the vowel is pronounced. This is typically reduced
to three or four steps: close, where the dome of the tongue is very close to the
roof of the mouth; mid, where the dome of the tongue is in an intermediate
position (when four steps are used, a distinction is made been close-mid and
open-mid); and open, when the dome of the tongue is as far away from the
roof of the mouth as possible. Since the absolute distance is not observable
without instruments, in practical terms this can be reduced to how open the
16 Introductory material

TABLE 2.2 Descriptions of the English consonants

Symbol Voicing Place Manner Example

p voiceless bilabial plosive pea

b voiced bilabial plosive bee
m voiced bilabial nasal me
f voiceless labiodental fricative fee
v voiced labiodental fricative vie
θ voiceless dental fricative thigh
ð voiced dental fricative thy
t voiceless alveolar plosive tie
d voiced alveolar plosive die
n voiced alveolar nasal knee
s voiceless alveolar fricative sue
z voiced alveolar fricative zoo
l voiced alveolar lateral lie
ʧ voiceless post-alveolar affricate chew
ʤ voiced post-alveolar affricate Jew
ʃ voiceless post-alveolar fricative shoe
ʒ voiced post-alveolar fricative leisure
ɹ voiced post-alveolar approximant rye
j voiced palatal approximant you
k voiceless velar plosive key
ɡ voiced velar plosive ghee
w voiced labial-velar approximant woo
h voiceless glottal fricative who

TABLE 2.3 Some other consonant sounds

Symbol Voicing Place Manner Example

β voiced bilabial fricative Spanish abuelo

ç voiceless palatal fricative German Sicht, Scottish
Brechin
dʑ voiced alveolo- affricate Japanese oji ‘uncle’
palatal
ɦ voiced glottal fricative English behave
ɱ voiced labio-dental nasal English emphasis
ɲ voiced palatal nasal French agneau, Italian
gnocchi, Spanish mañana
ɾ voiced alveolar tap English three; Spanish pero
tɕ voiceless alveolo- affricate Japanese chichi ‘father’
palatal
ʋ voiced labio-dental approximant Dutch week ‘week’
x voiceless velar fricative German Bach, Spanish jamon
χ voiceless uvular fricative French théâtre
 The phonetic background 17

mouth is: quite wide open for open vowels, with the lower teeth near the up-
per teeth for a close vowel. Close vowels are sometimes called high vowels,
and open vowels are sometimes called low vowels. An example of a close
vowel is the vowel [iː] in heed, an example of a mid vowel is the vowel [ɜː] in
bird, and an example of an open vowel is [ɑː] in father. The different degrees
of opening should be visible in a mirror if these vowels are pronounced in
sequence.
The second piece of information we need is whether the vowel is pro-
nounced with the dome of the tongue well forward in the mouth (more or
less opposite the palate where a consonant like [j] is pronounced), well back
in the mouth (more or less opposite the velum where a consonant like [k] is
pronounced), or somewhere between these extremes. A vowel like the [e] in
bed is a front vowel, a vowel like [ɔː] in thought is a back vowel, and a vowel
like [ɜː] in bird is a central vowel. To get a feel for the front–back dimension,
put a lollipop on the tongue and then pronounce these three vowels. The lol-
lipop should recede into the mouth as the vowel moves from [e] to [ɜː] to [ɔː].
The third piece of information that is required is the lip position. This is
simplified as rounded versus unrounded (or spread), although rounding af-
fects close vowels more than it affects open vowels. The vowel in thought in
English is a rounded vowel, and the vowel in heed is an unrounded (or
spread) vowel. This difference in lip position should be easily observable in a
mirror. Speakers of languages such as Danish, Finnish, French, German
and Mandarin will probably find that the difference between rounded and
unrounded vowels is very clear in their own languages. In English, the differ-
ence is usually far less clear-cut than in these other languages, with values for
rounding being neither tightly rounded nor clearly spread.
Other dimensions for describing vowels are sometimes needed for de-
scribing individual languages. The distinction between long and short vow-
els in English will be discussed at the start of Chapter 4.
With this set of labels, the stressed monophthongs in English can be de-
scribed as in Table 2.4.
The lexical set listed in the right-most column of Table 2.4 is a handy
reference word-list to refer to a group of words which contain the same
vowel sound. The kit lexical set contains not only the word kit but also a
great many other words such as bin, chip, film, gym, inch, limb, milk, ring,
snitch and trip. The names for the lexical sets are designed to be as unam-
biguous as possible. To avoid misunderstanding, we can refer, in a short-
hand, to the kit vowel rather than the vowel in bit (which might be confused
with bet, for instance).
Vowels can be divided into monophthongs and diphthongs. Monoph-
thongs sound just the same all the way through (apart from minor transi-
tions to adjacent sounds). In a diphthong, the tongue or the lips (or both)
18 Introductory material

TABLE 2.4 Description of some English vowels

Symbol Front/Back Vowel height Rounding Example/Lexical Set

ɪ front close unrounded kit

e front mid unrounded dress
a front open unrounded trap
ʌ central open unrounded strut
ɒ back open rounded lot
ʊ back close rounded foot
iː front close unrounded fleece
ɜː central mid unrounded nurse
ɑː back open unrounded start
ɔː back mid rounded thought
uː backa close rounded goose

Note:
a
Although this is a back vowel in conservative usage and in some varieties of English, it is
more often central in modern SSBE usage.

must move in the course of the vowel, which must be in a single syllable. In
SSBE, vowels such as the [ɪ] vowel in kit and the [ɑː] vowel in staff are
monophthongs, whereas the [əʊ] vowel in goat and the [aɪ] vowel in sight are
diphthongs. The transcription of diphthongs shows the approximate start-
ing point and the approximate end point of the diphthong.
Vowel sounds are often described in terms of a vowel chart, such as the
one in Figure 2.2. The chart is supposed to indicate the space within the

Front Back
Close

Close-Mid

Open-Mid

Open

FIGURE 2.2 A vowel chart.

 The phonetic background 19

mouth within which the dome of the tongue can move to produce a vowel
sound. If the high point of the dome of the tongue moves outside this area,
a consonant sound is produced (or the articulation is physiologically impos-
sible). The vowel chart is a simplified and idealised view of the relationship
between articulation and vowel quality, but it has been shown to be a practi-
cal tool for learners over the last century. These charts are not used in this
book, for two reasons, both practical. The first is that they are wasteful of
space and difficult to produce accurately. The second is that they give a false
impression of accuracy, to a degree that cannot be exploited in this book,
by providing a relatively small area within the vowel chart to position the
vowel. We know that, when multiple speakers are considered, the amount
of variation in vowel production is larger than this suggests. Many phoneti-
cians now prefer to use acoustical measures of vowel quality, averaged over
many speakers. Such figures are also more specific than is needed (or realis-
tic) for the goals here. However, sources that present vowel charts can be
used to compare vowels in terms of their height or frontness and backness,
which can be valuable for some students. The vowel chart does not show lip
rounding.

2.3 Phonemes
Natural speech is a stream of continuously changing sound. We rarely pause
in the middle of a word to concentrate on a particular sound, and we rarely
pronounce individual sounds in isolation unless we have a word made up of
just one sound (e.g. awe [ɔː]); our lips, tongue and other organs of speech
just glide seamlessly through a number of configurations in a way that is
then perceived as being meaningful. However, this ever-changing stream is
not an entity that lends itself easily to teaching. Instead, we teach language
as if it were made up of a succession of steady states, linked together in or-
der to make them easier to say. For instance, instead of viewing the word rest
as being made up a continuous flow from a configuration where the tongue
tip is raised slightly behind the teeth and then moves downward, the tongue
becoming grooved and then losing that groove at the end of the word, and
the tongue moving up to form a complete closure with the alveolar ridge,
while the front of the tongue takes up a position approaching the hard pal-
ate, we see the word as arising through the concatenation of four steady
states, which we can think of as [ɹ], [e], [s] and [t]. Although this is a fiction
(there is no steady state), it is a useful fiction, and it is a historically well-es-
tablished fiction, supported by alphabetic writing systems.
These fictional steady states are called ‘phonemes’. Phonemes can be
thought of as ideal versions of the sounds that we think we are making when
we speak (or in the case of L2 learners, the sounds that competent speakers
20 Introductory material

of the language that they are learning think they are making when they
speak). When we listen to a language in which we are competent, we hear
that language in terms of strings of phonemes and tend not to hear places
where the ideal version is not pronounced. Consider the phoneme which we
here transcribe with [ɹ]. In SSBE, this is pronounced in most places as [ɹ],
with the tip of the tongue turned up slightly behind the front teeth (in
American English, there may be a different norm). In the word through,
though, we may not say [ɹ], but replace it with [ɾ] (i.e. pronounced like a very
fast [d] sound and technically known as a tap). And in the word tree, what
we actually say may be pronounced with a lot more noise and no or little
vibration of the vocal folds, so that it is whispered. We tend to know that we
have pronounced the appropriate phoneme but not to be aware of these in-
stances where what we actually produce is rather different from the ideal [ɹ]
sound that is the default or the usual one.
These non-ideal variants, like the voiceless [ɹ] or the tap [ɾ], are called
bound variants of the phoneme (sometimes termed allophones, but the label
is less explanatory). Most textbooks distinguish between a phoneme and a
bound variant typographically. They enclose the phoneme within obliques
(e.g. /ɹ/) and the bound variants with square brackets (e.g. [ɾ]). This some-
times causes problems when trying to transcribe more than one sound at a
time, such as when transcribing words or utterances: what is the difference
between a default [ɹ] and an [ɹ] which is conditioned to be in that particular
place? How do we write the conditioned variant of a sound for which there
is no variation? Although we will keep this notation in reserve if it is useful,
mostly we will enclose all our speech sounds (actual or idealised) in square
brackets. However, to contrast with the sounds of speech, the letters of the
written language will be shown by the use of angle brackets, so that we can
say <c> is pronounced [s] before <e> (the letter C is pronounced with the
sound [s] before the letter E). Written words used as examples in the text will
be written in italics, so that we can say write is pronounced with initial [ɹ].
In this book, the technical terms phoneme and allophone will not be
used. Experience shows that learners often find them confusing, despite their
precision. Rather, talk will be of distinctive speech sounds and of bound
variants tied to a particular context. Distinctive sounds are those which can
be used to distinguish words, as can be illustrated by the use of minimal
pairs (pairs of words that differ in just one distinctive sound) such as cat and
rat, cat and cut, and cat and cap. Because we get a different word by substi-
tuting one sound for another, it proves the distinctiveness of the sound.
Minimal pairs will be used to illustrate distinctiveness and may form the
basis of useful exercises for learners. The names of the lexical sets intro-
duced in Table 2.4 can be thought of as unambiguous labels for the distinc-
tive vowel sounds.
 The phonetic background 21

2.4 Transcription
In general terms, the point of a phonetic transcription is to provide one
symbol for each distinctive sound. This ideal may be broken when writing
bound variants of those sounds, as described in the previous section, but it
avoids issues such as having one distinctive sound spelt in multiple ways, as
with the sound [iː] in see, sea, receive, thief and so on. In this book, the sym-
bols of the International Phonetic Alphabet (IPA) are used, but that does
not guarantee that all transcriptions of SSBE will always look the same.
Usually, the transcription of consonant sounds will be consistent across
transcriptions, but vowels are a different matter. A number of dictionaries
and other reference works do not use the IPA, and for some consonants,
there are non-IPA alternatives, which may be familiar to teachers or stu-
dents. The obvious ones are listed in Table 2.5.
The symbols used to transcribe vowel sounds when the IPA symbols are
not used are so disparate that listing them is unlikely to be helpful. However,
what may be confusing are the different transcription systems that use the
symbols of the IPA in rather different ways. Although the transcriptions dif-
fer, they are all representations of the pronunciation of the same variety of
English but give more or less importance to different features: some use
fewer symbols, some use more precise symbols (but more of them), some
distinguish between vowel qualities, other distinguish vowel lengths, and yet
others do both; some stick as closely as possible to ordinary Roman alpha-
bet symbols, and others feel free to use more exotic-looking symbols (e.g.
Greek letters). All of the transcription systems in Table 2.6 use the IPA, but
slightly differently. Any one of these transcription systems may be found in
reference works, and the list is not exhaustive. The transcription system in
column 1 of Table 2.6 is the one used here. It is useful for teachers to recog-
nise transcription systems used in individual reference works, but they
should not mix transcription systems from various sources if using them for
students or even for their own reference.

TABLE 2.5 Some non-IPA transcriptions

IPA Example Alternatives

ʧ cheese č
ʤ join ǰ
θ thigh th, þ
ð thy dh
ʃ leash sh, š
ʒ liege zh, ž
j you y
22 Introductory material

TABLE 2.6 Alternative IPA transcription systems for English vowels

Vowel 1 2 3 4 5 6

kit ɪ i ɪ i i ɪ
dress e e ɛ e ɛ ɛ
trap a a æ æ æ æ
strut ʌ ʌ ʌ ʌ ʌ ʌ
lot ɒ o ɒ ɒ ɒ ɒ
foot ʊ u ʊ ʊ ʊ ʊ
fleece iː iː i ij ij i
nurse ɜː əː ɜ ə əː ɜ
start ɑː aː ɑ a ɑ ɑ
thought ɔː oː ɔ ɔ ɔ ɔ
goose uː uː u uw uw u
face eɪ ei eɪ ej ej e
price aɪ ai aɪ aj aj ʌi
choice ɔɪ oi ɔɪ oj ɔj ɔi
goat əʊ ou əʊ ow o o
mouth aʊ au aʊ aw aw ʌʊ
near ɪə iə ɪə ɪə ɪə ɪə
square eə eə ɛə eə eə ɛː
cure ʊə uə ʊə ʊə ʊə ʊə
comma ə ə ə ə ə ə

The labels used for the vowel sounds in Table 2.6 are the names of the
lexical sets already introduced in Table 2.4.
Although Table 2.6 shows a lot of variation in the symbols used, note
particularly the symbol [a] used for the trap vowel. This symbol represents a
relatively new pronunciation with SSBE, and earlier transcriptions (and
transcriptions of American English) typically prefer the symbol [æ]. Like the
other variant symbols, this is just a choice between IPA symbols to represent
the distinctive sounds of English.

2.5 Beyond consonants and vowels

Pronunciation does not consist exclusively of consonant and vowel sounds.
Word-stress, intonation, rhythm, and voice quality are also important aspects
of pronunciation. We can term these features collectively ‘prosody’. Prosody is
harder to teach than the segments (consonants and vowels) because the dis-
tinctions are less clear-cut and so less clearly right or wrong for any given
model. They nevertheless have an important role to play in making any lan-
guage comprehensible and in making the learner sound as if they are part of
the target language community. Prosody is dealt with more fully in Chapter 5.
 The phonetic background 23

Highlights
Both consonants and vowels can be described in terms of a three-term label,
although the terms are different in the two cases. These labels can be seen as
recipes for producing sounds or as an analysis of what sounds share and
how they differ. Speakers of a language tend to hear that language (and
think of themselves as producing it) as a sequence of distinctive sounds, but
in some environments, the actual sound produced may differ from the ideal
that speakers perceive. In some cases, this becomes important for teaching
natural-sounding English. While phonetic transcriptions of English tend to
agree on how to transcribe the consonants, they do not agree on how to
transcribe the vowels. The system used depends on what factors the tran-
scriber gives priority to.

What should the teacher do?

Teachers who are not used to reading a transcription should practice this
skill – it is a transferable skill and has value in making it easier to break away
from the spelling, which in English can be confusing from time to time. They
should also practice what it means to make a plosive, a fricative and affri-
cate, a close vowel, a back vowel and so on and think about how to encour-
age students to devoice a fricative and how to make a fricative rather than a
plosive (e.g. to produce a [θ] rather than a [t]) or a plosive rather than a frica-
tive (e.g. to prevent Spanish learners from saying [ð] rather than [d] between
vowels). Those dealing with a uniform set of students can find out or work
out what consonants and vowels the students have and where these are likely
to cause problems in learning English (for more detail, see the start of
Chapter 7). The problems may be because students do not have a corre-
sponding sound, because they do have a corresponding sound which is lim-
ited to a certain environment, or because their nearest sound is a
monophthong rather than diphthong, is too close for the English equivalent
(e.g. [i] instead of English [ɪ]), is too far back for the English equivalent (e.g.
a French [u] for the English [uː]) and so on (see the remaining chapters in
Part II). To be able to teach students to produce individual sounds, teachers
need to make themselves aware of what they are doing with their tongue,
lips, and so on so as to see and hear what errors are being made and to be
able to give accurate guidance. Teachers may like to practise drawing a ver-
sion of Figure 2.1 and drawing in the tongue in different positions in order
to illustrate what is happening in the mouth if learners find this useful. This
is another strategy in the teacher’s set of skills and probably can be used
sparingly.
PART II

Mostly phonetics
3
CONSONANTS

In this section, we turn to the description of the sounds of English, looking

at those aspects which may provide difficulty for the teacher of English and
considering, by comparison with what happens in other languages, the types
of errors that are likely to occur. The various types of consonants (plosives,
fricatives, etc.) will be dealt with individually. Comments will be made about
points of difficulty and places where variable pronunciations are allowable.

3.1 Plosives
English has six distinct plosives, as set out in Table 3.1.
Although the plosives are labelled ‘voiced’ and ‘voiceless’, physical voic-
ing is not the primary way of distinguishing the pairs. In initial position, the
major difference lies in aspiration (or affrication). The so-called voiceless set
are followed by a short puff of air (an [h] sound) which comes before the
voicing for the following segment begins. In pain, for example, this can be
transcribed as [pʰeɪn] or as [pe̥eɪn] (i.e. with a voiceless part of the following
vowel), in train this can be transcribed as [tɹ̥eɪn] (i.e. with a voiceless part of
the following [ɹ]). For many speakers, the voiceless plosives are affricated
rather than aspirated; they are followed not by an [h] but by a voiceless frica-
tive articulated in the same place of articulation as the plosive. Strong affri-
cation is particularly associated with the English of Liverpool and the
surrounding region, but a small amount of affrication, especially for [k], is
normal. This means that coat can be transcribed as [kxəʊt]. If [t] is affricated,
there is a short [s]-type sound, the precise nature of which depends upon the
way the [t] is articulated. We can transcribe tie as [tsaɪ] when this happens.

DOI: 10.4324/9781032637020-5
28 Mostly phonetics

TABLE 3.1 English plosives

Voicing Place Initial Medial Final After word-initial

[s]

p voiceless bilabial pain, pea sopping, cap, rip spy, spare

nipple
b voiced bilabial bane, bee sobbing, cab, rib
nibble
t voiceless alveolar tie, train utter, catty hat, moult sty, stare
d voiced alveolar die, drain udder, had,
caddy mould
k voiceless velar coat, bicker, knack, sky, scare
Kate sacking brick
ɡ voiced velar goat, gate bigger, nag, brig
sagging

Even if [p] is affricated, we do not hear the friction much, and we can ignore
this possibility. Where the voiceless plosive is preceded by [s] in the same syl-
lable, there is no aspiration or affrication, what is transcribed as one of [p],
[t] or [k] in Table 3.1 can sound more like [b], [d] or [ɡ] respectively.
Speakers of Danish and Mandarin have similar distinctions based on as-
piration rather than voicing, and their versions of these are suitable for use
in English, although Danes have to avoid strong affrication after [t]. Speak-
ers of other Germanic languages have a similar lack of contrast between [p]
and [b] and the other pairs after [s]. The plosives in such positions are not
aspirated. Speakers of Spanish must avoid adding an initial vowel in words
like spy, sty, sky and Spanish. Speakers of Dutch, French, Polish, Spanish,
Russian and so on, where the difference between [p] and [b] and so on is
genuinely one of voicing, must reduce the force behind the so-called voiced
series. Speakers of Hindi, Thai and Vietnamese, for whom aspirated and
unaspirated plosives are distinctive, must take care not to exaggerate the
distinction. Speakers of Vietnamese must not equate their [ɓ] and [ɗ] with
English [b] and [d]: to English ears, their stops will sound overly voiced.
In medial position, the difference between the voiced and voiceless se-
ries is more a matter of voicing; in particular, [b, d, ɡ] are rather more
voiced here. There may be a small amount of aspiration for the voiceless
plosives. For some speakers, but not for speakers of Standard Southern
British English (SSBE), [t] between vowels may become voiced. Although
speakers will say that it is pronounced as [d] under these circumstances, the
sound is usually shorter than a normal [d], and a better transcription is [ɾ],
which will sound like an R-sound for speakers of Italian, Spanish and many
other languages.
 Consonants 29

Speakers of Danish, Hebrew and Spanish must note that the voiced plo-
sives remain plosives in medial position and do not become fricatives or
approximants.
In word-final position, the so-called voiceless plosives are increasingly
pronounced with a glottal stop. The articulation for the original consonant
may remain, or the glottal stop may be all that is left of the stop. In England,
the amount of use of the glottal stop has social implications – greater usage
of glottal stops is seen as less standard, even though it is very common.
Where the glottal stop alone is found, flip, flit and flick, for example, may all
sound the same. If there is no glottal stop or if the original articulation of
the plosive remains as well as the glottal stop, there may be some short aspi-
ration at the end of the word, especially before a pause. The so-called voiced
plosives retain a plosive pronunciation in final position. Also note that
voiced plosives do not become devoiced in final position, as is typically the
case throughout Germanic and Slavic. The biggest phonetic difference be-
tween [p] and [b] (and the other pairs) in final position is that what precedes
the plosive is shorter before [p] than before [b]. Other languages show similar
effects, but the effect is greater in English than in most other languages.
The plosives [p] and [b] have counterparts in a wide range of languages
which often transfer well to English. Note that Vietnamese has [ɓ] rather than
[b] and that Fijian, among other languages, has pre-nasalised voiced stops
(e.g. [mb]). The plosives [t] and [d] are typically pronounced with the tongue
against the alveolar ridge unless there is a following [θ] or [ð] (e.g. eighth or
breadth). While a dental plosive is unlikely to cause problems of comprehen-
sion, it is indicative of the organs of speech being in the wrong position for
first-language (L1) English. In the long term, it is likely to sound childlike.
The use of a post-alveolar or retroflex articulation in [t] or [d] is characteris-
tic of the English spoken in the Indian subcontinent and again is an indica-
tion of the wrong overall position of the organs of speech. It also sounds
intrusive and should be avoided; speakers from these areas need to be trained
to keep the tongue tip down. The tip of the tongue may articulate against the
alveolar ridge or the blade of the tongue (giving rise to tongue-tip up or
tongue-tip down articulations). Either is permissible, though tongue-tip
down is described in most textbooks. The plosives [k] and [ɡ] show variation
in how far back or forward on the velum they are pronounced, depending on
the following vowel, as is typical of these sounds across languages. However,
a language such as French has much greater variation than English does.
This can be seen as part of an overall laxer articulation in English than in
French. The plosives [k] and [ɡ] are less stable in many languages than in
English, where they must retain a plosive articulation before front vowels
(contrast, for instance, Norwegian and Swedish) or in general terms (con-
trast, for instance, Dutch). In a language like Turkish, <g> can be used as an
30 Mostly phonetics

TABLE 3.2 Uses of the glottal stop

Replacing [p], [t] or [k] word- Becoming more common but not standard,
finally, [ɹeɪʔ] for rape, rate, rake except if there is an immediately following
consonant, as in outcome
Supporting word-final [p], [t] or Very common and acceptable
[k], [ɹeɪʔ͡p] [ɹeɪʔ͡t], [ɹeɪʔ͡k]
Supporting or replacing [t] before Normal
[n], [bʌʔ͡tn] or [bʌʔn] for button
Replacing [t] between vowels, Associated with particular dialects, not widely
[maʔə] for matter acceptable
Initially before a stressed vowel, Acceptable, but not necessary; speakers of
[ʔɔːfəl] for awful Arabic, German or Thai must not do this as
much as in their native languages. Speakers
of Indonesian must note that [ʔ] is not used
in English to separate prefixes from bases.

orthographic representation of something other than [ɡ], and care may be

needed. For speakers of many languages (including Italian, Japanese, Man-
darin, Spanish and Vietnamese), it is important that plosives be pronounced
in word-final position.
Although it is not yet fully contrastive, we can add the glottal plosive, or
glottal stop, [ʔ]. This is used in a number of different places, often with
marked social implications. This is set out in Table 3.2.

3.2 Fricatives
English has nine distinct fricatives, as set out in Table 3.3.
Unlike the plosives, the fricatives are distinguished by voicing but also by
force of utterance, the so-called voiced series being less energetically pro-
nounced than the voiceless series. This contrasts with what happens in some
other languages, where voicing is the main point for distinguishing the two.
Some languages have only one set of fricatives, which may then not be clearly
voiceless or voiced. Such a language was Old English, where voicing de-
pended upon the environment.
Unlike the situation in languages as diverse as Dutch and Hindi, both the
[f] and the [v] are real fricatives, with the top teeth touching the bottom lip
(although just where on the bottom lip the contact occurs may differ – it may
not always be visible). Learners who have difficulty with this can practice
making their lip tingle when they say the sound. Hebrew, Spanish and Japa-
nese learners have to ensure a difference between [v] and [b]. There are mini-
mal pairs (illustrated in [X1]) which can be used to confirm the importance of
the distinction. English speakers may occasionally hear [β] as [b] if the fric-
tion is minor but will usually hear [β] as [v] and may fail to understand.
 Consonants 31

TABLE 3.3 English fricatives

Voicing Place Initial Medial Final

f voiceless labio-dental fie, fine safer, luffing safe, belief

v voiced labio-dental vie, vine savour, loving save, believe
θ voiceless dental thigh, tanhb ether, earthy mouth (noun),
sheath
ð voiced dental thy, than either,a worthyc mouth (verb),
sheathe
s voiceless alveolar sue, sap racing, looser hiss, face
z voiced alveolar zoo, zap raising, loser his, phase
ʃ voiceless post-alveolar sheet Confucian, leash, gauche
Aleutian
ʒ voiced post-alveolar gited confusion, liege, beigee
allusion
h voiceless glottal hate behave –
Notes
a
Either can be pronounced [aɪðə] or [iːðə]. The minimal pair works only with the second of
these.
b
Tanh, a technical mathematical term, has various pronunciations, of which [θan] is one – and
an unexpected one.
c
Obviously not a minimal pair: there are no minimal pairs.
d
Like so many words with initial or final [ʒ], this is a French loan word, and alternative pro-
nunciations can be heard.
e
Again, not a minimal pair, and for the same reason

(X1) 
bail / veil, ban / van, bane / vein, bat / vat, bet / vet, beer / veer, best
/ vest, bide / vied, bile / vile, boat / vote, bow(ed) / vow(ed), bowl /
vole, buy (bye) / vie; cubby / covey, hobble / hovel, lubber / lover,
ribbon / riven, robing / roving, Sybil / civil; dub / dove, grebe /
grieve, jibe (gibe) / jive, lobes / loaves, robe /rove, sib /sieve

Although they occur in languages like modern Greek and Icelandic, [θ]
and [ð] are considered relatively rare sounds, and a lot of time tends to be
spent on teaching them. This is particularly true since [ð] appears in so many
grammatical words and is therefore frequent. It is thus worth considering
the fact that many L1 speakers never use either. The difficulty with this is
that the alternatives are often considered socially inferior in England. Nev-
ertheless, it is worth considering the options, which, in England, are [f] and
[v] or [t] and [d]. These alternatives can be used in many cases, despite the
possible minimal pairs (see [X2]) as long as the sounds are not given any
emphasis: the moment the learner gives them full weight, their non-standard
nature comes to the fore. It is noteworthy that [s] and [z] are not alternatives
32 Mostly phonetics

for [θ] and [ð] within English, although many second-language (L2) learners
make the substitution.

(X2)

[θ] ≠ [s] sink / think, worth / worse

[ð] ≠ [z] bathe / baize, breathe / breeze
[θ] ≠ [f] loath /loaf, sheath / sheaf
[ð] ≠ [v] loathes / loaves, sheathe / sheave
[θ] ≠ [t] math / mat, thin / tin
[ð] ≠ [d] breathe / breed, there / dare, they / day, though / dough (doe)

There are two major ways of articulating [θ] and [ð]. The first is for the tip
of the tongue to touch the back of the front incisors; the second is for the tip
of the tongue to be visible between the front teeth. The first is most usually
used in England, the second is most usually used in the US. Either is permis-
sible. If students have difficulty in learning these sounds, teaching them to
bite the tip of their tongue (not too hard!) is one way to get the place of ar-
ticulation right.
Because [ð] is used initially only in grammatical words, it is possible to
give a complete list of the words that begin with [ð]. They are listed in (X3).
Any other words that begin with a dental fricative must start with [θ]. Note
that there are words which are spelt with <th> but pronounced with [t] (e.g.
Thai, Thames, Theresa and Thomas.)

(X3) Determiners; the, this, that, these, those

Pronouns: thou, thee, thy, thine, thyself; they, them, their, theirs,
themselves
There: there, therefore and other compounds such as thereabout(s),
thereafter, thereof, thereon
Conjunctions: thus, though, then, that (conjunction), thence and
rare compounds such as thenceforth, thenceforward
Than
Note: Even though through is a preposition, it starts with [θ];
though (like although) is pronounced with [θ] in some kinds of
English. Thon, as a determiner, is pronounced with initial [ð]
but is rarely heard these days except in dialectal usage.

Almost every language that has any fricatives at all has [s], and [z] is the
voiced equivalent. This does not mean, though, that the [s]-sounds in all
languages are the same. Modern Greek and Spanish notoriously have [s]-
sounds with a deeper pitch (more [ʃ]-like) than English. There are two ways
of articulating [s] and [z] in English: (1) with the tongue tip down and the
 Consonants 33

blade of the tongue close to the alveolar ride or (2) with the tongue tip
slightly up and the tip close to the alveolar ridge. Either works well in Eng-
lish, although the tongue-tip-down version is considered normal in England,
and the tongue-tip-up version is more likely to give a slightly whistled effect
or sound slightly lisped. A fully lisped [θ] for [s] should be avoided by L2
learners if at all possible: it often sounds childish, perhaps deliberately so.
The fricatives [ʃ] and [ʒ] behave less obviously as a pair than the other sets
of fricatives. This is because [ʒ] is a relatively new sound in English, arising
from French loans and the merger of [z]+[j] (e.g. in a word like pleasure).
This lack of commonality shows up particularly in the distribution of [ʒ],
which is rarely found word-initially or word-finally (and when it is found,
often alternates with [ʤ], so that genre and beige can be [ʤɒnrə] and [beɪʤ]
respectively). Despite this, [ʃ] and [ʒ] share noticeable lip-rounding, which
often distinguishes these sounds in English from similar sounds in other lan-
guages. Danish and Greek tend to use something approximating to [sj],
which can be improved by the use of lip-rounding.
There is a recent increase in the use of [ʃ] rather than [s] before [t] in words
like strong and Australia (with a following [ɹ]) and student and stupid (with a
following [j] or where the [t]+[j] gives [ʧ]). The social value of this is not yet
clear, but it appears to pass unnoticed much of the time. Although there is
little sign of the same happening before [p] (as in spring) or [k] (as in scream),
this might be expected to happen in the next few years. It is not clear whether
it will spread to words like still and stalk where a vowel follows the [t].
The sound [h] counts as a consonant in English because we say a hat and
not *an hat (and so forth). Phonetically, though, it can be considered to be a
voiceless version of the following vowel. Most of the time, this does not need
to be taught explicitly (although it may have some value for speakers of lan-
guages with no [h]), but it has some ramifications. Specifically, the sequence
[hi:] is often pronounced [çiː], especially when stressed, where the [iː] vowel
may be close enough to give rise to friction when voicing is removed. Al-
though we might expect to find [xuː] for who in similar cases, this is rare.
Speakers of Mandarin, Russian or Spanish, who typically use [χ] or [x] in
place of [h], should be discouraged if possible since these fricatives sound
intrusive in English. Treating [h] as a voiceless vowel may help here.
Although [h] is used in SSBE, it is avoided in most varieties of British
English (it is present in Scottish English, Tyneside English, and standard
varieties of North American English and New Zealand English). A word
like hill, in those varieties that ‘drop’ the [h], is treated as if it begins with a
vowel (so that people say [ən ɪl] for a hill and [ðiː aʊs] for the house, for ex-
ample). All varieties of English drop the [h] with pronouns like he, him and
her and with have, had and has when these are unstressed. If L2 learners
copy h-dropping (which, though common, is often seen as a sign of lack of
education), they should not replace the [h] with a glottal stop.
34 Mostly phonetics

The sound [h] is never found at the end of a word, but it can occur in the
middle of a word, where it may be slightly voiced, as in [bɪɦeɪv] behave. This
pronunciation is never necessary, though.
Some words that are spelt with an <h> are not pronounced with [h]: these
include heir, honest, honour, hour and (in the US but not in Britain) herb.
The use of an before a pronounced [h] in a few words, including, most fre-
quently, hotel and historical, is now slightly pretentious and can safely be
ignored.
As with the plosives, voiceless fricatives in word-final position make the
previous syllable shorter, and voiced fricative make it longer. Peas, therefore,
has a longer vowel than peace. This is often the main guide to the voicing of
the fricative.

3.3 Affricates
An affricate is something which acts as a single consonant, though it has
both a plosive and a fricative part which are articulated at the same place of
articulation and with the same voicing. We can say that English has just two
affricates: [ʧ] and [ʤ]. Although you can hear [ts] in words like nets in Eng-
lish, for example, it is considered to be a sequence of two sounds, not just a
single sound, as it might be in German or Russian. Similarly, [tɹ] and [dɹ]
and [tθ], although they otherwise fit the definition of affricates, are consid-
ered to be sound sequences in English.
The affricates can be found initially, medially and finally, as in Table 3.4.
Although the transcriptions of [ʧ] and [ʤ] seem to imply that they start
with a [t] and [d] sound respectively, the tongue makes contact further back
behind the alveolar ridge than it typically does for [t] and [d], in the position
where [ʃ] and [ʒ] are articulated. Like the post-alveolar fricatives, [ʧ] and [ʤ]
show noticeable lip-rounding. Many languages (including Italian and Span-
ish) have sounds which transfer easily to these English affricates. But in lan-
guages like Japanese, Vietnamese and some varieties of Chinese, there are
nearly equivalent sounds that are articulated slightly further forward in the
mouth. These are then transcribed [tɕ] and [dʑ]. For speakers of such lan-
guages, a better approximation to the English sound can be achieved by us-
ing the lip-rounding in English.

TABLE 3.4 English affricates

Initial Medial Final

[ʧ] choose, chin catches, beseeching rich, larch

[ʤ] Jews, gin cadges, besieging ridge, large
 Consonants 35

Although many speakers of SSBE distinguish between [ʧ] and [tj] and
between [ʤ] and [dj], there is an increasing tendency for the two to sound the
same, so that [ʤuː] can be either due or Jew, [ʤuːn] can be either dune or
June, and tune and tutor are frequently pronounced [ʧuːn] and [ʧuːtə]. In the
middles of words, such as nature, the affricate is now the norm. Before an [ɹ],
affricates are starting to replace [t] and [d], so that trap becomes [ʧɹap] and
drink becomes [ʤɹɪŋk].
As with the plosives and the fricatives, word-final [ʧ] makes the previous
sound shorter, and word-final [ʤ] makes the previous sound longer, so that
the vowel is shorter in batch than in badge, and this often determines which
of these sounds we perceive.

3.4 Nasals
English has three distinct nasals (or, more explicitly, ‘nasal stops’), as set out
in Table 3.5.
Because [ŋ] is derived historically from a sequence of [n]+[ɡ], where the
[n] was pronounced as [ŋ] before the velar consonant (just as it is in modern
think [θɪŋk]) and then the [ɡ] sound disappeared, [ŋ] never occurs in initial
position.
When there is a nasal consonant at the end of a word, any vowel imme-
diately preceding that consonant becomes nasalised, so that we hear [ɹãm].
Different speakers and different varieties of English differ in how far this
nasalisation spreads and whether it also spreads to following vowels. SSBE
allows only modest spread, and some varieties of Australian English allow
nasalisation to spread a long way. Because nasalisation of vowels does not
change meaning (as it does in French, Hindi, Polish, Portuguese and so
on), the nasalisation of the vowel (or indeed a fricative or an approximant)
does not cause misunderstanding. Although speakers of languages which
distinguish nasal vowels might sound rather unnatural if they do not na-
salise vowels adjacent to nasal consonants in English, it will not prevent
communication.

TABLE 3.5 English nasals

Initial Medial Final

[m] mat, mail simmer, dimming ram, sum

[n] gnat, nail sinner, dinning ran, sun
[ŋ] - singer, dinging rang, sung
36 Mostly phonetics

Although the default is to find [m] and [n] (and this is what happens at the
beginnings of words, for example), there is a large amount of variation in the
place of articulation for these two nasals. Instead of [m], we find [ɱ] (i.e. a
nasal consonant with the top teeth touching the bottom lip, that is a labio-
dental nasal) when there is a following [f] or [v], as in comfort and symphony.
This also happens when the nasal is written with an <n>, as in infinity. The
actual articulation may involve the incisors touching the centre of the bottom
lip or making contact on the inside of the bottom lip (in which case the con-
tact may be invisible); for what we think of as [m], there may also be a simul-
taneous bilabial closure, so that we get [m͡ɱ]. Instead of alveolar [n], we find a
dental articulation, [n̪] before [θ] or [ð], as in enthusiasm, and slightly ret­racted
versions of [n] before [ɹ] or before [(t)ʃ] and [(d)ʒ] in words like enrol, enchant
and range. L1 speakers of English are not aware of this variation, which is
automatic, but the changes may confuse speakers of other languages. These
changes will also occur over word boundaries as in come forward, in the end,
in Russia, and can choose. Many other languages show similar variation.
Both [m] and [n] can also form syllables on their own without any sup-
porting vowel, in words like spasm and button. In cases like button where
there is a preceding [t], the [t] is often replaced by [ʔ].
As with [k] and [ɡ], [ŋ] shows variation in place of articulation dependent
of the preceding vowel, but this is not a major factor in its pronunciation,
and it should never be replaced by a palatal [ɲ], which speakers of French
and Italian are likely to do.

3.5 Liquids: [ɹ] and [l]

The sounds [ɹ] and [l] are classified together as liquids within the larger group
of approximants.
The lateral alveolar approximant [l] has two major variants in SSBE, and
these are treated in different ways in other varieties of English. The type of
[l] that occurs before vowel sounds and [j], which we can transcribe [lʲ], is
called a ‘clear’ or ‘palatalised’ [l]. The front of the tongue behind the alveolar
ridge is raised towards the hard palate behind the alveolar contact while this
sound is articulated. Another way of saying this is that [l] and [ɪ] are pro-
nounced simultaneously. Speakers of Danish, French, German, Italian and
Spanish will find that this is their usual variant of [l], and the difficulty is not
using it in all positions. The non-pre-vocalic variant of [l] (i.e. not before a
vowel), which we can transcribe as [lɣ], has the back of the tongue raised
towards the soft palate or velum behind the alveolar contact (the [l] is pro-
nounced simultaneously with [uː]). We can call this ‘dark’ or ‘velarised’ [l].
Speakers of Dutch and Portuguese will find that this sounds rather like their
usual [l] sound, although the English version is not quite so ‘dark’, and again
 Consonants 37

the difficulty is in not using it in all positions. Some varieties of English use
only a single variety of [l]: Irish English, Welsh English and Tyneside English
use clear [lʲ] to the exclusion of dark [lɣ]; Glasgow English, Australian Eng-
lish and some varieties of American English use dark [lɣ] all the time or, at
least, use darker versions of [l] in all positions than does SSBE. In some
kinds of English, notably the English of the South East of England, types of
Australian and New Zealand English, the dark [lɣ] is replaced by a vowel. We
can transcribe that vowel [o], although its precise quality is quite variable.
This is then called ‘vocalised’ [l] (i.e. [l] made into a vowel). Speakers who
have a vocalised [l] will use the vowel in pull but a consonantal [l] in pull away
where there is a following vowel (even over a word boundary as here). Many
Chinese learners of English use a vocalised [l] in their English without it
sounding unnatural.
In a consonantal [l] sound, the airflow escapes over the side or sides of the
tongue. Some speakers have a unilateral [l], others have a bilateral [l], and the
difference is unimportant. The place where the air is flowing can be felt if the
articulators are held in position for the [l], and the speaker breathes in
through the mouth. When a consonantal dark [lɣ] is pronounced before a [θ],
as in health, the contact between the tongue and the upper roof of the mouth
is made against the upper incisors instead of against the alveolar ridge. Fol-
lowing [p] or [k] where these would be aspirated, the [lʲ] is partly or fully
devoiced, as in [pl̥eɪ] play.
The post-alveolar approximant [ɹ] is one of the hardest of the English
consonants to deal with, for various reasons. First, there is a matter of its
articulation. In SSBE, its default pronunciation involves the tip of the tongue
being curled very slightly back behind the front teeth while being far enough
away from the alveolar ridge not to cause friction. Many Americans have a
different articulation for the [ɹ], which nonetheless is auditorily equivalent.
Although [ɹ] functions as a consonant in English (e.g. we say a rabbit and not
*an rabbit), it may be easier to treat it as a vowel with the tongue turned up.
Many American phoneticians treat [ɹ] in this way, and though it not usual for
British phoneticians to use such descriptions, it makes some sense, and it may
resonate for Mandarin learners of English. This vowel is then non-syllabic in
pre-vocalic position (e.g. in rabbit) but syllabic post-consonantally (e.g. in
mastery – which we would then have to transcribe [mɑːstɹˌi]). Since this par-
ticular type of [ɹ] sound is rare in the languages of the world, most learners
of English will have to try to acquire some approximation to it, and starting
from a vowel may give better results than starting from some other kind of
sound represented by an orthographic <r>. It is important, however the [ɹ]
sound is approached, that a certain amount of lip-rounding be involved.
Small L1 children who cannot yet say [ɹ] approach it with something that
sounds to adult ears like a [w], where the lip-rounding is obvious.
38 Mostly phonetics

There is some slight variation in the pronunciation of [ɹ] depending on

the context. Following an aspirated plosive, as in pry, try and cry, the aspira-
tion is realised as a partial or full devoicing of the [ɹ], so that we get [pɹ̥aɪ],
[tɹ̥aɪ] and [kɹ̥aɪ] respectively. Following [θ], we may find [ɾ] rather than [ɹ], as
in three and throw. This variant is not obligatory but is common. Although
SSBE once allowed [ɾ] between vowels, as in very [veɾi], this is no longer cur-
rent and should not be used.
The big problem caused by [ɹ] is its distribution. This is best understood
if we consider its history. Originally [ɹ] was pronounced everywhere <r> is
written (the precise phonetic nature of the sound is in dispute but does not
matter a great deal). By the 16th century, word-final [ɹ] was no longer con-
sistently pronounced in the East of England, although it was retained in
other parts of the country, as it is today in many parts of England. Because
the first settlers from England in North America came from the West of
England (sailing from Bristol), the pronunciation of [ɹ] was maintained in
the American settlements and eventually became the most widespread vari-
ant there. Because Australia was settled after the loss of [ɹ] became stan-
dard and because so many Australian settlers came from the London area,
Australian English does not have [ɹ] at the ends of words. Although this loss
has been characterised here as ‘at the end of words’, it was also lost before
consonants in the same word, as in harp, smart and form. We can generalise
this by saying that non-pre-vocalic [ɹ] was lost (i.e. any [ɹ] that does not oc-
cur before a vowel). In many cases, this loss was compensated for in some
other way. In words like fore, sore and mar, farm the vowel became longer,
so that the vowel in ore became the same as the vowel in awe, and the vowel
in farm became the same as the vowel in father. In words like here, there and
cure, the [ɹ] was replaced by [ə], and the vowels became long vowels. (Origi-
nally, as in modern American English, the vowels here were like the vowels
in hill, hell and bull respectively and thus short.) The vowel [ɜː] arose from
sequences of [ɪ]+[ɹ] (first), [e]+[ɹ] (herd), [ɒ]+[ɹ] (word), and [ʌ]+[ɹ] (curse).
The result is that the vowel patterns in British English became much more
complicated than those in American English, and such a situation remains
to this very day, despite extra complications having been added to the
American system over the years. One of the results of all this is that speak-
ers lost the [ɹ] in more, where it was final, or in more bread, where it was
before another consonant in the stream of speech, but retained the [ɹ] in
more often where it was before a vowel. Similarly, inside the word, the [ɹ] is
pronounced in forest but not in fort and retained in snoring but not in snore.
This gives rise to a problem because snore, in isolation, rhymes with draw in
isolation, but in snoring there is [ɹ] while in drawing there was never an [ɹ].
Some speakers keep the two distinct ([snɔːɹɪŋ] vs [dɹɔːɪŋ]), while others
made the two the same, by analogy ([snɔːɹɪŋ] and [dɹɔːɹɪŋ]). When an [ɹ] is
 Consonants 39

used in drawing, it is called an ‘intrusive’ [ɹ], and the same label is used over
a word boundary as in China and India [ʧaɪnə ɹ ənd ɪndjə] or saw a film [sɔː
ɹ ə fɪlm]. Historical [ɹ] that is heard before a vowel in the next word (as in
more often and a fair idea) is called a ‘linking’ [ɹ]. If we just look at the way
in which words are spoken in isolation, [ɹ] can never occur at the end of a
word (or, within a word, before a consonant). If we look at the way in which
phrases are spoken, [ɹ] can occur between words where the second starts
with a vowel. In the transcriptions given above, this [ɹ] has been set apart
from both words, but it is usually associated with the first of the two words,
and in pronunciation dictionaries for British English, you will find some
kind of symbol to indicate that certain words are pronounced with final [ɹ]
only when a vowel follows.
Varieties of English which retain the historical distribution for [ɹ] are
called ‘rhotic’ ([ɹəʊtɪk]) varieties of English. They include standard US Eng-
lish and Canadian English, Scottish and Irish Englishes, and many dialects
of English in England, including most of the southwest of England (west of
about Reading) and parts of Lancashire. Non-rhotic varieties of English
include SSBE, standard Australian and New Zealand English, South Afri-
can English, and in England the dialects of Eastern and Central England.
L2 learners have to decide to use (or have modelled for them) either rhotic or
non-rhotic patterns. The benefits of using a non-rhotic pattern are that the
descriptions in British textbooks, pronunciation dictionaries and the like
will make sense, and rhotic varieties are not all alike. The benefits of using a
rhotic variety are more widespread:

• Rhotic usage is better reflected in the spelling.

• Non-rhotic speakers can understand rhotic speakers, while rhotic speak-
ers tend not to understand words like harp spoken with no [ɹ].
• There are fewer vowels in a rhotic variety.
• Linking and intrusive [ɹ] are not problems in rhotic varieties.

Although these advantages are clear, a choice is usually made on social or

geographical grounds rather than on grounds of logic.
Speakers of Mandarin and Vietnamese, where the sound written <r> in
Roman script is pronounced rather more like [ʒ], need to take particular care
in learning this sound.

3.6 Other approximants
The two other approximants to be considered here are [j] and [w], often
termed ‘glides’ or ‘semivowels’.
40 Mostly phonetics

Although there are transcription systems where what is here annotated as

[eɪ] is transcribed as [ej] (and similarly for a number of other vowels – see
Section 2.4), in the systems used here [j] does not occur word-finally. If learn-
ers have problems with this sound, they can think of it as a rapid [i] sound.
(This will make sense to speakers of Italian and Spanish, for instance.)
Though [j] is usually voiced, when it follows [p], [t] or [k] in a stressed syllable
(e.g. in pewter, tunic and cuter) it is often partly or wholly devoiced. This is
then the realisation of the aspiration of the plosive. The sequence [h]+[j] is
replaced by [ç] (e.g. in words like huge and humour), especially when uttered
forcefully. The [j] alone may be replaced by [ç] when it is devoiced (as de-
scribed immediately above). In a word like acute, the [k] may be replaced by
a palatal at the same time, so that we can get [əcçuːts], but such variation
should not be exaggerated.
When any of the vowels [i], [iː], [eɪ], [aɪ] and [ɔɪ] is immediately followed by
another vowel, either in the same word or in the next, a slight [j] links the
two. This happens in expressions such as see it, saying and fly over. Although
this link is clearly heard, it is not just the same as an ordinary [j], and say ‘S’
is not just the same as Say ‘yes’. Learners do not need to learn to produce
this but should not be discouraged if it emerges spontaneously in fluent
speech.
The major difficulty with [j] lies in its distribution. Although [j] can follow
many consonants at the beginnings of syllables (e.g. beauty [bjuːti], fuse
[fjuːz], enthuse [ɪnθjuːz], new [njuː] and view [vjuː]), there is a lot of variation.
It is now rare to find [j] following [l] or [s] (as in lewd, sleuth, supreme and
suitor), although it was used in such positions within living memory. In a few
dialects, [j] is avoided after all consonants, but this is rare. It is, however, fre-
quent to find that [j] is not used after [t], [d], [n], [z] and sometimes [θ] so that
we hear [tuːn] tune, [duːn] dune, [nuːz] news, [zu:ɹɪk] Zurich and [ɪnθuːz] en-
thuse. This so-called ‘yod-dropping’ is particularly noticeable in US English
(much less so in Canadian English), where it has become standard and occurs
after alveolar consonants and sometimes dental consonants. It is also found,
sometimes systematically and sometimes more variably, from speakers from
other places, too. The patterns of deletion are not the same in stressed and in
unstressed syllables, which along with the variability that is found makes it
difficult to provide a thorough description of this area. Learners can proba-
bly imitate what they hear around them if they are resident in an English-
speaking environment, but either pronunciation pattern is completely
acceptable. For some speakers, [s] and [z] are replaced by [ʃ] and [ʒ] when
there is an immediately following [j], as in assume and presume. In such in-
stances, most speakers lose the [j], but some still retain it, so that [əʃuːm] and
[pɹɪʒuːm] and [əʃjuːm] and [pɹɪʒjuːm] can both be heard. The problem is a
 Consonants 41

minor one, given that it affects very few words, and the less complex articula-
tion is probably preferable for learners.
Just as [eɪ] is transcribed [ej] in some systems, so [aʊ] is transcribed [aw],
and a similar solution is used with other vowels. Otherwise, [w] does not oc-
cur at the ends of words. The approximant [w] has a complex articulation,
requiring both lip-rounding and an articulation between the back of the
tongue and the velum. Speakers who have difficulties can think of it as a
quick [u] sound, though this will not work for Japanese speakers, for whom
[u] is not rounded. Speakers of Dutch and German tend to be misled by the
spelling systems of their own languages and to replace [w] by [ʋ] or [v]. Nei-
ther works well, although [ʋ] is less awkward. Speakers from the Indian sub-
continent often use [ʋ] for both [v] and [w]. The sounds [v] and [w] often
contrast in English, as illustrated in (X4).

(X4) 
vain (vane) / wain, vary / wary, veal / weal (wheel), veer / weir,
veggie / wedgie, veil / wail, vending / wending, vent / went, verse /
worse, vest / west, vial (vile) / wile (while), Vicki / wiki, vie / Y
(why), vine / wine, viper / wiper, voe / woe, volley / Wally, vow /
wow

When [w] follows [t] or [k] at the beginning of a stressed syllable (as in
twin and acquire), it is often devoiced, that devoicing corresponding to the
aspiration of the plosive. There are no native words that being with [pw], but
in a borrowed word like pueblo, the same thing happens. In pronouncing [w],
the lips must be rounded, and the teeth should be well out of the way.
When any of the vowels [u], [uː], [aʊ] or [əʊ] is immediately followed by
another vowel, either in the same word or in the next, a slight [w] links the
two. This happens in expressions such as go away, grew it, and how awkward
(as long as there is no initial [ʔ] in awkward). Although this link is clearly
heard, it is not just the same as an ordinary [w], and knew it is not just the
same as new wit. Learners do not need to learn to produce this but should
not be discouraged if it emerges spontaneously in fluent speech.
For a decreasing number of speakers of SSBE, there is a contrasting voice-
less version of [w], written <wh>, which can be transcribed [hw] or [ʍ]. For
speakers who use this, there is a difference between witch and which, wen and
when, wear and where, Wales and whales, and so on. Mostly, the spelling is a
reliable guide to which you find, though note that who, whom, whose, whoever
and so on begin with [h], as does whooping cough. The [hw] pronunciation is
used by some US speakers, speakers of Scottish English, and a minority of
speakers elsewhere. It is never necessary.
42 Mostly phonetics

Highlights
There is a good deal of variation in the way consonants are pronounced.
Some of this variation means that the teacher can be more relaxed, but in
some cases the potential variation is likely to lead to pronunciations which
may confuse L1 listeners. Similarly, many bound variants arise naturally in
context, but some of them (such as clear and dark [l], aspirated voiceless
plosives, including voiceless plosives followed by devoiced sonorants in
words like play and trouble) are really important for making English sound
natural and comprehensible.

What should the teacher do?

Teachers need to decide what variation they will allow and what their teach-
ing priorities are. Vital questions are what distribution of [ɹ] they will teach
and what distribution of [j] they will teach. Some of this is dependent on the
L1 of the students, so that the teacher needs to think about the first language
or first languages of the students as well as about English. Teachers may also
have different goals for students at different levels of learning. The aspira-
tion of voiceless plosives in relevant environments is crucial, and even if
students are not taught to distinguish all of the consonants, they must be
taught to distinguish [ɹ] and [l] (both frequent sounds in English) if this is
new to them. At beginning levels, exercises with minimal pairs allow for
teaching students to listen for things that are the same or different. Non-
sense words can be used in such exercises (some nonsense words are pro-
vided in Chapter 7). The sound [ð] is common in texts (mostly because of the
word the) but occurs in few words, and [ʒ] is rare in both. The balance be-
tween frequency and lexical distribution may be a factor that teachers want
to consider in planning what to teach.
4
VOWELS

The vowels will here be divided into four sub-systems: the system of short
vowels, the system of long monophthongs, the system of diphthongs and the
system of unstressed vowels. To understand this division, it must be under-
stood that vowels differ in length. Some of this difference is a matter of how
open the vowels are (more open ones tend to be longer) or how stressed they
are (stressed vowels tend to be longer), but some of it is a matter of the inher-
ent length in the vowel, which is what is important here. The monophthongs
of English can be divided into two length classes (which we will call ‘short’
and ‘long’) and that length difference is maintained as long as the vowels are
in identical environments. For example, between [b] and [t], the vowel [ɪ] is
shorter than the vowel [iː] (as indicated by the length mark in the second
symbol) so that most speakers of Standard Southern British English (SSBE)
can hear that bit is shorter than beat. The difference may disappear when the
vowels occur in different environments, so that the difference in length be-
tween bid and beat is far less clear. In English (but not necessarily in other
languages, such as Icelandic), all the diphthongs count as long vowels. This
distinction goes back many hundred years to a period when there were long
vowels and short vowels with the same vowel quality, distinguished purely by
length. These days, in most cases, the quality of the short vowels does not
match the quality of any of the long vowels; the most likely exception is that
[ʌ] may have the same quality as [ɑː], so that bud and bard may have very
similar-sounding vowels. Moreover, the quality is more important for the
perception of first-language (L1) speakers than the length. Not all varieties
of English retain the length distinction, but SSBE does. In Scottish English,
length differences are no longer relevant in the same way. In US English, the

DOI: 10.4324/9781032637020-6
44 Mostly phonetics

differences may not always match with the ones we hear in SSBE, and they
tend to be less clear. Speakers of languages like Finnish and Japanese are
used to having long and short vowels but not used to having them differ in
quality. Speakers of languages like Russian and Spanish are not used to hav-
ing a difference between long and short vowels and may have to learn to
make some vowels longer than others. Speakers of French may have some
long vowels but may not be used to hearing them as long. Speakers of Dutch
and German, for example, have a situation similar to that found in English.
Other labels for ‘long’ and ‘short’ are found in the literature. Some schol-
ars write about tense (long) and lax (short) vowels, though it not always clear
that tension matches length in English. Others prefer unchecked (long) and
checked (short) because in one-syllable words the short vowels must always
be followed by a consonant (so that the vowel is checked in the syllable),
while unchecked vowels (the long ones) and diphthongs can occur finally in
a word, as in see, say, sue and sow.
Vowels are now standardly referred to by the names of lexical sets (these
were introduced in Section 2.2). We talk of the lexical set of words which
contain the kit vowel, for example, meaning the set of words that has the
same vowel in it as the word kit has. In a resultant shorthand, we then talk
about the kit vowel. The names of the lexical sets are used in the tables
below.

4.1 The short monophthongs

There are six distinct short monophthongs in SSBE, as illustrated in
Table 4.1.
The strut vowel and the foot vowel both come from the same older
source, and most of the relevant words now have [ʌ] so that the foot vowel
is rarer than might be expected. In many dialects of the North of England,
there is no distinction drawn, and all the relevant words have [ʊ]. This should
not be imitated unless the learner lives in the one of the relevant areas.

TABLE 4.1 The short vowels

Vowel Lexical Set Examples

ɪ kit pit, fit, din, piss, miss, wit, hit, sit

e dress pet, den, mess, wet, het, set
a trap pat, fat, Dan, mass, hat, sat
ʌ strut putt, phut, done, pus, muss, hut
ɒ lot pot, don, moss, watt (what), hot, sot
ʊ foot put, foot, puss, soot
 Vowels 45

There are many languages which have only one vowel with qualities like
English [iː] and [ɪ], and learners from these language backgrounds have dif-
ficulty in distinguishing the qualities. This incudes, but is not restricted to,
speakers of French, Italian, Japanese and Spanish. Speakers of Danish,
Dutch and German have vowels of approximately the right quality, although
the match is not perfect. The Russian [i] after a hard consonant is too open.
Although Australian English and the English spoken in Birmingham, Eng-
land, have qualities for the kit vowel which is much nearer [i], these sound
rather extreme for most L1 speakers of English. Since the quality of [ɪ] is
important for L1 speakers, learners may need to be taught to make the vowel
sound more relaxed or nearer to [ə].
There is considerable variation in the pronunciation of the dress and the
trap vowels. This is due partly to change in the pronunciation of these vow-
els over the last century and partly to differences between varieties of Eng-
lish. Australian English and, even more, New Zealand English are notorious
for having a dress vowel which sounds rather like the kit vowel for speakers
of other varieties, while the current quality for the dress vowel in SSBE is
rather close to what was the quality of the trap vowel in earlier times. This
has the great benefit for learners that they probably have a vowel in their first
language which is in the area of permissible variation for English dress.
German learners, in particular, have been taught for many years that the
English trap vowel sounds rather like their vowel written with <e> or <ä>.
This was always something of a caricature but is, in any case, based on the
pronunciation of English in the 1940s or 1950s. Today, the use of their vowel
spelt with <a> is a better approximation to English trap. Singaporean and
Hong Kong English have no distinction between the dress vowel and the
trap vowel. Given minimal pairs such as man / men, the surprising thing is
how few problems this raises and how little comprehension is impeded. I
suspect that it is because the vowel that is used could fit within the range of
variation of either and that L1 speakers of the majority of varieties that
make a distinction simply predict which vowel they will hear. Where redun-
dancy is more important, such a solution cannot be recommended, but it
does suggest that precise articulation of distinct vowels need not be a high
priority.
The strut vowel is open and central and in many languages corresponds
to a vowel written with <a> in the Roman alphabet. In SSBE, it is now very
similar to the trap vowel, although the two are kept apart, even though at
one point it looked as though they might merge. In varieties of English away
from the South East of England, it often sounds more [ə]-like, and this or a
slightly more retracted version can be used to keep strut and trap apart, or
trap may be slightly longer. There is some confusion with lot, where the
vowel is written <o>. The word donkey has [ɒ] and the word monkey has [ʌ].
46 Mostly phonetics

There are several words where there is variation between the two, and either
may be heard. In particular, one is [wɒn] in the North of England (and none,
correspondingly, [nɒn]), and nothing is [nɒθɪŋ]. But speakers may use either
in Abercrombie, accomplish, Comrie, constable, Covent Garden, Coventry and
Montgomery and less frequently in combat, comrade and conduit. The pre-
ceding velar in all of these is presumably not coincidental. Speakers of Dan-
ish, who have difficulty in distinguishing strut and lot, can be encouraged
to round their lips for the latter.
The lot vowel is very back and very open. Although it is rounded in
SSBE (and also in Southern Hemisphere Englishes), it is not rounded in the
dialect of the West of England, Irish English or standard North American
Englishes. This means that lot and start merge in North America. On con-
fusion with strut, see the previous paragraph. A few speakers of standard
English and rather more dialect speakers from the South East of England,
especially London, have [ɔː] instead of [ɒ] in words with a following voiceless
fricative, such as cloth, off and (a)cross and occasionally with a following
nasal, as in gone. Learners should not imitate this very conservative usage.
The foot vowel was, until recently, a back, rounded vowel. In the last
half-century or so, it has tended to become, all round the world, a central-to-
front unrounded vowel, closer in pronunciation to the kit vowel than to the
goose vowel. The conservative pronunciation can still be used; there are few
languages which have anything that approximates to the innovative pronun-
ciation. But anything approximating to a back [u]-vowel sound should be
avoided.

4.2 The long monophthongs

Although the long monophthongs are so called because they used to be pro-
nounced as monophthongs and most of them still can be, most of them
(with the exception of [ɑː]) can also be heard with diphthongal pronuncia-
tions, but the diphthongisation is more likely when the vowel is lengthened
(in word-final position or before a voiced obstruent) and typically involves
fairly restricted tongue or lip movement or both. There are five distinctive
long monophthongs, as set out in Table 4.2.
We can divide these vowels into two subsystems: fleece and goose, which
historically have often worked as a pair, and the others, which are tied up
with non-prevocalic [ɹ]. The nurse vowel arises through the influence of a
following [ɹ] (except in the word colonel, which is homophonous with kernel,
even in North American English). In rhotic varieties of English that have a
nurse vowel (Scottish English does not), there is always a following [ɹ]. His-
torically speaking, the nurse vowel arises through the lengthening of a short
vowel in compensation for the loss of the [ɹ] sound. Scottish English retains
 Vowels 47

TABLE 4.2 The long monophthongs

Vowel Lexical Set Examples

iː fleece key, me, bead, beet, lean, leach, ease

ɜː nurse cur, myrrh, bird, Bert, learn, lurch, errs
ɑː start car, mar, bar, Bart, larch
ɔː thought core, more, boar, bought, lawn, oars
uː goose coo, moo, boo, boot, loon, ooze

the short vowels; North American English has a long nurse vowel. With the
vowels that have been labelled start and thought above, we can see two
distinct types that have merged in modern SSBE. The first, which we could
then call start and force, arise through the deletion of an [ɹ]. The second,
which we could call palm and thought, never had a following [ɹ]. These
subtypes remain distinct in North American English and in SSBE give rise
to the distinction between intrusive and linking [ɹ] (see Sections 3.5 and 8.3).
For some speakers, a long force vowel can sometimes be heard as a diph-
thong (which we might then transcribe as [ɔə] – the [ə] part of this is a rem-
nant of the following [ɹ]). The start vowel has many sources. The obvious
one is that it arises before [ɹ], so that its development is parallel to that of
[ɔː]. But we also find what in SSBE is the same vowel arising in a number of
other places (as in [X1]). These different developments have given rise to
important dialectal differences across English. In Scottish English, there is
no distinction between trap and start.

(X1) Places where [ɑː] is found in SSBE

Environment Examples Notes

before [f], [θ], [s] bath, class, Not in the North of England,
laugh where the vowel stays short,
as in North America
before [n] and another aunt, dance, Not in the North of England,
consonant command North America or Eastern
Australia
before an earlier [l] which is half, calm
now no longer pronounced
in foreign words memoir, sonata, Note the [eɪ] in the North
tomato American pronunciation of
tomato
48 Mostly phonetics

There are a number of words written with <as> where the lengthening of
[a] to [ɑː] never happened, so that ass, gas, lass, mass ‘material, matter’, pas-
senger and (for most, but not all, speakers) plastic retain [a]. Mass ‘Christian
rite’ can be heard with either the long vowel or the short vowel. Similarly,
finance, romance and occasionally telegraph have a short vowel.
The fleece vowel is usually described as a monophthong, but it is unlike
the vowels of French, Spanish, German and so on in that it is not as close
and in that it tends to be slightly diphthongised, perhaps to something like
[ɪi]. In Australian English, the diphthongisation may be really noticeable.
The nurse vowel has a range of possible pronunciations. In SSBE, it has
spread lips, and speakers of French, who tend to use the vowel in heureux for
this sound, may need to unround their lips (smile more). Rounded variants
may be heard in Birmingham, England and in New Zealand, but they do not
transfer well to other varieties. Any mid-central quality will do, but relatively
open varieties are the norm. Speakers of rhotic varieties sometimes think of
this as a long [ɹ] acting as a vowel.
The start vowel is virtually never diphthongised. In SSBE, it can be pro-
nounced a long way back, although if it is too far back or if it becomes
rounded, it will sound South African. More fronted variants are possible but
may tend to sound Australian. If the quality matches that of the strut
vowel, the length distinction must be maintained. Dutch speakers have to
remember that their short vowel is further back than their long vowel, which
is not true of English.
The thought vowel has been gradually becoming closer in SSBE, so that
any back rounded vowel is likely to be in the envelope of permissible varia-
tion. It is never as open as the lot vowel, and it is never close enough to be
confused with [uː]. When it is lengthened because it has no following conso-
nant in the same syllable, it may be diphthongised to [ɔə].
The goose vowel can create problems for learners of English. Although
the symbol used for its transcription, [uː], may make it look as though it
is equivalent to the French vowel in vous, it is now typically a lot further
forward and may sound more like the vowel in tu, particularly in Scottish
and Southern Hemisphere Englishes. Similar problems may arise for
speakers of other languages that have the two vowels, including Danish,
Dutch, Finnish and German. Speakers from languages that have no close
rounded vowel other than [u] may need to relax the articulation of the
English [uː] slightly and, especially, make it less clearly a back vowel. Un-
like the Japanese vowel written with <u> in transliteration, it requires
slightly rounded lips. Many varieties have a slightly diphthongised ver-
sion, in Australian English sometimes as much as [əu] but more usually
something more like [ɨu].
 Vowels 49

4.3 The diphthongs
What defines a diphthong is movement of the tongue or lips or both during
the articulation of what is considered to be a single sound. There are eight
relevant diphthongs in SSBE, as set out in Table 4.3.
The face and the goat vowels act in many ways as a pair, and they can be
discussed together. Although both are diphthongs in SSBE, they are pro-
nounced as monophthongs (of different qualities) in many dialects of the
North of England and in Scottish English. In Standard US English, they are
diphthongs when fully stressed but become monophthongs in places where
stress is reduced. Where they differ is in the range of variation they show.
With the face vowel, the variation is largely in the vertical position of the
onset. A range which is likely to cover the full range of vowels across lan-
guages spelt with an <e> in the Roman alphabet can be found, though some
values have social connotations. Very open varieties tend to be regional in
Britain or indicative of Australian and New Zealand accents. The variation
for the goat vowel is largely on the front–back axis (with some variation in
lip-rounding). A very conservative option has a rounded back onset, which
is similar to the general US variant. A neutral variant in Britain has a central
and unrounded starting point (hence our transcription of [əʊ]), and a more
fronted starting point may sound deliberately refined.
There is some danger of overlap between the face vowel and the price
vowel, with an open starting point for face sounding like a more conserva-
tive starting point for price. This is typical of vernacular London English
and Southern Hemisphere varieties. In such varieties, price typically has a
rather retracted starting point, so that it could be transcribed [ɑɪ]. Dutch
speakers have a vowel which is half-way between face and price and which
L1 speakers may not recognise as either. A raised starting point for price
may sound self-consciously refined.

TABLE 4.3 The diphthongs

Vowel Lexical Set Examples

eɪ face hay, pay, phase, cane, lane, neigh

aɪː price high, pie, guide, kine, line, nigh
ɔɪ choice hoy, poi, coin, loin
əʊ goat hoe, po, goad, foes, cone, loan, know
aʊ mouth how, pow, now
ɪə near here, pier, geared, fears, near
eə square hair, pare, fares, cairn
ʊə cure poor, gourd
50 Mostly phonetics

Both price and mouth typically have central starting positions and are
like front and back versions of each other. In conservative SSBE, the start-
ing point for price is further forward and that for mouth is further back,
but pronunciations with the same starting point for both will sound unex-
ceptional. Some varieties have the back onset for price and the front onset
for mouth (again vernacular London and Southern Hemisphere varieties),
so this has social connotations in Britain. A raised starting point for mouth
sounds lower-class in Britain. Both price and mouth can be followed by [ə]
in a triphthong in older varieties of SSBE, in words like fire [faɪə] and power
[paʊə]. Note that this is called a triphthong because it is a single syllable
with three qualities in it. In rather less standard varieties, words like fire and
power become disyllabic, which is recommended for learners. Alternatively,
a compromise diphthong like [faə] and [paə] will serve. For some speakers,
the vowels in both of these (and occasionally in mouth as well) can all
merge as [ɑː].
The choice vowel is a relatively late addition to the English language,
coming in from French. The transcription [ɔɪ] indicates a slightly old-fash-
ioned pronunciation, and a much closer onset is common.
The near, square and cure vowels are all involved with the problem of
post-vocalic [ɹ]. The cure vowel is increasingly replaced by the thought
vowel (and is, in any case, rare). This does create some extra homophones –
poor and pour (pore) sound the same, and moor and more (and maw) sound
the same – but it can be used to cut down on the learning burden. In New
Zealand English (and occasionally elsewhere) near and square merge, and
in Liverpool square and nurse merge, but neither of these can be recom-
mended for learners. In US English, these three diphthongs are sequences of
short vowels ([ɪ], [e] and [ʊ]) + [ɹ], which is their history). In British varieties,
the difference between fleece and near (and between dress and square)
gets lost before [ɹ] (ferries and fairies may end up sounding the same), and
square is, in any case, increasingly pronounced as a monophthong [ɛː] (as it
is in Australian English). For those for whom the cure vowel is not the same
as the thought vowel, goose and cure may not be distinct before [ɹ] (which
means the monophthong can be used).

4.4 The unstressed vowels

There is a lot of theory involved in any description of the unstressed vowels.
The first piece of theory is that the unstressed (or ‘weak’) vowels are not part
of the same system as the stressed vowels, and even if we find vowels with the
same symbol in both series, they should not be equated. This is hard for
students to understand (and even for linguists to deal with clearly). The sec-
ond theoretical point is that none of these vowels has a unique articulation
 Vowels 51

point across the community. There is a range of possible ways of pronounc-

ing these things, which certainly vary from place to place and can vary from
individual to individual and even within the speech of the same individual
(either over time or from utterance to utterance). The potential range of
unstressed vowels is given in Table 4.4, using the lexical set notation where
that is available and indicating the variation.
These vowels are here termed ‘unstressed’ vowels because they occur only
in unstressed syllables. The alternative label ‘weak’ focuses on the idea that
these vowels are not as forcefully articulated as the others. This is a particu-
lar problem for those many learners for whom the notion of an unstressed
syllable is either incomprehensible or very different from what it means in
English. Even if learners do not abbreviate unstressed vowels in their Eng-
lish, the use of an unstressed vowel in relevant positions will make the Eng-
lish easier for L1 speakers of SSBE.
If we consider the data in Table 4.4 purely logically, the learners could use
a comma vowel for the first two, a fleece vowel for happy and a foot vowel
for the last three (provided that they still use a relatively retracted foot
vowel), and this system would work. But learners are likely to hear other
vowels as well. Learners who have no [ə] will certainly need carefully to prac-
tice this vowel (which can be useful as a hesitation vowel, too).
Part of the trouble with [ə] in English is that it does not correspond to any
specific orthographic representation. Any of the vowel letters can be pro-
nounce as [ə] in some words (not all of the possibilities are given in Table 4.4;
see Section 14.5). Moreover, with any particular morphological element, it is
not necessarily predictable for the typical learner where the [ə] vowels will
fall. In telegraph, the second <e> can be the horses vowel (which means it
can be [ə]), whereas in telegraphy, it is the first <e> which has that value.
While linguists and language teachers may be able to predict this, based on
large amounts of information, it remains obscure to learners until they have
a wide enough range of vocabulary items from which they can draw parallels.

TABLE 4.4 Unstressed vowels

Symbol Lexical Set Overlap with Examples

ə comma strut about, bigger, ten to three, actor,

suffice, cupboard
ɪ horses kit, comma loses, wanted, quality, chocolate
i happy kit, fleece silly, industrious
ʊ foot comma stimulus
u goose, foot influence
o vocalised [l] goose, foot bottle, smile, help
52 Mostly phonetics

This means that students need to learn where the unstressed syllables are as
well as where the stressed syllables are when they learn words with more than
one syllable, and this needs to happen right from the very beginning.
A second problem is that, in SSBE and those varieties that work like it, [ə]
can represent a vowel with a following <r> letter or not. Data and dater, pica
and piker, and vela and velar sound the same, and data and crater end with
same phonetic sound (and are perfect rhymes). There is a story that Ameri-
can visitors to Britain do not understand why a Shoppa bus is so called: they
do not relate shoppa to shopper in the way that the locals do.
The open variant of the comma vowel occurs in word-final or phrase-final
position, so that data in isolation may have a more open vowel than in data
bank. In a few fixed phrases like See you, the you is unstressed and so has the
comma vowel, but it is phrase-final and so has the open variant which is then
perceived as strut or even start. Some Australian varieties have very long
and open vowels in such positions, so that koala may sound – even to L1
speakers – as if it has the same vowel in the last two syllables.

4.5 Comparing vowels across systems

As noted earlier, many languages have only one vowel in the area where both
English fleece and English kit are pronounced. Speakers of these languages
then tend to produce a single vowel sound to cover both. Although the two
differ in length, it is the quality which is crucial for comprehension for L1
speakers.
The same is true for goose and foot, although here the fronting (or diph-
thongisation or both) of goose and the unrounding of foot can be ways of
helping to make the distinction.
The dress and face vowels also differ in length, and the diphthongisation
of face ought to be enough to keep them apart. This suggests that speakers
of languages such as French and Italian may need to be taught to diphthon-
gise face clearly. Unfortunately, the vowel which Italians seem to consider
most comparable to face sounds like price to the ears of L1 speakers of
English, and starting from a vowel such as that in Standard Italian tre and
making it longer may prove a better route to a good pronunciation.
With the monophthongisation of square, dress and square become long
and short variants of each other. The dress vowel has to be kept short and
not diphthongised. Ferry and fairy sound different in SSBE, although they
may merge elsewhere. In US English, Mary, merry and marry may all sound
the same, but in SSBE they are all different, with square, dress and trap
respectively.
The lot and thought vowels are a long way apart phonetically, and there
is a tendency towards making them more different in English, but because
 Vowels 53

few other languages have a rounded open vowel with the quality of lot,
something approximating to thought is often used. Learners must be taught
to open their mouths wide for lot and to keep the length in thought. For
speakers of some languages, such as Spanish, goat also comes into the equa-
tion. Again, the difference between goat and thought seems large, but
Spaniards do not have vowels which differ in similar ways, and they pro-
nounce them all the same way. This leads to lack of comprehension. Con-
sider the data in (X2).

(X2)

lot thought goat

Aus[tralia] oars owes

cock cork coke
col call coal
con corn cone
- form foam
non Norn known
on awn own
pock pork poke
shone shown shorn
Sol Saul sole
stock stalk stoke
was wars woes
wok walk woke

Highlights
While, if we consider English as a world-wide language, there is tremendous
variability in vowel qualities, the skill is in reducing the amount of permis-
sible variability so that students sound reasonably coherent in the type of
English they produce. The short vowels are the most stable and create a
good starting point. The distinction between stressed and unstressed vowels
is vital, and students should be taught the unstressed vowels from the very
beginning. Teaching these as vowels involving little effort or force, or as be-
ing relaxed, is suggested as a way to start. The spelling does not mark un-
stressed vowels, so they have to be seen as part of what students need to
know about words. Most vowels are not stressed but are still pronounced as
full vowels (i.e. not specifically unstressed ones). There is some room for
reducing the number of contrasts students are taught, but this should not
be exaggerated.
54 Mostly phonetics

What should the teacher do?

Teachers need to prepare by thinking what set of vowels they must teach
carefully and which can be either ignored or left for later consideration. The
intersection between vowels and the pronunciation of [ɹ] is important here
since the dropping of non-prevocalic [ɹ] leads to more distinctive vowels.
When teaching vowel sounds in English, teachers need to focus on the qual-
ity of the vowel rather than the length: a short fleece vowel will still sound
like a fleece vowel rather than a kit vowel. Teachers also need to think of
the differences between the vowels of the L1 and English: some can transfer,
many cannot. In the latter case, the teacher needs techniques for achieving
the desired pronunciation (drag it out, snap it off, close your mouth a bit,
swallow it a bit more). In the early stages, perception exercises (perhaps us-
ing minimal pairs or even nonsense words) of the same-different or which-
am-I-saying type may be helpful for pairs which learners find difficult to
distinguish.
5
PROSODY

The label ‘prosody’ is used here to cover word-stress, rhythm, intonation and
voice quality—that is, those parts of pronunciation which are not just con-
sonants and vowels.

5.1 Word-stress
The perception of word-stress in English is based on a number of factors,
including pitch movement, relative pitch, length, loudness, vowel quality
and muscular effort. While the most important of these are those associated
with pitch as far as the perception of first-language (L1) speakers is con-
cerned, that is not necessarily the best way to teach it.
Length and loudness can both be seen as coming from greater muscular
effort, and this is probably how to approach stress in the second-language
(L2) classroom for those speakers who do not have something correspond-
ing to English stress in their L1. I have seen teachers encouraging beginning
students to clap on the stressed syllable or even having students standing in
a boxer’s pose and punching on the stressed syllable. Shouting stressed syl-
lables can illustrate the difference in energy between stressed and unstressed.
In written exercises, students can be encouraged to write stressed syllables in
capitals or to write big blobs over them.
Pitch movement is always perceived as giving greater stress. This is of
particular importance with contrastive stress, stress that implies contrast be-
tween what is said and the actual state of events. For instance, It wasn’t my
brother who crashed the car can be said with low pitch on both my and brother
(in which case it is a statement of fact – it could have been a total stranger),

DOI: 10.4324/9781032637020-7
56 Mostly phonetics

but if there is a pitch fall on my the implication is that it is your brother who
caused the crash, and if there is a pitch fall on the broth of brother, the im-
plication is that it is some other member of my family. L2 learners often have
difficulty with this fundamental use of pitch in English, and whether we
want to call it stress or intonation, it is an important part of speaking and
interpreting spoken English and is worth spending some practice time on.
Full vowel quality is not likely to be a problem for speakers of languages
which do not have specific unstressed vowels, a process also known as vowel
reduction; rather, the problem is getting rid of full vowel quality in un-
stressed syllables. Speakers of languages which do have vowel reduction (e.g.
German and Russian) may have to take care where they put the stress and
where they reduce as well as the quality of the reduced vowels, although the
concept of reduction should be less problematic. One way to attack this is
first to consider [ə] as a hesitation vowel, spoken with minimal effort (but
with no rounded lips and with the mouth in a relaxed, slightly open posi-
tion), and then to begin with disyllabic words like abroad, actor, bottom,
Caesar, collapse, couple, data, lover, master and odour (the variation in the
spelling of the unstressed syllable is deliberate). The unstressed happy vowel,
and also unstressed vowels in trisyllabic or longer words, can be introduced
later. At first, the focus should be on getting the difference in energy between
the two syllables, the difference in loudness, and the use of [ə] in the un-
stressed syllable. Students can be encouraged to ‘mumble’ or ‘swallow’ un-
stressed syllables.
An intermediate stage between stressed and unstressed is where there is a
full vowel but no full effort. This becomes important with tri-syllabic words,
where contrasting patterns can be found, although every word must have
one fully stressed syllable.

(X1) 
Stress patterns in trisyllabic words

Stressed -Unstressed-Unstressed Stressed-Unstressed-Full vowel

amazon, calendar, capital, crockery, Clitheroe, domicile, hesitate, legislate,
generous, influence, measuring macintosh, navigate, paragraph,
parallel
Unstressed-Stressed-Unstressed Full vowel-Unstressed-Stressed
banana, eleven, external, generic, coalesce, entertain, jubilee,
imagine, infrequent, merganser, Portuguese, rerelease, understand
regardless
Full vowel-Stressed-Unstressed
proactive, retsina, uncertain
 Prosody 57

Both disyllabic and trisyllabic words have instances where a syllable with
a full vowel comes next to a stressed syllable. Some examples are presented
in (X2).

(X2)

Stress on the first Stress on the second

Disyllabic arrow, franchise, mobile, bamboo, curtail, pastiche,
ozone unknown
Trisyllabic auspicious, Diana, half-hearted

Words with more than three syllables are rare in the early days of lan-
guage learning and when met can be taught as the addition of one pattern to
another.
In the early stages of learning, the stress has to be taught along with the
word; it has to be part of knowing the word. It is not that there are no rules –
these are dealt with elsewhere (see Chapter 10) – but that students cannot be
shown the generalisations until such time as they have a big enough vocabu-
lary of relevant words. Even in the early stages, with few words longer than
two syllables, care must be taken to include words with second-syllable stress,
as well as the preferred pattern of initial stress, to encourage the idea that
stress in English is variable.
Even for students who, because of the nature of their L1, are predisposed
to give every syllable equal weight, stress can be taught not only as a loss of
effort for the unstressed syllable but in terms of the pitch of the stressed syl-
lable (so that the stressed syllable is typically higher in pitch – though lower
pitch can also work; in Danish, dialects differ as to whether the stressed or
the unstressed syllable is higher in pitch).

5.2 Rhythm
Rhythmic patterns across the languages of the world are roughly classified
into two types: stress-timed and syllable-timed (although the classification
hides a great deal of variation). Although English is generally said to be a
stress-timed language in which the stressed syllables are pronounced on a
regular beat, this cannot be taken too literally. Even if stress-timing remains
a goal of English language teaching, it must also be recalled that many vari-
eties of English are, to a greater or lesser extent, syllable-timed (the syllables
are pronounced on a regular beat) and that as long as the stressed syllable
remains differentiated, this is not necessarily a problem for comprehension.
58 Mostly phonetics

In what follows in this chapter, the stressed syllable will be marked with
the symbol ‘|’ before it, and the phonetic material between the start of one
stressed syllable and the next will be called a ‘foot’. This is derived from, but
not identical with, the use of the term ‘foot’ in scansion.
Teaching rhythm can begin with a simple counting exercise:

|one, |two, |three, |four, |five, |six, |seven, |eight, |nine, |ten

This part can be used to teach either stress-timing or syllable-timing since

there is only one syllable to a foot (except with seven). L1 stress-timing speak-
ers will either make seven a single syllable [sebm] or [sem] or reduce the un-
stressed syllable in [sevən] so that no extra time is taken up. This can then be
extended:

|twenty, |thirty, |forty, |fifty, |sixty, |seventy, |eighty, |ninety

Seventy can again be made to fit with the pattern of the others, by saying
[sebmti] or by making the middle syllable in [sevənti] extra short. (The teens
could be used instead of the tens; either will work.) Students can be encour-
aged to run these two exercises together, keeping the disyllables on the same
beat (with the same timing) as the monosyllables. A metronome can be used,
or students can clap as they say the words, clapping at the beginning of each
foot.
As a next step, simple poems can be used. A start can be made with very
regular poetry, and less regular poetry (with different numbers of unstressed
syllables) can be added as a next step. Nursery rhymes often provide suitable
material.

|Peter, |Peter, |pumpkin |eater

|Had a |wife and |couldn’t |keep her;
|He put |her in a |pumpkin |shell
And |there he |kept her |very |well.

The potential dangers of this type of exercise are illustrated with this
particular example: the form imposes the stresses in the third line (where He
|put her would seem more normal), and the vocabulary is often dated or
awkward (pumpkin shell is not a fixed expression).

|Pease |porridge |hot

|Pease |porridge |cold,
Pease |porridge in the |pot
|Nine |days |old.
 Prosody 59

Pease porridge is no longer a recognised dish, and pease (which used to be

uncountable, as here) is now countable pea / peas. On the other hand, the
rhythm is regular, despite different numbers of unstressed syllables.

|Jack |Sprat could |eat no |fat,

His |wife could |eat no |lean,
And |so be|tween them |both you |see
They |licked the |platter |clean.

As |I was |going |up the |stair

I |met a |man who |wasn’t |there.
He |wasn’t |there a|gain to|day
I |wish that |man would |go a|way.

More serious verse can also be used, but there is something to be said for
starting with simple ideas (including nonsense!) and relatively simple rhyth-
mical patterns.
At a much more advanced level, a piece of dramatic prose can be used –
possibly learned by heart – and what is said between pauses can be kept at
one rhythm even if some of groups are said more slowly than others.

5.3 Intonation
We can think of intonation as being the tune of a breath group (what is said
between pauses). In any breath group, a single syllable is marked out from
the others by pitch movement, occasionally by having a level pitch. The syl-
lable can be called the ‘nuclear syllable,’ and the pitch movement can be
called the ‘nuclear tone’. Nuclear tones can have rising pitch, falling pitch,
level pitch or more complex pitch movements, such as fall and then rise or
rise and then fall. The movement is sometimes spread over two (rarely more)
adjacent syllables. Typically, the nuclear tone (or the first part of it) falls on
the last syllable which has word-stress in the breath group: moving the nu-
clear tone to an earlier syllable will usually imply importance or contrast, as
discussed in the section on stress (Section 5.1), or will imply that material
after the nucleus is already known and thus of little importance. Syllables
that are completely unstressed will either follow the pitch of the surrounding
more stressed syllables or have low pitch.
All languages have intonation. In some languages, such as many African,
Southeast Asian and South American native languages, there are contrastive
tones on individual syllables, and in these languages, the pitch on individual
syllables tends to feel more important than the overall tune. Learners of
English from such backgrounds will need to pay attention to the overall tune
60 Mostly phonetics

and less attention to the relative pitch of the individual syllables. In rough
terms, this means smoothing out the overall pitch, so that the pitch variation
applies to an entire breath group. Conducting the flow of the breath group
with a baton or a pennant can sometimes help students achieve this. Focus
on pitch in the individual syllables will make English sound more staccato
than it usually is, which may make the speaker sound unpleasant, angry or
aggressive.
Some languages are sometimes said to have a ‘flat’ intonation with rela-
tively little pitch movement; others tend to use more variation in pitch. Eng-
lish is somewhere in the middle. Using exaggerated pitch movements tends
to make English sound emotional; using extremely little pitch movement can
make the speaker sound bored. Stereotypically, women use more pitch
movement than men do (but like all stereotypes, there are plenty of excep-
tions), and varieties of English vary in how ‘flat’ or otherwise their intona-
tion patterns are. Americans often feel that British speakers sound more
emotional than they would themselves.
Attempts are often made to correlate the direction of the pitch movement
and the extent of the pitch movement with specific meanings. Within limits,
this can be helpful, but exposure to real usage will show that it is mostly too
simplistic. For instance, a correlation that is often made is that a fall shows
completion while a rise shows that the material is incomplete. Questions are
included in the ‘incomplete’ set since they require an answer to finish the
flow of information. However, questions can have falling intonation (espe-
cially questions beginning with who, what, why, when, where or how) and,
increasingly, statements are heard with rising intonation where the speaker
wants to ensure that the listener is following or to draw attention to a high
point in a narrative. Furthermore, even if rising intonation can show that a
speaker has not yet finished a conversational turn, speakers with high social
power can use falls more often without being interrupted because of their
status. Interrogations in a court of law or by police may therefore often be
heard with falling intonation on questions.

5.4 Voice quality
Voice quality deals with a number of different but semi-permanent settings
that affect the overall sound of the way we speak rather than the individual
speech sounds. Because voice quality is so ubiquitous in our speech – whether
it is a quality which lets other people recognise us as individuals or recognise
us a members of a particular speech community – it is very difficult to
change. Since, however, it can affect people’s perceptions of us and our per-
sonalities, it can be important to think about it in places where it may have
deleterious effects for the speaker. For instance, our stereotypes of Italians
 Prosody 61

as excitable people stem largely from voice quality (relatively loud, plenty of
pitch movement, delivery at a rather rapid pace). Some Chinese speakers can
sound as if they are permanently angry (harsh laryngeal setting, fairly loud
speech, quite clipped syllables, tone perceived as making the presentation
choppy). German and Dutch speakers can sometimes sound rude or bossy
(usually a matter of intonation, with fewer intonational rises than English
speakers would typically use). Many of our perceptions have been moulded
by the media. Examples are the mock Transylvanian accents of early vam-
pire movies, the ‘bad guy’ Russian accents of Cold War movies, or the ro-
mantic associations with some French accents. Needless to say, these
perceptions, however they are arrived at, are not necessarily accurate but are
nevertheless based on features of pronunciation.
The way we use our larynx can lead to differences such as a sexy breathy
voice, a harsh controlling voice, or what is sometimes called ‘creaky voice’ or
‘vocal fry’, which used to be used to express sympathy, or to indicate domi-
nance in males, but today is much more prevalent in the speech of Anglo-
phone women. Some of these settings are used for tones in some Southeast
Asian languages.
Where the tongue is held in the mouth is another feature which can carry
unexpected meanings. For example, holding the tongue near the front inci-
sors can sound childish (and is part of a little-girl voice used by adults),
whereas holding the tongue well back in the mouth (especially if accompa-
nied by a fairly slow delivery) can make the speaker sound rather relaxed and
easy-going.
A generally lax articulation can equally sound easy-going or even lazy,
whereas a very precise articulation can sound controlling, bossy or cross.
How much air escapes through the nose (and how tightly that is related to
the presence of nasal consonants) is often a feature connected with different
varieties of English. Birmingham (England) and Liverpool varieties tend to
have very little airflow through the nose, whereas western movies often char-
acterise their cowboys with quite a lot of airflow. Speakers of Indonesian
often have very little nasalisation in comparison with English speakers, and
speakers of languages in which nasal vowels contrast with non-nasal vowels
(languages like French, Hindi, Nepali and Portuguese) tend to restrict na-
salisation to the affected vowels and not let it spread.
Speed of delivery usually varies from breath group to breath group, so
that it is hard to make generalisations about individual languages being
faster or slower than others (though unfamiliar languages typically sound
faster to us as we struggle to work out what the words are). Extreme
speeds, though, can be disconcerting; fast speech gives an effect of urgency
or excitement, and slow speech conveys the feeling of not paying much
attention at all.
62 Mostly phonetics

A lot of these matters, along with matters of assimilation (see Section 8.1)
and intonation, can be dealt with under fluency training.

Highlights
Although most pronunciation teaching focuses on the individual consonants
and vowels, prosodic features are also important. Because it is less clear that
there is a ‘right’ or ‘wrong’ pronunciation in these regards, partly because
there are no minimal pairs, much of this is hard to teach.

What should the teacher do?

Teachers will probably want to concentrate on those parts of prosody which
are most different from what is found in the students’ L1 and which might,
as a result, be misleading in some way. Some clues on teaching parts of this
are given in the text. Voice quality and intonation may be approached by
thinking about how to sound – kind, friendly, sympathetic, cross, officious,
rude and so on – and then avoiding these when they are not intended.
6
PHONOTACTICS AND SYLLABLES

Syllables are phonetic units between the segment and the word. When we
scan poetry, we count syllables in English. The word word has one syllable (it
is monosyllabic), and the word wordy has two (it is disyllabic). The word
polysyllabic has five syllables. The syllable is a familiar unit in many lan-
guages; many use writing systems that write in syllables, for instance. In or-
der to demarcate syllables, we need to consider phonotactics, the ways in
which individual segments combine within words. Since both of these no-
tions are based on a preliminary notion of sonority, the presentation will
start there.

6.1 Sonority
Sounds which, at the same volume, are easier to hear at a distance are said
to be more sonorous that sounds which do not carry as well. Vowels are
more sonorous than consonants, and in both cases, the more sonorous
sounds are pronounced with less obstruction of the vocal tract. Although
there is some discussion of the uses of sonority and the relative sonority of
particular sounds in the literature, for our purposes the hierarchy of sonor-
ity in Table 6.1 will be all that we need.

6.2 Phonotactics of monosyllabic words

To make things simpler, we will consider the phonotactic make-up of mono-
syllabic words. The assumption will be that if we know how segments can
combine in single syllables, we can deduce how they will combine in multiple

DOI: 10.4324/9781032637020-8
64 Mostly phonetics

TABLE 6.1 Relative sonority of English sounds

Decreasing >
Obstruents

Nasals

Liquids ([ɹ] and [l])

Sonority

Semi-vowels ([j] and [w])

< Increasing
Close vowels

Open vowels

syllables. This works quite well in English, though it creates minor problems
in some other languages.
The overarching principle of phonotactics in English is that sounds are ar-
ranged in monosyllabic words so that the segments decrease in sonority towards
the edges of the words. For example, in a word like smelt, the most sonorous ele-
ment is the vowel [e]. Before the vowel, the [m] is more sonorous than the [s],
which is the least sonorous segment on that side of the vowel. After the [e], the
[l] is more sonorous than the [t], so that, once more, the least sonorous sound is
at the periphery of the word. Of course, that does not mean that words always
begin or end with obstruents: the initial sound in a word can fall anywhere on
the scale of sonority, but there cannot be a less sonorous segment between the
first sound and the most sonorous segment. That is, we cannot have monosyl-
labic words in English like *[ɹdɪs], *[wdel] or *[hasn]. This is not true of all lan-
guages: Russian has a monosyllabic word [rdetʲ] meaning ‘to flush, to turn red’.
But this overarching rule is not sufficient to explain the patterns that are
found (or, equally important, are not found) in English. For example, French
has a monosyllabic word pneu [pnø] ‘tyre’ and Russian has a word [ɡdʲe]
‘where’, neither of which is possible in English. Many Greek loanwords in
English are written with sequences of consonants which were both pro-
nounced in Greek, but one of which is silent in English: bdellium, mnemonic,
pneumatic, psychology, pterodactyl and so on.

6.2.1 Initial consonant clusters

Consonant clusters (i.e. sequences of consonants within the same syllable)
are limited in English, and those that begin words are limited to two-conso-
nant clusters and three-consonant clusters. Although there is overlap, these
will be dealt with separately. Table 6.2 sets out possibilities and limitations.
A few other combinations are found in loanwords but are rare because of
the rarity of the words concerned. The fundamental pattern here, subject to
gaps and occasional extra forms, is that plosives, nasals (that is, all the stops)
 Phonotactics and syllables 65

TABLE 6.2 Initial two-consonant clusters

First position Second Excluded in Examples Comment

position second
position

p, t, k; b, l, r, j, w plosives, plead, pray, pew, [pw] and [bw] are found

d, ɡ fricatives, tree, twin, only in loanwords;
nasals clean, cream, [tj] and [dj] are
queue, queen; missing in some
bleed, bread, varieties; [tl] and [dl]
draw, Gwen are not found.
m, n, l, h j nasals, muse, new, lewd, [lj] is now
liquids, [w] hew old-fashioned.
f, v l, r, j plosives, flew, free, few, [vl] and [vr] are
fricatives, view marginal, occurring
nasals in loanwords like
Vladimir or in vroom
θ ɹ, j, w plosives, three, Thule, [θj] is increasingly rare
fricatives, thwack and has very few
nasals words in which it
might occur.
ʃ ɹ, w plosives, shrimp, schwa [ʃ] makes clusters with
fricatives other consonants in
loanwords such as
schlep, schtick.
s p, t, k, m, spill, still, skill, [sf] is possible in Greek
n, l, j, w smile, snail, words like sphere, [sv]
slat, sue, swim appears in svelte
(which may have [sf]
instead), [sj] is
increasingly rare, as
the [j] is lost, [sɹ] is
sometimes heard in
Sri Lanka, but initial
[ʃɹ] is more common.
z j Zurich [zj] is becoming
old-fashioned; [zl]
can be heard in zloty.

and voiceless fricatives can be followed by the approximants [l, ɹ, j, w]. The
sound [s] is exceptional in allowing many consonants to follow it.
Initial three-consonant clusters are formed by adding [s] to the beginning
of the two-consonant clusters, usually those which start with a voiceless plo-
sive. The possibilities are set out in Table 6.3. Note the lack of [stl] which
follows from the lack of [tl], the lack of [spw] which follows from the lack of
[pw], and the lack of [stw] which is not explicable in this way. The clusters
with [j] are rare because the possible following vowels are so restricted.
66 Mostly phonetics

TABLE 6.3 Initial three-consonant clusters

Cluster Examples Comment

spl splay, spleen

spɹ spray, spring
spj spew
stɹ stray, strife
stj stew
skl Occurs only in sclera and sclerosis
skɹ screw, scruff
skj skew
skw squid, square
slj sleuth This is now nearly always [sluːθ].
smj smew This rare bird name is the only example.
snj This occurs only in non-standard dialectal forms
of snow, namely snew.

Many languages have no or very few consonant clusters at the beginnings

of words, and speakers of such languages will have to pay particular atten-
tion to clusters. For instance, speakers of Thai often simplify two-consonant
clusters in their own language and will need to learn not to do something
similar in English.

6.2.2 Final consonant clusters

Final consonant clusters are made more complex by the fact that a lot of
suffixes are made up of single consonants: final [s] and [z] can be the third-
person singular of the present tense of a verb or the plural or genitive form
of a noun (and some other rare instances such as adverbs), final [t] and [d]
can mark the past tense or the past participle, and final [θ] can mark a
noun (in warmth) or an ordinal number (in ninth). If we ignore this compli-
cating factor, the system is considerably simplified, particularly in the long
clusters.
The other rule to note is that, with a few minor exceptions which are often
overruled, voiced and voiceless obstruents cannot occur in the same cluster.
This leads to some morphophonemic alternation which is discussed in Sec-
tion 9.1. When this is added to the general decrease in sonority as we move
away from the vowel, the limits on clusters are quite strict. Note that because
we are dealing with SSBE, we do not find [ɹ] after the vowel. If we were deal-
ing with American or Scottish or Irish English, an extra set of clusters would
be needed. An overview of two-consonant final clusters excluding suffixes is
given in Table 6.4.
TABLE 6.4 Final two-consonant clusters (no affixes)

First Second consonant

consonant

p t k b d ʧ ʤ f v s z m n

p apt apse
t quartz
k act axe
d adze
m lamp lymph
n want hand lunch lunge pence lens
ŋ sank

Phonotactics and syllables 67

f aft
s clasp last ask
l help colt silk alb field filtch bulge elf valve else film kiln
68 Mostly phonetics

Three-consonant final clusters are usually two-consonant clusters with a

final affix; for example, acts has final suffix -s after the cluster [kt]. Some of
the three-consonant clusters, though they have the same form as one of
these, nevertheless occur without an affix (unaffixed mulct ends in the same
sound sequence as affixed milked; unaffixed prompt has the same final cluster
as affixed limped). (The same is true of some of the two-consonant clusters
such as those in quartz, contrast quarts and adze, compare adds.) Such con-
sonant clusters are often simplified, as discussed in Section 8.2.
Four-consonant clusters are always the same as three-consonant clusters
with extra affixes added. Five consonant clusters occur only in very explicit
pronunciations and always have a four-consonant variant which is more
likely in less formal settings. Examples come in words like sixths ([sɪks(t)θs])
and twelfths ([twelf(t)θs]). Even the four-consonant variants are often sim-
plified, so that sixths may be heard as [sɪkss].
Note that there are generalisations to be made about what can cluster
with what: plosives do not cluster together in initial position and in final
position are usually the result of an affix being added (apt and act are excep-
tional); nasals do not cluster together; fricatives cluster together only when
an [s]-affix is added (although note initial [sf] and [sv] which occur only in
loanwords); no consonant can cluster with itself (identical consonants, or
geminates, may occur together if there is an affix or a compound, as in green-
ness, unneeded or house-style).
There are many languages which have no final consonant clusters. Speak-
ers of these languages will have to pay attention to produce clusters at all, let
alone the wide range of clusters found in English.

6.3 Phonotactics of polysyllabic words

The general assumption when dealing with polysyllabic words is that each
syllable has the same phonotactic patterns that are found in monosyllabic
words. This implies that we can tell where syllables begin and end, and that
will be considered below. This assumption is almost certainly false given the
way in which syllable boundaries are determined. Rather monosyllabic
words should be taken to show maximal clusters in stressed syllables. Un-
stressed syllables tend to have less complex phonotactic structures than
stressed syllables, though this is partly the result of method.

6.4 Syllables
The number of syllables in a word can, on the whole, be determined by the
number of peaks of sonority that are found in the word. If the phonemes in
a word are plotted against the levels of sonority provided in Table 6.1, the
peaks of sonority can be easily seen. Some examples are given in Figure 6.1.
 Phonotactics and syllables 69

FIGURE 6.1 Syllables as peaks of sonority.

There are some problems with this general procedure, but they are not
particularly important. First, this way of operating will show that prism and
spasm are disyllabic words. For many speakers this is true, but for others
these are monosyllabic words. For people in the latter category, it may be
necessary to establish a minimal level of difference between sounds before a
syllable is identified. Another classic problem is provided by the pair hidden
aims and hid names for those speakers who have no [ə] in hidden. Then hidden
aims is [hɪdn eɪmz] and hid names is [hɪd neɪmz], with the same sequence of
consonants and vowels, but three syllables in the first and two in the second.
The moral here is that the system works within the word, not in sequences of
words. Third, the system does not deal well with sequences of vowels, such
as that in reality [rialɪti], where a single peak of sonority corresponds to two
syllables.
Unfortunately, knowing how many syllables there are in a word does not
necessarily tell you where the syllables begin and end. In some languages this
is not difficult; in English it is very difficult. The problem can be illustrated
with the word astride. This is disyllabic, but is it [ə.stɹaɪd], [əs.tɹaɪd] or [əst.
ɹaɪd] (the full stop/period indicates a syllable break)? To some extent, bound
variants are helpful. The [ɹ] in astride is not devoiced (see Section 3.5), so
that the syllable cannot begin with [tɹ] which would lead to a devoiced [ɹ]. To
some extent the morphology is helpful: conscript has the prefix con, and the
syllable boundary follows the morphological boundary, so that we get [kɒn.
skɹɪpt]. This suggests [ə.stɹaɪd] in our first example. The following set of
rules will cope with most of the cases. They have to be applied in order.

(SY1) A full short vowel must be followed by at least one consonant in the
same syllable: happy must be [hap.i], not [ha.pi].
70 Mostly phonetics

(SY2) Assume that syllables follow morphological boundaries where it

makes a difference: breaking is [breɪk.ɪŋ], not [breɪ.kɪŋ].
(SY3) Assign as many consonants to the stressed syllable as possible: this
is another argument for [ə.stɹaɪd] rather than the alternatives and also for
the analysis in (SY2).
(SY4) Single consonants between vowels are assigned to the beginning of
the following syllable rather than to the end of the previous syllable, ex-
cept when already assigned by (SY1): lady is [leɪ.di], not [leɪd.i]. If the
consonant is [ŋ], this cannot hold.

In most cases, this will predict the correct pattern of bound variants as well
as giving an appropriate position for splitting a word over a line-end where
that is required (these days it is mostly done by computer, with occasionally
odd results: textbook should be [tekst.bʊk] and not [teks.tbʊk], for instance).
It should also give rise to a minimum of problems with syllabification. Note
that one of the consequences of the rules for syllabification is that there is
more material in stressed syllables than in unstressed syllables. This goes
some way to explaining why stressed syllables in English take up more time,
and unstressed syllables less time, leading to the rhythmical pattern of
English.

Highlights
Recognising syllables is a matter of recognising levels of sonority, and the
way in which sounds are strung together in words also reflects sonority.
Stringing consonants together at the fringes of syllables is subject to a num-
ber of constraints, so that although English has many consonant clusters
(some of them quite long), the possibilities are severely restricted.

What should the teacher do?

When teaching consonant sounds, teachers should also teach them in clus-
ters. Final clusters that include affixes are particularly important for gram-
matical purposes (marking number, tense and superlatives, for example) and
must be included where relevant. Note that clusters are pronounced more
carefully in slow speech – which is often the speech of learners. The only
reason for teaching students to recognise syllables is to allow them to pro-
nounce longer words one syllable at a time. This is less useful in learning
English than in learning, for example, French, where the rules of syllabifica-
tion are easier to apply.
7
EXTRA ADVICE FOR TEACHERS

Second-language (L2) learners of English, like L2 learners of any language,

are strongly influenced by their first language (L1). Nonetheless, it is danger-
ous to assume that they will speak their L1 in precisely the way set out in
textbooks: people from Lancashire rarely speak English in the way set out in
the textbooks on English pronunciation. Thus, teachers have to observe
what the people they are teaching actually do in their pronunciation, and
while a good description of the learners’ L1 is often extremely helpful, teach-
ers have to be flexible enough to work around deviations from textbook
descriptions.
The most immediate influence from the L1 is that learners will literally
hear the L2 in the categories set down by the L1. Russian learners of English
will hear the two [l]-sounds in lull as belonging to different categories, and
Spanish learners of English will hear [b] and [v] as being ‘the same’. Nearly
all learners of English will have trouble with the English [ɹ], because there is
no corresponding sound in their L1. This is likely to cause particular diffi-
culties with vowel sounds since learners from most language backgrounds
will be familiar with far fewer vowels than are needed in English. People who
are used to only five monophthongs (which is the pattern for most languages
in the world) will have difficulty hearing (and thus in pronouncing) the ten
different monophthongs of English. Where the L1 has more than five vow-
els, it often has front rounded vowels (e.g. Dutch, Finnish, French, German,
Hungarian and Swedish) or back unrounded vowels (e.g. Burmese, Khmer
and Vietnamese) which English does not use, so that the problem is not al-
ways reduced simply because the L1 has a greater number of vowel sounds.
Note the point that, while there are exceptions, most learners will find it hard

DOI: 10.4324/9781032637020-9
72 Mostly phonetics

to produce sounds that they cannot hear, so that getting students to distin-
guish auditorily between unfamiliar sounds is often a preliminary to getting
them to produce them accurately. Exercises with minimal pairs can be useful
in supporting this (although the words in minimal pairs are often rare, so
that some care is required). It should also be noted that it is not always clear
what L1 sound the L2 sound will be taken to resemble. Traditionally, French
speakers use [z] instead of English [ð], whereas Turkish speakers tend to
prefer [d], although the reasons for the difference are not immediately obvi-
ous. Sometimes, a sound to be learned is heard in terms of its components
rather than as a new single sound. English speakers learning Spanish tend to
reproduce Spanish mañana ‘tomorrow’ as [manjɑːnə] with the Spanish pala-
tal nasal [ɲ] replaced by a sequence of nasal and palatal. The French word
debut has the French close front rounded vowel [y] replaced by a sequence of
front and (back) rounded, giving [deɪbjuː] for English speakers. Phonotactic
patterning also plays a role. English speakers learning Māori have difficulty
at first with words like ngaio (tree sp.) that have initial [ŋ]. English has word-
final [ŋ] (in sing, for example) but never word-initial [ŋ], which is therefore
replaced with [n], a familiar sound in that context.
Furthermore, although spelling is an interpretation of the spoken lan-
guage, literate learners are influenced by the spelling of their L1 in pro-
nouncing the L2. For English speakers, a word-initial <s> spells [s], but for
a German, a word-initial <s> spells [z] when followed by a vowel. This leads
to mispronunciations in both directions. It might seem that such influences
would not be felt when switching between the Roman alphabet (in which
English is written) and the kanji of Japanese, the Arabic script, the Hebrew
script, the Hindi script and so on. However, if a particular English sound is
always equated in the L1 script with a symbol that is not pronounced in the
same way as the English sound, then this will add pressure to the tendency
to mispronounce the English sound.
We tend not to notice intonation and voice quality in our own language
and transfer both to a new language, although they may be misleadingly dif-
ferent. Because these are not written and because we have rather poor
vocabulary for talking about them, it can be hard to practice them. We also
have to realise that the idealised forms that are dealt with in textbooks are
usually formal and sometimes unsuited to dialogue. Imitating the accent of
an English learner of one’s own language can be a way to become aware of
the relevant parameters.

7.1 Consonants
In general terms, if you are teaching learners with the same L1, you can
make a list of the consonants of that language and compare it with the list
 Extra advice for teachers 73

of consonants in English (or better still, make a chart like Table 2.2 in
Chapter 2). Lists of consonants and their variants can be found in many
textbooks or by experience of listening to the learners. The Journal of the
International Phonetic Association regularly carries phonetic descriptions of
other languages, including many rare ones.
If the consonants have the same articulation and the same set of variants
depending on the environment and they occur in similar environments, then
transfer from the L1 to English is straightforward. For example, French
learners are unlikely to have problems with English [f]. The teacher may de-
cide that some minor articulatory differences are not important enough to
require teaching and may treat the sounds as though they belong in this
class. For example, Danish [t] may sound close enough to English [t] to be
allowed to transfer, even though the standard Danish [t] is more affricated.
If an English consonant sound occurs only in specific environments in the
L1, breaking the link between the sound and the position will require overt
teaching. For example, Spanish speakers have a [ð] sound, written <d>,
which occurs between sonorants and word-finally, while [ð] in English regu-
larly occurs initially in grammatical words.
If the English sound has variants that the L1 does not have, the variants
will need to be taught overtly. For example, German speakers are likely to
have problems with dark [lɣ].
If there is a sound in the L1 which is written with the same letter of the
alphabet or which is easily associated with an L1 sound in some other re-
spect, but where the articulation is different, overt teaching will be required.
For example, Italian [r] and English [ɹ], both written with the letter <r>, will
cause difficulties for learners.
If English has a consonant which has no equivalent in the L1, overt
teaching will be required. For example, Greek students will need to be taught
[ʧ] overtly.

7.1.1 Plosives
Apart from [h], the obstruents come in voiced and voiceless pairs and can be
taught in this way.
The place of articulation for [p], [t] and [k] and their voiced equivalents is
not likely to be a problem, except that [t] may be made as a dental instead of
an alveolar. If a learner does not have [p] in their L1, start from [m]; if they
do not have [k], teach them to draw the tongue back by saying a sequence of
plosives or ejectives from palatal back towards uvular.
Teach aspiration by starting with [p] (as described just below); the others
can be affricated. Teach [b], [d] and [ɡ] as unaspirated versions of [p], [t] and
[k] when in initial position. In final position, teach students to make the
74 Mostly phonetics

vowel shorter or snappier before a voiceless plosive. Use voicing as a clue

only between vowels.
English has a well-known nonsense rhyme which can be used to practice
aspiration of [p].

Peter Piper picked a peck of pickled peppers

A peck of pickled peppers Peter Piper picked.
But if Peter Piper picked a peck of pickled peppers
Where’s the peck of pickled peppers Peter Piper picked?

A peck is an old measure equivalent to two imperial gallons. Since you

cannot pick peppers that are already pickled, there is a nonsense element to
this little verse. Aspiration, especially with [p], can be taught by holding a
sheet of paper a few centimetres from the mouth and trying to blow the
sheet enough to make it bend with each stressed [p]. This is much harder to
do with [t] and [k].
Any little verse of this kind can also be used to practice rhythm.
There is another verse which, on the surface, allows for the practice of [b]
and which can be used for that purpose. Students need to be aware of the
relatively lax articulation of the [b], meaning amongst other things that it is
not fully voiced.

Betty Botter bought some butter,

‘But,’ she said, ‘This butter’s bitter.
If I put it in my batter
It will make my batter bitter.’
So she bought some better butter,
Better than the bitter butter
And she put it in her batter
And her batter was not bitter.

This rhyme can also be used to practice short vowels: [ɪ], [e], [a], [ʌ] and [ɒ].
Since Botter is not a standard surname, it seems to have been made up to
introduce the [ɒ] vowel. The reason for the rhyme for L1 speakers, however,
is to practice medial [t], which, depending of the dialect of the speaker,
might be replaced by [ɾ] or [ʔ], both seen as socially undesirable in Britain.

7.1.2 Fricatives
Teach fricatives, if these provide difficulty, by telling students to blow through
the consonant or by telling them to make the consonant longer. Do not allow
dental [t] to replace [θ] unless it is part of your chosen model. The fricatives [s]
 Extra advice for teachers 75

and [ʃ] have a groove in the tongue, which is hard to feel. If these are difficult,
suggest trying to whistle. The difference between [s] and [ʃ] is largely a matter
of [ʃ] being further back, but rounding the lips will often have a similar effect.
If [f] and [v] create difficulties, get students to bite their lower lip. Teach [h] as
a voiceless (or whispered) version of the following vowel.
The voiceless fricatives are more energetic than the voiced ones, especially
at the beginnings of words. Word-finally, the length of the preceding vowel
is often the most important perceptual clue for L1 speakers. Voicing should
be used as the major clue only between vowels.
Since the so-called ‘voiced’ and ‘voiceless’ fricatives in final position are
distinguished as much by the length of the preceding material as by the voic-
ing, learners may need to be trained in perception of fricatives in final posi-
tion. The pairs in (X1) can be used as the basis of a number of exercises. The
distinction may be particularly important for speakers of languages like
German and Russian who do not have final voiced fricatives.

(X1)

Voiceless Voiced
brief breve
course cause
gorse gauze
grief grieve
leaf leave
mace maize
mouth (noun) mouth (verb)
peace peas
pence pens
pierce piers
rice rise
ruche rouge
sauce sores
sheaf sheave
sheath sheathe
shelf shelve
staff starve
teeth teethe
wolf’s wolves
wreath wreathe

Since alcohol consumption leads to loss of fine motor control, [s] can be
difficult to say for someone who is drunk. In English, this is stereotypically
76 Mostly phonetics

replaced by [ʃ] in the speech of the inebriated. The following can thus be used
to test for sobriety (though it is quite difficult to say even for the sober) or to
practice the distinction between [s] and [ʃ] (e.g. for speakers of Korean or Japa-
nese for whom [s] and [ʃ] are just bound variants of ‘the same sound’, namely
the same phoneme). Note that there is an assumption that shore and sure are
homophones. Because this is difficult to say, teachers should start with two-
word parts of the rhyme (e.g. sea-shells), concentrating on the frontness of the
[s] and the lip-rounding of the [ʃ] before trying to expand to larger parts.

She sells sea-shells on the sea-shore

The shells she sells are sea-shells, I’m sure.

There is a brief line used for L1 speakers of English to practice saying [h]
(which is often lost in non-standard varieties of British English).

In Hertford, Hereford and Hampshire, hurricanes hardly ever happen.

Hertfordshire [hɑːtfədʃə], Herefordshire [herɪfədʃə] and Hampshire are

names of old English counties.

7.1.3 Sonorants
Sonorants are all voiced by default (although they may have voiceless bound
variants after voiceless plosives). They are all sounds you can sing or keep
going for some time (though [j] and [w] will sound like vowels if extended).
The sounds [m] and [n] are likely to be familiar to most learners. If [ŋ] is
a problem, try having students say [anɡ] (where the [n] will very often be-
come an [ŋ] and then you can drop the [ɡ], as happened historically in Eng-
lish). This will not work with, for example, Poles, who have sequences of [nɡ]
in their own language. Speakers of French and Italian may try to replace [ŋ]
with [ɲ], which does not really work in English.
For [l], start by getting the central contact on the alveolar ridge and, while
keeping that, blow over the edge of the tongue. Then add voice. Alterna-
tively, for the dark variant of [l], start with a vowel like [o] and add tongue
contact with the alveolar ridge. Remember to devoice [l] after [p] and [k] (e.g.
plot and clot).
There are at least three ways of trying to get to English [ɹ]. The first is to
treat it as a vowel, something like [ə], and curl the tip of the tongue up. The
second is to start from [ʒ] and to curl the tip of the tongue up. The third is to
start from [tɹ] (as in tree) or [dɹ] (as in drive), where the place of articulation
of the [t] or [d] is with the tongue turned up on the back part of the alveolar
ridge. In contrast with [l], word-initial [ɹ] should have slightly rounded lips.
 Extra advice for teachers 77

There is a phrase for practicing the sound [ɹ], although it does not show
[ɹ] as part of a cluster.

Round and round the rugged rocks the ragged rascal ran.

Ragged is pronounced [ɹaɡɪd] and means ‘wearing worn-out clothes’. The

[ɹ] should be an approximant (no contact between the tongue and the alveo-
lar ridge area). Learners are unlikely to be able to produce [l] and [ɹ] unless
they can reliably hear them. Pairs like those in (X2) can be used to train
perception. The number of such pairs indicates the importance of distin-
guishing between the sounds. Such pairs can be used as the basis for a range
of class or individual exercises.

(X2)

[ɹ] [l]
crowded clouded
free flea
mirror miller
prate plate
pray play
rain lane
raj large
read (infinitive) lead (go ahead)
red lead (metal)
right light
road load
row (a boat) low
royal loyal
rue loo
sprat splat
wrong long

Although some English speakers use [ʋ] instead of [ɹ], it is viewed either
as defective or as pretentious in England, and the articulation is not notice-
ably easier to master. It is also often confused with [w] in English. Some
rounding of the lips is important for English [ɹ].
The sounds [j] and [w] can be treated as quickly articulated vowels (either
vowels like [i] and [u] or vowels like [e] and [o]). In Italian, uomo (‘man’) is
pronounced with initial [w], spelt <u>, and Spanish anchoa (‘anchovy’) has
[w] spelt as <o>. Similar phenomena are found in other languages. In some
Southeast Asian languages, the [j] or [w] is considered to be part of the vowel
78 Mostly phonetics

rather than to be a consonant, but the link with any preceding voiceless stop
is important in English.
Although there is no well-known phrase or rhyme to practice [w], teach-
ers can invent their own easily enough. For instance:

Waiting for William to wash the queen’s hair in warm water.

Similarly, where [v] and [w] need to be contrasted, we can invent something
like

Very well, we will put the vaccine in a vacuum flask, and wash the vessel
with water vapour.

7.2 Vowels
Where vowels are concerned, you cannot rely on the symbols that are used in
a phonetic transcription of the learner’s language to provide a guide to the
equivalence or non-equivalence of the pronunciation of the learner’s lan-
guage and English. There are many reasons for this, which are of minor im-
portance in this context. All we need to know is that the same transcription
symbol does not automatically imply the same sound. We can see that if we
compare various transcriptions of English, as was seen in Section 2.4. Any of
these transcription systems would provide a valid transcription within the
framework of the International Phonetic Association. The fact that the same
vowel can have so many alternative transcriptions (although one cannot be
substituted for another across systems) makes the point that the symbol
alone is not sufficient to indicate an accurate quality of pronunciation.
A better guide to pronunciation can be obtained by considering phoneti-
cians’ descriptions of the languages involved, with the use of vowel charts to
indicate the vowel qualities. Even these should show a rather large area of
pronunciations that are possible in both languages, and the variation be-
tween the way the vowel space is filled up in the two languages may not be
the same. Also, relatively small differences in lip position can alter the sound
for vowels in the same area of the vowel chart. Such descriptions can be use-
ful as first approximations if read carefully, but they will still need some in-
terpretation. They can give some useful information, though. For example,
they can show that Japanese and Turkish have vowels which do not appear
in English. This should not be a problem for the English teacher unless the
spelling leads learners from those backgrounds to misinterpret the English
spelling system. They can also show when English has a vowel in part of
vowel space which has no near equivalent in the learner’s language. An
 Extra advice for teachers 79

example might be English [ɪ]. Such vowels will always require particular care
in teaching.
What all this means is that the teacher will have to determine what is or is
not an acceptable vowel quality for a given vowel in the learner’s English.
This is trickier than it sounds. If teachers have been teaching English in the
country where the learners come from, they will have become accustomed to
the accent of that group of people and will be more likely to accept margin-
ally valid pronunciations than people without experience of that particular
set of learners’ typical errors. It is therefore always a good idea to be more
conservative in the pronunciation demands made on the students than the
teacher feels is absolutely necessary.
Because our abilities as imitators are limited and because English has so
many different vowel sounds, getting a new vowel sound right can be diffi-
cult. One technique is to glide from a vowel sound on one side of the target to
a vowel sound on the other. To get to [ɪ], glide from [i] to [ə], for example; to
get [ʌ], glide from [a] to [ɑː]. The teacher has to stop the student at the appro-
priate point along the trajectory. Where lip-rounding is concerned, a mirror
can be used to see the change in lip position. To produce [ɒ], start from [ɑː]
and round it. The difficulty is to avoid changing tongue position along with
lip position, or vice versa. The starting positions for the glide or for the change
of lip position can come from the target language or from the students’ L1.
To get more open or more closed vowel qualities, tell students to open or close
their mouths. To avoid a very close [iː] (as is found in French or Spanish), tell
them to smile less; for [ɜː], tell them to smile more if they produce a rounded
[ø] or [œ] type of vowel (as French or German students are likely to do). For
diphthongs, slow down the glide process between two points. The end point is
rarely crucial in English (though it may affect the impression given: Northern
English accents often have a vowel like [aɛ] in place of [aɪ], for example), but
get the starting point, and the direction of the glide, as right as possible. Make
the point that movement is required in diphthongs.
Teachers can always invent sentences or rhymes that allow students to
practice individual vowel sounds or contrasts between vowel sounds. For
example, to practice the foot vowel, we could invent a sentence such as the
following:

A good butcher should have a good woman to put sugar in his pudding.

To distinguish between the strut vowel and the lot vowel, we could come up
with something like:

On Monday in London and Gloucester the monkey and the donkey had
nothing to eat.
80 Mostly phonetics

As it happens, English has a set of such sayings to allow practice of indi-

vidual vowels, usually the ones that are socially salient in England, and they
can be pressed into use by teachers. For a rhyme to practice the short vowels,
see the rhyme about Betty Botter in the discussion of plosives (Section 7.1.1).
Other well-known instances are set out in (X3).

(X3)

Face: The rain in Spain stays mainly in the plain.

Foot and strut How much wood would a woodchuck chuck if a
woodchuck could chuck wood?
Horses and happy Moses supposes his toeses are roses.
But Moses supposes erroneously.
For nobody’s toeses are posies of roses.
As Moses supposes his toeses to be.a
Mouth: How now, brown cow.
Note:
a
Apart from the fact that toeses is not the correct plural of toe, but a joke form, this rhyme
was obviously created for speakers for whom the happy vowel and the horses vowel were
the same. There are few such people left. This rhyme can also be used in an exercise to
make the students laugh. The stressed vowel [əʊ] can be replaced by any other stressed
vowel of English, and the rhyme repeated. This gives rise to a lot of nonsense words but
allows repetition of a range of distinctive sounds in non-meaningful environments.

These days, there are computer programs which allow students to match
their vowels to those of L1 speakers. These have value in allowing students
to practice without the presence of a teacher, but it is unrealistic to expect
students to copy a vowel sound accurately unless they can perceive that
sound (it can happen, but it is not good pedagogy). Students should thus be
trained to perceive unfamiliar vowel sounds before they are trained to pro-
duce them. Also note that some accommodation may be required to take
account of the size of the vocal tract of the student.

7.3 Connected speech
Once students get beyond learning the individual sounds, there are many
exercises for developing further aspects of pronunciation.

1 Develop consonant clusters (or sequences) (see Chapter 6). Practise ini-
tial clusters before final clusters. Practise perception before production.
2 Get students to perceive sounds (or stresses). You can use nonsense words
(some examples are given as an appendix to this chapter) so that they are
not misled by thinking they must know the words. There are various
 Extra advice for teachers 81

possible exercises, such as getting students to say which words begin (or
end) with the same sound and which end with a different sound or to say
which two of three words are the same and which are different. Say one
word (or play a recording of a speaker saying it – this allows for a genuine
repetition if students fail to hear the correct sound) from a set of five
which might be confused. Use the same set of words again, producing a
different word from each list.
3 With persistent problems of pronunciation of individual sounds, talk
about the way the sound is pronounced, draw a diagram of the position
of the tongue, and contrast with the sound the student is actually making.
Talk about the changes you need before getting the student to produce
the sounds. Practise difficult sounds in nonsense words; later try minimal
pairs (perception before production).
4 Learn stress and intonation by taking a short sentence and lengthening it
one word (or two words) at a time: The teacher arrived – The teacher ar-
rived in class – The teacher arrived in class late – The teacher has arrived in
class late; The boys couldn’t see – The boys couldn’t see me – The little boys
couldn’t see me – The little boys couldn’t see me behind them – The little
boys couldn’t see me behind them in the garden.
5 Learn contrastive stress by getting students to provide the question to
which a sentence like I didn’t know that your sister liked strawberry ice-
cream might be the answer with different words getting the tonic syllable.
How might the person continue (e.g. I didn’t know your sister liked
STRAWberry ice-cream, I thought she only liked chocolate)?
6 Teach intonation patterns by meaning, such as intonation in lists, intona-
tion when doubtful, intonation when enthusiastic, and intonation when
using a name to attract attention or direct a message.

7.4 Fluency
Fluency is a matter of speaking without unnecessary breaks for thought
(e.g. to find the right word and to compute the syntactic pattern). The pro-
duction does not have to be fast (though it may be), but it has to be smooth.
Like all skills, this improves with practice. We hear three-year-old children
speaking without fluency, which they typically achieve by the age of five.
One way to practice this is to arrange the class into two concentric circles,
so that each student is facing a partner. One student tells their partner some-
thing on a prearranged topic (what I did last week, what happened in a film
I have seen, what I want to be after I finish my education, etc.). The prepared
narration should be short – no more than two minutes. Then the students
move on to meet a new partner and tell the new partner the same thing.
As they move on, the narration should become more fluent as it becomes
82 Mostly phonetics

more familiar. The teacher can let the class run or can intervene to correct
intonation, stress and assimilations. The teacher can also shorten the time
allowed on each repetition of the narration.
Teachers can also prepare dialogues of varying length and get students to
learn a role in the dialogue. Some examples of different types are given be-
low for illustration. In the first example, there are variants with different
degrees of formality. Students can also write their own or converse without
a script. Students and teachers can also pay attention to phrases that allow
time for thought (in my opinion, I wondered whether, the thing is, it depends
on, and so on), to politeness formulae and to words which allow students to
skip gaps in their vocabulary (thingy, what’s it, and whatchemacallit).

Good morning! Hello! Hi!

How are you today? How’s things? How are you doing?
I’m fine, thank you. Fine, thanks I’m good.
Are you going to school? Going to school? You heading my way?
Yes. You? Yeah. Sure. If you’re going to school.
Can I give you a lift? Do you want a lift? Jump in!
That would be great. Thanks! Yes, please. Thanks a lot.

A: Good afternoon, sir. What can I get you?

B: A cappuccino, please. (To C) What are you having?
C: I’d like a green tea, please. And do you have any cakes?
A: There’s a selection in the cabinet, if you’d like to make a choice.
B: I’ll have a look, too.
A: I’ll bring the drinks right over.
A: Have you done the history homework?
B: Yes. Well, most of it.
A: What did you put for the question on the causes of the war?
B: I just said that I thought that Archduke Ferdinand was just an excuse,
not the real cause.
A: That’s good. So what was the real cause, then?
B: That’s why I haven’t finished. The real cause was too complicated.
A: Why do you want this job?
B: For the time being, until my English is better, I don’t think I can do a
job where I face the public. But I need a full-time job to pay the rent
where I live. I’m healthy and strong, which you ask for, so this job looks
like a good fit for me.
A: Do you have references from previous jobs?
B: I have only been in this country for a short time, and I have been learning
the language. I can give you references from people in my home country.
 Extra advice for teachers 83

A: You seem to be very well educated for a job of this kind. You realise we
are only paying the minimum wage?
B: Well, I wouldn’t look for a job like this if I were somewhere where my
qualifications were recognised. You will find that I am a quick learner,
and highly motivated. And I am keen to have a job that suits my level of
English.
A: How soon would you be able to start if we offered you the job?
B: As soon as you like. Monday? Tomorrow if you prefer.

The use of literary texts remains a possibility (citing poems or speeches

from plays), but such texts are usually too complex for learners before rather
advanced levels, and care has to be taken that the vocabulary and syntax of
the passages are modern enough to provide a useful model.

7.5 Some statistics
Various counts were carried out on the most frequent 1000 words in the
Corpus of Contemporary American English, lists of which are easily ob-
tainable. Some care is required, however. The corpus lists words of different
parts of speech separately, so that answer is listed both as a noun and as a
verb, even though they are pronounced the same way. Use is also listed as a
noun and a verb, but this time there is a difference in pronunciation. A pair
like present (noun and verb), were they both listed, would differ in stress as
well as in the vowels.
Of the first 1000 most frequent words of English in this list, 529 are mono-
syllables. Of these, 321 contain long vowels or diphthongs. Of these long
vowels and diphthongs, 71 are long by virtue of having an earlier [ɹ] which is
present in the spelling (e.g. park and shear); 69 have the length shown by hav-
ing a vowel, a single consonant and then an <e> (e.g. make); only 19 have the
length shown by a double vowel letter (e.g. meet), which is less than half the
number that use a digraph to indicate vowel length.
Of the disyllables, 219 have initial stress, 111 have second-syllable stress
(and one, ok, has variable stress). Final stress is often predictable by means
of some other rule, such as the presence of a prefix (e.g. ago and remove).
Reflexive pronouns (myself and so on) are an interesting sub-class with final
stress. Of the words with three syllables, 59 have initial stress, 37 have sec-
ond-syllable stress, and only three have stress on the last syllable (employee,
represent and understand). Four- and five-syllable words are much rarer in
the frequent word-list, and more variable, though initial stress is rare (mili-
tary). Very often in these longer words, stress can be predicted by suffixes
(e.g. words in -ation).
84 Mostly phonetics

Appendix: Some nonsense words

Sometimes the teacher may prefer not to use real words of English (where
the meaning may interfere with concentration on the pronunciation) but to
use nonsense words instead. (X4) provides some monosyllabic nonsense
words, which can be used as seems appropriate.

(X4) Monosyllabic nonsense words

berb [bɜːb]
bloss [blɒs]
chold [ʧəʊld]
coove [kuːv]
croob [kruːb]
cupe [kjuːp]
curɡ [kɜːɡ]
eents [iːnts]
fadge [faʤ]
fet [fet]
firch [fɜːʧ]
flasp [flɑːsp]
froy [fɹɔɪ]
fusk [fʌsk]
gatch [ɡaʧ]
groosh [ɡɹuːʃ]
hane [heɪn]
harch [hɑːʧ]
hulm [hʌlm]
jast [ʤast]
laitch [leɪʧ]
lote [ləʊt]
loze [ləʊz]
lurm [lɜːm]
mang [maŋ]
mose [məʊz]
nair [neə]
nilt [nɪlt]
nise [naɪz]
nish [nɪʃ]
parge [pɑːʤ]
peth [peθ]
preef [pɹiːf]
reen [ɹiːn]
rog [ɹɒɡ]
 Extra advice for teachers 85

roke [ɹəʊk]
sadge [saʤ]
shathe [ʃeɪð]
slong [slɒŋ]
smine [smaɪn]
spink [spɪŋk]
thade [θeɪd]
thook [θʊk]
trell [tɹel]
verg [vɜːɡ]
vore [vɔː]
wibe [waɪb]
yuzz [jʌz]
zere [zɪə]
zonch [zɒnʧ]

These nonsense words (or pseudo-words) are not real words of English,
but the spelling-to-pronunciation link is regular. Depending on the purpose,
teachers may prefer to work with the spelling or the pronunciation. Because
of the regular spelling, some sounds are not represented. Thook provides an
interesting case, as it is one of the few words illustrating [ʊ]. Nearly all words
that end in ook in English contain [ʊ]ː book, cook, crook, nook, rook, shook,
took and so on. Spook (a Dutch loanword), snooker and, for some speakers,
snook (either a fish or a rude gesture) are the exceptions, and none of them
is very common. The pronunciation [θuːk] for thook would thus be possible
but unlikely. The sound [ʒ] does not appear in any of these words, because it
cannot be unambiguously spelt in monosyllabic words (where it hardly ever
occurs).
The nonsense words in (X5) are all disyllables stressed on the first syllable.
In many cases, the form of the word demands the initial stress.

(X5) 
Initially stressed nonsense disyllables

blisure [ˈblaɪʒə]
chiffing [ˈʧɪfɪŋ]
cortin [ˈkɔːtɪn]
creasure [ˈkɹeʒə]
fennow [ˈfenəʊ]
hambered [ˈhambəd]
iffling [ˈɪflɪŋ]
larted [ˈlɑːtɪd]
midger [ˈmɪʤə]
86 Mostly phonetics

mishless [ˈmɪʃlɪs]
muning [ˈmjuːnɪŋ]
mushle [ˈmʌʃəl]
neever [ˈniːvə]
plinner [ˈplɪnə]
satre [ˈseɪtə]
scussy [ˈskʌsi]
skimmy [ˈskɪmi]
thunkure [ˈθʌŋkjə]
toiple [ˈtɔɪpəl]
torrie [ˈtɒɹi]
wampry [ˈwɒmpɹi]
yishes [ˈjɪʃɪz]

Finally, in (X6), disyllabic nonsense words whose form demands stress on

the second syllable are presented.

(X6) 
Disyllabic nonsense words with second-syllable stress

anoled [əˈnəʊld]
beglines [bɪˈɡlaɪnz]
caheave [kəˈhiːv]
condrew [kənˈdɹuː]
despain [dɪˈspeɪn]
dismote [dɪsˈməʊt]
inflort [ɪnˈflɔːt]
percede [pəˈsiːd]
reclude [ɹɪˈkluːd]
unpesht [ʌnˈpeʃt]
PART III

Phonology
8
SOUND CHANGES INFLUENCED
BY ADJACENT SOUNDS

This chapter is mainly concerned with making the transitions between

words easier to pronounce by making adjustments to pronunciation at the
boundaries.

8.1 Assimilation
In assimilation, one consonant sound is modified to be more like an adjacent
sound (typically another consonant sound). Some scholars distinguish bet­
ween similitude (variation in the phonetic nature of sounds that arise within
a word and are fixed) and assimilation (changes in the phonetic nature of a
sound that arise when words are juxtaposed). Despite the many commonali-
ties between the two, this distinction is maintained here. Similitude can be
learnt as the way in which a particular word is pronounced; assimilation is
more likely to need overt teaching. This raises some questions of coverage
when affixes are added to words or when compound words are made: is the
distinction between [n] and [m] entrain and emplane a matter of similitude or
assimilation? Is the difference between [t] in hatstand and the frequent use of
[p] in hatband ([hapband]) a matter of similitude or assimilation? Despite
problems, these will mostly be treated as instances of similitude.
Where assimilation occurs, the sound that is changed by assimilation may
not sound identical to a normal version of the sound it becomes: there may
be small phonetic differences. Nevertheless, where assimilation changes one
distinctive sound to another, it is clear that an important change has oc-
curred, even if its phonetic nature is not obvious at first hearing. This may

DOI: 10.4324/9781032637020-11
90 Phonology

mean that assimilation is better taught as a matter of interpreting what

speakers say rather than as a recipe for speaking.

8.1.1 Assimilation of voice
In some languages, voiceless segments can become voiced next to a voiced
segment. This does not happen in English, where the rule is that voicing as-
similation leads to an increase in voicelessness.
There is one possible exception, which applies only in those varieties of
English that have a tapped [ɾ] between vowels.

(A1) 
Word-final [t] after a vowel can become [ɾ] before a vowel (e.g. get
eggs [ɡeɾeɡz]).

In some British English dialects, the [t] may become a normal [ɹ], not a
tap (e.g. get off [ɡeɹɒf]).
This rule is needed only for advanced-level students.
Plosives tend to keep their voicing, but word-final voiced fricatives may
become voiceless when the next word begins with a voiceless consonant.
This happens in fluent speech and usually where the two words are closely
linked syntactically. This close syntactic link will be a recurrent pattern in
this chapter; it means that the two words that are adjacent are frequently
collocated and can be viewed as a single syntactic chunk.

(A2) 
A word-final voiced fricative may become voiceless before a
closely linked voiceless consonant (e.g. have to /haftə/ and of
course /əf kɔːs/).

This rule can be treated as part of learning a few fixed phrases rather than
as a general rule. It is of value for intermediate students.
Where words like has and is are reduced to a fricative, the fricative follows
the same rules that govern the voicing of the plural, third-person singular
and genitive forms (see Section 9.1).

(A3) 
When is or has is reduced to just a fricative, the fricative agrees in
voicing with the final sound before it (e.g. she’s here /ʃiːz hɪə/ and
the cat’s gone /ðə kats ɡɒn/).

This rule can be taught along with the reduction of is and has. If students
have internalised the word-internal pronunciation of, for instance, the plu-
ral, this should not cause any difficulties.
 Influence of adjacent sounds 91

8.1.2 Assimilation of place
Assimilation of place in English is perhaps better thought of as the instabil-
ity of final alveolar consonants. Of these consonants, the nasal is the most
unstable and the fricatives are the most stable, so that this can be taught as
rules applying to final [n] and rules applying to final [t, d]. Because of the
instability of final [t] in modern English anyway (see Sections 3.1 and 3.4),
final [t] may be replaced by a glottal stop [ʔ] (so-called ‘glottal replacement’)
or accompanied by a glottal stop [ʔ͡t] (so-called ‘glottal reinforcement’). This
is not discussed here.

(A4) 
A word-final [n] takes on the place of articulation of a following
consonant in a closely linked phrase (e.g. gone back [ɡɒm bak] and
been quick [biːŋ kwɪk]).

Note that [ŋ] is possible here, even though it does not occur after long
vowels in isolated words. This rule can be taught as part of fluency training
from intermediate levels onwards.
In the examples given above, the change of place of articulation gives rise
to a different distinctive sound, but the change of place occurs even where
there is a non-distinctive change in pronunciation. Since [h] is, phonetically
speaking, a voiceless vowel, and the final [n] is not affected by a following
vowel, no change occurs before [h]. Examples are gone home [ɡɒn həʊm],
seen Fred [siːɱ fred], been jolly [biːɲ ʤɒli] and in the [ɪn̪ ðə].

(A5) 
Word-final [t] and [d] take on the place of articulation of a
following consonant in a closely linked phrase (e.g. that man [ðap
man], bad boy [bab bɔɪ], at Christmas [ək kɹɪsməs] and good girl
[ɡʊɡ ɡɜːl]).

As with the nasal discussed under (A4), new bound variants may arise
here, as in that thing [ðæt̪ θɪŋ] and bad joke [bad̠ ʤəʊk]. These changes are
never obligatory, but fluent spoken English without them will sound overly
meticulous.

(A6) 
[s] and [z] tend to become post-alveolar before a post-alveolar
consonant or [j] (e.g. as usual [əʒ juːʒʊəl], this shop [ðɪʃ ʃɒp] and
these chips [ðiːʒ ʧɪps]).

This assimilation is used less than those cited above and may need to be
taught only for listening and not for production.
92 Phonology

(A7) 
Rarely, [θ] and [ð] may become [s] and [z] respectively after [s] or
[z] (e.g. has the [haz zə] and what’s this [wɒts zɪs]).

This need never be taught but may be observed in fluent speech. In such
cases, the [θ] or [ð] may be elided completely or remain in some extra length
on the alveolar fricative.

8.1.3 Assimilation of manner
Assimilation of manner is rare in English but does occur in a few places,
mainly where nasals are involved.
Rarely, a voiced word-final plosive followed by a nasal in the following
word may become a nasal. The new nasals will take on the relevant place of
articulation determined by (A4). Because [b] and [ɡ] are more stable than [d],
this applies exclusively to [d].

(A8) 
Word-final [d] may become a nasal before a following nasal (e.g.
good morning [ɡʊm mɔːnɪŋ]).

This so rare that it can be thought of as restricted to the phrase good

morning and taught as a fluent way of pronouncing that phrase. Even there,
it is by no means obligatory.
Occasionally, a fricative can become a nasal adjacent to another nasal.
This happens mainly where the fricative is in an unstressed syllable.

(A9) 
A voiced fricative may become a corresponding nasal adjacent
to another nasal (e.g. have mine [ham maɪn] and in there [ɪn
neə]).

This is never required and occurs in very few phrases. It is more useful for
understanding than for production. It will sound wrong if the new nasal is
pronounced too precisely.

8.2 Deletion
Deletion occurs in order to make something easier to say. This means that it
does not occur in the most explicit and careful pronunciation; it occurs only
in fluent speech, especially when the style is slightly relaxed. The implication
of this for the teacher is that deletion processes do not need to be taught at
elementary levels but are part of teaching fluency. Deletion processes may
also occur within words, but such examples are not discussed here.
 Influence of adjacent sounds 93

(D1) 
If a sequence of three or more consonants arises over a word
boundary, delete the least sonorous provided that (a) it is between
consonants and (b) it is not a word-initial consonant, (c) that the
third consonant is not initial [h] and (d) that the deleted consonant
is an obstruent. If there is a choice between a plosive and a
fricative, delete the plosive. Examples are can’t believe [kɑːn bɪliːv]
(this may become [kɑːm bɪliːv] by assimilation), last night [lɑːs
naɪt], asked them [ɑːst ðəm] (note that the [k] rather than the [t] is
deleted here), last spoke [lɑːs spəʊk] and toast bread [təʊs bɹed].
Note also the lack of deletion in East Ham [iːst ham], turn black
[tɜːn blak] and caused havoc [kɔːzd havək].

Note that in most of these examples, it is [t] or [d] which is deleted, so this
phenomenon can be seen as part of the instability of final alveolars. One
result of this is that a final [t] or [d] in a past tense or past participle form is
often deleted, which can make it sound as though there is a base-form verb.
This is why what used to be written mashed potato is now often written mash
potato. Other examples are grabbed some [ɡɹab səm], reached Rome [ɹiːʧ
ɹəʊm] and finished product [fɪnɪʃ pɹɒdʌkt].
While the deletion of a consonant to simplify a three-consonant sequence
is widespread, deletion of one of the consonants in a two-consonant se-
quence is also found in casual speech. This is most obvious in a few fixed
expressions, and it might be taught as a matter of fluency with those expres-
sions or as a matter of understanding natural spoken English. Note that
some of these fixed expressions have a normalised spelling for the version
with deletion. Examples are provided in (X1) below.

(X1)

got to gotta [ɡɒtə] or [ɡɒdə]

let go leggo [leˈɡəʊ]
should have [ʃʊdə]
want to wanna [wɒnə]
would have [wʊdə]

In most British dialects, [h] is never used. In some, [h] is found only where
it does not occur in Standard Southern British English (SSBE): thus, a name
like Harry Awdry may be pronounced [aɹi hɔːdɹi]. However, SSBE retains
[h], at least in stressed syllables.
94 Phonology

(D2) 
Initial [h] is often deleted in grammatical words which are
pronounced without stress (e.g. that’s him [ðats ɪm]; I gave her a
present [aɪ ɡeɪv ə ɹ ə pɹeznt], which could also represent I gave her
her present; and why did he do it [waɪ dɪd i duː ɪt]).

This phenomenon can be considered as giving rise to weak forms of pro-

nouns rather than deletion, but these particular weak forms arise through
deletion.

(D3) 
The words is and has are often reduced to just a fricative, which
agrees in voicing with the previous sound (e.g. What is that? [wɒt
s ðat] and Mum has gone out [mʌm z ɡɒn aʊt]).

Again, this form of deletion can be seen as forming weak forms, and this
rule can be used even by elementary learners. This rule was covered as as-
similation in (A3).
A [ə] is often deleted before any of [ɹ, l, n] and rarely before [m]. In SSBE,
any such [ɹ] has to be followed by a vowel; otherwise, no [ɹ] would be present.
In some places, this is shown in the spelling; in other cases, it is not. Where
it is not shown by the spelling, it is typically variable rather than obligatory.
This is treated as deletion here, where the [ə] is assumed to be present in a
clear pronunciation (simply because this is often easier for second-language
[L2] learners), but because it happens so frequently for many SSBE speakers,
some sources treat this as [ə]-insertion.

(X2) [ə] deletion

Before [ɹ]: enter/entry [entə]/ [entɹi], February [febjuəɹi] ~ [febjʊrɪ],
temporal [tempəɹəl] ~ [tempɹəl] ~ [tempɹl], and temperature
[tempəɹəʧə] ~ [tempɹəʧə]
Before [l]: arrival [əɹaɪvəl] ~ [əɹaɪvl], assemble/assembly [əsembəl]/
[əsembli], detectable/detectably [dɪtektəbəl] ~ [dɪtektəbl] /
[dɪtektəbli] (and other such pairs), and environmental
[ɪnvaɪɹənmentəl] ~ [ɪnvaɪɹənmentl] (and other such words)
Before [n]: puritan [pjʊəɹɪtən] ~ [pjʊəɹɪtn], solution [səluːʃən] ~
[səluːʃn] (and other such words), and sultan [sʌltən] ~ [sʌltn]
Before [m]: prism [pɹɪzəm] ~ [pɹɪzm] and spasm [spazəm] ~ [spazm]
The rule does not apply when there is a full vowel immediately
before the /ə/: protozoan /pɹəʊtəzəʊən/.

We now turn to instances where the deletion seems to involve an entire

unit, not just an individual sound. In many of these cases, an alternative
 Influence of adjacent sounds 95

analysis is possible where the deletion is the result of phonetic processes or

where the so-called deleted material was never there in the first place. We will
look at two cases, the first where random material is apparently deleted and
the second where apparent affixes are deleted. We will consider neither clip-
ping (e.g. flu from influenza) nor back-formation (e.g. surveille from surveil-
lance) where the deletion is the mark of a particular word-formation
process.
One of the simplest kinds of deletion arises where processes of word-
formation are expected to bring two identical sounds together and only one
of them remains.

(D4) 
Geminate (or double) consonants and vowels do not occur inside
a morpheme (base or affix) (e.g. pattern [patən] and coffee [kɒfi])
and are unusual at morpheme boundaries.

The last two words of (D4) require some further clarification. Prefixes
fairly reliably allow geminates (e.g. misspell [mɪsspel], post-tonic [pəʊsttɒnɪk]
and unnatural [ʌnnaʧərəl]), but even here some simplification occurs (e.g.
dissimulation [dɪsɪmjʊleɪʃən] and other words with dis‑). With suffixes, the
rules are less clear, but frequency seems to be important: sequences of [nn]
can occur where the second [n] is initial in the suffix ‑ness (suddenness), and
sequences of [ll] occur with ‑less and ‑like; the suffix ‑ly tends to function in
the same way after a stressed syllable (e.g. drolly [drəʊlli] and solely [səʊlli]),
but gemination is variable in wholly [həʊlli] ~ [həʊli] and unusual in fully
[fʊli]. In the sequence ‑ally, there is no gemination (chemically [kemɪkəli]).
Other suffixes may show gemination on the rare occasions when the condi-
tions are met: shelfful [ʃelffʊl] and embalmment [embɑːmmənt]. Where gemi-
nation is found, it shows a boundary between base and affix. Gemination is
sometimes avoided over word boundaries, as in last time [lɑːstaɪm] and Prime
Minister [praɪmɪnɪstə], although this is usually put down to a more general
phenomenon (see [D1]).
Inside a base, sequences of the same vowel are not found. Sequences of
the same vowel, as in co-ownership [kəʊəʊnəʃɪp], always indicate a boundary
between word elements, but so few occur that the overall rule can be taken
to hold.

(D5) 
Sequences of vowel sounds are generally avoided within a word
element; exceptions include flower, Malvolio and naive. Vowel
sequences thus typically show a border between elements.
Examples are given in (X3).
96 Phonology

(X3)

Prefix Example with following vowel

co- coexist
pre- preamble, pre-eminent, pre-industrial, preoccupy, preordain
pro- proactive
re- react, re-enact, re-arm, reinforce, reopen, reorient
Suffix Example following a vowel
-al (adjective) arterial, factual, professorial
-al (noun) avowal, betrayal, renewal, trial, withdrawal
-an Herculean, labyrinthian
-at-ion pronunciation
-ee employee
‑er (adjective) dryer, emptier
‑er (noun) bricklayer, carrier, church-goer, courtier, drier, quantifier,
town-crier, valuer, viewer, widower
‑ing allowing, being, doing, seeing, showing, staying, thawing
-ism altruism, archaism, Dadaism, heroism, Hinduism,
McCarthyism, Monroeism, Naziism, refugeeism, Taoism,
theism
-ist altruist, atheist, canoeist, copyist, Dadaist, oboist, theist

In order to avoid such vowel sequences, vowels can be deleted or conso-

nants can be inserted. Here the deletions are considered. However, there are
also deletions which do not meet this goal. Many of them are borrowed
from other languages. Some examples are given in (X4).

(X4)

Deleted Examples
a (/ə/) amoebic, areal, Chinese, patinise
-ate appreciable, navigable, nominee
i (/i/) Wahabism
i (aɪ/) alkalify
-ic anaesthetist
is emphasise, epidermal, meningitic
le syllabify
on phenomenal
-um optimal, scrotal, spectral, velar
-us phallic, radial, rhomboid, stimulate, typhoid
y (/i/) jurist, memorable, philosopher, philatelic, summarise,
ventriloquist
 Influence of adjacent sounds 97

8.3 Insertion
The main kind of insertion is found in SSBE but not in so-called ‘rhotic’
kinds of English such as American, Canadian, Irish, Scottish kinds of Eng-
lish and many British dialects. In a rhotic pronunciation of English, the let-
ter <r> is pronounced before a consonant, as in car, firm and girl. In SSBE,
these words have no [ɹ] in them: [kɑː], [fɜːm] and [ɡɜːl]. The first rule here,
therefore, apply only to kinds of English which resemble SSBE in this regard
(including Australian and New Zealand English).

If a word ends in one of the vowels [ə, ɑː, ɜː, ɔː, ɪə, eə, ʊə] and the
(I1) 
following word begins with a vowel sound, insert an [ɹ] between the
two (e.g. Africa and Asia [afrɪkə ɹ ənd eɪʒə], far away [fɑː ɹ əweɪ],
wear out [weə ɹ aʊt], fair enough [feə ɹ ɪnʌf], I saw it [aɪ sɔː ɹ ɪt] and
the idea of [ðiː aɪdɪə ɹ əv]).

The literature aimed at speakers whose dominant language is English dis-

tinguishes between those cases where there is a letter <r> in the spelling
(instances which are then called ‘linking R’) and those where there is no let-
ter <r> (which are then called ‘intrusive R’), but the general principle applies
in both cases and there is no strong need to distinguish. Inside a word, the /r/
is pronounced when written in such instances (e.g. caring [keəɹɪŋ]), but inser-
tion can often be heard where no <r> occurs (e.g. cawing [kɔːɹɪŋ]). This is
often considered to be non-standard. Although there are not many words
ending in [ə] when it is not spelt with a final <r> and fewer when such words
take vowel-initial suffixes, I have heard non-established vanilla-y [vənɪləɹi].
Very often some part of the base is deleted to avoid such cases (e.g. areal
[eəɹɪəl] to avoid /ɪəə/) (see Section 8.2).
This rule can be used by intermediate students who are just starting to
acquire fluency.
Because intrusive R is often thought to be not entirely standard, some
speakers insert a glottal stop [ʔ] in such cases rather than an [ɹ]. Learners do
not need to copy this but might need to be aware of it.

(I2) 
When a word starting with a vowel is spoken with emphatic stress,
it may be pronounced with an initial glottal stop [ʔ] (e.g. it was
awful [ɪt wəz ʔɔːfəl] and come out of there [kʌm ʔaʊt ə ðeə]).

This insertion is never obligatory, and speakers of languages where a glot-

tal stop before a vowel is much more common (German, some kinds of Ara-
bic, Tagalog and Thai) may have to be discouraged from overusing it.
98 Phonology

A particular kind of insertion, a type that seems partly to do with mor-

phology and partly to do with phonology, is provided by the case of extend-
ers. Extenders are sounds or sequences of sounds that are inserted between
morphological units. In some cases, they are borrowed from donor lan-
guages; in some cases, they may be useful in facilitating pronunciation; they
may or may not be meaningful, though most are not. Some of them are as-
sociated with bases (and they could be analysed as forming part of the base),
and others are usually considered to be part of the suffix in English, what-
ever their etymological origin. In either case, their distribution is often de-
pendent on the precise bases or affixes used, but their presence or absence is
not predictable in terms of modern English. In other words, extenders are
irregularities in formation. As such, they are probably not worth teaching
explicitly, though they are worth drawing attention to since they otherwise
confuse an analysis into base and affix. Illustrations are given in (X5). Some
extenders are concatenated, and such examples are illustrated in (X6).

(X5) Extenders

Extender Affix Examples Comment

at -ic lymphatic, phlegmatic,
systematic
ion idolisation (and other Many take ‑ation to be
words in ‑ation) the major form of this
suffix because it is the
productive form; there
are also cases where
the <at> represents a
suffix.
-ism conservatism,
phlegmatism
-ure curvature, musculature
c -able applicable, explicable There are many words in
‑icable, but in most the
<c> belongs clearly to
the base.
-ation simplification (and other
nominalisations of
-ify verbs)
-ism Briticism, witticism The border between the
base and the extender
in Briticism is not
clear.

(Continued)
 Influence of adjacent sounds 99

Extender Affix Examples Comment

-y captaincy, chieftaincy, Some treat -cy as a
normalcy separate suffix.
e -an Caesarean, labyrinthean Most words with this
pattern are extremely
rare; the <e> may be a
spelling variant of <i>.
-ous gaseous, righteous Note that the <e>
modifies the
pronunciation of the
preceding consonant.
i -al editorial, ministerial, This is often seen as an
professorial, vestigial unpredictable form of
the -al suffix. Though
common after -er and
-or, it is not always
used there. Note that
this extender adds to
the number of vowel
sequences.
-an Arabian, logician,
magician, reptilian
-ary stipendiary
-ate noviciate, substantiate Note that the two -ate
suffixes illustrated here
are not identical.
-er glazier, hosier
in -al attitudinal, criminal,
libidinal
-ate incriminate, paginate
-ous criminous, libidinous
it -ion addition, competition, Some treat -ition as a
rendition separate affix.
-ive competitive, repetitive,
sensitive
-ise sensitise
-or competitor, servitor
-ure divestiture, expenditure
l -ese Congolese
n -ese Balinese
-ic bubonic, Platonic
-ise Platonise
-ism Platonism
-ist tobacconist
od -ic spasmodic

(Continued)
100 Phonology

Extender Affix Examples Comment

s word sportsman, sports Because this <s> looks
jacket, yachtsman like a plural marker, it
is often interpreted
this way, even when no
plurality seems to be
involved.
-ion recursion, repulsion Some see ‑sion as a
separate suffix.
t -ic Asiatic, dramatic,
-ism dogmatism, stigmatism
-ist egotist Egoist is also found.
-ure fixture, mixture Some see -ture as the
basic form of the suffix.
u -al sensual, textual, usual Some see -ual as a
separate suffix.
-ous sensuous, spirituous,
tempestuous
ut -ion diminution
-ive diminutive

(X6) 
Concatenated extenders

Extender 1 Extender 2 Affix Examples

ac e -ous herbaceous, rosaceous
an e -ous instantaneous
c at ion vinification
-or applicator, falsificator,
vinificator
-ory modificatory
i at -ion pronunciation
n i -an Torontonian
v i -an Peruvian

8.4 Merger
Post-alveolar consonants arise through a merger of an alveolar consonant
and a palatal. This has happened historically to give us, for instance, /pleʒə/
for pleasure from an earlier /plezjʊər/ as indicated by the spelling.
 Influence of adjacent sounds 101

(ME1) 
Over a boundary, an alveolar consonant and a /j/ merge as a
post-alveolar in the following way:
t + j > ʧ e.g. Tell me what you see [tel mɪ wɒʧə siː]
d + j > ʤ e.g. Had you any? [haʤʊ eni]
s + j > ʃ e.g. Pass your plate [pɑːʃə pleɪt]
z + j > ʒ e.g. Has your friend come [haʒɔː freŋ kʌm]

When an alveolar appears before a post-alveolar, there is often assimila-

tion, as in has she [haʒ ʃi] (see [A6]), and this is similar, except that the [j]
vanishes. Note that for many speakers, a sequence of [t] and [j] word-inter-
nally becomes [ʧ] as in tune pronounced [ʧuːn], and due and Jew may be
homophones. This is the same phenomenon, though viewed as similitude
rather than as assimilation. Similarly, in some kinds of English, presume can
be heard pronounced [pɹɪʒuːm]. While the pronunciation [prɪʒuːm] may de-
pend on the kind of English you hear, [haʒjɔː] for has your is very common
in all but the most formal and explicit styles. None of these mergers is oblig-
atory in production, but students should expect to hear them and can use
them in advanced fluency.

8.5 Consonant capture
Consonant capture refers to the moving of a consonant at the end of one
word to the beginning of the next, with corresponding changes in phonetic
values. Like other processes discussed in this chapter, it has the effect of
simplifying clusters at the ends of words. In some cases, it also means that
word-final consonants are not produced with a glottal stop, which some
speakers find to be non-standard.
Consonant capture is sometimes considered to be non-standard pronun-
ciation but seems to be increasingly heard, even though it is never obligatory.
It can be heard in many collocations, although it seems to be more frequent
in a few phrases.

(CC1) 
A word-final consonant may be treated as initial in the next
word provided that it does not cause an illegal cluster to arise
(e.g. at all [ə tɔːl], at any rate [ə teni ɹeɪt] and look after [lʊ kɑːftə]).

In the examples above, the voiceless plosives become aspirated. Where

other consonants are concerned, the length of the consonant may be
affected.
The exception clause in (CC1) is occasionally broken, giving structures
like left out [le ftaʊt], but this is a rare set of examples in a phenomenon
102 Phonology

which is already rare. Note that in the case of left out, there is also a stressed
short vowel in word-final position, which makes it doubly odd.
For learners from some language backgrounds, the problem might be to
prevent overuse of consonant capture; for most, this is more important in
language perception than in language production.

Highlights
This chapter deals with ways of smoothing out transitions between words in
order to make word sequences easier to say. This may involve changing con-
sonants to be more like each other, adding or deleting sounds, or merging
sounds together. Most of the phenomena dealt with here are not needed for
beginning students but can be taught as part of more advanced fluency
training.

What should the teacher do?

Teachers need to be aware of these phenomena but do not need to teach
them overtly for beginning learners. For intermediate or advanced students,
teachers might point out the phenomena and discuss them in terms of for-
mality of style.
9
MORPHOPHONEMICS

Morphophonemics deals with the situation in which the same morpheme

(base or affix) has two or more phonemic representations (i.e. is pronounced
two distinctly different ways). Some of these alternations are represented in
the spelling (e.g. clear and clarity, long and length). Other types are not (e.g.
divine [dɪvaɪn] and divination [dɪvɪneɪʃən]). Some changes are dependent
upon changes in stress, such as abstract (noun/adjective) [ˈabstrakt] and ab-
stract (verb) [əbˈstrakt]. Others are not (e.g. tone [ˈtəʊn] and tonic [ˈtɒnɪk]).
Some alternations are optional, such as scene [siːn] and scenic [senɪk] or
[siːnɪk]; others are obligatory as in the other examples given above. Many
morphophonemic changes are residues from patterns in languages like Latin
or French, from which English has borrowed many words (e.g. divide and
division); some are productive and continue to be used today.
This means that much of the morphophonemic alternation that is found
in English can be seen as affecting individual words and can be taught as
being idiosyncratic – the irregular past tense of verbs is regularly taught this
way, although there are some patterns. In other cases, though, there is a sys-
tem to the variation, and this can be of value to students. In this chapter, the
focus will be on those cases where a useful pattern can be observed, although
other examples will be listed.
Note that since we are dealing with alternation between phonemes (dis-
tinctive sounds), we could use the technical notation and transcribe words
and sounds between slashes (/s/ and /ɹest/). However, the practice of the rest
of the book is maintained, and square brackets are used ([s] and [ɹest]).

DOI: 10.4324/9781032637020-12
104 Phonology

9.1 Crucial inflectional alternation

The most important piece of morphophonemic alternation in English af-
fects suffixes written (e)s and (e)d. These rules need to be acquired at
elementary levels, when the first examples of the relevant suffixes are
intro­duced. Fortunately, they make good phonetic sense and lead to natu-
ral pronunciations.
The first of these rules applies to the plural suffix -s, the -s that forms the
third-person singular of the present tense of verbs, the genitive, the pronun-
ciation of the reduced forms of has and is when they are pronounced with
just a fricative, the -s that forms adverbs in words like forwards and the -s
that forms diminutives, as in ducks (as an informal term of address) or Debs
(as a pet name from Deborah).

(M1) 
When a suffix written <s> or <’s> occurs after any one of [s, z,
ʃ, ʒ, ʧ, or ʤ], it is pronounced [ɪz] (sometimes [əz] in accordance
with variation in the horses vowel) (e.g. Lyons’s [laɪənzɪz],
passes (noun or verb) [pɑːsɪz], watches (noun or verb) [wɒʧɪz]),
judges (noun or verb) [ʤʌʤɪz], and pushes (noun or verb)
[pʊʃɪz]). Note that in such cases there is usually a written <e>
before the <s>.
Otherwise, if the -s immediately follows a voiceless consonant,
it is pronounced [s] (e.g. cats [kats], cups [kʌps], giraffe’s [ʤɪrɑːfs],
kicks (noun or verb) [kɪks] and (his) grief’s (overwhelming)
[ɡriːfs]).
Everywhere else, the -s is pronounced [z] (e.g. Debs [debz], goes
[ɡəʊz], breathes [briːðz] and hen’s [henz]).

The next rule applies to regular past tenses and past participles or the
same forms used as adjectives.

(M2) 
When a suffix written <(e)d> occurs after either [t] or [d], it is
pronounced [ɪd] (sometimes [əd]) (e.g. wanted [wɒntɪd], derided
[dɪɹaɪdɪd], hated [heɪtɪd] and handed [handɪd]). Note that there is
always a written <e> before the <d>.
Otherwise, if the -(e)d immediately follows a voiceless
consonant, it is pronounced [t] (e.g. hoped [həʊpt], wished [wɪʃt]
and collapsed [kəlapst]).
Everywhere else, the -(e)d is pronounced [d] (e.g. banned
[band], desired [dɪzaɪəd], phased [feɪzd], breathed [bɹiːðd] and
sowed [səʊd]).
 Morphophonemics 105

If the past tense or past participle is written with a <t>, it is

pronounced [t] (e.g. went [went], slept [slept] and spoilt [spɔɪlt];
spoiled is also possible).

There are some adjectives ending in -ed which are always pronounced with
[ɪd] even if these rules would lead you to expect another pronunciation. The
adjectives are rare and include aged, crooked, ragged and wretched. These adjec-
tives are not formed from verbs. There is a general tendency for these adjectives
to take on the regular pronunciation described in (M2), and [eɪʤd] is heard as
well as [eɪʤɪd], though not necessarily in all contexts. The sequence of -ed and
-ly as in advisedly, markedly and supposedly is also sometimes pronounced [ɪdli],
even where the form [ɪd] is not expected, if the stress falls on the last syllable of
the verb. The same holds true with -ed followed by -ness, as in markedness.

9.2 Other processes involving inflection

9.2.1 Consonants
The most important process involving inflection is alternation between
voiced and voiceless fricatives at the end of some plural bases. The alterna-
tion between [f] and [v] is noted in the spelling, which is mostly trustworthy.
The alternation between [s] and [z] applies only in a single instance and is not
marked in the spelling; the alternation between [θ] and [ð], which is not
marked in the spelling either, is not consistently applied.

(M3) 
Where the base of a frequent noun ends in one of [f], [θ] and [s],
that fricative can be voiced before a plural -s (e.g. house [haʊs],
houses [haʊzɪz]; path [pɑːθ], paths [pɑːðz]; and calf [kɑːf], calves
[kɑːvz]). Note that the plural marker is what it would regularly be
when following the voiced fricative.

In Scottish English and some dialects of English elsewhere in the world,

houses is pronounced regularly with [s].
The alternation between voiced and voiceless fricatives is found elsewhere
in English, where it has different functions, typically derivational functions.
Whether this is strictly morphophonemics is perhaps controversial, but the
overlap in form in striking.

(X1) 
A noun ending with a voiceless fricative corresponds to a verb
ending in the equivalent voiced fricative: abuse, advice/advise, bath/
bathe (with vowel change), belief/believe, breath/breathe (with vowel
change), calf/calve, cloth/clothe (with vowel change), device/devise,
106 Phonology

excuse, glass/glaze (with vowel change), grass/graze (with vowel

change), grief/grieve, half/halve, house, life/live (with vowel change),
loss/lose (with vowel change), mouth, proof/prove, relief/relieve,
serf/serve, sheaf/sheave, sheath/sheathe, strife/strive, thief/thieve, use,
wife/wive (rare), and wreath/wreathe. Other forms which may show
the same pattern are choice/choose (with vowel change), price/prize
(also found as a noun), and staff/stave (with vowel change and
obscure semantic relationship); the spelling of practice/practise
suggests the same relationship, but the two are homophonous (and
in American English only the first spelling is used).

9.2.2 Vowels (Umlaut and Ablaut)

The change of vowel quality in inflection can be termed apophony. Histori-
cally, Umlaut (a change of vowel conditioned by a vowel in a following syllable)
and Ablaut (with no phonetic conditioning) are distinguished, but since the
relevant conditioning vowels have long since vanished, the two can be treated
as equivalent. The patterns here are so irregular that it scarcely makes sense to
call them rules at all, and they will be exemplified rather than set out as rules.
The first set of patterns (historically patterns of Umlaut) are those found
in plural formations with vowel change. There are very few of these, and
they have to be learned individually. What regularity there is here can be bet-
ter seen in the spelling than in the phonemic structure, which suggests that at
some stage in the history of English it was more regular than it is now.

(X2)

Singular Plural Singular vowel Plural vowel Comment

foot feet ʊ iː
goose geese uː iː
tooth teeth uːa iː
louse lice aʊ aɪ
mouse mice aʊ aɪ
man men a e
woman women The apophony affects
both vowel sounds.b
Note
a
Some speakers have /ʊ/ here.
b
/wʊmən/, /wɪmɪn/. The two are often homonymous in New Zealand English.

There is also one such change in marking comparison on adjectives: old/

elder/eldest (note that older and oldest are now more common, although el-
der/eldest are still found). Elder can also be a noun.
 Morphophonemics 107

The most important set of alternations here is that in marking past tenses
and past participles (historically, this is termed Ablaut). As with the plural cases
illustrated in (X2), there might be some technical argument as to whether this
counts as morphophonemics, but the important thing in both cases is a change
of vowel in special morphological environments. There is some variation as to
which verbs use which particular patterns (both historically and in current Eng-
lish), and in some cases regular versions are also found. Where there is variation
between patterns, this is marked with the symbol ‘%’, and where there is varia-
tion with a regular pattern, this is marked with the symbol ‘®’. Rare verbs –
probably not needed for production before very advanced levels – are marked
with the symbol ‘#’. Derived verbs (gainsay from say or outdo from do) are not
included. The swung dash ‘~’ indicates an alternative form.

(X3)

Present Past Participle Example

vowel vowel vowel
a ɔː ɔː catch/caught/caught
a ʊ ʊ stand/stood/stood
a ʌ ʌ drag/drug/drug % ®, hang/hung/hunga
aɪ aʊ aʊ bind/bound/bound, find/found/found, grind/
ground/ground, wind/wound/wound
aɪ ɒ ɒ shine/shone/shoneb %
aɪ eɪ eɪ lie/lay/lain
aɪ əʊ əʊ dive/dove/dovec % ®, shine/shone/shoned %
aɪ əʊ ɪ drive/drove/driven, ride/rode/ridden, rise/rose/risen,
shrive/shrove/shriven #, smite/smote/smitten #,
stride/strodel, strive/strove/striven ® #, thrive/
throve/thriven ® #, write/wrote/written
aɪ ɪ ɪ bite/bit/bitten, chide/chid/chidden ~ chid ® #, hide/
hid/hidden ~ hid, light/lit/lit ®, slide/slid/slid
aɪ ɔː ɔː buy/bought/bought, fight/fought/fought
aɪ uː əʊ fly/flew/flown
aɪ ʌ ʌ strike/struck/struck
e əʊ əʊ sell/sold/sold, tell/told/told
eə ɔː ɔː bear/bore/bornee #, swear/swore/sworn, tear/tore/
torn, wear/wore/worn
eɪ e e say/said/said
eɪ əʊ əʊ break/broke/broken, wake/woke/woken
eɪ ʊ eɪ forsake/forsook/forsaken #, shake/shook/shaken,
take/took/taken
əʊ e e hold/held/held
əʊ uː əʊ blow/blew/blown, grow/grew/grown, know/knew/
known, mow/–/mownf, sew/–/sewn, show/–/
shown, sow/–/sown, throw/threw/thrown
(Continued)
108 Phonology

Present Past Participle Example

vowel vowel vowel
iː e e bleed/bled/bled, breed/bred/bred, cleave/cleft/cleft
% ® #, deal/dealt/dealt, dream/dreamt/dreamt
®, feed/fed/fed, feel/felt/felt, flee/fled/fled, keep/
kept/kept, kneel/knelt/knelt, lead/led/led, lean/
leant/leant ®, leap/leapt/leapt ®, leave/left/left,
mean/meant/meant, plead/pled/pled ®, read/
read/readg, sleep/slept/slept, speed/sped/sped ®
iː e iː eat/ate/eatenh
iː əʊ əʊ cleave/clove/cloven % ® #, freeze/froze/frozen,
heave/hove/hovej #, speak/spoke/spoken, steal/
stole/stolen, weave/wove/woven
iː ɔː iː see/saw/seen
iː ɔː ɔː beseech/besought/besought #, seek/sought/sought #
iː ʌ ʌ sneak/snuck/snuck % ®
ɪ a ɪ bid/bade/biddenk #
ɪ a ʌ begin/began/begun, drink/drank/drunk, sing/sang/
sung/, sink/sank/sunk, swim/swam/swum/
ɪ eɪ ɪ give/gave/given
ɪ ɔː ɔː bring/brought/brought, think/thought/thought
ɪ ʌ ʌ cling/clung/clung, dig/dug/dug, fling/flung/flung,
sling/slung/slung, slink/slunk/slunk, stick/stuck/
stuck, sting/stung/stung, string/strung/strung,
swing/swung/swung, win/won/won, wring/
wrung/wrung
ɔː e ɔː fall/fell/fallen
ɔː uː ɔː draw/drew/drawn
uː ɒ ɒ lose/lost/lost, shoe/shod/shod ®
uː ɪ ʌ do/did/done
ʌ a ʌ run/ran/run
ʌ eɪ ʌ come/came/come

Notes
a
The regular verb hang when it means ‘to execute by hanging’ is now old-fashioned.
b
Shine is regular when transitive.
c
This form, originally used in American English, is spreading.
d
This form is used in American English. See also note b.
e
Note the spelling born when used as an adjective. Bear in the sense ‘carry’ is rare.
f
The dash replacing a past tense form indicates a regular form.
g
Note that the spelling does not indicate the difference in pronunciation.
h
Ate is sometime pronounced [eɪt].
j
This heave is used only in nautical senses and is rare. Heave meaning ‘pull’ is regular.
k
Bade is sometimes pronounced /beɪd/. It then patterns with give. This bid means ‘ask’; the
verb meaning ‘make a bid’ does not follow this pattern.
l
Although stride would be expected to have the past participle stridden, it is rare, and
speakers are unhappy with any of strode, strid, stridden or strided and often avoid the form
when possible.
 Morphophonemics 109

The information in (X3) can be used to determine when to teach indi-

vidual verbs and when to teach patterns. Some of the alternatives in (X3) are
typical of American English or British English, and either is found in other
kinds of English.

9.3 Main processes found in derivation

9.3.1 Consonants
There is a small amount of voicing variation within words and often with
words that can be taught as individual forms. The simplest example is given
in (M4).

(M4) 
A voiced obstruent before the suffix -th which creates ordinal
numbers and fractions and before the -teen which creates cardinal
numbers between 13 and 19 or the -ty which creates numbers that
are multiples of ten becomes voiceless (e.g. fifth [fɪfθ], fifteen
[fɪftiːn] and fifty [fɪfti]). Note that this is recognised in the spelling.

The next examples are far more common but are restricted to the learned
part of the vocabulary.

(M5) 
An adjective ending in [t] may have a corresponding noun ending
in [s] (e.g. prevalent/prevalence and resident/residence and other
words ending the same way).

This rule is really a rule of Latin, one which has been borrowed into Eng-
lish with the relevant words. Although it functions well for many words (al-
beit with a very limited pattern), care must be taken, as illustrated in (X4).

(X4) 
Exceptions to (M5)

The two words are no longer (or never accident/accidence, comment/

were) related in meaning in English. commence, emergent/emergency
There is a [t] adjective with no [s] noun. repellent
There is an [s] noun with no [t] adjective. conference, performance, sentence
Rather than just [s], the preferred form adjacency, consistency, deficiency,
ends in [si] (written -cy). occupancy, pregnancy
(Continued)
110 Phonology

Both the [t] word and the [s] word are precedent/precedence, variant/
nouns. variance
The [t] word is a noun with no [s] word. recipient
The [t] word can be a noun or an attendant, resident
adjective.
The base is not Latin and the rule doesn’t allowance, defendant, guidance
work.

Rule (M5) also applies before the suffix -y (e.g. pirate/piracy /paɪrəsi/).
The next rule is also a Latin rule.

(M6) 
[t] is changed to [ʃ] before the suffixes -ion [ən], -ial, -ian and -ious
[əs] (e.g. evaluation [ɪvaljueɪʃən], evidential [evɪdenʃəl], dietitian
[daɪətɪʃən] and fictitious [fɪktɪʃəs]).

This rule is very common because of the frequency of the sequence -ation
(also -ition and -ution). The most notable exception is equation, usually
[ɪkweɪʒən] though the regular [ɪkweɪʃən] is also heard. There is a voiced
equivalent of this, as shown in (M7).

(M7) 
[d] is changed to [ʒ] before the suffix -ion [ən] (e.g. persuade/
persuasion [pəsweɪʒən] and elide/elision [ɪlɪʒən]).

This is not quite as regular since bases which contain the Latin elements
cede (also with other spellings) and tend have [ʃ] rather than [ʒ] (e.g. concede/
concession [kənseʃən], proceed/procession [pɹəseʃən] and distend/distension
[dɪstenʃən]). Note that these changes are recognised in the spelling. Rarer cases
with Latin pend and pand (suspend/suspension, expand/expansion) also use [ʃ].
This set is harder to teach because it depends on recognising Latin elements.
There is a related change, as shown in (M8).

(M8) 
[d] is replaced by [s] (or sometimes [z]) before the suffix -ive (e.g.
evade/evasive [ɪveɪsɪv] ~ [ɪveɪzɪv] and divide/divisive [dɪvaɪsɪv] ~
[dɪvɪzɪv]).

This rule sometimes applies before -ible, as in comprehend/comprehensible

[kɒmprɪhensɪbəl], but not always: extend/extendible ~ extensible, ascendible ~
ascendable. The spelling reflects the pronunciation, even when both are
variable.
Corresponding to (M6), we also find (M9).
 Morphophonemics 111

(M9) 
[s] is changed to [ʃ] before the suffixes -ian, -ion and -ial (e.g.
express/expression, concuss/concussion and circumstance/circum­
stantial [sɜːkəmstanʃəl]).

There are a few instances where we find [ʒ] instead of – or as well as – the
expected [ʃ], as in immerse/immersion [ɪmɜːʃən] ~ [ɪmɜːʒən]. We would expect
to find a corresponding change with voiced consonants, but it is very rare
before -ion, as in circumcise/circumcision [sɜːkəmsɪʒən] and revise/revision
[ɹɪvɪʒən]. This rarity seems to be the result of there being few possible inputs
to it, especially because the suffix ‑ise is followed by ‑ation rather than by ‑ion.
The change to voiced consonants does apply, though, before -ual and -ure, as
in usual [juːʒəl] and exposure [ɪkspəʊʒə].
The gap in the system set out above is a relationship with [d] and [z]
(which makes better sense phonetically than one between [d] and [s] since
both [d] and [z] are voiced). This relationship is found but is relatively
rare.

(M10) 
[d] in a base word corresponds to [z] (or sometimes [s]) before
the suffix -ory (e.g. provide/provisory [pɹəvaɪzəri] and delude/
delusory [dɪluːzəɹi] ~ [dɪluːsəɹi]).

This rule is very rare because there are not many relevant pairs. A similar
relationship is sometimes found before -ible, as in divide/divisible [dɪvɪzɪbəl].
Note, however, that defend/defensible has [s], not [z]: [dɪfensɪbəl].
Although there is much variation in all of this, the best way of predicting
the pattern of alternation in an unknown word is to look for known deriva-
tives of that word and to transfer the pattern of voicing to the new word. If
we know expand and expansive (with [s]), we can predict that expansible (it is
very rare), if met, will have [s]. In production, the use of the suffix -able as
opposed to -ible does not have the same pattern of morphophonemic change,
so that defendable and expandable (both of which are also found) retain the
[d] and avoid the problem.

(M11) 
[t] is changed to [ʧ] before the suffix -ure (e.g. moist/moisture
[mɔɪsʧə]).

This rule is quite specific and rare because of the rarity of -ure. It does not
appear to have a voiced equivalent.
There is a generalisation to the above changes, though the specifics vary
from case to case. The generalisation is that [t] and [d] become fricatives
112 Phonology

before specific suffixes, usually suffixes written with <i> or <y>. With some
of these suffixes, notably -ion, the fricative becomes post-alveolar rather
than alveolar. The <i> can be seen as the orthographic warning of this pho-
nological change. The changes are mainly restricted to words of Latin ori-
gin. Some of the variation in voicing is not easily explained from the
viewpoint of English.
The next pattern is common enough to have acquired is own name: ‘velar
softening’. We can see velar softening as the morphophonemic result of the
spelling system: the letter <c> is pronounced [s] before some vowels and [k]
in other places (see Chapter 13). The rule is provided as (M16).

(M12) 
A base-final [k] written <c> is changed to [s] when it occurs
before any of the suffixes -ian, -ity, -ist, -ism, -ise and -ify (e.g.
classic/classicist/classicism and historic/historicity/historicise).

Although there is no base with a [k], this is consistent with rule (M9) in
words like suspicion [sʌspɪʃən] and suspicious [sʌspɪʃəs]. The rule has a wider
application in a word like sufficiency (where the second <c> is pronounced
[s]), even if there is no [k] variant. With the suffix -ian as in music/musician,
the outcome can be seen as a mixture of (M12) and (M9). Note that in most
cases where (M12) is relevant, the base will end in the suffix -ic, and where it
does not, as in zincify and monarchism (where the [k] is written <ch>), (M12)
does not apply.
There is also a voiced version of velar softening, given in (M13).

(M13) 
A base-final [ɡ] is changed to [ʤ] when it occurs before -ian, -ic,
-ism, -ist, -ise or -y (e.g. monologue/monologic [mɒnəlɒʤɪk],
dialogue/dialogise /daɪəlɒʤaɪz/, fugue/fugist [fjuːʤɪst] and
analog(ue)/analogy [ənaləʤi]).

Note that the rule can be applied even when no basic [ɡ] is immediately
obvious, as in eulogy [juːləʤi] and mythologist [mɪθɒləʤɪst]. The rule would
be expected to apply before -ial, but there do not appear to be relevant estab-
lished examples.
The velar softening rules are better taught as part of the spelling sys-
tem of English than as independent phonological rules. As a result, they
can be taught relatively early, although the words in which they apply are
not frequent enough to warrant teaching before a reasonably advanced
level.
 Morphophonemics 113

There are several places where [ɹ] and [z] (sometimes [ɹ] and [s]) are in al-
ternation, and they are quite disparate.

(M14a) 
In the words was and were, we find alternation between [z] and
a written <r> pronounced [ɹ] when there is a following vowel.
This pattern goes back thousands of years and is found only in
this pair.

(M14b) [ɹ] alternates with [z] (sometimes with [s]) before -ive; this
extends to [ɹ] alternating with [ʒ] before -ion (as in M9) (e.g.
adhere/adhesive [ədhiːzɪv] ~ [ədhiːsɪv] and adhere/adhesion
[ədhiːʒən]).

(M14c) There is a modern way of forming pet names (hypocoristics)

where [ɹ] is changed to [z] and a final [ə] is added (e.g. Barry/
Bazza and Sharon/Shazza).

Of these, (M14a) can be treated as suppletion, (M14b) is rare and (M14c)

is slangy and specialised.
Variation between [s] and [z] has been pointed out already in (M3). It is
also found rarely in suffixing environments.

(M15) 
[s] is changed to [z] before -al (e.g. spouse/spousal [spaʊzəl]).

Note that this rule is not very general: universal is pronounced with [s],
not with [z], and with a different -al suffix, one creating nouns, we find dis-
persal with an [s], too. There are very few relevant words; the predominant
pattern is for a [z] before -al to remain a [z] (e.g. cause/causal) or an [s] writ-
ten <ss> to remain an [s] (e.g. abyss/abyssal). For practical teaching pur-
poses, this rule may be ignored.
There is a set of forms containing a sequence of two consonant letters
where sometimes both of the consonants are pronounced and some-
times only one of them is pronounced. We can think of the one that is
not pronounced as being deleted under certain circumstances. The cir-
cumstances are not entirely predictable, but there are trends. The rele-
vant sets of consonants are [ɡm], [ɡn], [mn], [mb] and [ŋɡ]. There is an
overarching generalisation that only a single consonant is pronounced
in word-final position, although the places where the sequence is pro-
nounced vary from case to case. In some instances, there are also vowel
114 Phonology

changes involved. The spelling is morphophonemic in the sense that

both consonants are written even if only one is pronounced and that any
vowel changes are not shown in the spelling.

[ɡm] (e.g. paradigm [paɹədaɪm] and paradigmatic

(X5) 
[paɹədɪɡmatɪk])
[ɡn] (e.g. sign [saɪn], signal [sɪɡnəl], design [dɪzaɪn] and designation
[dezɪɡneɪʃən])
[mn] (e.g. autumn [ɔːtəm], autumnal [ɔːtʌmnəl], solemn [sɒləm] and
solemnity [səlemnɪti])
[mb] (e.g. crumb [krʌm] and crumble [krʌmbəl])
[ŋɡ] (e.g. long [lɒŋ] and longer ‘more long’ [lɒŋɡə])

The same sequences can be found in places where there is no alternation

inside a morpheme and both consonants are pronounced, as in agma, lig-
nite, amnesty, amber and finger.
In a few cases, the single consonant is found in other positions, as in mne-
monic [nemɒnɪk].
The consonant sequence is not found before all affixes, and it is not al-
ways clear what determines whether the consonant sequence will occur or
not: consider assignment [əsaɪnmənt], designed [dɪzaɪnd], solemnly [sɒləmli],
bomber [bɒmə] and longer ‘one who longs’ [lɒŋə]. While individual affixes
tend to be consistent in whether they allow the sequence before them or not,
various attempts at theoretical statements on the subject (e.g. inflectional
affixes allow only one consonant, vowel-initial affixes allow consonant se-
quences, and French and Latin suffixes allow consonant sequences) have all
proved not to hold consistently. In some cases, affixes are not even consis-
tent: longer and stronger have the sequence [ŋɡ], but if we use wronger (and
we tend not to, preferring more wrong), it has only [ŋ].

9.3.2 Vowels
The first set of vocalic alternations to consider is the set of variations that
occur when stress is lost because of differing positions of the stress in de-
rived forms. Where stress is lost completely, the usual vowel is [ə], but there
are some exceptions to that which are set out together in the examples below.
The vowel [ɔɪ] is never fully unstressed and does not participate in the alter-
nation. The example of authoritarian, the best example I can find of /ɔː/ al-
ternating with /ə/, is not true for those people who pronounce authoritarian
without full unstressing of the initial vowel.
 Morphophonemics 115

(X6)

Unstressed vowel Stressed vowel Examples

ɪ~ə ɪ ˈacid/aˈcidic
ɪ~ə e ˈtelephone/teˈlephony, ˈprophet/proˈphetic
ɪ~ə aɪ ˈmime/miˈmetic
ʊ~ə ʊ ˈfull/ˈspoonful
ʊ ~ əː uː ˈexecute/eˈxecutive
ʊ~ə ʊə maˈture/matuˈration
ə e ˈmen/ˈfrogmen
ə a ˈpharynx/phaˈryngeal
ə ʌ ˈsubstance/subˈstantial
ə ɒ ˈidol/iˈdolatry
ə iː Eˈlizabeth/Elizaˈbethan
ə ɑː ˈparticle/parˈticular
ə ɜː conˈfer/ˈconference
ə ɔː ˈauthor/authoriˈtarian
ə eɪ exˈplain/explaˈnation
ɔɪ no examples
ə əʊ ˈmelody/meˈlodious
ə aʊ ˈmouth/ˈPortsmouth
ə ɪə ˈSpenser/Spenˈserian
ə eə deˈspair/despaˈration

The next set of alternations are between long vowels and short vowels (with
occasional examples of variation between long vowels). In the history of
English, vowels have fairly frequently lengthened and shortened, and the
long vowels have changed their quality so that they no longer look like long
versions of the short vowels (which might have been the case about 500 years
ago). Moreover, some new long vowels have arisen: [ɪə], [eə], [ʊə] and [ɜː]
with disappearance of [ɹ] after the vowel (and the diphthongs do not exist in
American English and Scottish English, where the [ɹ] is retained), [ɔɪ] has
been borrowed from French, and [ɑː] has spread to new places (e.g. in farm
and dance). At the same time, what used to be the short congener of [uː] has
split into two: [ʊ] in foot and puss and [ʌ] in strut and pus. This has left the
English vowel system looking rather messy, especially where traces of these
changes have been left in the morphophonemic system.
The first set (and perhaps the most important historically speaking if not
today) have arisen as a result of the Great Vowel Shift (GVS), which affected
the quality of the long vowels between the Middle English period and the
end of the Early Modern English period. Examples are provided in (X7). In
this set, we are looking at instances where stressed vowels alternate.
116 Phonology

(X7)

Long vowel Short vowel Examples

iː e extreme/extremity, impede/impediment, obscene/
obscenity
eɪ a cave/cavity, explain/explanatory, fable/fabulous,
inane/inanity, mendacious/mendacity, nation/
national, nature/natural
əʊ ɒ globe/globular, joke/jocular, provoke/provocative,
sole/solitude
aɪ ɪ crime/criminal, derive/derivative, divine/divinity,
reside/residual, sign/signature, sublime/
subliminal, type/typical
aʊ ʌ profound/profundity

In these examples, the short vowel occurs when it falls three syllables from
the end of the word after suffixation, and the process is sometimes called ‘tri-
syllabic laxing’ (although ‘antepenultimate shortening’ might be better). The
same vowel pairings can be seen again in free variation in Chapter 11, where
speakers, in effect, do not know whether to treat a word as containing a sound
derived from a Middle English long vowel or short vowel. The alternation is
restricted to certain suffixes, so that, for instance, -ness never causes it. In gen-
eral terms, it is restricted to words with suffixes from Latin or French or both.
The same alternations can be found on the second or fourth syllable from the
end of a word (e.g. cycle/cyclic, metre/metric, please/pleasant, pronounce/pro-
nunciation, renounce/renunciation and south/southern – this was once regular
but has lost a syllable), but it is the antepenultimate alternation which is seen
as the regular one. Some other alternations are found in similar environments
but have different origins (e.g. compel/compulsion and point/punctual).
It is worth noting that there are other instances, in apparently parallel
environments, where there is no alternation, either because different vowels
are involved or because the instances are exceptional (perhaps created too
late in history).

(X8)

Long vowel Examples

iː obese/obesity
eɪ alien/alienate, brave/bravery, danger/dangerous
əʊ motor/motorist
aɪ advise/advisory, hyphen/hyphenate, nice/nicety
 Morphophonemics 117

There are other vowel pairings which arise in connection with derivation
but which either do not match the antepenultimate pattern or show variation
between different vowels.

(X9)

Long vowel Short vowel Examples

ɪə e austere/austerity, sincere/sincerity
uː ʌ assume/assumption (and similar examples),
introduce/introductory
eə a hilarious/hilarity
ɪə a clear/clarity

There are some surprising results of the GVS. One of them concerns the
pronunciation of Latin by English speakers. Ecclesiastical Latin in England
underwent the changes to long vowels brought about by the GVS, but sub-
sequently the pronunciation of Latin was reformed in the light of scholar-
ship about the way that Latin was originally pronounced. The result is that
Latin in the church and in the legal profession often ends up with two (or
more) distinct pronunciations of the same words, and both can be heard
today. Some examples are given in (X10).

(X10)

GVS-affected Restored Latin Examples

eɪ ɑː apparatus, data
aɪ iː alumni, regina, via

Another is that when we borrow foreign words, we borrow them with

their spellings (there are languages which respell them in their own way,
but English tends not to as long as the original spelling is known). Words
borrowed before the GVS show the pronunciation that has been affected
by the shift; words borrowed since have a different spelling–pronunciation
link.
118 Phonology

(X11)

Borrowed Examples Borrowed Examples

pre-GVS post-GVS

iː employee eɪ déjà vu, fiancee

aɪ oblige, Valentine iː prestige, quarantine

In other loan words, there are also two distinct pronunciations of the
vowels: one long and one short. Often the long vowel is associated with
American English but spreads easily to other kinds of English, including
Standard Southern British English, although there is quite a lot of
variation.

(X12)

Long vowel Short vowel Examples

əʊ ɒ adios, baroque, Costa Rica, shalom, yoghurt
ɑː a Iran, Iraq, kebab, Luganda, macho, pasta, taco
uː ʊ buffo
eɪ e debris, Pedro, pieta, tête-à-tête
iː ɪ hasta la vista

9.4 Minor patterns of alternation

There is a small set of verbs where transitive and intransitive are distin-
guished by Ablaut. The set is not productive.

(X13)

Intransitive Transitive
fall fell
lie lay
rise raise
sit set

Similarly, there is a small set of nouns and verbs related by an Ablaut pat-
tern; again, this set cannot be extended.
 Morphophonemics 119

(X14)

Verb Noun
abide abode
bleed blood
breed brood
shoot shot
sing song

A single example, fill/full, illustrates an Ablaut link between a verb and an

adjective.
There is a minor pattern, but a productive one, where an unstressed vowel
in the final syllable of a word becomes long and stressed before the suffix
-ian. The quality taken on by the vowel is determined by the spelling.

(X15) 
Amazon/Amazonian [aməzəʊniən]
Aristotle/Aristotelian [aɹɪstətiːliən]
Babylon/Babylonian [babɪləʊniən]
Bacon/Baconian [beɪkəʊniən]
Caesar/Caesarian [siːzeəɹiən]
Christ/Christian [kɪstiən] (Note the short vowel in the derivative!)
Devon/Devonian [dɪvəʊniən]
Johnson/Johnsonian [ʤɒnsəʊniən] (and other names ending in
-son)
Lilliput/Lilliputian [lɪlɪpjuːʃən]
Venus (the planet not the goddess)/Venusian [vɪnjuːziən]
Venice/Venetian [vɪniːʃən]

A sub-type of variation before -ian is found where the base word ends in
/f/ or (and this confirms the spelling basis of the rule) a written <w>, where
/v/ replaces the last consonant (letter).

(X16) 
Aronoff/Aronovian [aɹənəʊviən]
Harrow/Harrovian [haɹəʊviən]
Shaw/Shavian [ʃeɪviən] (George Bernard Shaw)

Before the suffix -th which makes nouns (e.g. truth and warmth), there is
sometimes apophony (e.g. in long/length). The difficulty here is that not
120 Phonology

everyone agrees on what words ending in the letters <th> or /θ/ contain the
relevant suffix: does month, for instance? The number of words containing
this affix is small, and the number of those with apophony is even smaller.
The patterns are set out in (X17). Note that the patterns are not consistent:
truth and troth both come from true but possibly in different dialects; warm
does not give rise to *wemth. Also, the spellings suggest that the patterns
were less consistent in earlier forms of English.

(X17)

Base vowel Vowel before suffix Examples

iː e deep/depth, heal/health, steal/
stealth, weal/wealth
ɒ e long/length, strong/strength
ɔː e broad/breadth
aɪ ɪ wide/width
aɪ e die/death
ɪə ɜː dear/dearth
eə ɜː bear/birth
uː ʌ moon/month
ʌ uː young/youth

We also find a set of alternations which arise through historical accidents

of various kinds and which are not at all systematic. Just where the bound-
ary between real alternations and random phonemic make-up occurs is not
always clear. On the one hand, we would probably agree that hound and ca-
nine are not related except by meaning (though historically, they are related
by form, too, if you go far enough back) but that serpent and serpentine are
related and that liberty and libertine probably are related, even if we’re not
quite sure how. With grenade and grenadine, however, we can be in doubt as
to whether there is a link or not, and there is room for personal opinion to
differ. Some examples of these irregular formations are given in (X18).

(X18)

ɪ e stink/stench
ɪ ʌ thumb/thimble
ɪ aɪ dine/dinner
ɪ əʊ gild/gold
e ɑː France/French
e uː moon/menstural

(Continued)
 Morphophonemics 121

e eɪ break/breakfast
a ɒ coffee/caffein
a iː peace/pacifist/pacify
a əʊ holy/hallow
a ɪə clear/clarity
ʌ ɔɪ destroy/destruction
ɒ iː hot/heat
ɒ uː school/scholar
ɒ eɪ name/nominate
iː eɪ Norway/Norwegian
iː aɪ see/sight
iː əʊ Aberdeen/Aberdonian
ɑː ɔː psalm/psalter
ɑː eɪ glass/glaze/glazier
ɑː əʊ ghost/ghastly
ɔː eɪ slay/slaughter
ɔː ɔɪ law/lawyer, saw/sawyer
ɔː əʊ coda/caudal
ɔː aʊ flower/flora/floral
eɪ əʊ nose/nasal
p v receive/receptive (and other words ending in -ceive)
b v approve/approbation

Highlights
Morphophonemics deals with situations where there is a change from one
distinctive sound (phoneme) to another in the service of the morphology.
When we talk about morphophonemic spelling, we mean spelling which re-
flects the meaning rather than the sounds, and so one spelling is pronounced
in different ways in related words. Although there are common cases of mor-
phophonemic alternation in inflection, most of the relevant cases can be
treated as irregularities in the system.

What should the teacher do?

Most morphophonemic alternation can be treated as a matter of spelling
irregularity, but the teacher may want to be aware of the larger patterns.
Where there are useful patterns for the students, the relevant vocabulary is
likely to arise only for advanced students, so that the teacher does not need
to teach most of this overtly at lower levels.
10
STRESS RULES

In this chapter, we will not be particularly concerned with the phonetics of

stress, nor with placement of stress in the sentence caused by pragmatic fac-
tors such as contrast. It is not that these points are not worthy of discussion
or that they are not of value to the learner of English as an additional lan-
guage. Rather, it is because they take us beyond the bounds of phonology –
to phonetics on the one side and to pragmatics on the other. Instead, we will
consider the placement of the main stress in words.
One of the major achievements of Chomsky and Halle (1968) was the
demonstration that the placement of stress in English is largely predictable
if you have enough information about the word whose stress you wish to
predict. Among the information you need is information on the etymology
of the word; the quality and length of the vowels in the word; the origins of
the vowels in the word; the number, nature and position of consonants in the
word; and the morphology of the word. Most learners do not have this in-
formation, and if they did, the calculations necessary to predict the position
of the stress would be beyond them in a situation where they were actually
speaking the language. Something simpler is necessary for learners. Even the
information provided here is too much for practical use in most cases. It
may, nevertheless, be of value to students to be able to see that there are
generalisations and to be able to see patterns.
That is not to say that there are no exceptions to the generalisations. Some
of the exceptional cases will be noted, along with the generalisations which
they are exceptions to.

DOI: 10.4324/9781032637020-13
 Stress rules 123

10.1 Monosyllables
Since we are working under the assumption that stress is a phenomenon
which affects individual words, every monosyllabic word must carry
stress when spoken in isolation. And while this is true, it provides some
contradictions.
Many words are rarely spoken in isolation, and when they occur in speech,
they are typically not stressed. Accordingly, when they are spoken in isola-
tion and with stress, they are spoken rather unnaturally. Two examples will
make the case.
The first is the definite article the. In normal usage, it is unstressed before
the word to which it applies, and it is pronounced with the vowel [ə] when it
occurs before a consonant sound and [i] when it occurs before a vowel
sound. In isolation, it is usually pronounced [ðiː]. The stressed pronuncia-
tion, which normally occurs only in sentences like Not the Boris Johnson (in
response to I’ve just met Boris Johnson), is unusual except when the word is
pronounced in isolation, when it is the norm. Pronouncing the word the
with stress falsifies its normal usage. Something similar could be said for
many prepositions.
The complementiser that, as in I know that you told me not to, is invariably
unstressed in general usage. But if we have to cite it as a word of English, it
gains stress. We cannot pronounce [ðət] in a stressed fashion, because [ə] oc-
curs only unstressed, so we need a vowel to use when we mention that. Be-
cause of the spelling, we use [a] and pronounce the complementiser [ðat] in
the same way that we would pronounce the demonstrative that. But this is,
to some extent, a misleading picture of English. Demonstrative that is fre-
quently pronounced [ðat] and is frequently stressed, but complementiser that
is pronounced [ðat] only when spoken in isolation (or in pedantically articu-
lated English).
Textbooks often talk of [ðət] and [ðə] as reduced forms, but the reduced
forms are the norm, and it is the unreduced forms which are abnormal.
If we look away from these grammatical words which are typically un-
stressed, then monosyllabic lexical words (nouns, verbs, adjectives and ad-
verbs), which make up most of the monosyllables though not the most
frequent ones, do carry stress in the expected way.

10.2 Disyllables
When we talk about disyllables here, we are talking in the first instance about
words which have neither prefixes nor suffixes and which are not compounds:
that is, we ignore words like aside, goodness and blackbird for the time being.
The general rule here is that such words are stressed on the first syllable.
Words such as those in (X1) are typical.
124 Phonology

(X1) 
agile, anchor, bacon, body, civil, father, fellow, govern, limit,
meadow, orbit, person, pickle, pony, reason, sparrow, stable, suet,
supper and system

Exceptions to this general rule include enough and hello.

Note that in most of the words in (X1), the final syllable is completely
unstressed; in words ending in /əʊ/, there is a full vowel, though it does not
carry the stress. In some common words ending in /əʊ/, this is reduced to just
/ə/ in colloquial non-standard types of English.
Note also that although the words in (X1) have many origins, the ones
most likely to fit the rule are basic words which have a long history in
English.
Although the rule has been presented as stress on the first syllable, a more
accurate rule would be stress on the first syllable of the base. This means that
prefixes are typically not stressed. There are problems with this, however.
First, many English prefixes are no longer felt to be particularly obvious
prefixes; second, prefixes from French or Latin may not be easily recognisa-
ble. Nonetheless, where there are prefixes, they are usually unstressed, and
the stress must fall on the second syllable.
English prefixes include a- (as in ablaze and away), be- (as in beget and
bewitch), un- (as in undo and unkind) and with- (as in withdraw and within)
French and Latin prefixes include ad- (as in adhere but also as in accept,
append, arrest and attire), de- (as in depend, derive and deter), dis- (as in dis-
band and disease), ex- (as in exceed and expel), on re- (as in reject and remain)
and sub- (as in subsist and subtract). When these come directly from French
or Latin, the prefixes are unstressed, but when they are used in English (as in
de-tune, ex-wife and re-use), they have full vowels.
With about 100 pairs of words (which, because they change from person
to person and from time to time, cannot be listed), there is a pattern whereby
the noun gets stress on the first syllable and the verb gets stress on the sec-
ond. The noun is derived from the verb in these instances.
The general rule is that the syllable without the main stress retains a full
vowel when it is an original base and becomes unstressed when it is an origi-
nal prefix (or, in a few cases, something that seems to act in a parallel way).

(X2)

Spelling Verb Noun

abstract abˈstɹakt ~ əbˈstɹakt ˈabstɹakt
addict əˈdɪkt ˈadɪkt
annex əˈneks ˈaneks
(Continued)
 Stress rules 125

Spelling Verb Noun

augment ɔːɡˈment ˈɔːɡment
combat kəmˈbat ˈkɒmbat
compress kəmˈpɹes ˈkɒmpɹes
consort kənˈsɔːt ˈkɒnsɔːt
discount dɪsˈkaʊnt ˈdɪskaʊnt
escort ɪˈskɔːt ˈeskɔːt
export ɪkˈspɔːt ˈekspɔːt
extract ɪkˈstɹakt ˈekstɹakt
import ɪmˈpɔːt ˈɪmpɔːt
insult ɪnˈsʌlt ˈɪnsʌlt
permit pəˈmɪt ˈpɜːmɪt
produce pɹəˈdjuːs ˈpɹɒdjuːs ~ ˈpɹəʊdjuːs
prospect pɹəˈspekt ˈpɹɒspekt
recess ɹɪˈses ˈɹiːses
recoil ɹɪˈkɔɪl ˈɹiːkɔɪl
refund ɹɪˈfʌnd ˈɹiːfʌnd
segment seɡˈment ˈseɡmənt
subject səbˈʤekt ˈsʌbʤekt
torment tɔːˈment ˈtɔːment
transfer tɹansˈfɜː ˈtɹansfɜː
transport tɹanˈspɔːt ˈtɹanspɔːt
update ʌpˈdeɪt ˈʌpdeɪt

Although the pattern is common with these (mainly foreign) words, it

cannot be predicted. With the verb proˈtest, which gives rise to the noun
ˈprotest in the normal way, there is also a verb ˈprotest, which means ‘take
part in a protest’. The verb consent gives rise to a noun consent, but both are
stressed on the second syllable, similarly with command, decree and design.
Retail has first-syllable stress on both the noun and the verb (though it once
fitted the pattern in [X2]). The same is true of compass, prefix and promise.
Because of this general lack of predictability, these words have to be learnt
individually rather than as an overall pattern.
There are a few adjectives which show similar stress changes. Because
there are so few, again they need to be treated individually.

(X3)

Spelling Verb Adjective Comment

abstract abˈstɹact ~ ˈabstɹakt See also in (X2)
əbˈstɹakt
(Continued)
126 Phonology

Spelling Verb Adjective Comment

compact kəmˈpakt kəmˈpakt, There is also a noun, [ˈkɒmpakt]
ˈkɒmpakt
compound kəmˈpaʊnd ˈkɒmpaʊnd There is a noun with initial
stress, and a verb ‘to make a
linguistic compound’ with
initial stress.
frequent fɹiˈkwent ˈfɹiːkwənt The verb is now often
pronounced with initial stress

Although the rules given above are the dominant ones, there are many
exceptions. Most of these simply have to be learnt individually, but there are
recurrent themes to some of these exceptions, and we look at these now. It
must be recalled that the most general patterns (given above) often out-rank
the patterns shown in the exceptions: the exceptions to the exceptional pat-
terns are in line with the most general patterns.
Where nouns end in /iː/ or /uː/, these sometimes (but not reliably) attract
the stress, and the first syllable sometimes retains a full vowel.

(X4) With stressed final vowel

agree əˈɡɹiː
balloon bəˈluːn
bamboo bamˈbuː
bassoon baˈsuːn ~ bəˈsuːn
canoe kəˈnuː
Dundee dʌnˈdiː
Hindu hɪnˈduː
kazoo kəˈzuː
lampoon lamˈpuːn
machine məˈʃiːn
maroon məˈɹuːn
marquee mɑːˈkiː
payee peɪˈiː
Peru pəˈɹuː
settee sɪˈtiː
shampoo ʃamˈpuː
taboo təˈbuː
tattoo təˈtuː
 Stress rules 127

(X5) With unstressed final vowel

cashew ˈkaʃuː
cuckoo ˈkʊkuː
emu ˈi:mjuː
guru ˈɡuːɹuː
haiku ˈhaɪkuː
igloo ˈɪɡluː
lichee ˈlaɪʧiː
menu ˈmenjuː
sari ˈsɑːɹiː
voodoo ˈvuːduː

Many names, of both people and places, have second-syllable stress, al-
though again this is not a general rule, and there are many exceptions. Ex-
amples are given in (X6), and some examples of the regular pattern appear
in (X7).

(X6) 
Alsace, Argyll, Assam, Berlin, Braemar, Caithness, Cornell,
Dalkeith, Diane, Dumfries, Hong Kong, Iran, Iraq, Joanne,
Louise, Madrid, Penrith, Penzance, Quebec, Seville, Torbay and
Ukraine

(X7) Names with expected initial stress: Albert, Berwick, Bradford,

Burnley, Doris, Edward, Falmouth, Frances, Gertrude, Louis,
Mary, Olive, Oxford, Richard and Robert.

Many disyllables with second-syllable stress are foreign words, from a

host of languages. This includes some placenames. (Some such examples
have already been given above.) Again, this is not a general rule: there are
foreign words with initial stress, too.

(X8) 
Foreign words with second-syllable stress: adieu, amen, amok,
baroque, bizarre, carafe, champagne, ensemble, façade, galah,
kaput, motif, papoose, pastiche, petite, regime, tycoon, typhoon
and unique

(X9) 
Foreign words with initial stress: argot, ayah, bagel, banco, bimbo,
bonsai, Boris, chalet, chassis, codex, Dali, Datsun, Delphi,
dugong, fakir, gigot, kiosk, shaman, tulip, wigwam and yashmak
128 Phonology

10.3 Trisyllables and longer words

Many of the rules of English stress apply in much the same way to trisyllables
and to longer words as they apply to shorter words. Similarly, some of the rules
that are provided here for longer words can be thought of as applying equally
to shorter words. It is just that different rules provide more coverage in differ-
ent places. For example, the first-syllable stress rule (except with prefixes) will
apply to many trisyllabic words though less to longer words (see Section 7.5).

10.3.1 Rules with suffixes

Suffixes are important in determining where stress falls in longer words. To
apply these rules, though, students have to be able to recognise suffixes. This
is not necessarily as easy as it sounds. The difficulty arises not only because
English has so many suffixes and because there are many different suffixes
which are written the same way (e.g. the -er in bigger and the -er in helper and
the -al in personal and the -al in arrival) but also because there are strings of
suffixes which can be considered single affixes or sequences of affixes (e.g.
-ical in historical), because there are words in which things that might look
like affixes are not straightforward affixes (e.g. -itis in Mondayitis) and be-
cause there are words which might look as though they end in a suffix but
which do not. A simple example of this last point is provided by fiancee. It
looks as though it contains the suffix -ee that we find in employee, but it
doesn’t, because it is pronounced [eɪ] rather than [iː]. A word like bargee
(which is sometimes included in the set of suffix -ee words) probably does not
contain the same suffix, because barge is a noun, and most words with -ee are
built on verbs; in this case, though, it may not matter, because bargee is
stressed in the way that it would be if the -ee were the suffix found in em-
ployee. Quite apart from these problems, we also find occasional exceptions
to general rules. Despite all these problems, it is still worth looking at suffixes
as helping to determine stress position because in so many instances, they
provide vital information.
We can begin with suffixes which take the main stress of the word (these
are sometimes called autostressed suffixes). Relevant suffixes are listed in
(X10) with examples, and comments where necessary.

(X10)

Suffix Examples Comments

-(i)ana Victoriana This suffix is very rare.
-ation compilation, If -ation is seen as a sequence of suffixes, a
investigation different rule applies.

(Continued)
 Stress rules 129

Suffix Examples Comments

-ee addressee, employee
-eer auctioneer,
mountaineer
-ese Chinese, journalese
-esque Byronesque, Mostly used on proper names
picturesque
-ette cigarette, kitchenette
-teen sixteen, nineteen First syllable stress often heard when the
next syllable is stressed (e.g. nineteen people
[ˈnaɪntiːn ˈpiːpəl]). Although these words
are only disyllabic, they are included here
because of the stress behaviour of -teen.

Next we find suffixes which regularly cause stress to fall on the syllable
immediately before the suffix.

(X11)

Suffix Examples Comments

-ian Egyptian, musician The -i is sometimes an extender or part of
the base.
-ible digestible, impossible The suffix -able behaves less predictably
with regard to stress.
-ic historic, pedantic Does not apply to nouns like arithmetic,
Catholic and rhetoric.
-ine elephantine, A rare suffix
labyrinthine
-ion correction, conversion This accounts for the stress in words like
conseˈcration, parˈtition and revoˈlution
where the stress falls on what may be an
extender (see Section 8.3).
-ity maternity, personality
-ive coercive, objective
-ory derisory, refractory
-ure adventure, conjecture,
procedure

There are many suffixes which do not affect the stress. Given words such
as ˈaverage, conˈtribute and acquiˈesce, the stress on the derivative with one
of these affixes remains on the same syllable: ˈaveragly, conˈtributor and
acquiˈescingly.
130 Phonology

It is notable that all the inflectional affixes belong in the group that do not
affect the stress, as do many derivational affixes, particularly those that are
not of Latin origin. Suffixes in this set are given in (X12).

(X12) 
-dom, -ed, -en (adjective, participle and verb), -er (comparative),
-er (noun), -ess (there are a few exceptions), -est, -ful (adjective),
-ing, -ish (adjective), -ise, -ism, -less, -like, -ly (adjective and
adverb), -ment, -ness, -some, -ward(s) and -y (adjective)

The result of many of these rules is to put stress on the antepenultimate

syllable (syllable 3 from the end of the word). This is usually a good default
rule if other rules fail, although, again, there are exceptions.

Unexpected early stresses: ˈadmirable, ˈimpious (~ imˈpious),

(X13) 
irˈrevocable (~ irreˈvocable) and ˈpreferable (~ preˈferable)

10.3.2 Stress with prefixes

The difficulty in dealing with prefixes is that so many of them in English are
borrowed and are not necessarily transparent. Reside may contain an ety-
mological prefix, but in present-day English, we cannot divide it meaning-
fully into re- and side, for example. Even English prefixes may not be
transparent: behave and believe contain the prefix that we also find in be-
witch, but it can be given no meaning in these examples, and there is no
longer a relevant base independent of the prefix.
Where prefixes are transparent, the general rule is that they are not
stressed, but there are so many counterexamples, not only those that arise
through stress changes with verbs (illustrated in [X2] above), that it may
sometimes seem that even this simple rule is perverse. Moreover, the lack of
stress on a prefix may be overruled if a suffix in the same word requires the
stress to fall on the prefix, often by demanding an alternation between
stressed and unstressed before the suffix itself. It is not that there are no
rules here, it is simply a matter of diminishing returns: where stress falls on
a prefix, it is easier to learn that stress as an exception than to compute the
stress.

10.4 Compound stress
Dealing with stress in compounds in English is difficult for two reasons:
there is no general agreement about what a compound is, and there is great
 Stress rules 131

variability. Therefore, rather than attempt full coverage here, only some fun-
damental rules will be given.
The first thing to notice about English compounds is that the spelling is
inconsistent. We can find coffee pot, coffee-pot and coffeepot and rain forest,
rain-forest and rainforest. Despite this, there are some general principles
which we can observe.
The second question is just what constitutes a compound in English. Ex-
perts differ as to what they include under the label. Everyone agrees, though,
that there are some noun + noun structures that count as compounds. For
present purposes, we can assume that any noun + noun sequence that is
written with no gap between the nouns is a compound. The question is
whether things that are written with a space between the nouns are also
compounds. According to these rules, rainforest is a compound, but the sta-
tus of rain forest is unclear. This does not make much sense since the spelling
is fairly random. But we can nevertheless use this to help predict stress.

(CS1) 
Any noun + noun compound written without a space between
the nouns takes its stress on the first noun. The stress falls where
it would fall on the noun in isolation.

According to this, rainforest is stressed on rain, and coffeepot is stressed

on coffee, and that is an accurate reflection of usage. The use of hyphens in
English compounds is subject to fashion, and currently the fashion is not to
use many of them. Where they are found, they can be treated as not being a
space, so that coffee-pot and rain-forest are, by this rule, also stressed on the
first noun, which is still true. There is a major class of exceptions here: where
the two nouns indicate different aspects of the same entity and the hyphen
can be replaced by a slash (although the slash is less usual in English than in
some other languages) or when the hyphen can be replaced by an en-dash
and the nouns linked show the end points of a journey or collaborators
(sometimes opponents), the stress falls on the second noun. Examples are
given in (X14).

(X14)

Meaning Examples
two aspects of a single entity fridge-freezer, poet-painter, singer-songwriter
two places that are seen as a Alsace-Lorraine, Minneapolis-St Paul,
unit Napier-Hastings

(Continued)
132 Phonology

Meaning Examples
end-points of a journey London–Edinburgh, Paris–Rome,
end-points of a figurative French–-English, English–Japanese
journey
collaborators Mercedes-Benz, Creutzfeldt-Jakob
opponents Chelsea–West Ham, England–Australia

Many of these types are used first and foremost in attributive position:
The London–Edinburgh train, a French–English dictionary, Creutzfeldt-Jakob
disease and the England–Australia test match.
It is where there is a space that problems arise. For some, the space indi-
cates that the first noun is attributive but is not a compound; for others, rain
forest is a compound but one whose stress cannot be predicted from the
spelling; for yet others, a further criterion is required to determine whether
or not we are dealing with a compound. None of this is helpful. We will say
that, whatever the status of the noun + noun construction with an interven-
ing space, we need to think about the stress pattern.
No simple method is completely reliable. The most reliable way of pro-
ceeding is as in (CS2).

(CS2) 
Look for other collocations with the same second noun, and use
the same stress pattern.

Consider two simple examples.

(X15)

Stress to the left Stress to the right

Fifth Street, Leeds Street, Madison Fifth Avenue, Leeds Road, Madison
Street, Princes Street, Regent Street Road, Princes Crescent, Regent
Terrace
apple cake, cheesecake, cherry cake, apple pie, cheese pie, cherry pie,
chocolate cake, strawberry cake, chocolate pie, strawberry pie

In the first example here, all street names take left-element stress if they
include the word Street (which, at one stage in the history of English, was
hyphenated to the first noun in the collocation); all other street types (ave-
nue, crescent, road, terrace, way, etc.) take right-element stress. A similar
 Stress rules 133

distinction is made between types of cake and types of pie. Examples like
these seem to suggest that there is no principled division between two con-
struction types; it’s just that some take left-element stress, and others take
right-element stress. The principle does not always work. Sometimes differ-
ent kinds of English use different patterns (ˈice cream is usually thought of
as American English and ice ˈcream as Standard Southern British English
(SSBE) – although the division is no longer as clear-cut as that suggests).
Avocado oil, canola oil, sunflower oil and vegetable oil all take first-element
stress, but in SSBE, olive oil takes second-element stress.
One pattern gives a reliable stress pattern: where the compound is made
up of noun + verb+er, where the noun is the direct object of the verb, as in
ˈbookseller, ˈbus driver and ˈopera singer, lefthand stress is the norm, but not
in town ˈcrier where town is not the direct object of cry. Even with this pat-
tern, there are some exceptions, like party ˈleader.
We can give rules for structures with adjective + noun in a way which re-
flects the noun + noun patterns.

(CS3) 
Where adjective + noun is written as a single word, it is stressed
on the adjective (e.g. ˈblackbird and ˈhighland).

(CS4) 
Where adjective + noun is written with a space between the
words, stress falls on the noun (e.g. blue ˈwhale and pink ˈribbon).

As we might expect from what happens in the noun + noun cases, it is (CS4)
which leads to problems. First of all, there are many types with a space be-
tween the adjective and the noun, ranging from ordinary noun phrases (e.g.
cold water) to specialised classifying terms (e.g. red wine), from adjectives
which are gradable (e.g. pretty picture) to those which are not (e.g. canine
tooth), and from expressions which denote subtype of noun (all of the exam-
ples listed in this paragraph) to ones where the whole expression does not de-
note a type of the noun (e.g. big top ‘circus tent’ and round robin ‘letter sent to
many people’). Despite this, the rules in (CS3 and CS4) hold quite well. One
major set of exceptions, illustrated in (X16), has stress on the adjective when
the adjective is a learned one. Although there are some patterns here (consider,
in particular, the nouns involved), the examples have to be learnt individually.

(X16) 
Adjective + noun examples with stress on the left: dental
hospital, digestive system, dramatic society, medical school,
musical box, nervous system, primary school, professorial
board, secondary school and solar system
134 Phonology

There are some instances where it is not immediately clear whether the
first element is a noun or an adjective. The first of these involves words end-
ing in -ing. In the Empire State building, I like driving, and Swimming is good
for your health, the words ending in -ing are nouns. In an interesting book,
The result was surprising, and An enterprising young man, the words ending
in -ing are adjectives: they can be modified by very, for example, and can take
an adverbial -ly. When a noun is modified by an -ing adjective, the stress falls
on the noun (in line with [CS4]); when a noun is modified by an -ing noun,
the stress falls on that noun (a rule not given above). You have to consider
the meaning to see where the stress falls.

(X17)

Example 1 Gloss Example 2 Gloss

(noun) (adjective)
a ˈdriving ‘a licence for driving’ driving ˈrain ‘rain which drives hard
licence into things’
ˈlaughing ‘a gas which makes a laughing ‘the kookaburra, which
gas you laugh (nitrous ˈjackass sounds as though it is
oxide)’ laughing’
ˈshooting ‘a place for shooting a shooting ˈstar ‘a star which shoots
range at targets’ across the sky’
a ˈsitting ‘a room for sitting in’ a sitting ˈduck ‘a duck which is sitting
room still (mostly used
figuratively)’
a ˈtalking ‘a point to talk a talking ˈdoll ‘doll which can talk’
point about’
ˈvisiting ‘time set aside for a visiting ‘a professor in one
hours people to visit’ proˈfessor institution who is
visiting another’

Other adjectives/nouns can also look identical. Adjectives/nouns of na-

tionality are a case in point. We can have the French language and a French
motorway (French is an adjective), and we can have French is a beautiful lan-
guage and The French eat healthily (French is a noun). The phrase a French
teacher is ambiguous in writing. If it means ‘a teacher who is French’, French
is an adjective, and the stress falls on teacher; if it means ‘a person who
teaches French’, French is a noun, and the stress falls on French.
Adjectival compounds can have a noun in first position or another adjec-
tive. Adjective + adjective compounds take second-element stress, and noun
+ adjective compounds vary.
 Stress rules 135

(X18)

Pattern Examples
noun + adjective: stress on airtight, bloodthirsty, colour blind, fireproof,
the left homesick, seaworthy, threadbare, workshy,
worldwide
noun + adjective: stress on bone lazy, grass-green, lemon-yellow, scot-free,
the right sky-blue, sky-high, stone-cold, top heavy,
word-final; note the stress on the left when
followed by a stressed syllable: Her dress was
sky-ˈblue and She wore a ˈsky-blue dress.
adjective + adjective: stress blue-green, dark green, historical-philosophical,
on the right red-hot, spicy hot

Many of these adjectives with right-hand stress appear to change to left-

hand stress when they are used before a noun, as illustrated by sky-blue in
(X18).
Stress on other types of compound is variable, both in the same individ-
ual and within the speech community.

10.4.1 Highlights
Although stress in English words is, in principle, predictable, the rules are so
complex that they cannot be used for teaching purposes. As a result, we must
use some rather less reliable guidelines, some of which are more reliable than
others but none of which is perfect.

10.4.2 What should the teacher do?

The teacher needs to consider the words that are to be taught to a particular
cohort of students and work out which rules are likely to be of most help at
that level. Other guidelines may be used to draw attention to patterns when
they present, but most should not be given as rules. The rules which involve
suffixes are the most reliable, but they are less likely to be useful at elemen-
tary levels.
11
FREE VARIATION

Although the focus in earlier chapters has always been on places where the
variation between sounds has been driven by the context in which they are
found, we have also seen instances where speakers are apparently free to
choose between different phonemic renditions of particular words or se-
quences of words. This apparently free variation is usually not entirely free –
it partly reflects dialect, social class, age, gender and so on although not
entirely. Sometimes it determined in part by context: although neither can be
pronounced either [naɪðə] or [niːðə] in English, and I personally generally use
the former, in the expression me neither, I typically say [niːðə].
In the first part of this chapter, we consider alternations between sounds
which show this ‘free variation’. Often the sounds which show variation are
the same as those we have seen in earlier chapters though not always. In this
sense, the relationship between the phonemes is part of phonology, although
many would see this material as being extra-phonological. Although very
few examples of each type are presented, some patterns of variation are very
common, while others are extremely rare. The automatic cases of variation
(such as that between [ɪ] and [ə] in unstressed syllables) are not listed here.
Neither is variation with foreign sounds, as in Bach ([k] ~ [x]), nor variation
which distinguishes kinds of English, as with dance ([ɑː] ~ [a]). Alternations
which are not restricted to single pairs of phonemes (e.g. forehead [fɔːhed] ~
[fɒrɪd]) are also ignored. Incidentally, the examples given here cannot be ex-
haustive – you are likely to meet others.

DOI: 10.4324/9781032637020-14
 Free variation 137

(X1)

1 2 3 Examples
a ɑː graph (and words containing this element, such
as telegraph)
a eɪ patent
ɑː eɪ data
e iː economic, ecumenical, genus, haemaglobin,
Megan, oestrus, penalise, scenic, zebra
e eɪ again, ate
f v dwarfs/dwarves, nephew, roofs/rooves, wharfs/
wharves
iː aɪ either, neither
ɔː ʊə moor, poor, sure, tour
s z adhesive (and many other examples before -ive), us
t θ Anthony
ʃ ʒ Asia(n)
ʧ ʤ Dulwich, Norwich, sandwich, spinach
ʊ uː broom, roof, room
ʒ ʤ beige, garage, genre, gigolo
θ ð oaths, thither, though, truths, with
e ɪə stereo
ɒ ɔː ʌ because
e iː premature
a ɑː eɪ shaman
t Ø often

Until recently, the words prince and prints were pronounced differently –
as indicated by the spelling – with a [t] in the latter and not in the former.
Increasingly, a [t] is heard in both: for some speakers a very brief [t], for oth-
ers an ordinary [t] sound, so that prince and prints sound the same. This
change affects a large number of words, including chance, compliance, dance,
presence, reference and silence. In many cases, what were once audibly differ-
ent words have become homophones because of this. Many younger speakers
can now not hear the difference between prince and prints or between refer-
ence and referents even if they meet someone who makes the distinction.
A similar change, but one that does not lead to homophony, affects words
with [n] followed by a post-alveolar consonant. A word like French can be
heard as [frenʃ] or as [frenʧ], lunch can be [lʌnʃ] or [lʌnʧ], range can be [ɹeɪnʒ]
or [ɹeɪnʤ] and ginger can be [ʤɪnʒə] or [ʤɪnʤə]. In this set of alternations,
the use of the affricate is to be preferred.
Next, we consider free variation in stress patterns. Some of these are ap-
parently random, but many are motivated either by changes in English
138 Phonology

(often leading to stress falling on the antepenultimate syllable, especially

where it once fell earlier in the word) or by a need to make the word more
transparent for the listener. The differences are called ‘conservative’ and ‘in-
novative’ in the sense that there is usually a more traditional and a new pat-
tern which exist side-by-side. Individual speakers may use stress patterns
from either or both of these columns.

(X2)

Variation in stress patterns

Word Conservative Innovative Motivation

anchovy anˈʧəʊvi ˈanʧəvi antepenult
Caribbean kaɹɪˈbiːən kəˈɹɪbiən antepenult, spelling
cigarette sɪɡəˈɹet ˈsɪɡəɹet antepenult
comparable ˈkɒmpəɹəbəl kəmˈpeəɹəbəl transparency,
antepenult
complex (adjective) kəmˈpleks ˈkɒmpleks
contribute kənˈtɹɪbjuːt ˈkɒntɹɪbjuːt antepenult
exquisite ˈekskwɪzɪt ɪkˈskwɪzɪt
impious ˈɪmpiəs ɪmˈpaɪəs transparency
irrevocable iˈɹevəkəbəl ɪɹɪˈvɒkəbəl antepenult
jubilee ˈʤuːbɪliː ʤuːbiˈliː analogy with other
words in -ee
kilometer ˈkɪləmiːtə kɪˈlɒmɪtə antepenult
lamentable ˈlaməntəbəl ləˈmentəbəl transparency,
antepenult
necessarily ˈnesəsəɹɪli nesəˈseɹɪli antepenult
pejorative ˈpiːʤəɹətɪv pɪˈʤɒɹətɪv antepenult
primarily ˈpɹaɪmərɪli pɹaɪˈmeɹɪli antepenult
research ɹɪˈsɜːʧ ˈɹiːsɜːʧ
substantive (adj) səbˈstantɪv ˈsʌbstəntɪv antepenult

Highlights
There is a considerable amount of fairly random variation in the pronuncia-
tion of individual words, particularly in the quality of vowels and in the
stress patterns.

What should the teacher do?

The content of this chapter does not require any action on the part of the
teacher, only awareness of permissible variability.
PART IV

Spelling
12
ENGLISH SPELLING

English spelling has a long history, going all the way back to ancient Rome,
which is where our alphabet comes from. In the last two millennia, letters
have been added and letters have been lost, but the fundamental structure of
our system was first devised to deal with a language that had only five vowel
qualities (which could be long or short), whereas modern English has ap-
proximately twice as many vowels as that (depending upon the variety of
English you speak). If there were no other problems with English spelling,
this would be enough to cause headaches.
When Old English (sometimes called Anglo-Saxon) was written a thou-
sand years later, the need was felt for some extra letters to expand on the
letters available in the Roman alphabet. The letter <Þ, þ> (capital and lower
case respectively), called ‘thorn’, was added for the initial sound in the word
thorn, and <Ƿ, ƿ>, called ‘wynn’, was added for the sound [w]. The last trace
of thorn remains in the spelling <ye> in Ye Olde Tea Shoppe. The letter <Ð,
ð>, called ‘eth’ or ‘edh’, though the name is relatively modern, was originally
used to mean the same as thorn but was later distinguished. Thorn and edh
are still used in modern Icelandic to distinguish [θ] and [ð]. The letter <ȝ>,
an open <g>, called ‘yogh’, was originally used in place of all the various
sounds that were written with <g>, namely [ɡ], [j], [ɣ] (mostly leading to
modern English [w]) and [ʤ].
With the arrival of French-speaking people with the Norman Conquest,
a number of French spelling conventions were introduced. For example,
French <ch> replaced Old English <c> for [ʧ], <gu> was introduced to
mean [ɡ] in words like fatigue and guest, <qu> replaces Old English <cƿ>
(corresponding to a modern *<cw>), the vowel [uː] was represented by

DOI: 10.4324/9781032637020-16
142 Spelling

French <ou> rather than the Old English <u>, and this later became [aʊ] in
English with the Great Vowel Shift.
It was not until the second half of the fifteenth century that printing was
introduced, and the leading name here is William Caxton. Caxton grew up
in England but worked for thirty years in the Low Countries and brought
with him some Dutch/Flemish spelling conventions when he returned to
print English books (many of which he translated himself). Perhaps his most
famous innovation is the use of <gh> in the word ghost (on which ghastly
was later modelled). But he standardised the use of London English in print
and started the standardisation of English inflections.
In Shakespeare’s King Lear, Kent insults Oswald by calling him ‘Thou
whoreson zed! thou unnecessary letter!’ thus suggesting that, although <z>
had been used before that date, it was still not seen as a central part of the
English alphabet. That is the most recent addition to the English alphabet.
At the same time, the pronunciation of English has changed considerably.
For example, the final <e> in words like come, love and name was pro-
nounced as a [ə] sound at the time of Caxton – and was written with an <e>
as is still the custom in modern Dutch. More importantly, between the time
of Caxton and approximately the time of Shakespeare, the long vowels of
English underwent a massive change of pronunciation, known as the Great
Vowel Shift. The Great Vowel Shift is the reason that the French pronounce
the name for the letter <i> rather like our [iː], while in English the letter is
called [aɪ]. Quite apart from that major upheaval, there have been various
vowel mergers and splits. For example, the words puss and pus had the same
vowel at the time of Caxton and the words beet and beat had different vow-
els (as they still do in some kinds of Irish English), and our spelling reflects
the old situation rather than the new in these examples. The result is that our
spelling became outmoded before it was even standardised, and our spelling,
in its fundamentals, is still trying to write a kind of English that was spoken
some 700 years ago.
English spelling has not been helped by the fact that English has always
tended to use the spelling of the donor language when it borrows a new word
from elsewhere. We write <chauffeur> rather than <showfur>, and <bou-
tique> rather than <booteek>. We write <yacht> rather than <yot>, and
<schism> rather than <skizzem>. English does not always borrow the pro-
nunciation of the donor language (or even a close approximation to it) but
does like to use the original spelling where relevant. We pronounce <tsu-
nami> as [suːnɑːmi] rather than with an initial [t] (the [ts] in Japanese is a
variety of [t] occurring before [u]), and quinine (from Spanish) is pronounced
with an initial [kw] rather than the [k] we would expect from Spanish (con-
trast with quinoa where the Spanish initial [k] is retained).
 English spelling 143

All of these factors make English spelling difficult to deal with. Yet when
critics of the system complain about it, they almost invariably use two ex-
amples: how to pronounce <ough> and how to spell fish. The final <gh> in
words as varied as borough, plough, through and tough indicate a historical
pronunciation with [x], which no longer exists in Standard Southern British
English. Sometimes that <gh> consonant was lost, sometimes confused
with [f] (with which it was auditorily surprisingly similar – consider the Brit-
ish pronunciation of the name Van Gogh with final [f], coming from Dutch
[x]), and so on. The result in modern English is that the same written se-
quence <ough> can have several pronunciations. This is true, but the pro-
nunciations are not endlessly variable. The idea that fish could be spelt
<ghoti> in English is attributed to the playwright George Bernard Shaw:
<gh> as in enough, <o> as in women, and <ti> as in nation. This ignores the
system of English spelling, where <gh> is pronounced [f] only at the ends of
words, <o> is pronounced [ɪ] only in the word women, and <ti> is pro-
nounced [ʃ] only in the middle of Latin or French words, especially those
having the (originally Latin) ending -tion.
Although there is a certain amount of personal opinion involved in de-
ciding what is ‘regular’ in English spelling, recent investigations suggest that
English spelling may be something like 70% regular (some say even more).
This suggests that English spelling is not a hopeless case, that much of the
pronunciation of a word can be deduced from its written form (absolutely
not all), and that teaching phonics in schools has some value for teaching
first-language (L1) learners to read.
It is often assumed that an ideal spelling system reflects (more or less di-
rectly) the pronunciation. English spelling tells us about things as well. The
fact that <ch> is pronounced [ʧ] in church, [ʃ] in chauffeur and [k] in chemis-
try tells us that church is a basic English word, chauffeur is French and chem-
istry is Greek in origin. The fact that sign and signal contain the same
sequence <sign> tells us about the meaning rather than about the pronun-
ciation. The fact that know is written with an initial <k> tells us about the
history of the word rather than about its present pronunciation. How much
we care about these things might depend on our goal in using the spelling
system, but we should not criticise the system because it is not doing one
thing when it is doing another perfectly well.
We also need to think that initial <kn> is regularly pronounced [n] but
that [n] is not regularly written <kn>. That is, what counts as regular may
depend on whether we are reading or writing. For second-language (L2)
learners, whose vocabulary is often limited, getting from spelling to pronun-
ciation is often more important than being able to go the other way. Accord-
ingly, interpreting the spelling will be given greater focus in what follows.
144 Spelling

Highlights
The complex history of English spelling means that English spelling is often
doing more than simply indicating pronunciation. English spelling is more
regular than is often claimed.

What should the teacher do?

This chapter provides background only, but the teacher should note that some-
times the English spelling system is not simply indicating pronunciation.
13
SPELLING CONSONANTS

In the rules set out below, the most important rules for the spelling of Eng-
lish consonants are given. Important exceptions are noted. The fundamental
rules, which students need from the very beginning of learning English are
marked by bold type. Where the rules involve digraphs (i.e., a sequence of
two letters indicating a single sound, such as <sh> for the sound [ʃ]), the two
letters may be pronounced separately over a word boundary or affix bound-
ary: dishonest and mishear have a sequence of [sh] rather than [ʃ]. An attempt
has been made to order the rules so that the last rule applies if the conditions
for earlier rules have not been met.

B 1. <mb> at the end of a word or a possible word is pronounced [m], so the

<b> is silent: thumb, lamb, lambs, lambing. Thimble and lambaste have
the <b> pronounced in the usual way because the <b> is not at the
end of a possible word. (See [X5] in Chapter 9.)
2. <b> is silent in a few words like debt, doubt and subtle because [bt] is not
a possible combination except over word boundaries.
3. <b> and <bb> are pronounced [b]: abbey, amber, bat, brain, bulb, crab
and hob.
C 1. <ch> and <tch> are pronounced [ʧ] in basic English words: child, chip,
choose, church, fetch and match; <ch> is pronounced [ʃ] in French words:
chauffeur, cache, douche, gauche and moustache; <ch> is pronounced [k]
in Greek words: architect, chaos, chemist, mechanic, orchestra and
psychology and also in the English word ache; <ch> is pronounced [k] in
German words: Bach and Munich and also in some Scottish words where
the original is [x] or [ç]: Brechin, Buchan and Leuchars.
(Continued)

DOI: 10.4324/9781032637020-17
146 Spelling

2. <cc> in a few words from Latin (often via French) following <a> and
with the next letter either <e> or <i> is pronounced [ks]: accent,
access, accident and vaccine. There are not many words involved, but
the rule works well.
3. <c> is pronounced [ʧ] in Italian words before <e> or <i>: cello and
concerto; <c> is pronounced [s] before <e> or <i> or <y>: ace, advice,
cell, city, cycle, mercy and suicide; <c> and <cc> are pronounced [k] at
the end of a word, before a consonant letter or one of the vowel letters
<a>, <o> or <u>: account, act, cake, clutter, coal, cut, epidemic and
product; <c> is pronounced [ʃ] before <i> and another vowel letter:
ancient, delicious, official and social. This last option is often
dependent on the nature of the following affix: glacier can be
pronounced with [s] or with [ʃ].
4. <ck> acts as double <c> and is pronounced [k]: attack, heckle, lack,
packet and sick.
D 1. <dg> acts as a double <g> and is pronounced [ʤ]: badge, bridge and
budget.
2. <d> and <dd> are pronounced [d]: bad, dig, ladder, maiden and sudden.
3. The rules for the pronunciation of the suffix -ed are morphophonemic
and are dealt with in Section 9.1.
F 1. <f> and <ff> are pronounced [f] except in the word of, which is [ɒv], thus
contrasting with off: differ, foreign, from and scarf.
G 1. <g> is silent at the end of a word or possible word before <m> or <n>
and at the beginning of a word before <n>: align, diaphragm, reign,
sign, gnaw and gnome; any preceding vowel spelling usually contains
<i>, but champagne is a counterexample. There are some French
counterexamples where <gn> is [nj]: cognac and poignant.
2. <gh> is usually pronounced [ɡ] except at the ends of basic words that
end in <igh> or <ugh>: gherkin, ghost, spaghetti; enough, laugh, neigh
and sigh.
3. <g> is usually pronounced [ʤ] before <e>, <i> or <y>: general, genius,
gin, prodigy and knowledge. There are numerous counterexamples such
as begin, gear, geese, get, giggle, gill (of a fish), gird, girl, give,
gynaecology and tiger; some differences are predictable depending on
the source of the word (French or basic English), but this is hard for
the learner to see.
4. <ng> is pronounced [ŋ] at the end of a word, before an inflectional affix
or the suffix -er, before a consonant or when the following letter is not
<e>, <i> or <y> (when the <g> is pronounced [ʤ]; see G3): hangar,
length, ring, ringing and singer. It is pronounced [ŋɡ] in the middle of a
meaningful unit – anger, dangle and finger – and unexpectedly in
longer, stronger and younger.
5. <g> and <gg> are pronounced [ɡ]: bag, egg, go, grow, maggot, rugby and
wiggle. In a few words like suggest, rule G3 overtakes rule G5.
H 1. <ch> and <tch> are pronounced [ʧ]: choose and watch.
2. <gh> is usually pronounced [ɡ] (see G2 above)
(Continued)
 Spelling consonants 147

3. <ph> is pronounced [f] in words of Greek origin: alphabet, elephant,

photo, phrase and sphere.
4. <sh> is pronounced [ʃ]: clash, fashion, fish, ship, shoe and shop.
5. <th> is pronounced [θ] or [ð]. Initially, it is pronounced [ð] only in
grammatical words like the, this, there and though; in other words,
initial <th> is pronounced [θ]: thatch, theatre, think and throat. Where
<th> is followed by a silent <e> at the end of a word, it is pronounced
[ð]: breathe, clothe, loathe and writhe. Final <th> is usually [θ]: both,
cloth, myth and youth: note the [ð] in mouth (the verb) and with (where
either [θ] or [ð] can be found).
6 <wh> is pronounced [w]: what, when, whisker and why. Note who and
words derived from it and whooping cough, which are pronounced with
[h].
7. <h> at the end of a word, where [h] is never pronounced, is silent but
usually indicates a long vowel: shah and yeah.
8. <h> is pronounced [h]: behave, heaven, here and high.
J 1. <j> is pronounced [ʤ]: eject, jam, job, joke, just and major.
K 1. <k> at the start of a word or potential word followed by <n> is silent:
foreknowledge, knee, knife, knock and know.
2. <k> (or the rare <kk>) is pronounced [k]: akin, brekkie, joke, keep,
kitchen, leak and rebuke.
L 1. <l> is silent in a small number of common words: calf, calm, chalk, folk,
half and yolk.
2. <l> and <ll> are pronounced [l], except for <ll> in some Welsh names or
Spanish words: fill, flame, like, pillage, silent and wool.
M 1. <m> is silent initially in the word mnemonic.
2. <m> and <mm> are pronounced [m]: autumn, comma, hum, made, money
and summer.
N 1 <n> is silent after <m> at the end of a word or before an inflectional
suffix: autumn, column, damn, damning and hymn.
2. <ng> is pronounced [ŋ] at the end of a word, before an inflectional affix
or the suffix -er, before a consonant or when the following letter is not
<e>, <i> or <y> (when the <g> is pronounced [ʤ]; see G3): hangar,
length, ring, ringing and singer. It is pronounced [ŋɡ] in the middle of a
meaningful unit – anger, dangle and finger – and unexpectedly in
longer, stronger and younger.
3. <n> is pronounced [ŋ] before <c> pronounced [k], <g>, <k>, <q> and
<x>. This means before the sounds [ɡ] and [k]: anger, anxious, bank,
banquet, larynx, uncle and uncut. N2 and N3 work together, though
they are sometimes awkward: anger and manger do not rhyme (manger
comes from French), singer and linger do not rhyme, and longer ‘more
long’ and longer ‘one who longs’ – rare) are not homophonous.
4. <n> and <nn> are pronounced [n]: diner, dinner, hewn, lion, manna,
minor, new, nice and pun.

(Continued)
148 Spelling

P 1. <p> is silent before <n>, <s>, and <t>, usually in Greek words:
pneumonia, psittacosis, psychology, pterosaur and ptarmigan. The letter
<p> is also silent in a few French words (coup and corps) and a few
other words: cupboard and raspberry.
2. <ph> is pronounced [f], mainly in Greek words: alphabet, elephant,
diphthong, phrase and sphere. Note that nephew used to be pronounced
with [v] but is now more often heard with the regular [f].
3. <p> and <pp> are pronounced [p]: camp, capable, happy, play, professor
and put.
Q 1 <q> is pronounced [k] in a number of French words, notably those
ending in <que>: antique, bouquet, boutique, cheque, liqueur,
picturesque, quay and queue.
2. <qu> is pronounced [kw]: liquid, quality, queen, quick, quote and request.
R 1. <r> when it does not precede a vowel is often part of the way of writing
the previous vowel sound and is not pronounced in Southern
Standard British English (SSBE), though it is pronounced in
American English.
2. <r>, <rr> and <rh>, <rrh> are pronounced [r] (<rh> mainly in Greek or
Welsh words): carry, diarrhoea, free, merry, pray, red, rhetoric,
rhinoceros and street.
S 1. <s> is silent at the end of a number of French words and in a small
number of other words: chamois (also spelt shammy when it means a
cloth), corps and debris; aisle, isle and island.
2. <sh> is pronounced [ʃ]: fashion, friendship, mash, perish and shop.
3. <sch> is pronounced [sk] mostly in words from Greek and [ʃ] mostly in
words from German or Yiddish: scheme, schnozzle, school, schnapps
and Porsche. Schedule has a conservative pronunciation with [ʃ] in
Britain but is more generally pronounced with [sk].
4. <s> and <ss> are pronounced [ʃ] before <ur> or <ion>: admission,
censure, emulsion, pressure and passion. If the suffix has been added to
a word that ends with [z] or [d], we get [ʒ] rather than [ʃ]: closure,
collision, conclusion, exposure and pleasure; note treasure, which has
no base word.
5. <s> is pronounced [z] between vowel letters and in consonant clusters
with other voiced consonants and when required by the rules for
plural nouns, third-person singular verbs, genitives and so on (see
Section 9.1). This is a messy rule, with a lot of exceptions, but provides
a first approximation. Some of the exceptions can be avoided if it is
recognised that the prefixes dis- and mis- are always pronounced with
[s]. Some examples where the rule holds: chasm, deserve, egoism,
gosling, grisly, husband, organise (also spelt with <z>), phrase and
sarcasm. Note that in many words with the same spelling, there is a
noun-verb distinction, with the noun having a voiceless fricative and
the verb having a voiced one (e.g., in house) (see Section 9.2.1).
6. <s> and <ss> are pronounced [s]: assorted, bossy, mess, mumps, past, say,
sky, slop, stem and yes. Has and is are exceptions, having [z].

(Continued)
 Spelling consonants 149

T 1. <t> is silent at the end of French words and often in clusters in other
words: ballet, bouquet, buffet (‘array of food’ but not ‘shake’), fasten,
gourmet, listen, nestling, tsar and waltz.
2. <t> is pronounced [ʃ] before a few suffixes beginning with <i>: action,
devotion; cautious; dementia; palatial, potential; Egyptian.
3. <th> is pronounced [θ] or [ð]. Initially, it is pronounced [ð] only in
grammatical words like the, this, there and though; in other words,
initial <th> is pronounced [θ]: thatch, theatre, think and throat. Where
<th> is followed by a silent <e> at the end of a word, it is pronounced
[ð]: breathe, clothe, loathe and writhe. Final <th> is usually [θ]: both,
cloth, myth and youth: note the [ð] in mouth (the verb) and with (where
either [θ] or [ð] can be found).
4. <t> and <tt> are pronounced [t]: attend, cattle, mat, ten, tent, title, tree,
skate and what.
V 1. <v> and <vv> (which is very rare) are pronounced [v]: aver, behave, civvy,
knives, leave, navy, navvy, rev, very and view.
W 1. <w> is silent before <r>: wrath, wreck, wrist, write and wrong. It is also
silent in sword and in some place names ending in -wich or -wick.
2. When <w> is not part of the spelling of a vowel sound (which occurs
most obviously before a stressed vowels), <w> and <wh> are
pronounced [w]: beware, sweat, twinkle, watch, water, wet, where,
whisker and word. There are kinds of English where <wh> is
pronounced differently, as [hw], but this is now old-fashioned in SSBE.
X 1. <x> is pronounced [z] initially in Greek words: xenophobia and
xylophone.
2. <x> is pronounced [gz] in the sequence <ex> when the stress falls on the
following syllable (though there are exceptions): exam, example,
exempt, exist and exotic.
3. <x> is pronounced [ks]: axe, axis, box, execute, mixture, Saxon and sex.
Y 1. When it is not part of the spelling of a vowel, <y> is pronounced [j]
before a stressed vowel: beyond, yacht, yellow, yes, you and youth.
Z 1. <z> or <zz> are pronounced [z] except in some Italian and German
words: brazen, fizz, sizzle, sneeze, zip and zoo.

Highlights
This chapter presents a list of the most common spelling-to-pronunciation
rules for consonants.

What should the teacher do?

The teacher needs to be aware of the common spellings and the common
words which have unusual spellings. While many of these rules can be re-
versed (<ng> is pronounced [ŋ] and [ŋ] is spelt <ng>), the teacher needs to
be able to distinguish. The most basic rules should be taught first.
14
SPELLING VOWELS

The spelling of vowels is far more variable than the spelling of consonants,
for a whole lot of reasons, some of them historical, some to do with changes
in pronunciation that have affected English since about 1500, some to do
with patterns of borrowing (so that sometimes we use Latin spellings, for
example), and so on. We also have to cope with the fact that the spelling
system has to use the same set of alphabetical symbols to write short
monophthongs, long monophthongs, long diphthongs and unstressed vow-
els, sometimes using the same symbols in different ways in the different sys-
tems. Knowing which system a particular vowel belongs to is therefore
important though far from easy for a learner (whether a first-language [L1]
or a second-language [L2] learner) to absorb. In particular, the difference
between stressed and unstressed vowels is not shown in the spelling, but stu-
dents need to know which they are dealing with to interpret the spelling
properly. As elsewhere in this book, the systems will be dealt with individu-
ally in this section.

14.1 Short monophthongs
There are five vowel letters of English, six if we include <y>, and six short
monophthongs, so it is immediately clear that there is a problem. The prob-
lem here turns out to be [ʊ], which does not have its own letter. When L1
students are taught to read English, the short values for the letters are taken
to be the default values, and the long values are treated as secondary.
We need to think about how short vowels are specifically marked. There
is no overall system which guarantees this, but typically, at least for basic

DOI: 10.4324/9781032637020-18
 Spelling vowels 151

English words, if the vowel letter is followed by two consonant letters, it is

short. Digraphs (where two letters together are used to write a single sound,
like <sh> for [ʃ]) are usually included in this, so that dishy and richer are
shown to have short vowels in the first syllable. The letter <x> with value [ks]
also acts as two letters, so that fixing is marked as having a short vowel in the
first syllable. The letter <v> is very rarely doubled (usually only in modern,
slangy words, like navvy and revving), and <v> can follow either a short or a
long vowel: even has a long first vowel, seven has a short one, and lever has a
long first vowel in Standard Southern British English (SSBE) but a short one
in American English. Some consonant sequences act as double letters and
thus fit the rule: <ck> acts as double <c> or <k>, <dge> acts as double [ʤ],
and <tch> acts as double <ch>. Some examples of relevant words are pro-
vided in (X1).

(X1)

<a> [a] abstract, actor, alpenhorn, cabbage, mantle

<e> [e] enter, meddle, pendulum, sextet
<i> [ɪ] fiddle, fisticuffs, prismatic, vivid
<o> [ɒ] belong, bottle, obstinate, oxen, socket
<u> [ʌ] brushing, husband, luggage, plumper, sculpture
<y> [ɪ] hymn, pygmy, symbol, tympani

The same is true when two consonants end a monosyllable, but here not
all letters can be doubled. The examples in (X2) show the possible double
letters, and in some of these instances, the extra consonant is added so that
a content word will be at least three letters long, which means that the num-
ber of letters that are regularly doubled in this position is smaller than (X2)
implies.

(X2) 
ebb, add, off, egg, fetch, lack, ill, inn, repp (also spelt <rep>), err,
loss, sett, fizz.

Even if these double letters are possible, many of them are rare, and a
single consonant following a single vowel letter is often more common, as in
(X3).

(X3) cub, tic, mud, clef, bag, trek, nil, sum, sin, sap, gas, get and fez.
152 Spelling

Note, however, that some of the words in (X3) are loan words, not en-
tirely in tune with native English spelling norms. Only a handful of conso-
nants regularly appear doubled at the end of monosyllables:

<f> buff, cliff, huff, sniff, snuff, staff, stuff

<l> ball, bell, call, fill, mall, shell, still, stall, tell, wall
<r> burr, purr
<s> bass, bless, class, cross, hiss, kiss, -less, less, mass, mess, miss, moss,
muss, -ness, pass, puss, truss
<t> butt, mitt, putt, sett, watt
<z> buzz, fizz, fuzz, jazz, whizz

Despite the strength of the two-consonant (or double consonant) rule, it

has many exceptions, some of them systematic, some of them idiosyncratic.
Some of these are set out in (X4) where there is a long vowel despite the fol-
lowing consonants or a short vowel without double consonants. Part of the
difficulty here is that the double-consonant rule applies to basic English
words, not to Latin and Greek words, where a vowel followed by a single
consonant letter can be long or short.

(X4) 
find, mind, rind; chance, chant, glance
staff, calf, aft, craft; bath, path; class, glass, grass, pass, ask, blast
child, mild; Christ
cynic, penalty, rigid
anaesthetic, busy, manic, minute (‘60 secs’), orange, sugar, woman

What is important, though, is that a vowel spelt with a single letter and
followed by a single letter consonant at the end of a monosyllabic word is
never a long vowel unless the final consonant marks the plural, as in gos and
nos (both of which have alternative spellings).
Although the short vowels are mainly spelt with the expected letter as set
out in (X1), there are exceptions, where other letters are used. These are set
out in (X5).

(X5)

<a> is pronounced [e] in any and many.

<a> is pronounced [ɒ] after [w]: squad, swan, wallet, was, watch and what. There
are many counterexamples, including swam, wag and wax.
 Spelling vowels 153

<e> is pronounced [ɪ] in England, English and pretty.

<o> is pronounced [ɪ] in women.
<o> is pronounced [ʌ] in a lot of words, including above, come, cover, love, oven,
monkey and onion.
<o> is pronounced [ʊ] in bosom and woman.
<o> is pronounced [uː] in womb.
<u> is pronounced [ɪ] in busy and business.
<u> is pronounced [e] in bury.
<u> is pronounced [ʊ], especially where <p> or <b> precede: pull, push, puss,
put, bull, bush, butcher, cushion and full.

Largely because of historical changes in vowel length, some of them idio-

syncratic, there are places where short vowels are written with digraphs. Ex-
amples are given in (X6). For practical purposes, these can be divided into
common exceptions to rules and rare words which need not be taught until
they are encountered.

(X6)

[ɪ] <ee> breeches

<ie> sieve
<ui> built, guild
[e] <ai> said
<ea> bread, deaf, feather, heavy, meadow, pheasant, ready,
weather
<ei> heifer, leisure
<eo> jeopardy, leopard
<ie> friend
[a] <ai> plaid, plait
[ʌ] <oe> does
<oo> blood, flood
<ou> couple, touch, trouble
[ɒ] <au> cauliflower, sausage
[ʊ] <oo> book, foot, good, hooray, wood, wool
<ou> could, should, would

14.2 Long monophthongs
Just as there are ways of marking a vowel as being short, there are ways of
marking one as being long, at least in the basic English words.
With monosyllabic words, the vowel can be put finally. Since a short vowel
cannot occur in a stressed syllable at the end of a word, any vowel ending a
154 Spelling

monosyllable must be long, at least when the word is said in isolation. When
the word occurs in speech, it may be unstressed (some of them are typically
unstressed), in which case the rules for unstressed vowels apply. Examples
are given in (X7).

(X7) 
be, do, flu, go, he, hi, I, lo, ma, mi, mo, mu, no, pa, phi, rho, she,
ski, so, spa, the, to, we

Note that there is not always a regular correspondence between the vowel
letter and the pronunciation, depending on the source of the word.
A vowel can be marked as long by being written with a digraph. We have
already seen that there are some cases where a digraph is used for a short
vowel but that most of these are idiosyncratic, so that the digraph implies
a long vowel rule is a reasonably good predictor of a long vowel. Some of
the most common digraphs for the long monophthongs are given in (X8).
Rare digraphs (e.g. <ui> in fruit and <ue> in blue) can be taught as
exceptions.

(X8)

<au> and <aw> are pronounced [ɔː]: author, drawing, faun, lawn, saucer and
taught. Note aunt, which has [ɑː] in SSBE.
<ea> is pronounced [iː]: beaver, clean, defeat, dream, eagle, measles, please,
season and treacle.
<ee>is pronounced [iː]: asleep, beetle, canteen, cheese, feeble, needle, reed and
succeed.
<ei> is pronounced [iː], especially in the sequences <ceit> and <ceive>: caffeine,
ceiling, deceive, protein, receipt and seize. The one rule of English spelling that
the English know is ‘I before E except after C’, and this rule is the ‘after C’
part of that and the exceptions to the rule. The rule is not very useful.
<ie> is pronounced [iː]: achieve, believe, brief, field, niece, piece, retrieve, siege
and thief.
<oo> is pronounced [uː]: bamboo, boot, booze, cartoon, doodle, food, fool, moon
and swoop.

The case that is most likely to cause problems of interpretation here is

<ea>, which is a regular representation of both [e] and [iː]. The infinitive
read has [iː], while the past tense read has [e]; lead ‘to go before’ has [iː], while
lead the metal has [e] (as a result of these two, many L1 speakers are con-
fused as to how to spell led, the past tense of lead); mead has [iː], while the
semantically related meadow has [e]; and so on. There is, unfortunately, no
way to distinguish accurately except by experience.
 Spelling vowels 155

Three of the long vowels – [ɜː], [ɑː] and [ɔː] – are regularly spelt with an
<r> following a vowel letter and not preceding another vowel letter in the
same word, creating a rather different set of digraphs. These spelling-to-
sound correspondences work in SSBE and similar kinds of English but do
not all work in American English and do not work in Scottish English. The
regular patterns are set out in (X9).

(X9)

<ar> is pronounced [ɑː]: car, chart, farm, harmony and martyr. After [w], we find
[ɔː]: quartet, swarm and warm.
<er> is pronounced [ɜː]: fern, kernel and mermaid.
<ear> is pronounced [ɜː] before a consonant: earn, heard, hearse, learn and
search.
<ir> is pronounced [ɜː]: bird, fir, firm, first and sir.
<or> is pronounced [ɔː]: border, cork, corn, corner, orchid and pork. After <w>,
we find [ɜː]: word, work and worse.
<oar> is pronounced [ɔː]: board, hoarse, oar and roar.
<ur> is pronounced [ɜː]: burn, curt and fur.
<yr> is pronounced [ɜː] (this is rare): myrrh and myrtle.

Note that from the point of view of the spelling, [ɜː] usually acts as a
short vowel because the relevant vowels were short when the system was es-
tablished. The spelling <purr>, for example, leads us to expect a short vowel,
and <hearse> gives conflicting information: a digraph for the vowel implies
long, but the final two consonant letters suggest short.
The third way of marking a long vowel is to follow the symbol used to
represent the vowel by at most one consonant letter and then another vowel
letter. When L1 learners are taught to read, this is called ‘magic E’ or the ‘E
that makes the letter say its name’. The letter <e> is picked out here because
the <e>, in modern English, is not pronounced. But when the English spell-
ing system was established, the so-called ‘magic <e>’ was pronounced as a
vowel sound, and the rule still works when other vowels follow, not just <e>.
The fundamental rule is given in (V1), although there are exceptions to this
rule.

(V1) 
A vowel sound written with a single letter is pronounced long
when in a stressed syllable and followed by a single consonant
sound written with a single consonant letter and a following vowel
letter. Examples: behave, came, cute, mode, rise and serene.
156 Spelling

Since the long pronunciation is ‘the name of the letter’, namely [eɪ], [iː],
[aɪ], [əʊ] and [juː], most of the letter names are diphthongs, not monoph-
thongs. Here the relevant examples are [iː] written <e> in the context of a
following consonant letter + a vowel letter and [juː] written <u> in the same
environment (although, depending on the preceding consonant and the va-
riety of English spoken, the [j] may or may not be present).
There is a general exception to this rule when the vowel is followed by two
unstressed syllables. Often this can be reduced to being followed by a few
specific affixes, such as -ity. Serene has a long vowel in the final stressed syl-
lable, following (V1), but serenity has a short vowel, [e]. This is the rule of
antepenultimate shortening. Obesity, with a long [iː], is an exception to the
exception (see Section 9.3.2).
Some examples of the regular case, with different vowels following, are
given in (X10), and some exceptions are given in (X11). Note that both (X10)
and (X11) include words which are not basic English words. The point is that
the rules sometimes seem to apply there, but their application is not guaran-
teed. In British English, pedal has a short [e] but bipedal has a long [iː].

(X10) 
convene, ego, extreme, famous, gamey (also written <gamy>),
hoping, opus, pony, requital, rising, tutor and visor

(X11) agony, famish, honey, money, onion, leper, punish and sinuous

In foreign words, long <i> is sometimes [iː], long <a> is sometimes [ɑː]
and long <u> is sometimes [uː]. Some examples are given in (X12). These
patterns are not reliable. Khaki is pronounced with [ɑː] in British English,
but [a] in American English; cicada can have either [ɑː] or [eɪ]; alpaca has [a].

(X12)

long <i> as [iː] bikini, boutique, casino, kilo, kiwi and mosquito
long <a> as [ɑː] avocado, banana, cantata, drama, mascara, pyjamas,
soprano and tomato (British)
long <u> as [uː] amuse, brute, consume and introduce

14.3 Diphthongs
Because, from a historical point of view, long monophthongs have turned
into diphthongs and some diphthongs have changed into monophthongs,
the spelling system does not make a clear distinction between the two series.
 Spelling vowels 157

We even find cases where the same digraph can be pronounced with a
monophthong or a diphthong. In this section, we have to look at the diph-
thongs individually.
We have already seen that with ‘magic <e>’, [eɪ] is the expected pronun-
ciation of long <a>, [aɪ] is the expected pronunciation of long <i>, and
[əʊ] is the expected pronunciation of long <o>. These are exemplified in
(X13).

(X13)

long <a> agent, made, name, rave and wave

long <i> filing, line and smile
long <o> broke, coping, home, note and open

All of these diphthongs are also written with digraphs. Where [eɪ] is con-
cerned, the digraphs <ay> and <ai> are the most frequent, with <ay> usu-
ally at the end of a word and <ai> when there is a following consonant letter
in the same potential word. Examples are given in (X14).

(X14) 
<ay> and <ai> pronounced [eɪ]: clay, complain, day, faith, mail,
paint, plain, saying, stay, trays, waist and waive.

Other digraphs for [eɪ] are rare but include some words with final <gh>
which are irregular. Some examples are given in (X15).

(X15) 
<et> is pronounced [eɪ] in French words, usually at the end of
the words and most often not stressed. An example with stress is
croquet, pronounced either [ˈkrəʊki] or [krəʊˈkeɪ].
<eigh> is pronounced [eɪ] in eight, neigh, weigh and weight.
<ei> is pronounced [eɪ] in beige, feign, reign, rein, veil and vein.
<ao> is irregularly pronounced [eɪ] in British English in the word
gaol, also written <jail>.

There are fewer potential spellings for [aɪ]. They are exemplified in (X16).

(X16) 
<ie> and <y>, <ye> are pronounced [aɪ] when final in a
monosyllable: die, dye, lie, pie, shy, sky, stye and tie.
<igh> is pronounced [aɪ]: delight, flight, high, sigh and sight.
158 Spelling

<ei> is pronounced [aɪ]: either, height and neither.

<y> is pronounced [aɪ] in Greek words: cycle, rhyme and thyme.

The difference between <y> and <ie> in stressed final position in monosyl-
lables can be seen as a way of keeping a minimum of three letters in a lexical
word. The spelling <ye> is an alternative solution to the problem but far less
common. Either and neither both have alternative pronunciations with [iː].
One change is so regular that we can promote it to the status of a rule,
although there are exceptions.

(V2) 
A word-final <y> is often replaced by <ie> before an affix as long
as the <y> is not part of a digraph for a vowel sound (e.g. baby/
babies, cleanly/cleanliness, comedy/comedian, copy/copies [noun or
verb], fallacy/fallacies, silly/sillier and specify/specified. Note dryer
~ drier [as a noun, not as an adjective] and shyer).

The vowel [əʊ] has two basic spelling alternatives apart from the one with
magic <e>. They are illustrated in (X17), and both are mostly found in
word-final position, <oe> finally in monosyllables, but <ow> anywhere.
There are also some rare spellings, also listed in (X17), which can be treated
as exceptional.

(X17) 
<ow> is pronounced [əʊ]: below, flown, know, own and show.
<oe> is pronounced [əʊ]: doe, floe and toe.
<oa> is pronounced [əʊ]: boat, coal, coach and loaf.
<ew> is pronounced [əʊ]: sew.
<ough> is pronounced [əʊ]: dough and though.

Even though it is not fully stressed, <ow> at the ends of words (especially
two syllable words) is regularly pronounced [əʊ], which sometimes gets re-
duced: billow, shadow and tomorrow.
The diphthong [ɔɪ] is spelt <oy>, or <oi>, and <oy> is usually found at
the end of the word (though there are exceptions). Examples are annoy,
boy, coin, loiter, oil and toy. Note buoy [bɔɪ], pronounced [buːi] in some
places.
The diphthong [aʊ] is spelt <ou> with <ow> finally or before vowels.
Examples are allow, compound, cow, house, mountain, mouth, now, power,
stout and thousand. The spelling <ow> is occasionally found before a conso-
nant (e.g. in brown, clown, crowd and owl), which can be treated as irregular.
Bough and plough show an exceptional spelling.
 Spelling vowels 159

Note the problem that <ow> can be pronounced [əʊ] or [aʊ] and that
<ei> can be pronounced [eɪ] or [aɪ] and even, occasionally, [iː]. To some ex-
tent, these conflicts can be sorted out by looking at the following consonant
or at the meaningful element in which the digraph occurs, but while such
analogies might work for the experienced reader, it is probably too compli-
cated to try to introduce them into overt teaching. A better approach is likely
to be to take the most common spelling–pronunciation link and to treat
other patterns as exceptions to the general rule. While foreign words are
regular causes of exceptional spellings, identifying such words is a skill in
itself, and it cannot be expected that speakers of non-European languages
will be able to do this without some specific training.
The diphthongs [ɪə], [eə] and [ʊə] all arise from the deletion of [ɹ], which
is regularly present in the spelling. They are awkward because the spellings
for [ɪə] and [eə] often overlap (here and there do not rhyme), because [ʊə] is
often replaced by [ɔː] (with an intermediate stage [ɔə]) and because, in
places where an [ɹ] remains (because there is a following vowel), there is
often variation between the diphthong and a monophthong. Increasingly,
[eə] is a monophthong in SSBE and in Australian English: a long counter-
part of [e]. In varieties where the [ɹ] is pronounced in this position, the
vowel is short.
Spelling conventions are listed in (X18) along with examples and
comments.

(X18)

Spelling Sound Examples Comment

<air(e)> [eə] chair, fair, millionaire
<ar(e)> [eə] bare, canary, fare, scarce
<ear> [ɪə] fear, hear, year Note tear ‘liquid from the
eye’.
<ear> [eə] bear, swear, wear Note tear ‘rip’.
<eer> [ɪə] beer, sneer, engineer
<ere> [eə] ampere, there, where
<ere> [ɪə] here, severe, sphere
<oor> [ʊə] boor, moor, poor
<our> [ʊə] gourmet, tour
<ure> [ʊə] cure, sure

Other spellings, such as those in bier, heir and idea, have to be treated as
irregular.
160 Spelling

14.4 Full vowels that are not stressed

Consider a word like excruciating. The main stress is on the syllable contain-
ing the <u>, pronounced [uː], as in (X12). But there is another vowel with
full vowel quality, namely the one in the syllable written <at>. Following the
standard rule for long <a>, this is pronounced [eɪ]. As with this example,
most of the full vowels will have pronunciations which also follow from the
spelling-to-pronunciation rules that have been set out above. However, the
double-letter rule is even less reliable here than with the stressed vowels. If
there is a double letter or a sequence of two consonant letters, the vowel will
be predictably short, as in the <o> in paradox or the first <e> in Tennessee.
But a single following consonant letter does not guarantee that the vowel
will be long. The first <e> in telepathic is short (no doubt because the <e>
in telepath is short), as is the first <e> in mediocrity (where mediocre has a
long vowel for many).
Words in compounds retain the pronunciation they have in their individ-
ual elements, even if the main stress is removed.

14.5 Unstressed vowels
The difficulty with the unstressed vowels is that their spelling is so variable
and that the spellings for these vowels can also be used for full vowels. You
thus have to know that you are dealing with an unstressed syllable. This is
not trivial – even L1 speakers used full vowels in unstressed syllables if they
are reading aloud or trying to speak particularly clearly. We also have the
problem that the quality of the unstressed vowels is variable and, to some
extent, overlapping (see Section 4.4). The various vowels will be treated
separately.
The most important of these is the comma vowel (pronounced [ə] ~ [ʌ] ~
Ø, with the [ʌ] only in final position and Ø, i.e. not pronounced, as an option
before [l], [n], [m]). This vowel can be spelt with every vowel letter and, in
SSBE, can also be spelt with <r> following the vowel letter. Examples are
provided in (X19).

(X19) 
Spellings for comma
Examples Comments
<a> about, comma, data, Initially, often in prefixes; finally, often
elephant, vendetta in Latin and Greek plurals and other
loans. In text, the indefinite article
before a consonant is regularly [ə].
<ar> burglar, lunar, mustard In the suffixes -al, -ar, -able and -wards
(Continued)
 Spelling vowels 161

Examples Comments
<e> witness, turtle, peroxide In turtle, the [ə] ~ Ø is before the [l]. In
text, the before a consonant is
regularly [ðə].
<er> bigger, cover, eager, driver,
perturb
<re> centre, litre, metre Many such words are spelt with <er> in
America.
<ia> confidential, initial The written <i> is a sign that the
preceding <t> is [ʃ].
<io> nation, perturbation, The written <i> is a sign that the
fruition, etc. preceding <t> is [ʃ].
<o> commence, daffodil, exodus,
handsome, kingdom, pilot,
police, today
<or> actor, forget, sailor
<our> colour, favour Usually spelt <or> in America.
<u> substantial
<ur> murmur, surprise
<y> pyjamas Also spelt <pajamas>; can be
pronounced with the horses vowel.
<yr> martyr

The horses vowel ([ɪ] ~ [ə], with [ə] increasingly used) can be seen as merg-
ing with the comma vowel and treated as a variant of it. Its spelling is much
simpler. The spelling for this vowel is given in (X20).

(X20)

Spellings for horses

Examples Comments
<e> horses, wanted In regular inflections, as set out
in Section 9.1
<i> bikini, engine, internal,
rabbit
<y> bicyclist

Spellings with <i> seem to retain the pronunciation with <ɪ> longer than
those in inflections, which are more usually reduced to [ə].
162 Spelling

The happy vowel has (relatively) recently changed its pronunciation, so

that most L1 users now equate it with [iː], while it used to be equated with
the [ɪ] vowel. Some northern accents of British English still use an [ɪ] pronun-
ciation, and some conservative speakers of SSBE do the same. In final posi-
tion, it has a range of possible spellings, as set out in (X21).

(X21)

Spellings for word-final happy

Examples Comments
<e> apostrophe, karate, Phoebe, Mostly in Greek words but also in
posse, sesame, simile other loans.
<ea> Chelsea, guinea Rare.
<ee> coffee, toffee For those who use [iː], this looks like
the regular stressed vowel spelling.
<ey> abbey, attorney, honey,
kidney, valley
<i> chilli, Iraqi, kiwi, spaghetti, Mostly in loan words.
taxi
<ie> cookie, eerie, lassie, movie,
quickie
<y> doggy, happy, fortunately, Sometimes <ie> is preferred as a
friendly, silly spelling to <y> in nouns: <auntie>
is now more usual than <aunty>.

The happy vowel, when it occurs word-internally, usually occurs before

another vowel, but there are exceptions. The spellings for word-internal
happy are given in (X22).

(X22)

Spellings for word-internal happy

Examples Comments
<e> alveolar, atheist, creation, eolian, meander,
meteor, stupefy
<i> alias, curious, polio, trivia
<y> polyester
 Spelling vowels 163

The sound [u] ([ʊ] ~ [uː]) is always spelt <u> before another vowel, as in
fruition.

14.6 Problems with vowel spellings

As has already been made clear, the same spelling can have multiple read-
ings, some of which are, for L1 readers with suitable experience, unambigu-
ous in the relevant word but which are not necessarily transparent to anyone
who does not already know the word. In any case, there are instances where
the same spelling can be pronounced different ways (see [X23] where differ-
ent pronunciation of consonant letters is ignored), where different spelling
have the same pronunciation (see [X24]), or where distinct pronunciations of
the same word can co-exist in the community – sometimes with regional
preferences (see [X25]). This is before we consider the variability in the indi-
vidual vowels. All of these sets show a few examples of much wider
phenomena.

(X23)

Different pronunciations of the same spelling – homography

Word 1 Meaning 2 Meaning

lead [liːd] ‘go in front’ [led] ‘heavy metal’
read [riːd] present tense [red] past tense
row [ɹaʊ] ‘noise’ [rəʊ] ‘propel with
oars’
tarry [taɹi] ‘wait’ [tɑːɹi] ‘covered in tar’
tear [tɪə] ‘water from eye’ [teə] ‘rip’

(X24)

The same pronunciation with different spellings – homophony

Sound 1 Meaning 2 Meaning 3 Meaning

[beə] bare ‘naked’ bear ‘carry; large
animal’
[bɪə] beer ‘alcoholic bier ‘on which a
drink’ dead body is
carried’
(Continued)
164 Spelling

Sound 1 Meaning 2 Meaning 3 Meaning

[bɔː] boar ‘male pig’ bore ‘drill; be
tedious’
[ðeə] their ‘of them’ there ‘yonder’ they’re ‘they are’
[feə] fair ‘just; fare ‘payment for
amusement transport’
place’
[flaʊə] flower ‘blossom’ flour ‘ground meal’
[ɡambəl] gamble ‘play games gambol ‘to skip like a
of chance’ lamb’
[ɡeɪt] gait ‘way of gate ‘opening in a
walking’ fence’
[kɹuːz] crews ‘teams’ cruise ‘sail for
pleasure’
[laɪə] liar ‘one who tells lyre ‘old musical
lies’ instrument’
[meɪd] maid ‘serving girl’ made ‘created’
[miːt] meat ‘flesh of meet ‘come mete ‘measure’
animals’ together’
[mɔː] maw ‘jaws’ more ‘additional’ moor ‘poor
unculti­
vated
land’
[ɔː] awe ‘wonder’ ore ‘rock from or ‘alterna­
which metal tively’
is extracted’
[ɔːltə] altar ‘table for alter ‘change’
sacrifices’
[paɪ] pi ‘Greek letter’ pie ‘food in
pastry’
[peə] pare ‘cut the skin pear ‘fruit’ pair ‘two’
off
vegetables’
[ɹəʊ] row ‘propel a roe ‘fish eggs’ rho ‘Greek
boat’ letter’
[ɹuːd] rood ‘cross’ (old) rude ‘impolite’ rued ‘pitied’
[sɜːf] serf ‘lowest class surf ‘white water’
of worker’
(hist)
[siː] sea ‘body of salt see ‘perceive with
water’ the eyes’
[sɪmbəl] cymbal ‘percussion symbol ‘sign’
instrument’
[stɔːɹi] storey ‘floor of a story ‘tale’
building’
 Spelling vowels 165

(X25)

Alternative pronunciations of the same word

Word 1 2 Comment
been bɪn biːn [biːn] is more common in
SSBE; [bɪn] may be an
unstressed form.
data dɑːtə deɪtə
economic ekənɒmɪk iːkənɒmɪk There are many such examples,
including the name Megan.
fault fɒlt fɔːlt There are many similar cases,
with <aul> and a consonant
letter.
hoof hʊf huːf One of a small number of
cases of this alternation
(others are roof, room and
tooth); the long vowel is
more usual now.
patent patənt peɪtənt Another common pattern
year jɪə jɜː
yoghurt jɒɡət jəʊɡət The former is more usual in
SSBE.

Highlights
The spelling of vowels is more variable than the spelling of consonants, not
only because of borrowed spellings and historical changes but also because
the rules for unstressed vowels are different from those for stressed ones. The
result is that often the same sound can have different spellings and the same
spelling can represent different sounds. This is not as chaotic as it might
sound since the options are limited.

What should the teacher do?

Since short vowels are more regularly spelt than long vowels and diphthongs,
an early focus on such vowels can have advantages. Consonant doubling on
basic English words is a fundamental rule of English spelling, even if it does
not extend to more learned words. The alternation between <y> and <ie> is
also something that can be taught early. It is the lack of stress which is im-
portant for the unstressed vowels rather than any link between spelling and
vowel sound, and the precise qualities are relatively predictable.
15
NAMES AND OTHER DIFFICULT
WORDS

There are various reasons why names are difficult to spell and to pronounce.
At their easiest, names work the same way that ordinary words do: Baker,
Brown, Freda, Harry, James, Martha, Mary, Smith, Toby, Tommy, Vanessa
and Vincent are all spelt and pronounced perfectly regularly (and, where
relevant, identically to the same series of letters in a word that is not a name).
However, names are different from ordinary words in that (a) they may have
very conservative spellings which do not correspond to the modern pronun-
ciation, (b) they may contain extra letters (or use <y> rather than <i>) in
ways that we would not expect in other words, (c) they may be foreign, and
their pronunciation may be anglicised or still foreign, and (d) they may ex-
ploit different ways of pronouncing a particular letter or set of letters. These
comments apply to place names as well as to given names and surnames.
Names with conservative spellings include place names such as Leicester
([lestə]), Leominster ([lemstə]), Milngavie ([mʌlɡaɪ] near Glasgow) and
Mousehole ([maʊzəl] in Cornwall), Worcester ([wʊstə]) and surnames such as
Beauchamp ([biːʧəm] also spelt Beecham), Featherstonehaugh ([fanʃɔː]) and
Marjoriebanks ([mɑːʧbaŋks]). These are extreme examples, and some of
them are famous for their eccentricity, but they make the point that pronun-
ciation may not follow from the spelling.
Where the additional letters are concerned (or the use of <y>), Mr Smyth
may or may not pronounce his name [smɪθ] ([smaɪθ] and [smaɪð] are also
possible), but the eccentric <y> makes it clear that we are dealing with a
name. We find double letters that are not used in ordinary words in surnames

DOI: 10.4324/9781032637020-19
 Names and other difficult words 167

like Bartlett, Briggs, Crabbe, E(l)liott (also spelt Eliot), Kerr (which may be
[kɜː] or [kɑː]), Smollett and Truscott and an extra <e> in Burke, Dunne (also
Donne, Dunn), Hoare, Moore, Paine and so on.
Foreign names are a minefield. First, we have place names which are not
the names used in the place concerned. Some examples are given in (X1). In
some cases, these names are borrowed from a third language (often French).

(X1) 
Athens, Belgrade, Cologne, Copenhagen (both the German
pronunciation [kəʊpənhɑːɡən] and the British [kəʊpənheɪɡən] are
heard), Dublin, Florence, Jerusalem, Lisbon, Lyons, Moscow
([mɒskəʊ] though the American [mɒskaʊ] is also heard), Munich,
Naples, Prague, Rome, Turin, Venice, Vienna and Warsaw

When foreign placenames are used, they are usually pronounced follow-
ing English rules (so when the preferred transliteration was changed, Peking
became Beijing with a new pronunciation; Bombay became Mumbai when
the spelling was changed). Helsinki, Madrid, Oslo, Paris, Stockholm and Tri-
este, for example, are all pronounced as English words ([helˈsɪŋki], [məˈdɹɪd],
[ˈɒzləʊ], [ˈpaɹɪs], [ˈstɒkhəʊm] and [tɹiːˈest]). On the other hand, some attempt
is made to copy the original pronunciation in Calais [ˈkaleɪ], Ibiza [ɪˈbiːθə],
Marbella [mɑːˈbeɪə], Schleswig [ˈʃlezvɪɡ] and some others.
We find names that have varying pronunciations (for different places with
the same name or different people with names written the same) depending
on how the spelling is interpreted. Aaron can [eəɹən] or [aɹən]; Bernstein can
be [bɜːnstiːn] or [bɜːnstaɪn]; Cheviot is [ʧiːviət] in Scotland, but [ʧeviət] in
New Zealand; Helena can be [heˈliːnə] or [heˈleɪnə] or [ˈhelənə]; Lisa can be
[liːzə], [laɪzə] or [liːsə]; Gillingham is [ʤɪlɪŋəm] in Kent, but [ɡɪlɪŋəm] in Nor-
folk; Stour may be [stʊə], [staʊə] or [stəʊə] depending where in England you
are.
Then we find multiple spellings of the same name (pronounced the same
way). For example, Ann, Anne; Bridget, Brigit; Catherine, Katherine, Kath-
ryn; Eleanor, Elinor; Euan, Ewan, Ewen; Jane, Jayne; Lynn, Lynne; Rachel,
Rachael; and Sean, Shaun, Shawn. Spellings such as Meegan may be used to
indicate the pronunciation of the name more usually written Megan.
Since English has long accepted names from Ireland, Scotland, France
and Greece (to name but a few) and increasingly there are names from other
places in the world found in English-speaking countries, names like Ahmad,
Aphra, Asif, Bianca, Caitlin, Dipak, Fleur, Kumar, Kylie, Niamh, Padraig
and Surayya may not be easily interpretable from the spelling since they are
based on foreign patterns.
168 Spelling

Since it is easy to be in doubt, you can always ask, ‘How do you spell/
pronounce your name?’
In teaching reading to first-language (L1) learners, there has been con-
tinual tension between two schools of thought: look and say and phonics.
Those in the look-and-say school think that English spelling is so irregular
that you simply have to recognise words and then be able to say them. Those
in the phonics school believe that English spelling contains so many regu-
larities that children should be taught to spell out the word and calculate
what it must say. Although the battle lines often seem to be drawn in these
extreme terms, the best way of dealing with English spelling lies between
them. If you can read only those words that you recognise, you cannot read
any new words, which is not helpful. On the other hand, there are some
words whose spelling is so irregular (or where the number of common words
with that spelling pattern is so small) that it is more efficient simply to learn
that this is how a particular word is said or spelt. Here, we look at those
words which are probably best learnt according to the look-and-say
system.
We begin with the celebrated cases with <gh> in them. The pattern of
<ight> in bight, fight, light, might, right, sight and tight contains so many
common words that it can be treated as a regular pattern. But words ending
in <ough> are really problematic and have to be learned as individual words.
The options are set out in (X2).

(X2) 
Pronouncing <ough>

Relevant words

[ʌf] chough, enough, rough, slough ‘cast off skin’, sough*, tough
[ʌp] hiccough (also spelt hiccup)
[ɒf] cough, trough
[ɒk] hough (more usually spelt hock), lough (Irish usage)
[aʊ] bough, plough, slough ‘marsh, boggy place’, sough*
[əʊ] although, dough, though
[ɔː] bought, brought, fought, nought, ought, sought, thought, wrought
[uː] through
[ə] borough, thorough

Some of the words in (X2) are so rare that they can safely be ignored
(chough, hough, lough, sough and slough), but only the series with a following
<t> set up anything resembling a pattern. Names containing the sequence
<ough> are not necessarily predictable.
 Names and other difficult words 169

Some other unexpected and unpredictable spellings are given in (X3),

along with comments as necessary. Most of these will simply need to be
learnt as irregularities, but sometimes they might not be needed at all
(sleight); sometimes a regular form of pronunciation can be used without
problems. Names are not included here. These words are genuine candidates
for the look-and-say method however much of a phonics fan the teacher
may be.

(X3) Irregular spelling–pronunciation links

Example(s) Comment
<a> = [a] have The final <e> makes this look like a long
vowel.
<a> = [e] any, many
<ai> = [e] again(st), said
<ai> = [a] plaid, plait
<al> =[al] shall Usually <al> is pronounced [ɔːl] as in all,
also, call, tall and so on.
<ao> = [eɪ] gaol The spelling <jail> avoids the issue.
<au> = [ɑː] aunt, laugh Both have [a] in North American English
and the North of England; laugh has
<gh> pronounced [f].
<au> = [ɒ] cauliflower, sausage
<ay> = [iː] quay
<ay> = [i] always, Note that the <ay> in these examples is
Monday,Tuesday unstressed; with some stress, regular [eɪ]
(etc.) is possible.
<ay> = [e] says Regular [eɪ] is heard but indicates lower
social standing.
<c> is silent indict
<e> = [ɪ] England, English,
pretty
<eigh> = [aɪ] height, sleight Sleight is homophonous with slight.
<eigh> = [eɪ] eight, neigh, sleigh, Can be viewed as showing silent <gh>
weigh(t)
<eo> = [iː] people
<ew> = [əʊ] sew
<ey> = [iː] geyser, key geyser has [aɪ] in New Zealand (the
Icelandic original has something that
sounds much more like [eɪ])
<eye> = [aɪ] eye Homophonous with I and aye
<f> = [v] of
<i> = [ɪ] give The spelling suggests a long vowel, but
we find a short one.
(Continued)
170 Spelling

Example(s) Comment
<i> = [aɪ] child, kind, mild Before two consonants, we would expect
a short vowel.
<ie> = [e] friend
<is> = [aɪ] island, isle
<kn>, <gn> know, knew, knee, This is regular for the reader but irregular
= [n] kneel, gnat for the writer.
<mb> = [m] comb, crumb, thumb Regular for the reader in final position
but not for the writer
<o> = [ɪ] women
<o> = [əʊ] both, most, old There is a long vowel sound followed by
two consonants; only may be included
here.
<o> = [ʊ] bosom, woman
<oe> = [ʌ] does, come, love Present tense of do; the plural of doe is
regular; come leads us to expect a long
vowel, as in home.
<oe> = [uː] shoe
<o> = [wʌ] one, once
<o> = [uː] do, to, lose, move, Although these represent a regular
who, womb development for long <o>, they are
unusual; we would expect a short vowel
in womb.
<oul> = [ʊ] could, should, would
<our> = [ɔː] your you’re may be homophonous or
pronounced [jʊə].
<ph> = [v] nephew Regular [f] is possible and more modern
<s> = <s> dose, grouse, house, We would expect [z] between vowels but,
louse, mouse, here and in some rarer words, get [s].
spouse Note that the verb house has [z] in
Standard Southern British English
(SSBE) (compare use where we find the
same pattern), as does the plural houses.
<s> = [ʃ] sugar
<t> = Ø often, soften [sɒfən] Often is sometimes heard with a [t].
<tw> = [t] two The [w] is pronounced in twin and so on.
<u> = [e] bury
<u> = [ɪ] busy, business
<u> = [uː] truth
<ue> = [uː] blue, true
<ue> = Ø tongue <o> for [ʌ] is also awkward but relatively
common.
<ui> = [ɪ] build, built, guild, If the <bu> and <gu> represent [b] and
guilt(y) [ɡ], then these are regular.
<uoy> = [ɔɪ] buoy The pronunciation [buːiː] is not SSBE.
(Continued)
 Names and other difficult words 171

Example(s) Comment
<uy> = [aɪ] buy, guy If the <bu> and <gu> represent [b] and
[ɡ], then these are regular.
<wh> = [h] who, whom, whose, Note also that final <o> in who is [uː] and
whoever (etc.) similarly for the others.
<wr> = [ɹ] wrap, wren, wrist, Regular for the reader but not for the
write, etc. writer

Highlights
Some words are spelt too irregularly to allow their pronunciation to be cal-
culated from their spelling. Many of the common ones are listed here.
Names, while they may be regularly spelt, are often spelt unpredictably.

What should the teacher do?

Teachers should not try to generalise from irregularly spelt names and
should simply provide the pronunciation. With the irregularly spelt common
words, they should teach them as irregularities.
READING AND REFERENCES

There are many good books available that treat phonetics in general and
English phonetics in particular. A good introductory text to phonetics in
general is Ladefoged’s Course in phonetics, in multiple editions, the most re-
cent being Ladefoged and Johnson (2014). A more advanced discussion of
phonetics is provided by Laver (1994), which is particularly good on rhythm
and on voice quality. Abercrombie (1967) provides a classic introduction.
For English phonetics, the best work by far, in my opinion, is Gimson’s,
again in many editions, the later ones revised by Cruttenden. The most re-
cent edition at the time of writing is Cruttenden (2014). This book discusses
the phonetics of Standard Southern British English. It is encyclopaedic
rather than introductory, so for those who want something that is easier to
read, the textbooks by Ashby (1995) and Ogden (2009) might make better
starting places. The notion of lexical sets can be found in Wells (1982).
Specialised books on English stress include Kingdon (1958) and Fudge
(1984). They both provide a great deal of data, but our understanding has
improved since they were written, and alternative patterns may be more ex-
planatory. Chomsky and Halle (1968) provide a very different approach,
which is difficult to read but which has been theoretically very important.
Cruttenden (2014) provides another analysis, which may be easier to follow.
On consonant capture, see Baldwin (1995).
Books on phonology are often introductions to phonological theorising
rather than to phonological patterns of English. McMahon (2002) is an el-
ementary text that focuses on English. Carr (1999) covers both phonetics
and aspects of phonology. For the interplay of phonology (including stress)
and morphology, see Bauer et al. (2013). For long and short vowels in loan-
words, see Lindsey (1990).
 Reading and references 173

The obvious source for information on English spelling is Carney (1994).

There is a shorter, textbook version of this, Carney (1997). Upward and David-
son (2011) consider the subject from a historical perspective. For an alternative
pedagogical view of spelling, see the appendix in Nation and Macalister (2021).
Anyone with a serious interest in English pronunciation needs access to a
pronunciation dictionary. There are many of these on the market. I find
Wells (2008) particularly helpful, but that is a matter of personal preference.
Upton and Kretzschmar (2017) cover more words but give fewer variants.
The best established of such dictionaries is Jones et al. (2003).

Abercrombie, David 1967. Elements of general phonetics. Edinburgh: Edinburgh

University Press.
Ashby, Patricia 1995. Speech sounds. London and New York: Routledge.
Baldwin, John 1995. A ‘tenny’ rate. In Jack Windsor Lewis (ed.), Studies in general
and English phonetics, 301–309. London and New York: Routledge.
Bauer, Laurie, Rochelle Lieber & Ingo Plag 2013. The Oxford reference guide to Eng-
lish morphology. Oxford: Oxford University Press.
Carney, Edward 1994. A survey of English spelling. London and New York:
Routledge.
Carney, Edward 1997. English spelling. London and New York: Routledge.
Carr, Philip 1999. English phonetics and phonology. Oxford: Blackwell.
Chomsky, Noam & Morris Halle 1968. The sound pattern of English. New York:
Harper & Row.
Cruttenden, Alan 2014. Gimson’s pronunciation of English. 8th edition. London and
New York: Routledge.
Fudge, Erik 1984. English word-stress. London: Allen & Unwin.
Jones, Daniel, Peter Roach, James Harman & Jane Setter 2003. Cambridge English
pronouncing dictionary. 16th edition. Cambridge: Cambridge University Press.
Kingdon, Roger 1958. The groundwork of English stress. London: Longman.
Ladefoged, Peter & Keith Johnson 2014. A course in phonetics. 7th edition. Stamford
CT: Cengage.
Laver, John 1994. Principles of phonetics. Cambridge: Cambridge University Press.
Lindsey, Geoff 1990. Quantity and quality in British and American vowel systems.
In Susan Ramsaran (ed.), Studies in the pronunciation of English, 106–118. Lon-
don and New York: Routledge.
McMahon, April 2002. An introduction to English phonology. Edinburgh: Edinburgh
University Press.
Nation, Paul & John Macalister 2021. Teaching ESL/EFL reading and writing. 2nd
edition. London and New York: Routledge.
Ogden, Richard 2009. An introduction to English phonetics. Edinburgh: Edinburgh
University Press.
Upton, Clive & William A. Kretzschmar 2017. The Routledge dictionary of English
pronunciation for current English. 2nd edition. London and New York: Routledge.
Upward, Christopher & George Davidson 2011. The history of English spelling.
Chichester: Wiley-Blackwell.
Wells, J.C. 1982. Accents of English. 3 vols. Cambridge: Cambridge University Press.
Wells, J.C. 2008. Longman pronunciation dictionary. Harlow: Longman.
INDEX

Pages in italics refer to figures and pages in bold refer to tables.

Ablaut 106, 118; see also free variation; Caxton, William 142
vowel alternation central vowel 17
Adam’s apple 12 checked see short vowel
affix 95; see also prefix; suffix Chinese (unspecified) 34, 37, 61
affricate 13, 34–5 clear [l] 36
affrication 27 close-mid 15
allophone see bound variant close vowel 15, 64
alveolar 14 cluster 64, 81
American English (US or unspecified) compound 89, 130–5
4–5; consonants 33, 37, 38, 40, 41; connected speech 80–1
rhoticity 37, 38, 39, 47, 66, 97; consonant 13, 27–42, 72–8,
vowels 43, 46, 47, 49, 50, 52, 107–8, 145–9
115, 118, 151, 156, 169; spelling 106, consonant capture 101–2
155, 161; words 34, 46, 47, 167 creaky voice 61
antepenultimate shortening 116, 156
apophony see Ablaut, Umlaut Danish 17, 28, 29, 33, 37, 45, 46, 48,
approximant 16, 35–41, 77 57, 73
Arabic 30, 97 dark [l] 36–7, 73, 76
artificial norm see norm deletion 92–6
aspiration 27, 73–4 dental 14, 29, 36, 74
assimilation 89–92 derivation 9, 105, 109–21, 124
autostress 128 digraph 145, 151, 153, 154, 157
diphthong 17–8, 43, 46, 49–50,
back vowel 17 156–9
base 9, 95, 98, 124 distinctive sound 19–20, 89, 91, 103
bound variant 20, 69, 70, 76, 91 double consonant letter 151–3, 160
breath group 59, 60 Dutch: consonants 28, 29, 30, 36, 143;
breathy voice 61 prosody 61; spelling 41, 142; vowels
Burmese 71 44, 45, 48, 49, 71
 Index 175

edh, eth 141 hesitation vowel 51, 56

English varieties: Australian 6, 35, 37, high vowel see close vowel
38, 39, 45, 48, 49, 50, 52, 97, 159; Hindi 28, 30, 35, 61
Birmingham (England) 45, 48, 61; Hungarian 71
Canadian 39, 40, 97; Early Modern hypocoristic see pet name
English 115; Eastern 38, 39;
Glasgow 37; Hong Kong 45; Irish Icelandic 31, 43, 141, 169
37, 39, 46, 66, 97, 142; Lancashire Indonesian 61
39; Liverpool 27, 50, 61; London 46, inflection 9, 104–9, 130, 147
49, 50; Middle English 115; New insertion 97–100
Zealand 33, 37, 39, 45, 48, 49, 50, 97, instability of alveolars 90, 91, 93
106, 169; North of England 44, 46, International Phonetic Association
47, 49, 79, 162, 169; Old English 30, and its alphabet 8, 21, 73, 78
39, 141; Scottish 33, 39, 41, 43, 46, intonation 59–60, 72, 81
47, 48, 49, 66, 97, 105, 115, 145, 155; intrusive R 39, 47, 97
Singapore 45; South African 39, 48; Italian 30, 36, 40, 45, 52, 60, 73, 76,
South-East England 37, 46; 77, 149
Standard International English 6;
Tyneside 33, 37; Welsh 37; West of Japanese 30, 34, 41, 44, 45, 48, 76,
England 38, 39, 46; see also 78, 142
American English; SSBE
extender 98–9 Khmer 71
King’s English see SSBE
Fijian 29 Korean 76
Finnish 17, 44, 48, 71
fluency 81–3, 91, 92, 93, 101 labial-velar 14
foot 58 labio-dental 14
free variation 136–8 larynx 12, 61
French: consonants 28, 29, 36, 64, 73, lateral 14, 36
76; prosody 61; spelling 141, 143, Latin 103, 109, 110, 112, 116, 143, 146,
157; vowels 17, 35, 44, 45, 48, 52, 71, 152, 160; church and legal 117
79; words 72, 167 lax see short vowel
fricative 13, 30–4, 64–5, 68, 74–6, 90, lexical set 8, 17, 44
92, 105, 111–2 linking R 39, 47, 97
front vowel 17, 29 lip-rounding 33, 34, 37, 41, 75, 76
liquid 36–9, 64
German: consonants 30, 34, 36, 73, 75, lisp 33
97; prosody 61; spelling 41, 145, 148, loan words 31, 33, 41, 64, 65, 68, 85, 96,
149; vowels 44, 45, 48, 56, 71, 79; 98, 103, 109, 117–8, 127, 142, 150,
words 167 156, 160, 162, 167
Germanic 28, 29, 72 long vowel 17, 43, 46–8, 83, 115
glide see approximant low vowel see open vowel
glottal 14; glottal reinforcement 91;
glottal replacement 91; glottal stop Mandarin 17, 28, 30, 33, 37
13, 29, 30, 97, 101 manner of articulation 14
Great Vowel Shift 115–18, 142 Māori 72
Greek: consonants 31, 32, 33, 73; loans merger 100–1
64, 65, 143, 145, 147, 148, 149, 162; mid vowel 15, 17–8
spelling 152; vowels 160 minimal pair 20, 31, 32, 42, 54, 72, 81
GVS see Great Vowel Shift monomorphemic 9
monophthong 17–8, 43, 44–6, 46–8,
[h]-dropping 33, 93–4 150–6
Hebrew 29, 30 monosyllable 83, 154
176 Index

morpheme 9, 95, 103 short vowel 17, 43, 44–6, 115

morphophonemics 66, 103–21, 114 similitude 89, 101
Slavic 29
name 127, 166–8; see also pet name sonorant 13, 76–8
nasal 13, 35–6, 61, 64, 65, 68, 91, 92 sonority 63, 64, 68, 93
Nepali 61 Spanish: consonants 23, 28, 29, 30, 32,
non-pre-vocalic 36, 38 33, 36, 40, 71, 72, 73, 77, 142, 147;
nonsense words 42, 54, 84–6 vowels 44, 45, 48, 53, 79
norm, artificial 5–6 spelling 72, 73, 110, 112, 114, 119, 123,
Norman Conquest 141 131, 139–71
Norwegian 29 spread see unrounded
notation 8, 43, 58, 69, 103, 107, 160 SSBE 4, 8; consonants 13, 20, 36, 37,
nuclear syllable 59 38, 41, 93, 97, 170; prosody 133;
nuclear tone 59 spelling 148; vowels 18, 35, 43, 47,
49, 50, 94, 118, 151, 159, 162
obstruent 13, 46, 64, 66, 73, 93, 109 standard 4
open-mid 15 Standard Southern British English
open vowel 15, 64 see SSBE
Oxford English see SSBE stop 13, 64; see also glottal stop; nasal;
plosive
palatal 14, 36, 40, 72, 100 stress 55–7, 81, 83, 122–35, 137–8;
palatalised [l] see clear [l] antepenultimate 130, 138; compound
past tense 103 130–5; contrastive 55, 59, 81; in
pet name 104, 113 disyllables 123–7; in longer words
phoneme see distinctive sound 128–30; in monosyllables 123; loss
phonology 9, 87–135 114–5; with prefixes 130; with
phonics 143, 168 suffixes 128–30
phonotactics 63–8, 72 stress-timing 57, 58
place of articulation 14–5 suffix 9, 66, 68, 83, 95, 96, 98, 104, 116,
plosive 13, 27–30, 38, 65, 68, 73–4, 76, 128–30
90, 93, 101 Swedish 29, 71
plural 100, 103, 105, 106 syllable 36, 57, 59, 63, 68–70, 116
Polish 28, 35, 76 syllable-timing 57, 58
Portuguese 35, 36, 61
post-alveolar 14, 100–1, 112, 137 Tagalog 97
prefix 9, 95, 124, 130 tap 20, 90
prosody 22, 55–62 tense see long vowel
Thai 28, 30, 66, 97
reduced form 90, 113, 123; see also vowel thorn 141
rhoticity 39, 46, 48, 97 tone 59
rhythm 57–9, 70, 74 trachea 12
Roman alphabet 21, 39, 41, 45, 72, 141 transcription 21–2, 78
rounded vowel 17, 41, 46, 48, 79 triphthong 50
RP see SSBE tri-syllabic laxing see antepenultimate
rule 8 shortening
Russian 28, 33, 34, 44, 45, 56, 64, 71, 75 Turkish 29, 72, 78

segment 22, 27, 63–4, 90 Umlaut 106

semivowel see approximant unchecked see long vowel
Shakespeare, William 142 unrounded vowel 17
Shaw, George Bernard 119, 143 unstressed vowel 50–2, 160–3
 Index 177

variety 7–8 vowel 13, 43–54, 78–80, 150–65;

velar 14, 46 reduction 56
velum 14, 15, 17, 29, 36, 41 vowel alternation 14–21, 137
velar softening 112 vowel length see long vowel, short vowel
velarised [l] see dark [l] vowel chart 18
verb forms 107–8
Vietnamese 28, 29, 30, 34, 71 weak forms 94; see also unstressed
vocal cords see vocal folds vowels
vocal folds 12–3, 15 Welsh 147, 148
vocal fry see creaky voice whisper 13, 20, 75
vocal tract 13 word 9; basic 124, 143, 146, 150, 153
vocalised [l] 37 wynn 141
voice quality 60–2, 72
voiceless 13, 41, 64, 66, 90, 104 Yiddish 148
voiceless vowel 33, 75, 91 yod-dropping 40, 156
voicing 13, 27, 90 yogh 141