(Understanding Language) Kate Burridge - Alexander Bergs - Understanding Language Change-Routledge (2017)

Understanding Language Change
Understanding Language Change offers a complete and accessible introduction to

historical linguistics. Assuming no prior knowledge, this engaging book introduces
core concepts through English examples; builds on them with a rich variety of illus-
trations from across the Germanic, Romance, Pacific and Asiatic languages; and
demonstrates their applicability to real-life changes taking place in languages today.
Key features include:
• A step-by-step approach introducing language change at the sound, word and

sentence levels, covering shifts that are currently underway in addition to his-
torical changes;
• Coverage of topics such as pidgins, creoles and mixed languages, as well as lan-
guage death and how to reconstruct lost stages of languages;
• Breakout boxes showing the real-life applications and impact of theories, from
Shakespeare’s English to the speech of modern-day celebrities, and introducing
current research problems;
• A wide range of supporting material, including exercises, further reading and
research projects for more in-depth independent study.
Written by two experienced teachers, Understanding Language Change is essen-

tial reading for anyone studying historical linguistics or language change for the
first time.
Kate Burridge is Professor and Chair of Linguistics at Monash University, Australia.
Alexander Bergs is Professor and Chair of English Language and Linguistics at the Uni-
versity of Osnabrück, Germany.
Understanding Language series
Series Editors:
Bernard Comrie, University of California, Santa Barbara, USA
Greville Corbett, Surrey Morphology Group, University of Surrey, UK
The Understanding Language series provides approachable, yet authoritative,
introductions to major topics in linguistics. Ideal for students with little or no prior
knowledge of linguistics, each book carefully explains the basics, emphasising
understanding of the essential notions rather than arguing for a particular theoreti-
cal position.
Other titles in the series:

Understanding Pragmatics
Gunter Senft
Understanding Child Language Acquisition
Caroline Rowland
Understanding Semantics, Second Edition
Sebastian Löbner
Understanding Syntax, Fourth Edition
Maggie Tallerman
Understanding Phonetics
Patricia Ashby
Understanding Phonology, Third Edition
Carlos Gussenhoven and Haike Jacobs
Understanding Morphology, Second Edition
Martin Haspelmath and Andrea D. Sims
Understanding Language Testing
Dan Douglas
Understanding Second Language Acquisition
Lourdes Ortega
Study Skills for Linguistics
Jeanette Sakel
Understanding Language Change
Kate Burridge and Alexander Bergs
For more information on any of these titles, or to order, go to
www.routledge.com/series/ULAN
Understanding
Language
Change
Kate Burridge and
Alexander Bergs
First published 2017
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
711 Third Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2017 Kate Burridge and Alexander Bergs
The right of Kate Burridge and Alexander Bergs to be identified as authors of this work has been
asserted by them 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.
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
Names: Burridge, Kate, author. | Bergs, Alexander, author.
Title: Understanding language change / Kate Burridge and Alexander Bergs.
Description: Milton Park, Abingdon, Oxon; New York, NY: Routledge, [2017] | Series:
Understanding Language series | Includes bibliographical references and index.
Identifiers: LCCN 2016015297 | ISBN 9780415713382 (hardback) | ISBN 9780415713399 (pbk.) |
ISBN 9781315463018 (ebook)
Subjects: LCSH: Linguistic change. | Historical linguistics.
Classification: LCC P142 .B87 2017 | DDC 417/.7—dc23
LC record available at https://lccn.loc.gov/2016015297
ISBN: 978-0-415-71338-2 (hbk)
ISBN: 978-0-415-71339-9 (pbk)
ISBN: 978-1-315-46301-8 (ebk)
Typeset in Minion Pro
by Apex CoVantage, LLC
Contents
List of figures ix
List of maps x
List of tables xi
Acknowledgementsxiii
Publisher’s acknowledgements xiv
List of abbreviations xv
1 Setting the scene 1

Introduction1
1.1 Change occurs at all levels 5
1.1.1 Phonetics and phonology 5
1.1.2 Morphology 8
1.1.3 Syntax 9
1.1.4 Semantics 10
1.1.5 Pragmatics 11
1.1.6 Change across different levels 13
1.2 Variation and change – two sides of the same coin 13
1.3 Evidence of change in progress 18
1.4 Attitudes to change 21
1.5 How this book is structured 23
Summary24
Further reading 24
Exercises25
2 Changes to the lexicon 28

Introduction28
2.1 Gaining words – lexical addition 29
2.1.1 Compounding 31
2.1.2 Affixation 33
2.1.3 Backformation 34
2.1.4 Conversion 35
2.1.5 Abbreviation 35
2.1.6 Acronyms 36
2.1.7 Blending 37
2.1.8 Commonization 38
2.1.9 Reduplication 38
vi Contents
2.1.10 Borrowing 39
2.1.11 Sound symbolism 41
2.1.12 A final word on the processes 42
2.2 Losing words – lexical mortality 43
2.2.1 Obsolescence 43
2.2.2 “Verbicide” 44
2.2.3 Reduction 44
2.2.4 Intolerable homonymy 44
2.3 Etymology – study of the origin of words 46
Summary47
Further reading 48
Exercises49
3 Changes to the semantics 52

Introduction52
3.1 Consequences of semantic change 54
3.1.1 Broadening 54
3.1.2 Narrowing 55
3.1.3 Shift 55
3.1.4 Changing values – amelioration and deterioration 55
3.1.5 Chain reaction changes 56
3.2 Why words change their meanings 57
3.2.1 Socio-cultural factors 57
3.2.2 Psychological factors 58
3.2.3 Linguistic factors 61
3.3 Regularity in semantic change – a more explanatory account 66
Summary69
Sources and further reading 70
Exercises70
4 Changes in sound structure 75

Introduction75
4.1 Phonological processes 76
4.1.1 Old sounds drop out – loss 77
4.1.2 New sounds appear 80
4.1.3 Old sounds are modified – assimilation, dissimilation, metathesis 81
4.2 Phonetic versus phonemic change 90
4.2.1 Phonetic change – changes in accent 90
4.2.2 Phonemic change – changes that alter the sound system 91
4.3 On exceptions in sound change 94
4.4 Why do sounds change? 97
4.4.1 Simplicity 98
4.4.2 Contact induced change 98
4.4.3 Structural pressure 98
4.4.4 Social change 100
Contents vii
4.4.5 Frequency factors 100

Summary101
Further reading 101
Exercises102
5 Changes in word structure 106

Introduction106
5.1 Reanalysis and actualization – reinterpretation of structure 106
5.2 Analogy – attraction to structure 109
5.3 Typology – change in morphological type 121
5.4 Why – explaining morphological changes 127
Summary130
Further reading 131
Exercises131
6 Changes in sentence structure 134

Introduction134
6.1 Change in word order 137
6.1.1 Grammatical change is gradual 141
6.2 Typology and word order change 143
6.2.1 The contribution of Joseph Greenberg and others 144
6.3 Creating grammar 148
6.3.1 Case study – the evolution of not in English 151
6.3.2 Where is English negation heading now? 154
Summary156
Sources and further reading 157
Exercises157
7 The spread of change 162

Introduction162
7.1 Diffusion within the linguistic system 163
7.2 Spread in social structures – the speech community and the individual 167
7.2.1 Wavy, gravity, cascady . . . 168
7.2.2 Social factors 171
7.2.3 Social networks 182
7.2.4 Communities of practice 185
Summary186
Further reading 186
Exercises187
8 Languages in contact 189

Introduction189
8.1 Types of contact 190
8.1.1 Language maintenance 190
8.1.2 Language shift 192
viii Contents
8.1.3 The creation of new languages 195

8.1.4 The linguistic Sprachbund – a special case of language convergence 201
8.2 Case study: Pennsylvania German and English in contact 202
8.2.1 English to Pennsylvania German – superstratum influence 203
8.2.2 Pennsylvania German transfers into English – substratum influence 204
8.3 Language death 207
8.4 Language planning and policy 209
8.5 The dangers of purism 210
Summary211
Further reading 211
Exercises212
9 Relatedness between languages 215

Introduction215
9.1 Languages, dialects and standards 216
9.2 Establishing genetic relationships 217
9.3 The family tree model 220
9.4 The wave model 222
9.5 Quantitative approaches to language divergence 226
9.6 Long-distance comparison and Proto-World 229
9.7 Reconstructing the past 233
9.7.1 Written sources as evidence of change 234
9.7.2 The comparative method 237
Summary243
Further reading 244
Exercises244
10 An end on’t 247

Introduction247
10.1 Five questions for any theory of change 248
10.2 Additional theoretical perspectives 253
10.2.1 Corpus linguistics 254
10.2.2 Historical pragmatics 256
10.3 Historical sociolinguistics 260
10.4 The rate of change – diversity and stability 264
10.4.1 The effects of new and emerging media 266
10.4.2 Social change 270
10.5 Where to from here? 272
Further reading 273
Exercises274
References277
Index288
Figures
1.1 Change in apparent time in Ghazvin (an) 19

1.2 Percentage of do forms in various sentence types, 1400–1700 20
4.1 Complete merger 91
4.2 Partial merger 92
4.3 Primary split (split + merger in Old High German) 92
4.4 Secondary split 93
4.5 Secondary split in early English 94
4.6 The Great English Vowel Shift 99
5.1 Communication process (simplified) 107
5.2 The “wug” test 111
5.3 From fusional to isolating to agglutinating and back again 126
7.1 The S-curve of diffusion 165
7.2 Percentage of -s vs. -th in the third person present tense 166
7.3 The development of periphrastic do, 1400–1700 167
7.4 (a:) in Norwich 173
7.5 (r) in New York City 174
7.6 (ai) and (au) on Martha’s Vineyard, centralization index
stratified by age 177
7.7 Martha’s Vineyard 40 years later 178
7.8 (ing) in Philadelphia 181
7.9 Devoicing of [ʒ] in Buenos Aires by gender and age 182
7.10 The expanded “dots-and-lines model” 184
9.1 Family tree of the McFly family (Back to the Future)220
9.2 Family tree for the Germanic languages (simplified) 221
9.3 Family tree for the Romance languages (simplified) 222
9.4 The Indo-European family tree 223
9.5 Wave model of diffusion 224
9.6 A wave diagram of the Germanic language family 225
9.7 The Nostratic family tree 231
9.8 A simple language family tree 238
9.9 A simple language family tree for the word ‘hand’ 238
10.1 The funnel view of language history 264
Maps
7.1 The Rhenish fan 169

8.1 Map showing location of Gurindji and Warlpiri 199
Tables
1.1 Some formal correspondences between different Romance languages 4

1.2 The development of nasalized vowels in French, ca. 1200–1800 6
1.3 The inflectional forms of Early Modern English verbs 8
2.1 Examples of English borrowings into German 41
4.1 Compensatory lengthening in English 79
4.2 Medial consonant excrescence in English 80
4.3 Final consonant excrescence in German 81
4.4 Total assimilation in Italian 82
4.5 Palatalization in Russian 82
4.6 i-mutation in Germanic 83
4.7 Umlaut in German 84
4.8 Changes to the ‘stone’ paradigm 86
4.9 Some typical paths for lenition 86
4.10 Final consonant devoicing in German 86
4.11 Examples of dissimilation in English and German 88
4.12 Metathesis in English and German 89
4.13 Metathesis in Hanunoo 89
5.1 Māori passive formation 108
5.2 Analogy and irregular verbs in the history of English 112
5.3 Basic and past forms of [ai:t] verbs in English 117
5.4 Latin inflection and word order 122
6.1 World Atlas of Language Structures Online frequencies for
Subject, Object and Verb 143
6.2 Word order correlations 145
7.1 Grimm’s Law (First Germanic Consonant Shift) 164
7.2 Variable pronunciations of -Vry words 165
7.3 Lexical stem of hoagie in Philadelphia telephone listings 170
7.4 Diffusion of hoagie to Pittsburgh from 1961 to 1966 171
7.5 Five types of variation 179
8.1 Pennsylvania German-English contact – direction of influence 205
8.2 Number of Aboriginal languages originally spoken in Australia 208
8.3 Comparison of the 2005 and 2014 National Indigenous
Language Surveys 208
9.1 Lexical correspondences between different Romance languages 215
9.2 Comparison of four words in seven languages 217
9.3 Grimm’s Law 218
xii Tables
9.4 List of 100 basic words 227

9.5 Nostratic word list 231
9.6 Proto-World word list 232
9.7 Cognates in Romance languages 239
9.8 Sound correspondences for goat, dear, head, meat, dog240
Acknowledgements
We hope you will have as much fun reading this book as we had writing it! Usu-
ally, writing is a pretty lonely job, but working together as a team made this a really
enjoyable enterprise.
But even as a team, we had help from others, for which we are extremely grateful.
Our heartfelt thanks go to Nadia Seemungal and Helen Tredget, our editors at
Routledge, for their trust in our work, their advice, their patience – and those more
or less gentle nudges. To our series editors, Bernard Comrie and Greville Corbett,
we owe a special debt of gratitude. This book has benefited enormously from their
comments on earlier drafts, and they have saved us from quite a few blunders.
A special thanks to Felicity Meakins and Carmel O’Shannessy, who have been so
generous in sharing their work on mixed languages in Australia. Thanks also to the
students of Monash University and Universität Osnabrück for being such willing
guinea pigs when it came to trying out our ideas about language change. In particu-
lar, we are grateful to Andrea Tompros, our student test driver for an early manu-
script, for her help with style and language, and the creative artwork in Figures 5.1
and 5.2. We also appreciate the great help of Katrin Birgit Hänel and Tabea Jenner
with formatting the manuscript and compiling the references.
Last, but by no means least, our whole-hearted thanks also go to our families for
bearing with us for so many months. There is a life after the book. And special con-
grats to Julius Bergs (age 7 years and 11 months) for finishing his first book three
months before Dad finished this one.
Publisher’s acknowledgements
The author and publishers would like to thank the following rights holders for per-
mission to reproduce copyright material:
Map 7.1 by Hans Erren (2010) https://en.wikipedia.org/wiki/High_German_con-

sonant_shift#/media/File:Rheinischer_faecher.png reproduced under licence CC
BY-SA 3.0
Map 8.1 by Brenda Thornley, reproduced with kind permission of Felicity Meakins.
Figure 9.6 from Robert McColl Millar (2015) Trask’s Historical Linguistics, Third
Edition, Abingdon: Routledge. Reproduced with kind permission of Taylor and
Francis Ltd.
Abbreviations
.ACC accusative
.DAT dative
.NOM nominative
* reconstructed
—> process underway
> historical process (change took place)
BCE before the Christian era
C consonant
CE Christian era
COCA Corpus of Contemporary American English (See http://corpus.
byu.edu/corpora.asp.)
COP-PAST copula-past
DEF definite
GEN genitive
LMC lower middle class
LWC lower working class
MMC middle middle class
MWC middle working class
NFUT non-future
NML nominalizer
NONPAST not in the past
NP noun phrase
OED Oxford English Dictionary
PIE Proto-Indo-European
xvi Abbreviations
pl plural
poss possessive
S subject
sg singular
Skr. Sanskrit
TE-ASP tense-aspect
TOP topic
TRANS transitive
UMC upper middle class
UWC upper working class
V verb or vowel
VP verb phrase
X any syntactic function (e.g. object, adverbial)
1
Setting the scene
INTRODUCTION
Languages change – constantly, inevitably. Unfortunately, this is a little bit hard to

see when you just look at present-day English for example. The language you use
every day doesn’t change . . . Or does it? You’ll know the answer by the end of this
book.
Let’s start with a little experiment. There is a very famous English text about
1,000 years old, from the Anglo Saxon Chronicle.1 The text describes the arrival of
dragons in northern England in 793:
Ann. dccxciii. Her ƿæron reðe forebecna cumene ofer norðhymbra land. 7
þæt folc earmlic breʒdon þæt ƿæron ormete þodenas 7 liʒrescas. 7 fyrenne
dracan ʒæron ʒeseʒene on þam lifte fleoʒende.
‘A.D. 793. This year came dreadful fore-warnings over the land of the
Northumbrians, terrifying the people most woefully: these were immense
sheets of light rushing through the air, and whirlwinds, and fiery, dragons
flying across the firmament.’
You will probably recognize a word or two (like cumene ‘come’, land ‘land’ or dracan
‘dragons’), but it is very unlikely that you can read this original Old English text
without any formal training – even though this is supposedly the same language
you are reading right now!
Let’s fast forward a bit. Compare the example above to the experience of reading
a little bit of Geoffrey Chaucer, one of the most famous English poets of the later
Middle Ages (he died in 1400). Chaucer describes a scene in which a woman who
married five different men debates and laments the fact that apparently the Bible
disapproves of such behaviour:
“Thou hast yhad five housbondes,” quod he [Jesus],

“And that ilke man that now hath thee
Is nat thyn housbonde.” Thus saide he certain.
What that he mente therby I can nat sayn,
But that I axe why the fifthe man
Was noon housbonde to the Samaritan?
How manye mighte she han in mariage?
Yit herde I nevere tellen in myn age
Upon this nombre diffinicioun.
2 Understanding Language Change
Men may divine and glosen up and down,

But wel I woot, expres, withouten lie,
God bad us for to wexe and multiplye:
That gentil text can I wel understonde.
(The Canterbury Tales, The Wife of Bath’s Prologue)
“Now you have had five husbands,” he [Jesus] said,

“But he who has you now, I say instead,
Is not your husband.” That he said, with certainty,
But what he meant thereby I cannot say;
But I must ask, why the fifth man
Was no husband to the Samaritan?
How many [men] is she allowed to have in marriage?
I’ve never heard it said in all my time
How this number is defined;
Men may divine and gloss up and down,
But I know well, expressly, without lie
God bade us to increase and multiply –
That noble text I understand well.
The Chaucer text will probably look a lot more familiar and less confusing than
the Old English text. And yet you would still need to have some training in Middle
English or an amazing guessing ability in order to read and fully understand it. This
is clearly not present-day English yet. Let’s move ahead another 200 years and read
some Shakespeare, around 1600:
BEATRICE
I wonder that you will still be talking, Signior
Benedick: nobody marks you.
BENEDICK
What, my dear Lady Disdain! are you yet living?
BEATRICE
Is it possible disdain should die while she hath
such meet food to feed it as Signior Benedick?
Courtesy itself must convert to disdain, if you come
in her presence.
BENEDICK
Then is courtesy a turncoat. But it is certain I
am loved of all ladies, only you excepted: and I
would I could find in my heart that I had not a hard
heart; for, truly, I love none.
Setting the scene 3
BEATRICE
A dear happiness to women: they would else have
been troubled with a pernicious suitor. I thank God
and my cold blood, I am of your humour for that: I
had rather hear my dog bark at a crow than a man
swear he loves me.
BENEDICK
God keep your ladyship still in that mind! so some
gentleman or other shall ’scape a predestinate
scratched face.
BEATRICE
Scratching could not make it worse, an ’twere such
a face as yours were.
(Much Ado about Nothing, I.1)
At last we have reached a point where the language looks a lot more like present-
day English. It may still be hard to understand, but this is perhaps due to the style
of the author rather than the language itself. Here you probably only have to guess
the meaning of a few words and phrases (meet in meet food is an archaic word for
‘fitting’ or ‘convenient’, and humour in Shakespeare does not mean ‘banter, playful-
ness’, but rather refers to the theory of four body humours, or temperaments). In
contrast to the Anglo-Saxon Chronicle, you should be able to get an idea of what is
going on instead of just guessing a few words!
So this little experiment shows nicely that languages change. English under-
went some drastic and radical changes between ca. 700 CE and 1600 CE (a period
of almost 1,000 years!). In fact, these changes seem to be so radical that some
people would even question whether Old English is related to present-day Eng-
lish – which is also why some people prefer to call it Anglo-Saxon rather than
Old English.
The same can also be said about Latin and its “daughter” languages Spanish,
French, Italian and Portuguese. All these languages go back to Latin as their source,
or “mother language”. Hence, this is called the family of Romance languages, as
Romance is derived from Rome, the capital of the Roman Empire, and not from
romantic. But it is difficult to see the history and connections (relations) between
all of these languages at first glance. The following example gives you a famous line
from the Ten Commandments (“Thou shalt not take the name of the Lord thy God
in vain”; Exodus 20:7, originally written in Hebrew) in Latin and in French, Italian,
Spanish and Portuguese (and English as a gloss). Latin is usually considered to be
the “mother” of these four modern “daughter” languages. The development of the
daughter languages also probably took more than 1,000 years, just like Modern
English.
Latin: non adsumes nomen Domini Dei tui in vanum nec enim habebit
insontem Dominus eum qui adsumpserit nomen Domini Dei sui
frustra.
French: Tu ne prendras point le nom de l’Eternel, ton Dieu, en vain; car
l’Eternel ne laissera point impuni celui qui prendra son nom en vain.
Italian: Non usare il nome dell’Eterno, ch’è l’Iddio tuo, in vano; perché
l’Eterno non terrà per innocente chi avrà usato il suo nome in vano.
Spanish: No tomarás el Nombre del SEÑOR tu Dios en vano; porque no dará
por inocente el SEÑOR al que tomare su Nombre en vano.
Portuguese: Não tomarás o nome do Senhor teu Deus em vão; porque o Senhor
não terá por inocente aquele que tomar o seu nome em vão.
English: Thou shalt not take the name of the Lord thy God in vain; for the
Lord will not hold him guiltless that taketh his name in vain.
(Exodus 20:7)
Just like with our English examples, some similarities between the different lan-
guages can be still be found, but many things have changed drastically. Latin seems
to have changed a lot on its way to modernity. Table 1.1 gives you a first overview.
Why don’t you try to find some examples for this table yourself?
You might still think now that language change is essentially a matter of the
past, something that affected languages more than 1,000 years ago, when we still
had knights, Vikings, gladiators and so on. The fact that we can still read Shake-
speare, but not Old English, even supports such a view. However, this is far from
the truth.
Languages are in a constant state of flux, today just as in the past – even though
we might not always notice it. There are a couple of words that used to be very
popular not too long ago in 20th century English: do you know what a church key
is? In the 1960s before pop top cans, this was a special opener to puncture the tops
of soda and beer cans. A fox in the 1970s was not only an animal, but also a sexy
looking woman or man. On the other hand, nobody in the 1960s or 1970s knew the
words DVD, google, or selfie.
Pronunciation has also changed. Words like applicable and primarily used to be
stressed on the first syllable: 'applicable and 'primarily. Today, we tend to stress them
as ap 'plicable and pri 'marily. Many varieties of present-day English are currently
changing from street to shtreet. And is it student or stjudent – or even shtjudent? Ask
or aks? To cut a long story short, there is good evidence that English and all other
Table 1.1
Some formal correspondences between different Romance languages
Latin French Italian Spanish Portuguese English (gloss)2
in vanum en vain in vano en vano em vão in vain

nomen nom nome nombre nome name
adsumes prendras usare tomarás tomarás take
Setting the scene 5
living languages are still changing and will continue to do so. You will find many
more examples for this throughout the book.
We should emphasize, though, that languages do not change at a constant rate.
As the overall history of English shows, there can be periods of speeding up and
periods of slowing down. From the time of the dragon in the Anglo Saxon Chronicle
through to the 17th century, and even the 18th century, changes were complex and
rapid. As a consequence, people in, say, the 1700s could not read with ease the
literature of three centuries earlier. The extract above from Chaucer would have
presented difficulties, just as it does today. And language from a still earlier time
prompted even Chaucer to make the observation: “[y]e knowe ek that in forme of
speche is change” (‘you know also that in (the) form of speech (there) is change’),
noting the “wonder nyce and straunge” (‘wonderfully curious and strange’) nature
of early English words (in Troilus and Criseyde II, 22ff). And yet Modern English
readers have little trouble reading texts of the 1700s. The language of Jonathan Swift
or Jane Austen is stylistically different and has some unfamiliar looking vocabulary,
but it is recognizably Modern English.
The fact that languages change is a great concern to many people, and we find
frequent complaints in the media that languages aren’t what they used to be.
People mourn the loss of culture and language standards, especially in children
and young adults. From this perspective, languages change (i.e. “decline”) because
younger generations lack the competence, precision and strictness to speak their
language properly. Kids are simply too lazy. Whether the advocates who see lan-
guage as being in decline are right or wrong is a question we will discuss later on
in this chapter.
1.1 CHANGE OCCURS AT ALL LEVELS
Language is a complex phenomenon, and most linguists today agree that we need
to distinguish between several layers or levels of linguistic structure: the sound level
(phonetics and phonology), the word level (morphology), the sentence level (syn-
tax) and the meaning levels (semantics and pragmatics). Language change occurs
in all of these. Let’s take a look at some examples.
1.1.1 Phonetics and phonology

Sound changes belong to the best studied phenomena of linguistic change. To give
you just one or two examples from segmental phonology, i.e. phonology that looks
at individual sounds and their features: In Old English, the two sounds [f] and [v]
(as in feel and love respectively) never occurred in the same position in the word.
We call this complementary distribution. So, [f] (a voiceless labiodental fricative)
only occurred word initially and finally (or next to another voiceless sound), while
[v] was used in the middle of words, between two voiced sounds, such as vowels.
Thus, the Old English word for ‘dear, beloved’, lēof, was pronounced like [le:of],
while lēofan, the plural form, was pronounced like [le:ovan]. Since the two were in
complementary distribution we treat them as allophones of a single phoneme /f/.

Today, however, they do occur in the same position, and they clearly distinguish
meaning (as in feel and veal for instance). How did that happen? After the Norman
Conquest in 1066, many words from Norman French entered the English language,
among them words like veal, venison and even very. When they became part of
English we all of a sudden find word-initial [v] (which did not exist before in Old
English). And once [v] occurs word initially, we get minimal pairs like veal and feel,
or very and ferry. English also lost final vowels in words like save, which introduced
the voiced fricative word finally, so that it contrasted with the voiceless fricative; e.g.
the noun safe [seɪf] versus the verb save [seɪv]. This means that [v] and [f] are no
longer allophones but have become phonemes of English, or meaning distinguish-
ing sounds.
Another example comes from the history of French. Some vowels in French
are pronounced with a nasal quality when they occur before the consonants
n or m (in which case the consonants themselves are not pronounced): danse
‘dance’ [dãs],3 un ‘a, one’ [œ̴ ], vin ‘wine’ [vɛ̃], faim ‘hunger’ [fɛ̃], parfum ‘perfume’
|parfœ̴ ] etc. This, however, is a phenomenon that only developed slowly and
gradually in the Middle Ages and beyond. First, we simply see a vowel followed
by a nasal consonant (VN), then the vowel is nasalized (ṼN), then the nasal con-
sonant is lost (Ṽ), so (VN) > (ṼN) > (Ṽ). Between 900 CE and 1300 CE, the first
step occurred; i.e. vowels were nasalized. Between 1300 CE and 1600 CE alterna
tive spellings suggest a lot of variability in the presence or absence of nasal con-
sonants in this combination, so (ṼN) ≈ (Ṽ). In modern Standard French (from
1600 CE onwards), the nasal consonant is lost altogether (Ṽ). Second, this kind
of sound change did not affect all vowels [ɛ̃, œ̴ , œ̴ , ɔ̃ ] at the same time. While
they were all nasalized by the end of the 14th century, not all of them have lost
their following nasal consonant. Table 1.2 summarizes the state of affairs for the
different periods.
Apparently, as we can see in Table 1.2, vin and maison were the first to show vari-
ability in having (or losing) their nasalized consonants. These seem to be the leaders
of this change, and they were followed by the others. Between the 14th and the 16th
centuries, all vowel-consonant combinations showed this variability, and they all
lost their consonants by roughly 1700 CE. What this shows is that sound change can
proceed very gradually and can actually take several centuries.
Table 1.2
The development of nasalized vowels in French, ca. 1200–1800 (based on Sampson 1999)
ca. 13th century ca. 14th–16th century ca. 17th–18th century

a vend ‘sell’ [vẽnt vãnt] [vã(n)t] [vã]
b vin ‘wine’ [v ĩn vẽ(n)] [v ĩ(n) vẽ(n) vɛ̃(n)] [vɛ̃]
c maison ‘house’ [mæjzõn mɛzũ(n)] [mɛzõ(n) mɛzũ(n)] [mɛzõ mɛzɔ̃]
d humble ‘humble’ [ỹmblə] [ỹ(m)blə œ̃(m)blə] [œ̃blə]
Setting the scene 7
Grassmann’s Law
A very famous textbook example for sound change is Grassmann’s Law. It
was named after Hermann Grassmann, a 19th century German mathema-
tician and linguist, who discovered that in Ancient Greek and in Sanskrit
(the ancient language of India) the first of two aspirated stops (i.e. stops
accompanied by some breathing noise) within a word loses its aspiration
(the breathing noise). So, in the mother language of Sanskrit and Greek,
Proto-Indo-European, *dhi-dhē-mi ‘I put’ or ‘I place’ with two aspirated stops
(dh) turns into Sanskrit da-dhā-mi and into Greek ti-thē-mi with only one
aspirated stop. Interestingly, this so-called dissimilation process can be used
in historical linguistics to reconstruct earlier stages and changes in the lan-
guage. So, Greek now has two different forms for ‘hair’: thriks in the nomina-
tive singular, but trikhós in the genitive singular. How is that possible? Both
go back to forms with two aspirated consonants: *thríkh-s for the nomina-
tive singular and *thrikh-ós for the genitive singular. Why did the genitive
change as predicted (and lose its first aspirated consonant), while the nomi-
native behaved seemingly unruly and kept it? The answer is actually simple:
the aspiration of *kh in the nominative is before s and was lost (with *kh-s >
ks) before Grassmann’s Law kicked in. And since the new *thrík-s is not sub-
ject to Grassmann’s Law (it doesn’t have two aspirated consonants anymore,
but only one), the forms are not affected by the law. So they only appear to
be unruly in Modern Greek. In fact, however, all changes were completely
regular. The same happened to all other words with similar structures (e.g.
*thréph-s-ō ‘I will rear’ > thrép-s-ō, and *thréph-ō ‘I rear’ > tréph-ō, and many
more).4
Suprasegmental phonology (phonology beyond the segment, i.e. beyond the

individual sound, like rhythm, metre and stress) also undergoes change. We have
seen some examples above where English words today are stressed in different posi-
tions: bal 'cony (old) versus 'balcony (new) or Carib'bean (old) versus Ca'ribbean
(new). Note that the latter is still very much in flux, and most speakers are (still)
very confused about where to put the stress. Both stress patterns are listed in the
major dictionaries such as the Oxford English Dictionary (OED) (www.oed.com)
and Merriam-Webster (www.merriam-webster.com), but where English stress is
heading is towards the third last, or antepenultimate, syllable (so 'balcony won out
and we predict Ca 'ribbean will too). If you are interested, you may want to check
out the pronunciation of the word in Pirates of the Caribbean, the cruise company
Royal Caribbean, Billy Ocean’s song “Caribbean Queen” and Bob Dylan’s “Carib-
bean Wind” – among many others. As we will see below, variation like this is at the
heart of every language change process.
1.1.2 Morphology
In order to show an example of morphological change we come back to almost
present-day English. Remember Beatrice’s lines from Shakespeare’s Much Ado
about Nothing quoted above:
BEATRICE
Is it possible disdain should die while she hath
such meet food to feed it as Signior Benedick?
The last word in the last line is hath, but we would expect Modern English has
in that particular position. Early Modern English, the language of Shakespeare and
Elizabeth I, still had -th for the third person singular present tense; this is an inflec-
tion (an affix that gives grammatical information). A favourite pastime of most lin-
guists is to construct paradigms, i.e. tables which lay out sets of inflectional forms.
The verbal paradigm for Early Modern English roughly looked as in Table 1.3.
Table 1.3
The inflectional forms of Early Modern English verbs
1st person I sing

Singular 2nd person you, thou sing-est / you, thou sing
3rd person he sing-eth / he sing-s
1st person we sing

Plural 2nd person you sing
3rd person they sing
This, in turn, goes back to the much more complicated verbal paradigms of Old
English:
1st person ic· singe

Singular 2nd person þu singst
3rd person he-, hı-t, he-o
sing(e)þ
1st person we- singaþ

Plural 2nd person · - singaþ
ge
3rd person hı-e singaþ
Many of the verbal endings had been lost in the Middle English period already
(the plural, for instance, is longer signalled by -aþ in Shakespeare’s time). Some
other endings are still there (-st in the second person singular, -eth in the third
person singular), but these are variable for Shakespeare; i.e. they don’t occur all
the time, but only occasionally and not with all verbs. Wash-eth and teach-eth, for
instance, are already gone for Shakespeare and have been replaced by wash-es and
Setting the scene 9
teach-es. And, again, variation is what underlies linguistic change. Ultimately, we

lose the -th forms altogether so that in present-day English we only have -s for the
third person singular and no inflection for the second person singular.
Similarly, word formation, i.e. the creation of new words, is affected by linguistic
change. According to Bauer (1994: 38), abbreviations like CD (‘compact disc’), CIA
(‘Central Intelligence Agency’) and NATO (‘North Atlantic Treaty Organization’)
were very rare in the 19th and early 20th centuries. We find only few counterex-
amples such as American Telephone and Telegraph, which became AT&T in the late
1800s, or situation normal all fouled up, which became SNAFU in 1941 or 1942.
From the middle of the 19th century, onwards, however, we see a massive increase
in these abbreviated forms (LASER comes from light amplification by stimulated
emission of radiation, TASER is Thomas A. Swift Electric Rifle, and NSA/CSS stands
for National Security Agency/Central Security Service). Bauer documents an almost
six-fold increase in abbreviations in the Supplement to the Oxford English Diction-
ary, from just 2 (or 0.4% of all new word formation types) in 1880–1913 to 13 (or
2.5% of all new word formation types) in 1939–1982.
1.1.3 Syntax
An example from Modern German can help to illustrate changes on the syntac-
tic level. Modern German is usually considered a verb-second language in main
clauses; i.e. the verb always comes in the second position, no matter what is in the
first position in the sentence (note that English looks very similar, but is actually
Subject-Verb; i.e. the verb always follows the subject, but need not be in the second
position). This is illustrated in (1.1a and 1.1b) below.
(1.1) a Coco Chanel erfand das kleine Schwarze.

Coco Chanel invented the little black [dress]
b 1926 erfand Coco Chanel das kleine Schwarze.
1926 invented Coco Chanel the little black [dress].
‘In 1926 Coco Chanel invented the little black dress.’
However, in Standard German subordinate clauses the finite verb comes last, as
shown in (1.2).
(1.2) Es ist interessant, dass Coco Chanel das kleine Schwarze erfand.
It is interesting that Coco Chanel the little black [dress] invented.
‘It is interesting that Coco Chanel invented the little black dress.’
Interestingly enough, recent informal spoken varieties of German show a vari-

able pattern when the subordinate clause is introduced by weil ‘because’ or obwohl
‘even though’. (1.3a) shows the standard verb-final pattern, and (1.3b) illustrates
the informal innovation with what seems to be an English word order, i.e. sub-
ject followed by verb, and possibly preceded by other material, such as weil and
obwohl.
(1.3) a Hepburn wurde berühmt, obwohl Coco Chanel das kleine Schwarze erfand.
H became famous even though CC the little black [dress] invented.
‘H became famous even though CC invented the little black dress.’
b Hepburn wurde berühmt, obwohl Coco Chanel erfand das kleine Schwarze.
H became famous even though CC invented the little black [dress].
‘H became famous even though CC invented the little black dress.’
Here we see syntactic change in progress. German subordinate clauses seem to

be gradually losing their verb-final order and aligning somehow with main clause
word order (even though, strictly speaking, they are not verb-second). A number of
possible causes for this change have been identified. Some scholars believe that hav-
ing mixed word order in main and subordinate clauses is more difficult to acquire
and process. Other scholars believe in English influences (note that the English
and German subordinate clauses in (1.3b) are exactly alike). Some think that this
kind of variation in word order is actually very old and fairly stable (Selting 1999;
von Polenz 1999; compare Elspaß 2005); in other words, what we see today is not
change but ongoing variability. Yet other scholars have claimed that the verb-second
order in subordinate clauses began its life as a special, more subjective expression of
speaker perspectives (e.g. Keller 2003). For them, the subordinate clause in (1.3b)
expresses a more debatable, subjective reason for what happens in the main clause
than the alternative in (1.3a). Whatever the right answer might be, we find variable
word order in subordinate clauses, and this may be the ultimate source for change.
We will come back to this question in our chapters on semantics and on syntax.
A much less disputable example of syntactic change comes from English. Have
you ever wondered where prepositions like before and behind and conjunctions like
because come from? They actually stem from constructions like John is in front of/
at the back of something or Something is the cause of something else. In other words,
these prepositions and conjunctions gradually developed out of perfectly transpar-
ent lexical material into some grammatical elements like prepositions and conjunc-
tions. And the same happened to the Old English numeral ‘one’, ane. This, over
time, turned into the modern indefinite article a and an, as in I have a dog. Again,
something lexical became something more grammatical. This very important and
pervasive process is called grammaticalization and will be the subject of intensive
discussion in Chapter 6.
1.1.4 Semantics
Some of the changes mentioned above may go unnoticed by many speakers. For
example, most Germans claim never to have used or even heard of weil or obwohl
with verb-second order in subordinate clauses as in (1.3b) above! This is usually
very different for changes on the level of meaning, and semantics in particular,
which are very often noticeable. Speakers often realize that words seem to change
their meaning. English gay is such an example. According to the OED, it began its
life around 1100 CE as a borrowing from Norman French, and it meant something
Setting the scene 11
like ‘happy, bright, cheerful looking, noble, excellent’. Later on, around 1600 CE, it
also meant something like ‘hedonistic, promiscuous, dedicated to social pleasures’.
The meaning ‘homosexual’ only came up much later, in the English of the second
half of the 19th century. Many speakers today still know about the original mean-
ing, for instance when they say things like this sentence by Jane Campion in the
Village Voice: “There’s no prizes for being happy and gay in a noir movie – that’s fail-
ure” (OED, s.v. gay). Or consider the (American) Christmas song “Deck the Halls”:
“Don we now our gay apparel. Fa-la-la-la-la, la-la-la-la! Troll the ancient Yule tide
carol . . .” And yet most are aware of the changes that took place. Only think of the
various new meanings of troll in this case . . .
Similarly, English has two words for canines, namely dog and hound. Today, the
former refers to any particular kind of the canine species, while the latter refers to
hunting dogs in particular. In the Middle Ages, the situation was exactly reversed.
Dogge was a particular kind of dog, like a large hunting dog (i.e. bulldog, mastiff,
Great Dane), while hound was the general term for canines. This is still the situation
we find in Modern German: Hund means ‘dog’, while Dogge means something like
‘Great Dane’.
Salary is a word in English that was incorporated from Latin salarium. Salarium
was the allowance of salt (Latin sal) for Roman soldiers (which could then be traded
in for money). It obviously changed this meaning, and now salary does not have
anything to do with salt, but simply means (regular) income.
So, to cut a long story short, the fact that words change their meaning, become
more positive or more negative, or just change their reference, is something that
most people are quite aware of.
1.1.5 Pragmatics
Changes of meaning in context, i.e. in pragmatics, are more difficult to detect, and
many people don’t notice them. Some examples include swearing and the use of
taboo words. While in the Middle Ages speakers of English felt very strongly about
the use of religious terms (including, among others, I swear, Oh God!, Jesus!), these
are less frowned upon today. Present-day speakers in many English speaking com-
munities rather find words referring to body parts and bodily functions obscene
(the F-word is certainly among the most famous examples, and so is the C-word).
Current runners-up to this category are racial terms, and many speakers would
rather be caught saying f . . . than n . . . .
A second example from the history of English comes from address terms. In
Middle English (Chaucer’s time) speakers had to choose between thou (second per-
son singular) and you (second person plural). Two hundred years later, in Early
Modern English (Shakespeare’s time), speakers did not choose between singular
and plural when they used these pronouns, but rather between polite and impolite
when talking to singular addressees. If somebody used thou they were signalling
either solidarity (you were on a par socially) or social superiority (the speaker using
thou towards the addressee was more powerful than the addressee). If somebody
used you they were signalling social inferiority (the speaker using you towards the
addressee was less powerful than the addressee). But once social ranks were less vis-
ible and clear cut, speakers were sometimes unsure about which form they should
use. So to be on the safe side they opted for the more polite, socially upward looking
form – and used you. Eventually, thou was lost in this upward spiral of politeness.
(The thou form was also a more demanding pronoun, since it required an extra
ending on its verb. As Table 1.3 shows, the you form simply used the same ending
that all the plural pronouns took. As English was in the business of getting rid of
such endings, it suited speakers to dump thou for this reason also.)
Another process that historical linguists investigate is pragmaticalization, i.e. the
development of pragmatic markers out of other (usually lexical or grammatical)
material. This is a very common process, and the same can be observed in English
when you look at the history of like, for example, which developed out of a (lexical)
verb (1.4) and an adverb (1.5) into some element that helps to structure the dis-
course of the speaker, for example as a “quotative” (1.6), which introduces quoted
speech, much like inverted commas, or a careful hedge-like function that saves the
speaker’s face when making a request (1.7).
(1.4) I like Halloween.

(1.5) He is like his brother.
(1.6) And I’m like ‘You can’t go there, dude!’
(1.7) Could I like borrow your lecture notes?
Obviously, like in English today has many different functions, some of which are
lexical (1.4) and (1.5), while others are more pragmatic and discourse oriented (1.6)
and (1.7).
Japanese demo
One last example comes from Japanese. The conjunction demo has changed
from a regular, connecting element meaning something like ‘even’, as shown
in (a), into a clause-initial discourse marker, as shown in (b). In present-day
Japanese it can also signal ‘changing the topic’, ‘opening the conversation’ or
‘claiming the floor’.
a kono heya no denki hiruma demo itsumo tsuite-iru

this room GEN light day.time even always turn.on:TE-ASP:NONPAST
‘The light in this room is always on even during the day.’
b a, demo Aso itta n dattara jaa ano jooba toka wa uma ni
oh but Aso went NML COP-PAST if riding well horse- like TOP horse on,
back
riding
‘Oh, but if you went to Aso, then, well, did you go horseback riding?’
The development of demo is thus from less personal to more personal, or

from ideational (lexical) to expressive, and is thus yet another example for
pragmaticalization, i.e. the development of “extra” pragmatic material out of
formerly lexical or grammatical elements.
1.1.6 Change across different levels

What makes matters even more interesting (but also more complicated) is the fact
that there is some interaction between these different levels, so that change on one
level can trigger change on other levels. When, for example, phonetic erosion leads to
the loss of final syllables, this often has consequences for both morphology and syn-
tax, as (in the case of Germanic languages) final syllables are the place for inflectional
suffixes (which signal grammatical functions and relations). When inflectional suf-
fixes are on the way out, syntactic structures usually need to be adjusted (and fixed),
so that it remains clear what these functions and relations in the sentence are. As we
discuss more in Chapter 6, this is what happened in English. Old English was still
very rich in inflections, and you could usually tell what the subject and the object of
a given sentence were by looking at the inflection of the verb (which tells you if the
subject is singular or plural, and first, second or third person, and how many objects
there should be) and the inflection of the relevant noun phrases (this tells you which
one is the subject, and which ones are the direct and the indirect object, if any). But
final syllables and inflections are easily lost. They simply wear off! This is something
we observe in many different languages. For simple phonetic reasons speakers tend
to erode the pronunciation of word-final syllables. And if inflectional information
is signalled in word-final position, this also gets lost on the way. As the old way of
distinguishing subject and object is lost, new ways arise. The solution to the prob-
lem was word order, on the one hand, and prepositions, on the other. So instead of
morphological inflections, present-day English signals grammatical function by word
order and prepositions. Whatever stands in front of the verb is usually the subject,
and whatever follows it are indirect and direct object, respectively. And prepositions
help us, for example, to find and understand adverbials of all sorts.
1.2 VARIATION AND CHANGE – TWO SIDES OF THE SAME COIN
As we have mentioned above, language variation and language change are intri-
cately entwined – they are two sides of the same coin. Language change is thus not
very different from evolution in general. Charles Darwin coined the idea that evolu-
tion basically rests on variation (or mutation) and selection (the infamous survival
of the fittest). In language change, the story is not much different.
In contrast to everyday beliefs, language is actually a very dynamic and variable
activity. Just think about pronunciation. Hermann Paul, a very famous 19th century
linguist, compared this to hitting the bull’s eye with an arrow. You know quite clearly
what the word should sound like every time you want to say something. This is your
bull’s eye. But no matter how hard you try there is no guarantee that you will hit it
perfectly. Rather, you actually might miss it here and there. These are the teeny tiny
differences that we experience in our everyday pronunciation.
The fact that we miss the phonetic bull’s eye is hardly surprising, though. Our
phonetic articulation system is essentially a muscular one. Pronouncing a word is
like playing tennis or playing Bach’s Air on the G String on your violin. It will never
quite sound the same twice, simply because you can’t control your muscles per-
fectly. If you could, you’d never serve a fault and never fail to hit that C sharp in the
ninth bar. Similarly, you’ll never pronounce the word cat the same way twice. There
will always be minuscule differences, and these are some of the sources of language
variation.
A second source, related to the first but also more noticeable, has to do with the
fact that humans like to imitate. Countless studies in psychology and sociolinguis-
tics have shown that we like to adapt (the technical term is “to accommodate”) to
the people we talk to when we like them, and that, naturally, we dissociate if we don’t
like them. Only think about your behaviour when it comes to accents and dialects.
Many speakers catch themselves imitating (subconsciously) the accent of the
people in their favourite holiday resort. This is another source of language varia-
tion. Similarly, we try not to sound like the people we don’t like. There is a nice little
story about a linguistics professor from the south of Germany who was appointed
to a professorship in the very far north. The two dialects (and cultures) are very
distinct; northerners usually find it difficult to understand southerners, and vice
versa. For independent reasons, this professor had a very hard time in the north,
and he noticed that the more he did not feel at home, the more he used southern
dialect features in order to show that he was not accommodating to an environment
he didn’t like.
Another source of language variation is “analogy”. Douglas R. Hofstadter, the
famous author of Gödel, Escher, Bach, once described analogy as the fundamental
core of human cognition (https://www.youtube.com/watch?v=n8m7lFQ3njk). The
idea is very simple: humans like to build analogies. We can see that in our children
every day. For some kids at least, it’s clear that the past tense of teach is not taught
but teached. And a German speaking child, upon hearing the church bells ring,
might exclaim: Es hat geglockt!, lit. ‘It has belled’. The word geglockt from a verb
glocken does not exist in German today. And yet the fictitious verb glocken (lit. ‘to
bell’) in this case follows all the necessary morphological rules and makes perfect
sense – even for adult speakers. So why shouldn’t the child try to use it? He or she
probably does not know that the common word in this case is läuten as in proper
Es hat geläutet! lit. ‘It has rung’), which makes another verb like glocken ‘to bell’
superfluous.
And before you start laughing: what’s the plural of fungus? Hippopotamus? How
many of you still say theses? Traumata? We will come back to this in Chapter 5
when we talk about morphological change. In any case, it seems intuitively clear
that humans (both big and small) like analogies and that analogies can lead to
language variation and thus also to linguistic change – the past of climb once was
clomb, but climbed pushed out clomb, just as teached might well push out taught in
the long run.
Language acquisition is yet another point of entry for linguistic variation and
change. It is quite obvious that children are not perfect copies of their parents, nei-
ther physically, psychologically, nor in terms of their language. The job of small
children learning their native language is to decipher the endless stream of sounds
they hear from the people around them, to identify the words and phrases, and to
figure out what they mean and how they are put together. This job is usually done
with the help of a logical process called abduction. Sherlock Holmes used abductive
reasoning in his cases. For example, if the door is locked from the inside, but the
window is open (on the fourth floor), if there is some brown-orange monkey hair
clenched in the victim’s fist, and if the victim choked on a banana, we can assume
that the murderer could have been the orangutan named Rodney that escaped from
the local zoo two days ago. Note that the conclusion that Rodney is the murderer is
not strictly logical or necessary. There may be other explanations for the open win-
dow, the brown-orange hair and the banana, but Rodney as the potential murderer
seems like the most plausible explanation at this point. This, essentially, is abductive
reasoning. You start from the facts and think long and hard about possible sources
and mechanisms that could have led to this particular factual situation. But the
same kind of reasoning can also happen very quickly and intuitively. This is what we
find in first language acquisition (obviously, small children usually don’t think long
and hard about the linguistic input that they get): children who hear members of
their environment repeatedly uttering sentences in Subject-Verb-Object ordering
might assume that this is the underlying grammatical rule. If the rule they assume is
the same as the one that produced all these sentences, we get a perfect copy. Some-
times, however, the rules the kids develop might differ a little bit from those of the
previous generations. These new rules then produce almost the same output as that
of the previous generation, but every now and then these new rules might also lead
to tiny innovations and to language variation.
Yet another major source for language variation is contact in the widest sense,
i.e. contact between different languages, but also between dialects and accents of
the same language. (Note that distinguishing between dialects and languages is a
particularly difficult problem. We will come back to this in Chapter 9.) Let’s begin
with the first, language contact proper.
Ever since language diversified in prehistoric times (which is a complicated story
in itself; see Chapter 9), different languages have been in contact with each other.
To be more precise: the speakers of these languages were in contact with each other.
The possible outcomes of these contact situations are actually quite simple. First,
nothing productive might happen. Speakers meet and depart again without any
long lasting effects. But languages can also exist side by side, and for long periods
of time; Evans (2012) describes the remarkable linguistic diversity that exists in
New Guinea; the southern region alone (an area about the size of the Netherlands)
contains about 40 different languages from between five and six unrelated language
families.
Sometimes one group of speakers becomes so dominant that the other group
stops using their language. This is called language death, and unfortunately this
seems to be the most common scenario in language history. The age of colonial-
ism alone must have killed thousands and thousands of native languages in North
and South America, Africa and Australia, and many are still endangered. Of the
approximately 200 native North American languages still spoken today, more than
two thirds count as endangered.
As Chapter 8 describes, contact situations can trigger all sorts of changes. One
language may incorporate linguistic material from the other, and lexical material
is the most readily transferred. Scheler (1977) suggests that out of the ca. 500,000
entries in the Shorter Oxford English Dictionary, probably more than 350,000 words
(70%) come from other languages such as Latin, French, Spanish, Hindi, Greek and
many more. In tandem with the words, speakers may also incorporate sounds and
linguistic structures, such as morphological patterns. When the first man-made
object to orbit the Earth, the Russian Sputnik 1, launched in 1957, became vastly
popular, it also popularized the Russian (and Yiddish)5 morphological pattern with
-nik (a person who is connected to something). From 1957 onwards we thus find a
ton of new English words created on that basis. The OED lists a few: muttnik, beat-
nik (both 1958), flopnik, puffnik, stayputnik, kaputnik (all 1968), protestnik (1965),
computerniks (1973), conferenceniks (1989), no-goodniks, faroutniks, sickniks (all
1993).
Finally, when two or more languages are in contact, the speakers of these lan-
guages may develop yet another interim language in order to be able to communi-
cate. This new language is called a pidgin. Pidgins are typically very simple language
used solely by adults for purposes such as trading. Most pidgins are created by mix-
ing the lexicon of the “superstrate” (more powerful) language with some grammati-
cal structures of the “substrate” (less powerful) language. The results are simple (but
fully functional) languages. Most of these newborn languages quickly disappear
again when the trading situation is over or when one group learns the language of
the other, but some do survive. When pidgins are learned by children as their native
language, these pidgins turn into creole languages. These creoles rapidly grow in
complexity and develop all the features we know from already existing languages
such as English, Latin, Mandarin and so on. Today we have more than 100 such
creole languages spread all across the world (you can get an excellent first overview
by checking out the online Atlas of Pidgin and Creole Structures, APiCS: http://
apics-online.info). So, one possible outcome of language contact can be the birth of
a whole new language.
Contact between different dialects or accents, or even just with other speakers,
is another important source for language variation. In any given linguistic com-
munity we find regional differences in language use. Just think about your own
language and what people “from the North”, “from the South”, “from the West”
and so on sound like. When these different dialects come into contact with each
other, linguistic features may spread from one variety of the language to the other.
Essentially, the same can be found when any kinds of speakers are in contact. As
we mentioned before, no one talks in exactly the same way, and every speaker is
characterized by a personal idiolect: his or her individual use of language. Some

of these idiolects are easily noticeable, e.g. when somebody has a peculiar speech
feature or likes to use certain remarkable phrases a lot. But even when we don’t
notice it at first hearing, every speaker is characterized by particular features. As
described earlier, when speakers talk to each other, they tend to accommodate or
dissociate. Accommodation means that we (subconsciously) like to sound like the
people we talk to, especially when we like them or when they are some kind of
role model for us. Dissociation is what we notice when speakers (again, subcon-
sciously) try not to sound like the people they talk to because maybe they don’t
like them or do not want to be associated with them. We can accommodate and
dissociate in very small, almost unnoticeable features like vowel quality, but also
in larger and more noticeable domains such as whole accents, phrases, vocabulary
and much more. It is quite obvious that this is a huge source for linguistic variation
of any kind.
Finally, one last, but important source of language variation is simply playfulness.
Anthropologists and philosophers have sometimes characterized human beings as
homo ludens, or ‘playing man’. The idea is quite powerful: humans love to play, and
they need play in order to thrive and survive, and to organize their lives. It can be
argued that we also love to play with language in the widest sense. “Play” in this
case means that we like to invent new and fancy words, that we love to swear and
curse, that we like hyperbole (exaggerations) of any kind. This can form the basis
for linguistic change. Some very nice examples come from present-day English.
Consider (1.8) and (1.9):
(1.8) Rosa Luxembourg was a terribly nice name for a revolutionary.

(COCA, The Life and Work of Alphonse Kauders, 1995)
(1.9) It is a sick board and worth every penny of your hard-earned cash.
(2007, On Board, http://blog.oxforddictionaries.com/2011/05/
inverted-meanings-sick/)
In (1.8) we find a terribly nice name. But terribly comes from terrible, and ter-
rible means something like ‘very shocking and upsetting’. This, however, can’t be
the intention of using terribly in (1.8). Rather, it means something like ‘really’ or
‘very’. The same happened in (1.9), taken from a snowboarding and skateboard-
ing context. A sick board is not a board that is sick or used for being sick; it is ‘a
cool, very good looking board’. Sick in this context does not mean ‘unwell’ or ‘ill’
but ‘cool’. This development of so-called boosters (elements used for emphasizing)
from negative to very positive is a process that happens quite frequently, and it has
to do with playfulness of language users. Speakers like to use words in novel ways to
get attention, display their wit and prove their coolness. This in turn may give them
social advantages. Keller (1994) even described this as an evolutionary advantage.
He defines several maxims of successful language use, and one of them is, “Talk in
such a way that you are noticed” (if you want to be noticed). This has been the topic
of countless works of literature, from William Shakespeare (Benedick and Beatrice
in Much Ado about Nothing) to Edmond Rostand (Cyrano de Bergerac, who wins
his lady’s heart by writing powerful poetry, though he can’t show himself because
of his big nose). It is one of the key elements in classic rhetoric. The powerful use of
language convinces, or at least persuades, your hearers! So, in brief, using cool new
lingo makes you the cool kid on the block.6 In any case, it’s a source for novelty and
language variation.
One final word of caution: in the previous section we have shown that language
variation is an important prerequisite for language change. Most changes ultimately
go back to variation. However, not all variation leads to linguistic change. Sociolin-
guists have clearly documented that in some social contexts and with some variables
speakers do not accommodate or dissociate as we would predict. In English there
are two ways of pronouncing words like dancing: the more standard [dænsɪŋ] and
the less formal [dænsɪn]. The latter is also sometimes written as dancin’, and you can
hear it very clearly in Bruce Springsteen’s “Dancing in the Dark”. This variability has
been around for centuries. Jonathan Swift, for example, the author of Gulliver’s Trav-
els, could rhyme: “See then what mortals place their bliss in! / Next morn betimes
the bride was missing” (Phyllis, 1716, ll. 25–26). In the 19th century literary works
of Charles Dickens both -ing and -in spellings can be found, and they are often used
to distinguish working class from upper class speakers. In any case, the point is that
this sort of variation has more or less been stable for many, many years, and there is
no reason to believe that one form will oust the other any time soon.
1.3 EVIDENCE OF CHANGE IN PROGRESS
As we have mentioned before, language change is not something that only hap-
pened in the past. It is happening all around us, all the time. We just have to keep
our ears and eyes open. But this sounds much simpler than it actually is. Essentially,
there are two different pieces of evidence we can look at when we study linguistics.
The first is called change in apparent time; the second is called change in real time.
Let’s begin with the first.
When we study change in apparent time we assume that the kids of today will be
the adult speakers of tomorrow, and whatever we witness in the younger genera-
tions could be new forms of the language coming in, i.e. incipient linguistic change.
Figure 1.1 shows one such example. Modaressi (1978) studied Farsi as it is spoken
in Tehran and in Ghazvin, a city about 150 kilometres from Tehran. One variable is
(an), i.e. the raising of [a:n], to [o:n] or [u:n]. These new forms [u:n] and [o:n] are
very popular, and speakers are apparently very well aware of this variable, since the
new forms dramatically drop in frequency for the more formal speech styles (such
as reading style, word list style and minimal pairs). What is interesting, however, is
the use of this variable in the casual and careful styles in Ghazvin by two different
age groups. Apparently, in both these styles the younger age group (aged 10–29)
uses the new form somewhat more than the group aged 50+. Assuming that the
younger group will keep this feature when they grow up, we may conclude that this
is ongoing linguistic change, and that the new form is becoming more popular as
Figure 1.1
Change in apparent time in Ghazvin (an) (based on Modaressi 1978)
80
70
60
% raised (an)
50
40
30
20
10
0
Casual Careful Reading Word lists Minimal pairs
Tehran 10–19 Tehran 20–29 Tehran over 50

Ghazvin 10–29 Ghazvin over 50
younger speakers will promote its use and actually move towards innovative Tehran
language use. This, in a nutshell, is change in apparent time. We take a slice of lan-
guage at any given time (this is also called synchronic) by speakers from different
age groups, and then project their synchronic behaviour into the future.
While this method is generally well accepted and used in a large number of stud-
ies, there is also one crucial problem associated with it. This problem is called age
grading. It is a well-known fact that children and teenagers speak differently from
adults. Popular current (2016) language use by teenagers includes chirped, as in
Ms. Hubble yelled at me for using my cell in class. Man, I got chirped! It means some-
thing like being told off very clearly. Another example is ship, as a verb as in I ship
Brangelina so much, they are really cute! It means that you endorse and support this
particular relationship (and this in turn makes you a relationshipper, or simply ship-
per, of Brangelina in this case). There is very little doubt that these phrases prob-
ably won’t make it till 2020, if that long. Many popular words and phrases simply
disappear (just like church key and fox from the 1960s and 1970s, as mentioned
above). The principle behind all of these examples is called age grading. Certain
forms and expressions belong to certain age groups (and not just teenagers; there
are also language forms that rather belong to senior citizens – would you expect
teenagers to refer to their knickers?). Obviously, distinguishing between age grad-
ing and actual change in progress from an apparent time perspective can be very
difficult, and sometimes even impossible. In the case of Ghazvin mentioned above,
however, the evidence for actual change is still somewhat clear, with an age group
that goes beyond the teenage years, and a fairly robust pattern that matches Tehran
as a possible role model.
Real time is the second approach to studying linguistic change. In real-time stud-
ies, we take language samples from two different points in time (this is called dia-
chrony) and compare the two different language stages with each other. Whatever is
different from Time1 to Time2 must have changed in the time between. Figure 1.2
presents the results from a famous study carried out in 1953 by Alvar Ellegård.
Ellegård investigated the development of do in English questions and negatives
between 1400 and 1700 (we call this “periphrastic do”). In order to do that, he iso-
lated four different contexts in which do could be used:
Affirmative declaratives (statements): I eat cheesecake vs. I do eat cheesecake.

Negative declaratives: I eat not cheesecake vs. I don’t eat cheesecake.
Eat you cheesecake? vs. Do you eat
Affirmative questions (interrogatives):
cheesecake?
Negative questions: Eat you not cheesecake? vs. Don’t you eat
cheesecake?
When you look at Figure 1.2 at least two things become clear. First, the use of
periphrastic do increased on the whole from 1400 to 1700. Second, it did so at dif-
ferent rates for the different contexts. While the negative questions were very quick
and reached 90% by 1600, negative declaratives were quite slow and only had about
35% do in 1600. What we see here is a study in real time; i.e. Ellegård took several
slices of language at different intervals and compared them to each other. What we
see is change happening in real time, over 300 years. The positive aspects of this are
obvious: we don’t have to speculate what could happen in a few years, and whether
Figure 1.2
Percentage of do forms in various sentence types, 1400–1700 (based on Ellegård 1953)
100
90
80
70
60
% do
50
40
30
20
10
0
1400 1500 1600 1700
Negative Questions Affirmative Questions

Negative Declaratives Affirmative Declaratives
we see age grading or change in apparent time. On the other hand, we only get sev-
eral snapshots of the language at different times and don’t see exactly what is hap-
pening in between. We can speculate how the language got from Time1 to Time2,
but essentially all we have are two snapshots. Nevertheless, this is one of the most
widespread and accepted methods of catching change in progress. And the smaller
the intervals, the more details become visible, of course. On the other hand, we also
should not forget that the majority of the world’s 7,000 or so languages are not writ-
ten, and that our data hardly reaches back more than about 100 years! So studies in
real time are pretty hard, and we have to rely a lot on linguistic reconstruction (our
topic for Chapter 9).
1.4 ATTITUDES TO CHANGE
As we have mentioned before, we do not always notice linguistic change. But when
we do, many people actually do not approve. From a socio-psychological perspec-
tive, any sort of change often leads to worries, scepticism and even fears and anxi-
ety. And language change is no different in this respect. In a famous sociolinguistic
study, Milroy and Milroy (1999) have identified a so-called complaint tradition.
This tradition seems to reach back centuries, and it comprises two different aspects,
or complaint types. Type 1 complaints are rather “legalistic” and concerned with
correctness. They “attack ‘mis-use’ of specific parts of the phonology, grammar,
vocabulary” (1999: 31). Type 2 complaints are rather “moralistic” and “recommend
clarity in writing and attack what appear to be abuses of language that may mislead
and confuse the public” (ibid.). Both types of course interact and can feed into each
other. What concerns us here is the fact that linguistic variation and change often
lie at the heart of complaints of both types.
First, many speakers disapprove of linguistic variation and rather expect linguistic
question to have “right or wrong” answers. Try it for yourself: what is the past tense of
the verb dive? Dove or dived? Both are perfectly OK. The former is preferred in North
America, the latter in Great Britain. But educators of all kinds usually find this hard
to accept. Germany does not have an official language academy that defines what is
right or wrong in “Standard German”, but it has a dictionary which has almost acad-
emy status, the so-called Duden, named after its first author, the high school teacher
Konrad Duden (1829–1911). Over the last 20 years or so, German orthography, and
with it the Duden, has seen a large number of changes. Many of these changes called
the language guardians to the fore. So, the word ketchup can be spelled as either
Ketchup (old) or Ketschup (new). And what’s worse: it is either grammatically mas-
culine der Ketchup or neuter das Ketchup. Also, paragraph can be spelled as either
Paragraph or Paragraf. This kind of reform brought lots of protests in its wake. In a
very popular weekly newsmagazine, Der Spiegel, we find comments such as
Anything goes. Alt und Neu bestehen nebeneinander, viele Kann- und
einige Mussbestimmungen weichen die Sprachwirklichkeit auf, unzählige
Varianten kursieren – niemand weiß mehr, woran er ist.
‘Anything goes. Old and new coexist, many “can” and “have to” rules soften
and blur linguistic reality, countless variants circulate, nobody knows right
from wrong anymore.’
(http://www.spiegel.de/spiegel/print/d-31699578.html)
Just to clarify: most of the variants which are now acceptable in no way endan-
ger intelligibility or even bear the risk of ambiguity. It’s simply up to writers if they
like Biographie or Biografie better. And yet the reactions to this kind of freedom in
language use have been massive. And just as people usually don’t like variation of
any kind, they also don’t approve of change. Many linguistic changes are perceived
as decay, as a loss of standards, norms and values. According to Milroy and Milroy
(1999) this kind of decay is usually seen in tandem with a perceived (though maybe
nonfactual) decline of culture and tradition. For the complainers, speakers who
don’t know how to spell correctly anymore, who use verb-second order in German
subordinate clauses, who say “Peter his car” instead of “Peter’s car”, are also likely to
be sloppy in other aspects of their lives, such as punctuality, politeness, precision or
abstract thinking. So, for them, stopping linguistic variation and change and main-
taining language standards is a direct contribution to the maintenance of culture and
tradition. Needless to say, the job of linguists is not to be prescriptive in any sense;
i.e. we do not evaluate whether particular changes are good or bad. Rather, linguists
work descriptively and simply document whatever is happening in the language
without evaluating this from a “right or wrong” perspective. For linguists, language
is a natural (even if social) phenomenon, something that evolves and adapts and
can be studied objectively. But for many in the wider community, language is more
like an art form, something to be cherished and revered – and preserved at all costs.
The second type of complaint mentioned above can also be affected by linguistic
changes. In England from about the middle of the 16th to about the middle of the
17th century, we find a fierce debate about whether the English language should
incorporate more Latin and Greek terms such as eximious ‘excellent, distinguished,
eminent’ in order to increase its sophistication and expressive powers. Opponents
of these “inkhorn terms” as they were known (the inkhorn referred to the inkwell
made out of horn – the idea being that these words took up a lot of ink) argued that
the English language was doing fine and that these clumsy and opaque new terms
only led to confusion. Famous Tudor purist Sir John Cheke (1514–1557) claimed in
1557 that “our own tung should be written cleane and pure, unmixt and unmange-
led with borowing of other tunges; wherein if we take not heed by tiim, ever borow-
ing and never paying, she shall be fain to keep her house as bankrupt” (preface to
Sir Thomas Hoby’s translation of The Courtier). Hostile responses to linguistic exot-
ics continued even into the modern period, involving writers like Dickens and Ger-
ald Manley Hopkins right through to the modern day conlangers (or constructed
language users). All have sought to erase non-Germanic elements.
Obviously, much of what has been said in the previous paragraphs is based on
generalizations. At every point in time we also find the reverse, i.e. speakers who
embrace variation and change as chances for development and improvement. So, for
example, the inkhorn enthusiasts in the 17th century obviously asked for change,
and so do many people surfing the internet today. In many online communities it’s
a sign of wit, power and status to be able to play with language, to invent new words
or to coin new phrases. And whoever comes up with the wittiest new forms wins. In
these communities, maintenance and stability in the traditional sense are regarded
as boring, old-fashioned and conservative. The only constant here is change and
innovation. While this, at first sight, sounds very (post)modern and “internet gen-
eration”, we shouldn’t forget that similar phenomena can be found for almost any
other time as well. Shakespeare “invented” more than 1,500 new words or construc-
tions for the English language,7 and the slogan “Make it new” characterized the
work of many modernist writers such as Ezra Pound and Gertrude Stein. Hardly
conservative and culture preserving. So, while it still seems fair to say that on aver-
age many people do not particularly like variation and change, it is also important
to point out that there are other perspectives as well.
1.5 HOW THIS BOOK IS STRUCTURED
The following 10 chapters will introduce you to all central aspects of historical lin-
guistics, beginning with the lexicon and lexical changes. For most readers, these will
be the most familiar, obvious kinds of changes. After that, we will walk you through
changes in semantics and sound, word and sentence structure, respectively. Chap-
ter 7 will then focus on how changes actually spread in a language, i.e. how they
diffuse in the linguistic system across the different levels, but also how they spread
in the speech community, from individual to individual. Chapter 8 looks at lan-
guages in contact and how change develops in these contexts. Chapter 9 will then
go back in language history and present how we can establish relatedness between
languages, i.e. how languages can be arranged in family trees and how these trees
can develop over time. It introduces you to methods, problems and perspectives
in reconstructing the past of a language. We only have written evidence for lan-
guages no older than about 5,000 years, at best. But language itself is probably at
least 50,000 years old, maybe more. So what can we say about all the scarcely docu-
mented or even undocumented languages of our history? To conclude, Chapter 10
offers additional theoretical perspectives, i.e. insights from new developments in
historical linguistics (note that the discipline itself is one of the oldest in linguis-
tics!). In particular, we will focus on historical sociolinguistics, the sociology of lan-
guage, historical pragmatics and language evolution. Finally, we will discuss some
challenges and problems that may be new or particular for historical linguistics and
language change in the 21st century. Some of these problems concern the rate of
change, the media and the endangerment of language.
Every chapter in this book contains some breakout boxes, just like the present
one. These contain case studies or some extra, advanced material that challenges
the more daring readers. Moreover, every chapter concludes with a summary of the
most important terms and ideas, exercises and some further reading. We have tried
to keep the additional references in the book to a minimum in order to increase
readability. Needless to say, for every one of the topics discussed here you will find
shelves and shelves full of studies. In the further reading section, we will try to
introduce you to what we find the most relevant and/or accessible pieces of research
that will guide you deeper into the heart of the matter. Needless to say, these sug-
gestions are highly selective and subjective, and we can only encourage you to find
further readings yourself.
SUMMARY
This chapter introduced the two concepts of language variation and change. We
established that change in language is something natural, constant and unavoid-
able. We saw that change happens at all levels of linguistic structure, i.e. in phonet-
ics and phonology, morphology, syntax, semantics and pragmatics. It is crucially
intertwined with language variation; i.e. most change goes back to language varia-
tion, but not all variation necessarily leads to change. This is why evidence for lin-
guistic change can be hard to find. We introduced the two notions of apparent time
(when we take younger generations to be the seeds of future change) and real time
(when we study language at two or more points in time and compare the results,
thus capturing linguistic change).
In the discussion of linguistic variation, it became clear that language variation
in turn is something natural. It goes back to language acquisition, contact, play,
economy and other cognitive factors, such as analogy. However, we also pointed
out that despite their ubiquity, language variation and change are often the object
of public criticism. Many people (though not all) actually prefer black and white
over shades of grey, and change is often associated with decay rather than simple
development or even improvement.
FURTHER READING
There are a number of excellent textbooks on language variation and change that
introduce their readers to the key ideas and principles. Some of the gold standards
are still Lyle Campbell’s Historical Linguistics: An Introduction (1st ed. 1998, now in
its 3rd ed. 2013), Larry Trask’s Historical Linguistics (3rd ed. 2015, edited by Rob-
ert McColl Millar) and Terry Crowley’s An Introduction to Historical Linguistics
(4th ed. 2010, with Claire Bowern). This last book is particularly interesting, since
it presents not only a very hands-on approach but also tons of data from more
“exotic” (mostly Austronesian) languages.
A classic article is Weinreich et al. (1968). This paper is very long, with more than
100 pages, but it perfectly summarizes scholarship before the 1960s and it outlines
the research program for modern studies in linguistic change. Despite its veritable
age, many ideas presented here are still very topical.
Milroy and Milroy (1999) as well as Watts (2011) are very readable key texts for
all those who are interested in attitudes towards linguistic change, in the complaint
tradition, and the mythology associated with the development of national stan-
dard languages. Articles in the journal English Today 102, Vol. 26: 2 (June 2010) are
devoted entirely to topics on and around prescription.
Keller (1994) offers a very philosophical, non-technical and inspiring new
account of the whys and hows of language change. Drawing on the metaphor used
originally in economics, he introduces the notion of the “invisible hand”, which can
explain why changes in different individual speakers can actually gain momentum
and move in tandem in the same direction.
EXERCISES
1 Examples of language change

Identify and describe at least one linguistic change for every level discussed above
(phonology, morphology, syntax, semantics and pragmatics). You can do this for
English (past and present), but if you want to increase the challenge, every example
should come from a different language, so that in the end you will have discussed at
least five different languages.
2 Features of pidgins and creoles
Check out the Atlas of Pidgin and Creole Language Structures, APiCS: http://apics-
online.info
a How many pidgins and creoles have English as their lexifier (i.e. as the language
that contributes the vocabulary to the mix)?
b One of the English lexifier creoles has optional OSV (Object-Subject-Verb) word
order: which one?
c How many of the English lexifier creoles have modifying adjectives following
their nouns? Is this the exclusive order? Give one example from APiCS data.
3 Age grading
Identify 5 teenage words or phrases each from the 1970s, 1980s, 1990s, and today
(that’s 20 words in total). Have any of them survived? Have they made it into the OED?
Look at the current ones: do you think they have a chance to survive? Why or why not?
4 Attitudes to language change
Printer William Caxton, in his preface to the Eneydos (1490), describes how he
came upon an old book in his bookshelf. It caused him to reflect upon changes that
have occurred in English.
I [. . .] toke an olde boke and redde therin / and certaynly the englysshe
was so rude and brood that I coude not wele vnderstande it [. . .] And
certaynly it was wreton in suche wyse that it was more lyke to dutche [=
German] than englysshe I coude not reduce ne brynge it to be vnder-
stonden / And certaynly our language now vsed varyeth ferre from that
whiche was vsed and spoken whan I was borne / For we englysshe men /
ben [= are] borne vnder the domynacyon of the mone [= moon], which
is neuer stedfaste / but euer wauerynge / wexynge [growing] one season /

and waneth & dyscreaseth another season.
What is so striking about Caxton’s attitude to language change, especially compared

to those described in this chapter?
5 Research essay: the relevance of dictionaries in the 21st century
Write about the place of print dictionaries in modern times. Is it possible for these
publications to keep up with the changes in vocabulary and the speed at which
these changes are occurring? As part of your discussion consider the web based
Urban Dictionary (http://www.urbandictionary.com), and describe its approach.
What do you think might be the value of a dictionary that is regulated by volunteer
editors and rated by site visitors? We suggest you go to http://www.ted.com/index.
php/speakers/view/id/143 and also http://tedxtalks.ted.com/video/The-Power-of-
the-Dictionary-Sus to hear a lexicographer’s view.
NOTES
1 The text comes from the 11th century; the manuscript here is British Library Cotton
Tiberius B. iv. There are some letters here you won’t recognize: the consonant symbol
<þ> (called “thorn”) and <ð> (called “eth”), which were later replaced by <th>; <ƿ>
(called “wynn”), pronounced as [w]; <ʒ> (called “yogh”), pronounced as [j], [ç] or [x];
and the vowel symbol <æ> (called “ash”), which was pronounced like the vowel in ash.
The symbol 7 stands for ‘and’. Take a look at the original manuscript at http://www.bl.uk/
learning/timeline/item126532.html
2 Note that English does not belong to the group of Romance languages, of course, and
that it is only listed here for clarification. You probably wonder why the similarities
between the Romance languages listed here and English are so great. This is due to
the fact that English borrowed many words from Latin and French in the Middle
Ages.
3 The funny symbols used here (and in the following chapters) are phonetic transcriptions
based on the International Phonetic Alphabet (IPA), available at https://www.interna
tionalphoneticassociation.org/content/ipa-chart. The squiggly line above the vowel, for
example, shows you that the vowel is nasalized.
4 Unfortunately, the whole story is a bit more complicated. Ringe and Eska (2013: 136–
141) talk about this as a sound change that “has fragmented into several rather different
rules” (137). The story itself is very illuminating, and it tells you a lot about the intricate
details of sound changes generally. However, it might not be for the faint at heart because
of its complexity. Ringe and Eska offer a very good and readable account, so we can only
invite you to muster up courage and have a go at it.
5 The OED lists both a Russian and a Yiddish origin for the suffix, and some borrowings
from the 19th and early 20th century (e.g. alrightnik). However, it is still fair to say that
the Sputnik mission helped a lot in spreading the word.
6 The story is not that simple, unfortunately. While sociolinguistics has shown that “cool”
language use in male speakers may indeed be attractive for female hearers, this might
only apply to male speakers, at least in certain communities. “Cool language use” in
women might actually be a signal of power and strength, and may thus be unattractive
for (traditional) male members of society. Give it a think.
7 Well, strictly speaking, he probably did not personally invent them. But they are recorded
in Shakespeare’s plays for the first time – which also counts for something.
2
Changes to the lexicon
INTRODUCTION
Words are the most observable part of any language, and people are generally
fascinated by curious facts having to do with the ins and outs of the lexicon (or
the vocabulary) of their language. There are websites devoted to the “most irri-
tating words”, “favourite words”, “dead words”, “new words”, “peculiar words” and
“clichés” – and many other topics, all revealing people’s interest in vocabulary. There
is always considerable media attention when dictionaries announce their word of
the year. Articles flourish on the meaning of the winner, its origin, and even its wor-
thiness of the award. There is nowhere near the same excitement with other aspects
of the language; there were no breaking news stories when linguists announced
developments affecting the conjunction because (e.g. I’ve been missing out on sleep
because the “Breaking Bad” series or I missed the ending because I fell asleep). Dic-
tionary editors are almost the new celebrities, answering questions like: What is the
longest word in the language? Is there a word to describe someone who drinks their
own bathwater? How many words do speakers know? And perhaps the thorniest
question of all – when does a new expression enter the dictionary?
Dictionary making was much more straightforward for early lexicographers,
who sourced their new words almost exclusively from books. It was formal writ-
ten language that typically made it into dictionaries. The words were written on
cards each time a new instance of their usage was discovered, and when there was
a substantial collection of cards, it could be established that a word was in general
usage. So, these were largely respectable words, and anything else that managed to
sneak through would be well and truly branded (originally with symbols such as
the dagger (†), the double dagger (‡), the asterisk (*) and even the fleur-de-lis ( ),
and later with more precise usage labels such as “low”, “(im)proper”, “ludicrous”,
“barbarous” or “vulgar”, as appeared in Samuel Johnson’s 1755 dictionary). These
days, it is all very different. Lexicographers have to consider an array of different
text types, including newspapers, magazines, pamphlets, menus, memos, TV and
radio broadcasts and, of course, emails, chat-room discussions and blogs. And the
approach today is more democratic and more descriptive, even though, as works of
reference, dictionaries inevitably have an element of prescription (and dictionary
users may interpret descriptive usage symbols or labels as normative, turning lexi-
cographers – whether they like it or not – into censors).
The internet, particularly social networking platforms, makes it much easier for
dictionary makers to track a word and to test its currency, but it is also the trigger for
huge numbers of new words to be created and the reason that they are taken up so
Changes to the lexicon 29
quickly. Just like funny videos, celebrity gossip and other internet “memes”, within
a matter of hours they have worldwide visibility. There are hundreds of neologisms
specific to the internet – these include dignified specialist terms such as software,
network, and interface as well as slang such as twitterholic, twaddiction, celebritweet/
twit and twitterati – to give a few of the “tweologisms” that Twitter has spawned.
People love to play with language, and when communicating electronically they
have free rein (with an alleged average of 500 million tweets each day, Twitter has
considerable capacity not only to spawn new expressions but to spread them, as do
Facebook, Instagram and other social networking platforms).
So, how do we create new words? Rarely are they created from scratch, and it is
hard to find examples of true coinages (new expressions that haven’t been built on
pre-existing elements). The technical word quark is often cited as an example of a
coinage. It first appeared in James Joyce’s book Finnegans Wake and was later taken
over by physicists to describe some sort of elementary particle of matter. It’s not
based on quark ‘the call of a gull’ and doesn’t seem to have any links with any previ-
ous existing word. Such examples are rare. Usually they involve proper names and
so are peripheral to the language. But often there are lexical associations lurking in
the background of even these coinages. Take Kodak. George Eastman, who came
up with the name and the product, claims it is not based on any other word – the
inspiration was simply his love of <k>, which he described as “a strong, incisive sort
of letter”. However, Eastman’s biographer, reports that this fondness probably came
from his mother’s name Kilbourn, so the two appearances of this letter in Kodak are
not entirely accidental.
This chapter is concerned with etymology, the area of study that examines the
history of the forms and meanings of words such as Kodak (not to be confused with
entomology, the branch of zoology that studies insects). While Chapter 3 addresses
aspects to do with the meaning of words, here we cover the major methods lan-
guages have of creating expressions. Though our examples come largely from Eng-
lish, we emphasize here that these processes are found in languages around the
world, although there are usually differences in significance and liveliness when it
comes to word creation in individual languages. We will also consider the other end
of the life cycle of words – their disappearance.
2.1 GAINING WORDS – LEXICAL ADDITION
These days dictionary updates comprise a hotch potch of sedate terms-of-art (phar-
macovigilance ‘the monitoring of medical drugs after they have been licensed for
use’ and in silico ‘(of scientific research) conducted by means of computer simula-
tion’), boisterous slang (amazeballs ‘extremely good’ and FML ‘fuck my life’) and
lexical frippery (spit take ‘an act of spitting out liquid while being drunk in reaction
to something funny or surprising’ and douchebaggery ‘obnoxious or contemptible
behaviour’). Some entries appear so newly minted you might wonder at the wis-
dom of the editors in including them at all (adorbs ‘adorable’ and ship ‘to endorse a
romantic relationship’). But we know that these expressions have been scrutinized
within an inch of their lives – they wouldn’t be there unless they “had legs” (to quote
John Simpson, former editor of the Oxford English Dictionary (OED)). Small won-
der people are so fascinated by words.
People love creating new ones. In fact, a recent team of scientists has discovered
that learning the meaning of new words can stimulate exactly those same pleasure
circuits in our brain as sex, gambling, drugs and eating (the pleasure associated
region called the ventral striatum). Neologasm says it all – defined by the Urban
Dictionary as ‘the intensely pleasurable sensations generated by using, hearing or
coining a new word or phrase (that doesn’t suck)’. Leximania starts in childhood
and stays with us as we grow up. The majority of these creations are one-offs, spur-
of-the-moment and short-lived. But many people end up sending their inventions
off to dictionary editors, in the hopes they might make it onto their lists. However,
for this to happen there needs to be some indication of general usage. Telecrastina-
tion ‘the act of always letting the phone ring at least twice before you pick it up, even
when you’re only six inches away’, glarpo ‘the juncture between ear and skull where
pencils and pens are stored’ and sloovers ‘the remnants of soap too small to use, but
too big to throw away’, like all of comedian Rich Hall’s creations, fill a need, but they
haven’t yet made it – they remain sniglets ‘words that should be in the dictionary,
but aren’t’ (another of Rich Hall’s creations).
Hothouse words and mountweazels

Dictionary makers have also been guilty of leximania. “Hothouse words” are
ghost words that have never really existed but are the creations of lexicographers.
Many hothouse sprouts were cultivated in early English dictionaries, especially
during the 17th century. How tempting for compilers to invent words, plop
them into the dictionary and watch them grow. Perhaps they felt such erudite
sounding words as anatiferous ‘producing ducks’ and decacuminated ‘having
the top cut off ’ should exist, or perhaps they wanted to show off their knowl-
edge of Greek and Latin. No doubt some were also planted to catch plagiarists.
It was commonplace for lexicographers to pilfer entries from other works. In
fact, Jonathon Green describes plagiarism as a “lexicographical necessity” – it
wouldn’t make sense to start a dictionary from scratch. And once in the protec-
tive environment of a dictionary, these creations often survive. Authority after
authority repeats the erroneous word until no one would dare doubt its prov-
enance. Quite simply, it must be a word because it’s in the dictionary.
Hothouse words are a type of “mountweazel”, the general term for fictitious
entries in reference works, including encyclopaedias and even maps. The
name comes from a famous bogus entry in the New Columbia Encyclopedia in
1975 for Lillian Virginia Mountweazel (an American fountain designer and
photographer who accidentally blew herself up while researching an article
for the magazine Combustibles). In the words of one of the encyclopaedia
editors, “[i]t was an old tradition in encyclopedias to put in a fake entry to

protect your copyright [. . .] If someone copied Lillian, then we’d know they’d
stolen from us” (New Yorker, 29 August 2005). From then on, mountweazel
came to describe any sort of ‘ghost or false entry’. And it seems that mod-
ern dictionaries still occasionally use a mountweazel to flush out cheats. One
famous made-up word, esquivalience, appeared in the 2001 edition of the
New Oxford American Dictionary. The word was (appropriately) glossed: ‘the
willful avoidance of one’s official responsibilities [. . .] late 19th cent.: perhaps
from French esquiver, “dodge, slink away.” ’
In 1991 John Algeo completed a study of new words over a 50-year period (1941–
1991), sourcing his neologisms from the collection “Among the New Words” that
appears each year in American Speech, the journal of the American Dialect Society.
For each of the major types of word formation processes, he found the following
percentage of new words:
Type Percentage
Compounding 40
Affixation 28
Shifting 17
Shortening 8
Blending 5
Borrowing 2
Creating from scratch below 0.5
This study shows affixation and compounding as the major sources for the new
words, and blending and borrowing to be insignificant. As we discuss these pro-
cesses, consider whether you think Algeo’s breakdown still holds true. (And in the
exercises we suggest you replicate the study and find out.)
2.1.1 Compounding
Compounding is a word formation process that is found extensively in languages
around the world; indeed, Laurie Bauer (1983) claims there is no known language
that does not have compounds. The process involves the combination of two (occa-
sionally more) free-standing forms; for example hot dog, dog-collar and neckbeard. In
some languages this closeness is easy to spot because compounds are written as one
word. But as these examples show, English compounds often appear as two words
(with or without a hyphen). But you can still hear the “one-wordedness” because of
the main stress on the first element. It is easier to illustrate this with an example that
can occur both as a compound and as an ordinary string of two free words.
Take hot and dog. We can combine them and put the stress on the first element
only (we use underlining to indicate stress here) as in hot dog. In this case it has the
quite specific meaning of a ‘frankfurter served in a long soft bread roll’. On the other
hand, if we put stress on both parts as in hot dog, then it refers to any old canine
quadruped that just happens to be hot. This example also illustrates a second prop-
erty of compounds; the meaning is often more specific than just the sum of the two
parts, and it may often also become figurative. A dog-collar, for instance, is not just
the kind of collar that a dog would wear; it can also refer to the white collar worn by
some ordained clergy. Some compounds have highly idiosyncratic meanings – bag
lady, spaghetti western and black sheep. Again their unified meanings show that we
are thinking of these expressions as single units.
Often it’s the newer compounds, like air punch ‘act of thrusting a fist up into the
air’ and side-eye ‘a side-long glance of disapproval’, that appear as separate words or
hyphenated; well-aged compounds, like breakfast and cupboard, usually appear as
single words. Many now write hotdog solid. While there are some general guidelines
for how to write compounds, it remains a tricky (and often disputed) aspect of
English. Neckbeard ‘growth of hair on a man’s neck’ is a recent word, but it is writ-
ten solid. The semi-soft frozen dessert icecream has been around since the 1600s,
but the parts are still normally separated using spaces or hyphens, as ice-cream or
ice cream. Other languages are more consistent. In the German writing system the
parts of the compound are joined to form one word (sometimes even with a con-
necting element between the parts, as in Verbesserungsvorschlag ‘suggestion for
improvement’ from Verbesserung ‘improvement’ + s + Vorschlag ‘suggestion’).
The “one-wordedness” of compounds also becomes obvious when we look at the
way they behave in the grammar. For example, the compound still-life has a nor-
mal plural form (still-lifes) compared to the irregular plural of life (lives). Similarly,
people would say jack-in-the-boxes rather than jacks-in-the-box, which means they
are thinking of jack-in-the-box as a fused word. (Note how the heavy stress falls on
the first syllable and the following parts are all squashed together: Jack-in-the-box.)
Compounding has always been a major word formation process in Germanic;
1,000 years ago it was one of the most important sources of new lexical items
for English, and there are some wonderful examples, especially in poetry (bān
‘bone’ + hūs ‘house’ = ‘skeleton’, brēost ‘breast’ + cofa ‘cove, chamber’ = ‘heart, affec-
tions’). New creations are still appearing: seagull manager ‘a manager who flies in,
makes a lot of noise, craps on everything and then leaves’ (Urban Dictionary); hot
mess ‘when someone’s thoughts or their looks are in a state of disarray, but they
maintain an undeniable attractiveness’; humblebrag ‘a supposedly modest or self-
deprecating/critical statement but the actual purpose is to boast’.
The longest German word – a record set to be broken

The opportunity for compounding in German makes for some very long
words, sometimes known as Bandwurmwörter ‘tapeworm words’ (e.g. the
kitchen device known as the Eierschalensollbruchstellenverursacher lit.
‘egg-shell-necessity-break-spot-causer’). Indeed, in his 1880 essay “The
Awful German Language”, Mark Twain describes compounding as “one of

the most notable features” of German; he concludes that “[s]ome German
words are so long that they have a perspective”. One with a particularly
impressive perspective is Germany’s officially longest word: Rindfleischetik-
ettierungsüberwachungsaufgabenübertragungsgesetz (lit. ‘cow-meat-labelling-
monitoring-responsibility-delegation-law’). The word died when the law was
repealed after the European Union lifted a recommendation to carry out tests
on healthy cattle for BSE (bovine spongiform encephalopathy, the technical
name for ‘mad cow disease’, and itself not a bad compound). In the Telegraph
report of the news piece (27 July 2013), the journalist observed: “In theory, a
German word can be infinitely long. Unlike in English, an extra concept can
simply be added to the existing word indefinitely.”
He’s correct about there being no theoretical limit to the length of German
compounds. Eierschalensollbruchstellenverursachergerät shows Gerät ‘device’
tacked on the end, and we can keep doing this. But what about English? Cer-
tainly some varieties of English produce extremely complex clumps of words;
chemical compounds and nasty lung diseases (e.g. pneumonoultramicroscop-
icsilicovolcanokoniosis ‘a lung disease caused by inhaling very fine dust’), as
well as compounds found in super-literate varieties like Bureaucratese (e.g.
prototype crisis shelter development plans, occupational choice-vocational
interest congruency). And in English, as in German, there is also no actual
limit to the length of these “alphabetical processions” (Twain’s description).
English has anti-missile missiles and even anti-ballistic missile defence coun-
termeasures ‘strategic actions to evade anti-ballistic missiles’, but when a mis-
sile is deployed against anti-missile missiles, we have anti-anti-missile-missile
missiles. And we can keep playing this game (as comics have shown), as long
as we have one more missile than anti- (otherwise, there wouldn’t be anything
to blow up).
Theoretically, words and parts of words can combine and recombine in this
way to form an infinite number of different words. German and English here
illustrate the structural complexity and creativity that distinguish our com-
municative behaviour from that of animals. The infinite capacity to express
and understand meaning is probably not found in the language of any other
species (see Pinker 2000).
2.1.2 Affixation
Affixation is a similar process to compounding except that it involves parts of words
that can’t stand alone. They include prefixes that are added at the beginning (un-,
mis- and re- as in unhappy, misfortune and reapply), suffixes that are added at the
end (-ish, -ness and -ic as in blackish, happiness and linguistic) and (though rare)
infixes that occur somewhere in the middle of the stem (Homer Simpson’s -ma-
infix in words such as edumacation, sophistimacated and viomalin – the inspiration
for “Homeric infixation” probably comes from expressions like whatchamacallit

and thingamabob).
Affixes come and go. English has a long history of having fun with infixes and a
very early example is -de- (originally mock French) in flibberdegibbet ‘evil spirit’,
grizzledemundy ‘grinning stupid person’ and slubberdegullion ‘miserable worthless
wretch’. But these words are now found only in literature, and the infix is no longer
useful. In comparison, some other affixes become highly productive and achieve
almost voguish popularity. The prefix Mc- has now severed all ties from the fast-
food giant McDonalds to give us new (disparaging) expressions such as McJob
‘unstimulating, low paid jobs with few prospects’ (the first McCoinage), McBesity
‘fatness relating to the consumption of junk food’, McMansions ‘housing develop-
ments made up of large, cookie-cutter houses’, Baldy McMullet ‘mullet worn with
very little or no hair on the front and sides’, McGarbage ‘waste created by dispos-
able dishes, cutlery and packaging’, McBooks ‘quickly produced mass-market books’,
McMovies, McShopping, McMedicine, McPrisons and even McGod ‘the god of TV
evangelists’ (not all will survive, but the flourishing of such creations shows the suc-
cess of the affix). Another new prefix, uber- ‘super, mega’ (from German), is churn-
ing out neologisms – ubergeek, ubermodel, uberfan and uberdork are just some.
Some affixes have died out completely and survive only in relic form; and-
‘against’ is fossilized in a word like answer. Others may no longer be productive but
survive intact in words in common usage; -th was once used to form abstract nouns
like stealth, filth, wealth and truth. Still others appear to move in and out of fashion;
the old suffix -dom was raised from the dead in the last century, and new forms are
being created all the time: riddledom, filmdom, moviedom, blokedom, hackerdom,
professordom, parentdom, stuffed-shirtdom and lawnmowerdom are just a few of the
around 300 (currently appearing in lists such as Wiktionary). Some of these may be
one-off creations (or “nonce” words), but they attest to the liveliness of the suffix.
2.1.3 Backformation
Backformation is the opposite strategy to affixation. It can happen that words exist
with prefixes and suffixes, but not without them – so speakers backform them (back-
form has been created in this way from backformation). Empath ‘person or being
with the paranormal ability to perceive the feelings of another’ is a recent creation
from empathetic. In two-player games (such as chess or backgammon), the word ply
(from reply) is sometimes used for a turn that is taken by one of the players. This
process has given English many of its standard words, especially verbs like burgle,
shoplift, babysit, edit, afflict, enthuse, laze, aggress, grovel, televise, manhandle, eaves-
drop, househunt and jell.
In some of these cases, speakers have removed something they believe to be an
affix (but it isn’t) to create the new word. A fairly recent example is the verb to verse
(as in England is versing Australia), where speakers have reanalysed the word versus
as the verb form verses. So they have removed what looks to be a verb ending to
create a new verb. Such backformations are more likely to occur with very strongly
entrenched patterns, and they also have the effect of filling an apparent void – if
there is a noun butler then there should be a verb to butle (in fact there once was,
but it didn’t survive). In the same way, the word to beg was created from beggar, the
final -ar wrongly interpreted as the same -er suffix as in bake, baker (the word beg-
gar comes from Beghard, a member of a medieval Christian brotherhood).
2.1.4 Conversion
Conversion (or shifting, as it is labelled in Algeo’s chart) changes one part of speech
to another without anything being added; for example the new verb to toilet-paper
‘to cover (a building, trees etc.) with toilet paper’. In English, conversion usually
involves the major word classes: nouns, verbs, adjectives and adverbs. The most
common conversions are:
verb —> noun (a guess; a call; a think; a read; an ask);

noun —> verb (to bottle; to bridge; to trash; to network; to leaflet);
adjective —> verb (to better; to empty; to open; to total [a car]);
adjective —> noun (a roast; a weekly; a regular; a given; a nasty).
Conversion is extremely productive in languages like English because the basic

form of words is identical and so they can move with ease between different classes.
There are even examples of conversion from minor word classes like prepositions
and conjunctions. The prepositions up and down can transform into verbs (to up
the stakes / to down tools), nouns (on the up / to have a down on someone), adjectives
(he’s up today / a down experience) and adverbs (to go up and down).
Conversion is more difficult in languages whose words carry grammatical affixes.
Though English and German are close relatives, German conversion is restricted on
account of the fact that words have retained more inflections. The most common
conversions involve creating nouns from basic verbs; for example the noun das
Sprechen ‘the act of speaking’ from the verb sprechen ‘to speak’. Otherwise, affixa-
tion is the more common process; e.g. der Sprecher ‘the speaker’ and der Sprechende
‘the man who is speaking’.
2.1.5 Abbreviation
Abbreviation covers truncated forms or clipped words. The Oxford Dictionaries
“Word of the Year 2014” was to vape (from vapour or vapourize); it describes the
action of inhaling and exhaling the vapour produced by e-cigarettes (or e-cigs).
Sometimes shortened forms end up replacing the longer versions (e.g. mob for
mobile vulgus ‘movable or fickle common people’, cab for cabriolet and bus for omni-
bus ‘motor vehicle for paying passengers’). But often the longer and shorter forms
coexist as stylistic variants (vax alongside vaccine or vaccination), and with time
the meanings can diverge, as has happened with (mobile) app versus (computer)
application; old examples include hussy versus housewife, stroppy versus obstreper-
ous, grotty versus grotesque. The use of the shorter colloquial forms in casual con-
texts often leads them down separate semantic tracks, thus creating two radically
different words as the long and the short part company. Special cases of shortenings
arise when initial unstressed syllables are lost (cos from because), and this can also
give rise to separate words with very different meanings (fence – defence; cute –
acute; squire – esquire; ticket – etiquette).
Some shortened expressions involve other processes. Both adorbs (from ador-
able) and totes (from totally) show the addition of the diminutive (or hypocoristic)
-s ending. Like other hypocoristic endings, such as the -ie in breakie ‘breakfast’ and
-o in arvo ‘afternoon’ (the earmark of Antipodean varieties of English), the suf-
fix suggests informality and a friendly attitude. It shows an expressive use of the
original plural -s ending that first appeared in pet names (such as Legs, Susykins
and Cuddles) and nursery creations (such as dindins and beddie-byes). It now tags
many slangy creations like whatevs (from whatever), probs (< probably), awks (<
awkward) and fabs (< fabulous), and makes regular appearances in internet slangs
like LOLspeak (though often spelled <z>, as in muahz ‘kisses’).
2.1.6 Acronyms
Acronyms illustrate another kind of abbreviated expression, one that really took
off in the 20th century (see Chapter 1). This time, words are formed only from the
initials of other words – the word acronym comes from Greek acro ‘tip, point’ and
onym ‘name’. Dictionaries are incorporating new examples of these words all the
time: BOBO ‘burnt out but opulent’, YOLO ‘you only live once’, PAL ‘parents are
listening’, POS ‘parents over shoulder’. Technically, for something to be an acronym
the resulting word has to be pronounceable like other ordinary words in the lan-
guage. Examples that are pronounced as strings of letter names such as KPC ‘keep
parents clueless’, SMH ‘shaking my head’ and LMIRL for ‘let’s meet in real life’ are
not acronyms but rather initialisms (or alphabetisms).
Occasionally, acronyms are based on successive syllables of just single words, as in
the oldies TV from ‘television’ and PJs from ‘pyjamas’. Other variations incorporate
more of the words than simply the initial letters. This might involve, for instance,
using the first consonant and vowel, usually to make the acronym pronounceable,
as in the case the well-established acronym sonar from ‘sound navigation and rang-
ing’. Some acronyms use even larger chunks. Something like modem takes the first
syllable from the two words modulator and demodulator; hi-fi is something similar.
These words fall somewhere between being acronyms and blends (which are com-
ing up next). Hi-fi also smacks of a bit of wordplay (akin to reduplicated forms like
bow-wow, hob nob and nit wit).
Once they have been around for a while, acronyms lose their capital letters and
enter the language as ordinary words, such as sonar. Even where this hasn’t happened,
the original source words are usually forgotten – HIV stands for ‘human immuno-
deficiency virus’, yet people commonly refer to the HIV virus; compare ATM machine
‘automatic teller machine machine’ and PIN number ‘personal identification num-
ber number’. (Examples like PIN-Nummer in German have given rise to an acronym
to describe the practice: RAS-Syndrom (Redundantes-Akronym-Syndrom-Syndrom
‘redundant acronym syndrome syndrome’), a joke that works equally well in English.)
Backronyms
There has been an interesting twist in the formation of acronyms in recent
times. These expressions are backronyms (or reverse acronyms). WIMP and
MACHO, for example, are technical acronyms. WIMP stands for ‘weakly
interacting massive particle’ and MACHO for ‘massive astrophysical com-
pact halo object’. Of course, it could be a happy coincidence that the initials
in these phrases made for such apt and cute sounding words as WIMP and
MACHO, but more likely the creators of these acronyms fudged and fiddled
until they came up with the right sequences of words.
A lot of reverse acronymy goes on in the names of organizations and agen-
cies. A bunch of people might cook up a word that stands for something they
want their group to be associated with, let’s say, HOPE. Then, on the basis of
the letters that make up this word, they concoct a plausible sounding string
of words that is also appropriate to their activities and concerns. In this case,
it might be ‘Health Opportunities for People Everywhere’. Some linguistic
cookery undoubtedly went on when the Microsoft Corporation announced a
new program called Windows DNA standing for ‘Windows Distributed inter-
Net Architecture’; there is no doubt they were deliberately cashing in on the
famous initialism DNA (in this case, ‘deoxyribonucleic acid’). People have
a lot of fun with this kind of reverse acronymy. Clearly the phrase A CYA
Operation ‘a cover your arse operation’ was a deliberate pun on the US Cen-
tral Intelligence Agency (the CIA).
2.1.7 Blending
This process refers to the creation of new words from the combination of two (or
occasionally more) existing words. The new portmanteau word then incorporates
meaningful characteristics from both. While blending was an insignificant process
in Algeo’s study, the process has taken off in recent years. Dozens of new portman-
teaux are entering English on a regular basis, and blends like the following abound
in lists of new words: listicle ‘an article on the internet presented in the form of a
numbered or bulleted list’, zonkey ‘a hybrid (literally) of a donkey and a zebra’ and
mansplain ‘(of a man) to explain something (usually to a woman) in a condescend-
ing way’.
The nature of the mixing process is also changing. Rather than combining splin-
ters of words, the newer blends tend to be more like vodkatini – a complete word
with part of another. It might be that the front is intact, as in vodkatini and shoefiti
(‘the practice of decorating overhead wires with shoes’). Many of the modern blends
involve some sort of overlap, as in guesstimate ‘to estimate by guessing’. American
economic and cultural expansion is now often described as cocacolonization. This
is a clever fusion of part of Coca-Cola and colonization. Very occasionally, the full
word appears in the middle of the blend. In the case of ambisextrous ‘bisexual’, the
word sex is squashed and blended inside ambidextrous. These words are somehow
catchier and more playful because they overrun each other. Occasionally, the over-
lap is total, as in sexploitation ‘sexual exploitation’ and bagonize ‘to wait anxiously
at the baggage carousel for luggage to arrive’, which retain both words intact. But
more usually only parts coincide; affluenza ‘the disease of being too rich’ combines
a bit of affluence and a bit of influenza (the <fluenz> part is common to both source
words, affluence and influenza). In this way the blend echoes more effectively their
individual meanings.
2.1.8 Commonization
In Chapter 3 we describe special cases of semantic broadening, where proper names
extend from a specific case and end up referring generally to the whole class of
items. In terms of word formation, this is called commonization.
Personal names frequently lose their capital letter and enter the general lexicon as
household words, as in the mountweazels ‘fictitious entries’ mentioned earlier. There
are the usual eponyms (< epi- ‘upon’ and onym ‘name’) such as cardigans and sand-
wiches from the Earls of Cardigan and Sandwich (in fact, English has amassed more
than 35,000 such expressions), and new ones are appearing all the time, though
they don’t always survive. When Lorena Bobbitt famously cut off her husband’s
penis in June 1993 the verb to bobbitt suddenly appeared in newspapers around the
world to describe this and similar events, and continues to appear (perhaps because
the alternative depenistrate is a little less memorable). Peter Gilliver, the associate
editor of the OED at the time, stated that “[i]f ‘bobbitt’ turns up in several different
stories over a period of time, this suggests it should go in the dictionary” (cited in
The Age, 7 February 1994, p. 11). The word hasn’t yet secured an entry but continues
to make regular appearances in the media.
There are also eponymous phrases that arise spontaneously in everyday language.
Someone in a fringed rhinestone studded suit might be described as doing an Elvis.
Most are short-lived, it is true, but their pervasiveness speaks to the value of names.
Ours is a culture that promotes personal names.
Brand names also provide a common source for new words. Google has been
“verbed”; to google even refers to any search, not simply a Google search, so it illus-
trates both conversion and commonization.1 Place names also enter the language as
ordinary words in this way. Jeans have their origin in the town of Genoa, where a
type of heavy cotton fabric (resembling denim) was once made; denim itself derives
from Nîmes, the name of a city in southern France (originally serge de Nîmes ‘serge
(cloth) of Nîmes’).
2.1.9 Reduplication
Reduplication is a repetition process where all or part of the stem of a word is reit-
erated, and the resulting form is a kind of compound. This is a peripheral process
in English compared to some other languages (and we give examples from Māori
below). Nonetheless, over the years it has produced some thousands of words, and
occasionally new ones appear (cray cray ‘really crazy’ is a recent addition). In Eng-
lish, there are basically three types of reduplicated compounds. One involves rep-
etition of the whole stem (goody-goody, hush-hush). Many of these are confined
to nursery language (choo-choo, gee-gee). A second type involves the repetition of
the rhyme. Sometimes both stems are existing words (brain-drain, tin-grin), but
usually only one, or sometimes none, of the elements is independently meaningful
(argy-bargy, artsy-fartsy) – it can happen that the words become obscure through
sound change, or else they might simply drop out of use (willy-nilly derives from
the expression will I nill I based on earlier verbs willen ‘want’ and nillen ‘not want’).
A third (now rather rare) type of reduplicative compound involves some sort of
modification of the stem vowel (mish-mash, flip-flop).
In many languages reduplication has emotional functions; the repetition is more
expressive than ordinary speech. We repeat things to beef them up (I’m ok, I’m
ok); this can be for emphasis, to get across a sense of conviction or urgency. But
it gets more interesting than this. An example found more usually in American
English (and inspired by Yiddish) is “schm-reduplication”. It produces expressions
like school-schmool and fancy-schmancy, where the second part of the phrase is
a nonsense word, beginning with the same <schm> cluster. It’s fully productive
(almost any word can be cloned in this way), and the meaning is dismissive, the
linguistic equivalent of a snort or a sniff; school-schmool means something like
‘school – who cares!’
English also has something called “contrastive focus reduplication” (Ghomeshi
et al. 2004). You might say, “You mean he’s GONE gone” (there’s a heavy stress on the
first instance of the repeated word, and it contrasts with the second mention). The
question asked here is whether the person had actually gone for good, as opposed
to just ducking out for a short while. The copied expression always points to the real
or true meaning of the item referred to, and the expression can usually be rephrased
using modifiers (instead of he’s GONE gone, you could say he’s really gone).
Other languages are better known for their use of reduplication. Sometimes it
relates more to the grammatical life of the language; in Māori, for example, some
nouns lengthen the first vowel to indicate plurality (e.g. waahine ‘women’ from
wahine ‘woman’). More usually, however, it concerns expansion in the lexicon. Par-
tial reduplication of verbs can either strengthen the meaning (paki ‘pat’ becomes
papaki ‘slap hard’ and pakipaki ‘applaud’; kimo ‘blink’ becomes kikimo ‘close eyes
firmly’ and kimokimo ‘blink repeatedly’) or encode reciprocity (piri ‘stick, cling’
becomes pipiri ‘cling together’; patu ‘strike’ becomes papatu ‘beat each other’) (see
Harlow 2007).
2.1.10 Borrowing
The examples we’ve looked at so far have illustrated language internal processes; in
other words, ways languages enrich their vocabulary by drawing on sources already
available to them. Another way is to look to external sources – to expand the lexi-
con via what is usually called “borrowing”, the process whereby a language takes
and incorporates some linguistic element from another language. As described in
Chapter 1, lexical borrowing is the most usual, although any part of the grammar
can be “on loan”. English is an enthusiastic borrower and has adopted material from
as many as 120 different languages.
German is one that has been making important contributions over many years.
Recent borrowings include Kummerspeck (lit. ‘grief bacon’), a word to describe the
excess weight gained from emotional overeating, and Fachidiot (lit. ‘subject idiot’)
‘someone who knows a lot about their special area but little else’. The newness of
these borrowings is reflected in the fact that they haven’t yet lost the capital letter
required by German. Compare them to golden oldies in the area of food and drink:
noodle (< Nudel), pretzel (< Brezel), muesli (< Swiss German Müsli), Delicatessen (<
Delikatessen), gummy bear (< Gummibär), schnapps, lager and of course the frank-
furter and hamburger (adjectives/nouns formed from the place names Frankfurt
and Hamburg).
Sometimes an idea is borrowed, but not the actual word. English stole the seman-
tics of the German Ohrwurm (lit. ‘earworm’) to describe that really annoying little
bit of music that rattles around inside a person’s head, sometimes for days. This
expression earworm retains the German idiom but converts it into English. This
kind of borrowing where the meaning of a phrase is borrowed and expressed using
existing words in the borrowing language is known as a calque (or loan translation).
An earlier example is the expression power politics, which is a calque on the German
word Machtpolitik, a compound of Macht ‘power, strength’ and Politik ‘politics’.
Most of these lexical aliens have been naturalized – they fit the English sound and
spelling system (e.g. no capital letter), and they have been integrated into the gram-
matical system (they take the English plural ending -s). Sometimes this assimilation
process changes the loanwords beyond recognition. Hamburger is a good illustra-
tion. The first part of the word happens to correspond to a type of meat in English,
and the word has been reinterpreted or “reanalysed” as a compound ham + burger
(even though there is no ham involved). This false analysis has spawned many new
compounds such as Aussie lamburger, eggburger, cheeseburger, chickenburger, steak-
burger and so on.
The fact that English has expanded way beyond its original mother tongue coun-
tries has triggered a burgeoning of diversity in the form of hybrids, dialects, nativ-
ized varieties, pidgins and creoles, all influenced by the many different environments
and languages it has come in contact with; this has opened up many more potential
borrowing sources (for example the local vernacular languages, which may or may
not be the first language of speakers of these contact Englishes). For example, main-
stream dictionaries now contain words from Philippine English (such as carnap
‘to steal a car’ and presidentiable ‘a person who is a likely or confirmed candidate
for president’), and also loanwords that have come via Spanish (estafa ‘fraud’) and
Tagalog (barkada ‘group of friends’).
To see more closely what aspects of words can be borrowed, consider some of the
borrowings, this time from English into German, shown in Table 2.1.
You can see that some of these borrowed words are used in ways that most Eng-
lish speakers wouldn’t even recognize; what the German call a mobile (or cell)
phone shows a very different use for the adjective handy ‘useful’.
Table 2.1
Examples of English borrowings into German
What has been borrowed Words Translation

Form and meaning crashen ‘to crash (of a computer)’
Form and meaning chatten ‘to chat (online)’
Form and (almost) meaning Peeling ‘facial/body scrub’
Form Smoking ‘dinner jacket’
Form Handy ‘cell phone’
Meaning (=calque) Liebe machen ‘make love’
Meaning (=calque) Seifenoper ‘soap opera’
2.1.11 Sound symbolism

Finally we consider the creation of words that come about via some sort of sound
symbolism. All of you can likely think of examples of onomatopoeic words; words
like cuckoo, whoosh, cheep and plop, which all somehow echo their senses. It can
also happen that speakers start to associate certain sounds or sound sequences with
particular meanings, and these then can be used in the formation of new words
when the patterns get extended. Such sound sequences are called phonesthemes.
They occur either at the beginning of words (e.g. tw-, gl-, fl-) or in their rhymes (e.g.
-ash, -ump, -itter). They symbolize a certain meaning, although it is often not easy
to pin down exactly what this meaning is. Consider words that rhyme with stodge:
splodge (‘a thick, heavy, or clumsy splotch’), podge (‘short, stout, thick-set’), wodge
(‘a bulky lumpy mass’). Words with the -odge rhyme denote something solid, bulky
or lumpy, and this has been the inspiration for new words such as stodge (the word
that denotes a particularly thick, heavy, starchy, lethally fattening type of English
food). It is a good example of how speakers manipulate language for their own ends.
We could also include here imitative expressions such as ouch; in other words, those
that convey the sort of involuntary noisy responses we make to express some sort of
emotion. They include automatic noises like kerchoo, burp, hack, hiccup and so on.
Such words start life being rather inadequate attempts to represent the actual sound
being made; ha-ha, hardy-har, yuk-yuk, haw-haw and te-he don’t sound much like
actual laughter. But over time, as these written forms become more and more famil-
iar, they are conventionalized as ordinary vocabulary items. So, when we encounter
ha-ha or haw-haw in a text, it seems a perfectly normal representation of laughter.2
Yackety-yak seems a perfectly good way to represent someone talking. These words
then start functioning as ordinary nouns and verbs. For example They ummed and
ahhed; She hemmed and hawed; He pooh-poohed my memo. Once these expressions
develop their own spelling pronunciations, this removes them even further from the
original noises they’re supposed to represent. Tut-tut and tsk-tsk don’t sound much
like the curious clicking sound we make to express impatience or disapproval, and
yet tsk-tsk now seems a reasonable way to represent this noise. What’s more, so com-
pletely do we come to accept these as somehow natural, they then become the actual
noises we make. We might even say tsk-tsk when we’re ticking someone off.
2.1.12 A final word on the processes

Chapter 5 will discuss two important processes in morphological change: reanaly-
sis (the reinterpretation of structure) and analogy (or attraction to structure). The
focus in that chapter will be on the grammatical existence of words. But since both
processes are major driving forces in the creation of new words, we also mention
them briefly here.
Speakers often reinterpret boundaries within words (= reanalysis) to give their
structure more meaning. This can create new morphemes that then get extended
to form new expressions (= analogy). We saw this already with the restructuring of
hamburger. In what is sometimes referred to as folk etymology (see Chapter 5), the
“correct” morpheme boundary is ignored and reassigned to where it makes most
sense. The word bikini derives from the name of a Pacific Ocean atoll. The first part
of the word happens to coincide with the prefix bi- in English meaning ‘two’, and
of course the swimsuit happens to be made of two pieces. This is all the encourage-
ment speakers need to start forming new creations like mankini, microkini, mono-
kini, seaweed kini, macramé kini, Chanel’s eye-patch kini and even the dental floss
kini (where the bikini bits are held together by something resembling the thread
used to remove food from teeth). Kinis of all shapes and sizes now abound in lin-
gerie catalogues, everything from the G-kini with minimal coverage to the more
generously proportioned high-cut kini or hi-kini. Whether or not these new kini cre-
ations will survive is not clear; most will probably disappear. However, they nicely
illustrate the kind of inventive restructuring and generalization that is so much a
part of speakers’ wordplay. And many creations do endure.
A clear example is helicopter, from Greek helic ‘spiral’ > and the root pter ‘wing’
(the -o- is a linking device). Of course English speakers wouldn’t know what to do
with a word beginning with <pt> so they make the cut after heli-, giving rise to the
shortened form copter, as well as compounds such as helipad, heliport ‘terminal
for helicopters’ and gyrocopter ‘light single-seater autogiro’, hoppicopter ‘backpack
helicopter’ and most recently hexacopter ‘unmanned helicopter having six rotors’.
Reformations of this kind often result from an original blending process; if
there are enough blended combinations, it allows the speakers to reinterpret and
to restructure in the way we’ve just seen, and this gives rise to new morphemes
(either words or affixes), which can then extend (via analogy) to become productive
new affixes. For instance the -holic/-aholic/-oholic ‘addict’ suffix describes one who
habitually consumes whatever the first part of the word denotes, as in chocoholic,
footballaholic, golfaholic, newsaholic, junkaholic, icecreamaholic, potatochipoholic,
beefaholic, bookaholic, sleepaholic and so on. The source for these new formations
was 18th century alcoholic ‘pertaining to alcohol’ (alcohol + ic). The sense shifted
in the 20th century to ‘person addicted to alcohol’, and in the 1970s this led to the
blend formation workaholic/workaholism. Blends like these then allowed for the
reinterpretation of -oholic as a suffix.
Many new suffixes have evolved in this way: -athon originally from blends with
marathon (e.g. swimathon, bikathon, telethon and showerathon, which denote
events showing endurance and usually for charity); -gate originally from blends
with Watergate (e.g. choppergate, nipplegate, dianagate and prisongate, which denote
some sort of scandal); most recently -(a)licious originally from blends with delicious
(e.g. babelicious, bootylicious, funalicious, partylicious and scrumptilicious, which
denote something or someone very attractive).
Particularly interesting from the perspective of analogy and reanalysis is the
process by which phrases become set in the language, and eventually become new
words; for example the nouns wannabe ‘poser, follower’ (from want to be) and dru-
ther ‘preference’ (from (I)’d rather), verbs to don (from do on) and doff (from do off ),
and even the creation of more grammatical words such as the conjunction because
(from the prepositional phrase by cause). These are called amalgamations (some-
times also lexicalizations, though this label covers other types of word formation).
We will be revisiting examples such as these both in Chapters 5 and 6.
These amalgamations illustrate the reduction that also takes place in well-established
compounds. Consider the condensed pronunciation of breakfast [brɛkfəst] (as in
‘breaking the fast’) and cupboard [kʌbəd] (‘board for cups’). Here the spelling has pre-
served the original compounds, but this is not always the case – only the real word
enthusiast will be aware that nostril began life as a compound (Old English nosþyrel
‘nose thirl [= hole]’). Frequency is a driving force here (and we will revisit this theme
in many parts of this book). More unusual compounds don’t show the same sort of
reduction. Compare the full pronunciation of infrequent words such as handspike and
handstroke with the more common handkerchief [hæŋkətʃif].
2.2 LOSING WORDS – LEXICAL MORTALITY
We’ve just seen the many ways in which languages can extend their lexicon, but
there are many ways they lose words too. Dictionary editors of course have to
be aware of the endangered words. They need to make decisions all the time as
to whether they classify a word as “archaic” or “obsolete” or even whether they’ll
bother to include it at all. It’s a difficult decision – words may no longer be relevant
for modern speakers; yet they are important for people reading texts of the past.
Influential works of literature act rather like artificial life support systems for words
that otherwise disappeared from people’s active lexicons, sometimes hundreds of
years ago. Here are just some of the main reasons that words will drop out of use.
2.2.1 Obsolescence
This is probably rather obvious – if objects, ideas and institutions no longer form
a part of the speakers’ mental world then they will be forgotten. In areas such as
food, lexical obsolescence is probably a matter of course. We no longer recognize
medieval words like pottage (porridge-like dish of vegetables and/or meat), mor-
trews, buknade (pottages), civet (stew), frumenty, losyns (porridge-like dishes),
rapey, doucetes (desserts) and letelorye (savoury custard). The tendency in those
times to macerate, smash into pulp and spice food beyond recognition makes few of
the dishes appealing to modern palates. Clothing shows a similar high turnover of
vocabulary. Battle fashions have changed, and we no longer require medieval terms
for armour like vambrace, rerebrace, crinet and peytral. (Of course such words can
remain useful for members of “living history” societies.)
Disappearing words often tell of societal change. Most of us have given up the
habit of interpreting omens by the appearance of entrails or the behaviour patterns of
birds, rendering words such as augury ‘divination’ and more specifically pyromancy
‘divination by fire’ and tyromancy ‘divining by the coagulation of cheese’ no longer ter-
ribly useful. The traditional vocabulary of sin and virtue provides a more immediately
relevant example. As Geoffrey Hughes (1989) describes, words such as honour, virtue,
temperance, modesty, chastity and virginity are by no means dead, but (driven by chang-
ing mores and attitudes) they no longer form part of people’s active moral lexicon.
2.2.2 “Verbicide”
Examples like these also illustrate another fact of lexical life – words wear out. There
are certain areas of our vocabulary, like terms of abuse, that are more prone to
weakening than others. It’s no longer effective to insult someone by calling that
person a slubberdegullion druggel or a fondling fop, a blockish grutnol or a grout-
head gnat-snapper. Mangy rascal, drowsy loiterer, flouting milksop, base loon, scoffing
scoundrel and ruffian rogue just don’t pack much punch anymore. Expressive words
will become insipid, and alternatives have to be found. More recent disappearances
include bounder, cad and rotter – even ratbag, rogue, rascal, scallywag and scoundrel
(once highly offensive) are rarely heard. We see this in many areas of vocabulary.
Speakers are always on the lookout for new, exciting ways to express themselves,
and inevitably many expressions just fall away.
2.2.3 Reduction
It seems that words have to have a certain amount of phonetic saliency if they are to
function as a useful part of the vocabulary (what we are saying here refers to lexical
words – grammatical words, such he, the and of, are by their nature short). As we
will see more vividly in Chapter 4, sound change is generally reductive. Severe muti-
lation can reduce a word to a fragment of its former self, and it then simply drops
by the wayside. Old English ǣ ‘law’, ēa ‘river’ and īeg ‘island’ didn’t survive (except
among Scrabble players who find knowledge of such words as ai ‘a three toed South
American sloth’ quite handy). Interestingly, īeg got a new lease of life when speakers
expanded the word to īgland, modern island (the <s> here was introduced in the
1500s because people connected it to isle).
2.2.4 Intolerable homonymy

It can happen that through change two originally distinct words come to have the
same form but different meanings (= homonyms). Old English hrūm ‘soot’ collided
with rūm ‘room’ after the [h] dropped off, and it subsequently dropped out of use.
One of the most famous examples comes from Gascon (a variety of Occitan spoken
in south-western France). It involves the Latin words gallus ‘rooster’ and cattus ‘cat’.
In some rural dialects sound change meant that these two words ended up merg-
ing. Imagine the disastrous potential in a farming context – picture the poor farmer
unable to distinguish whether a cat or a rooster had got into the hen house (see
Hock 1991: 298).
It can even happen that words with completely contradictory meanings collide in
this way. Typically, speakers then end up replacing one of the homonyms. For exam-
ple, in Old English there were originally two verbs lāettan ‘permit’ and lettan ‘stop,
hinder’. Sound change left these two verbs homophonous: let ‘to permit’ and let ‘to
stop’. The second let has now disappeared except for relics like without let or hindrance
and let ball (in tennis). The same thing happened with Old English cleofian ‘to stick
together’ (compare modern related forms glue and clay) and clēofan ‘to split apart’
(compare modern cleaver and cleft). Sound change left the one word cleave with either
the meaning ‘to stick together’ or ‘to cut in half ’. The latter sense now predominates.
For collisions to cause problems, the words must usually belong to the same
sphere of ideas and occur in similar contexts. German arm ‘poor’ and Arm ‘arm’ are
different parts of speech and have very different meanings; they are not likely to be
confused. But very different contexts of use do not always ensure survival. In the
case of taboo, homonyms of taboo terms will quickly disappear. The phonological
collapsing of arse and ass in some varieties of English, for example, caused consid-
erable problems for the animal now generally referred to as a donkey. The Early
English word coney/cunny, meaning ‘rabbit’ (and rhyming with honey), dropped
out of use when it collided with the tabooed female body part cunt. Words can even
disappear if they sound a little too much like taboo words. Many single syllable
words beginning with <f> and ending in <k> have disappeared from the English
language. During the Victorian era we lost feck, meaning ‘efficiency’; feckless must
have sounded different enough that it lingered a little longer. Typically, we will drop
words like hot cakes if they sound too much like expressions that are offensive or
embarrassing. Such is the power of taboo.
Some disappear for no good reason

We’ve examined some of the reasons words drop out but have to admit that
huge numbers of English words disappear for no obvious reason at all. Fnast
‘to snort’ bit the dust, as did all other <fn> words, such as fnese ‘to sneeze’.
True, some words seem spectacularly useless (hothouse peristeronic ‘sug-
gestive of pigeons’ and gymnologize ‘to dispute naked’), and we can perhaps
understand why these didn’t endure. Others seem handier, like supernaculum
‘the act of drinking the very last drop from a glass or bottle’ and velleity ‘a
sense of desire that doesn’t come with any action’.
Over the years many seemingly useful expressions have disappeared with-
out a trace: multiscious ‘knowing many different things’ and nod-crafty ‘with
an air of great wisdom’; once people could have gome ‘wit, tact’, ruth ‘compas-
sion’, ert ‘skill’ and list ‘joy’, they could also be wieldy ‘agile’. The symposiast
‘the banquet lover’, the wine-knight ‘one who drinks valiantly’ and the gas-
trophilanthropist ‘benevolent purveyor for the appetites of others’ might be
conjubilant ‘filled with good cheer’ and with vitativeness and felicificability
or ‘love of life’. English has lost many effective insults too: the buffleheaded
booby, the cuddy clotpoll and clodplate, the jobbernowl jolthead and the noddy
ninnyhammer.
It is a fact of lexical life that words will wear out, some faster than others
(insults will lose their wounding capacity, swearing its pungency, and slang its
vibrancy). But like in the fashion industry, people want to change their language
(especially vocabulary), just as they want to change the hemlines on the trou-
sers and dresses they wear. There is a constant tug of war in language between
people’s desire for new, exciting ways of saying things and the tendency for
words and structures to become routine; this is most obvious in vocabulary
changes, but as we’ll see later in Chapter 6 it also happens in grammar.
2.3 ETYMOLOGY – STUDY OF THE ORIGIN OF WORDS
Etymology is the study of the origin and history of words. It’s a subject that fasci-
nates most people – and it’s often full of surprises. Dictionary makers and diction-
ary users can have quite different ideas about how an expression has come into
being. Many speakers have stories about the history of certain words, and they’re
often shocked (even irritated) when they find no mention of these in the dictionary.
Lexicographers do rain on people’s picnics with their cautious labels “of uncertain
origin” or “etymology unknown”.
The expression OK (or okay) is one that has spawned an extraordinary array of
imaginative etymologies based on languages from all over the world. One even
derives it from a boxing term KO (or kayo), an abbreviation for ‘knock out’. If the
boxer wasn’t kayoed, then he was OK. A good story but there is no evidence, and
the chronology is wrong – kayo appeared more than 80 years after OK. This time
it seems the story that is best supported by documentary evidence and wins the
approval of the dictionary makers is one that bases it on an acronym. OK originates
from a jokey misspelling oll korrekt in the 1830s of the expression all correct. Com-
pare other phrases like oll wright or OW.
The majority of lexical creations don’t endure – so how do we explain those that
do? Clearly they have to fill a need, but dictionary makers track thousands of new
useful expressions every year and only a fraction survive. Why did the expression
OK take off and not OW? It turns out that the prosperous creations often have mon-
grel origins; in other words, a number of influences come together to establish the
meaning of the form and to secure its currency.
It seems that OK was also used for many other jokey abbreviations, including out of
kash, oll koming and oll konfirmed. But what really helped to popularize the expression
was the fact that it was adopted as an election slogan by supporters of the Democratic
candidate Martin Van Buren (1782–1862). Born in Kinderhook (New York State), he
was dubbed Old Kinderhook, and his supporters then formed the Old Kinderhook
Club or the OK Club to solicit money for campaigns. People’s elaborate etymologi-
cal tales probably also have a place in the history of this word. West African, Greek,
German, American Indian and French origins have been proposed for OK – even
Scots och aye has been suggested as a likely source. The stories surrounding people’s
favourite words might be phony, but if an expression captures the imagination in this
way, it probably has a much better chance of becoming established and surviving.
The term OK took off in the 20th century to become a truly international word.
Etymological fallacies can have unfortunate consequences. In June 2003 a British
Government minister was severely criticized for his use of the phrase nitty-gritty at
a police conference because of its supposed racist overtones. He had told his audi-
ence apparently that it was high time to “get down to the nitty-gritty” in training
officers. It seems that the expression nitty-gritty is prohibited in the British police
service lexicon because people believe it to have originally been used in reference
to those in the lowest reaches of slave ships. This etymology is false. There is noth-
ing linking nitty-gritty with the early slave trade. It appears in fact to have entered
English only sometime during the 1960s, probably via Black English and initially as
a bit of popular music slang.
Since the 1990s there have been similar controversies sparked by the use of the
word niggardly. In 1999 an employee in the Washington, DC mayoral office, David
Howard, told his staff that, in light of cutbacks, he would have to be “niggardly”
with funds. Many connected this word with the taboo word nigger, and the uproar
that followed resulted in Howard’s resignation. In 2002 Stephanie Bell, a fourth
grade teacher at Williams Elementary School (Wilmington, Indiana), taught the
word niggardly to her students. At least one parent wanted her fired.
These examples are popular etymologies that have no linguistic basis. However, in
issues to do with language, it often doesn’t matter what the linguistic facts suggest –
what really matters to speakers is how they perceive their language to be. The reality
that nitty-gritty and niggardly have absolutely no etymological connections with the
N-word is of no consequence, and if people do make these etymological connec-
tions, then this will be the kiss of death for these words. Fuk ‘sail’ and feck ‘purpose’
had absolutely nothing to do with the F-word either, but that didn’t save them. And
although country shows no sign of falling to the power of the C-word, coney [kʌni]
‘rabbit’ has disappeared.
SUMMARY
This chapter has been all about changes to words. It has examined the various meth-
ods by which people expand the lexicon by creating new expressions, word for-
mation processes such as affixation, compounding, acronymy, blending and folk
etymology, conversion, backformation, wordplay and borrowing. Words are also

disappearing and for all sorts of reasons – obsolescence (of ideas, objects, institu-
tions and so on), sound change, intolerable ambiguity and taboo. Most of the time,
speakers don’t realize a word is on the way out. Ironically, they only become aware
of this when for some reason the word makes a re-appearance.
The transmission of linguistic forms is sometimes reminiscent of the spread of
thought contagions (or memes) – fads that spread from person to person within a cul-
ture. Expressions are particularly infectious, and successful ones disseminate rapidly
through speech communities, especially virtual communities. Celebrity endorsement
can help to propel them to success. The verb to dog didn’t take off until Shakespeare
used it, and neither did to twerk until Miley Cyrus “twerked into history”. Perhaps
we will now see a spike in the use of conscious uncoupling since Gwyneth Paltrow
and Chris Martin announced their wilful unyoking on 25 March 2014. Clearly social
media helps to spread these linguistic cults and trends – within a matter of hours they
can have a worldwide following. The internet gives them a cachet, a respectability. But
the progression of change is much more exciting and revealing than the introduction
of a few new expressions for reasons of fashion. One of the most exciting areas in
historical linguistics now involves the research that looks at the social use of language
and its role in language change, and this is something we focus on in Chapter 7.
Vocabulary is the most unstable aspect of a language, and dictionary makers are
constantly having to redraw the admission and exclusion boundary for marginal
vocabulary items. Meh is a modern interjection to convey indifference; it was in
use online in the early 1990s, but it was almost 15 years before it appeared in main-
stream dictionaries. Yeah-no is about the same age, but it is still missing from most
dictionaries. It is almost impossible for printed dictionaries to keep up with the
protean nature of vocabulary these days – not just new expressions, but old expres-
sions changing their appearance, their meaning and their grammatical behaviour,
and this is all happening at breakneck speed. We live in interesting linguistic times.
FURTHER READING
Specific treatments of English words and word formation are offered by Bauer
(1983), Katamba (1994), Stockwell and Minkova (2001) and Carstairs-McCarthy
(2002); and cross-linguistic perspectives are offered by Haspelmath and Sims
(2010), and several new handbooks on word formation (e.g. Štekauer and Lieber
2006; Lieber and Štekauer 2011, 2014; Müller et al. 2015–16).
If you are interested in specifically slang and the creativity of the lexicon, some
recent studies include Adams (2009) and Coleman (2012); Kwon and Round (2015)
deals specifically with phonesthemes. For historical accounts of English words see
Hughes (1989) and Liberman (2005), and for an overview of etymology and the key
debates we recommend Mailhammer (2014).
Much has been written recently on the relevance of dictionaries in the 21st cen-
tury. We suggest you log on to www.ted.com/index.php/speakers/view/id/143
and also http://tedxtalks.ted.com/video/The-Power-of-the-Dictionary-Sus to hear

lexicographers Erin McKean and Susan Butler talk about the place of print dic-
tionaries in modern times. Historical linguist Anne Curzan also appears on
TED and talks about how and when words change: https://www.ted.com/talks/
anne_curzan_what_makes_a_word_real.
EXERCISES
1 English (or any other language you choose) – word formation processes
Many websites devoted to new words have recently appeared for English (the OED,
for example, updates four times a year and publishes lists of the new words each
year). There will also be dictionaries of new words in your library. Consult any of
these sources to come up with an example for each of the processes we’ve identi-
fied in this chapter for creating new words (examples could also come from your
own slang, but make sure you explain them). Note: Give the word together with the
process. You could choose examples from a language other than English; just make
sure you explain these examples so that the word formation processes are clear (to
a non-speaker).
You might also consider word formation processes found across languages and
how they differ in importance and liveliness. For example, take two languages (one
can be English) and compare the strategies by which new lexical items are formed
in each language. Outline any similarities or differences in word formation pro-
cesses. Give plenty of examples (make sure you explain these examples so that the
word formation processes are clear to non-speakers).
Another way to go about this exercise is to examine product names; for example,
go out and browse through the supermarket shelves and collect examples of prod-
uct names that illustrate each of the processes we have looked at in this chapter.
2 Attitudes to lexical change

In 1712 Jonathan Swift published what is his famous A Proposal for Correcting,
Improving and Ascertaining the English Tongue, of which the following is an extract.
The English Tongue is not arrived to such a Degree of Perfection, as, upon
that Account, to make us apprehend any Thoughts of its Decay: And if it
were once refined to a certain Standard, perhaps there might be Ways to
fix it for ever [. . .] I see no absolute Necessity why any Language should be
perpetually changing; for we find many Examples of the contrary [. . .] But
what I have most at Heart, is, that some Method should be thought on for
Ascertaining and fixing our Language for ever, after such Alterations are
made in it as shall be thought requisite [. . .] What Horace says of Words
going off, and perishing like Leaves, and new ones coming in their Place,
is a misfortune he laments, rather than a Thing he approves: But I cannot
see why this should be absolutely necessary.
Briefly (in around 300–350 words), describe Swift’s attitude to language change
expressed here. As part of this discussion, include the goal that Swift sets up. How
desirable is it? How practical?
3 Linguistic fossils
When linguistic changes occur they often leave behind some sort of trace – a relic
of the original set-up. Words don’t just disappear; there is typically something left
over. This is what prompted the Dutch linguist Van der Tuuk to once describe lan-
guage as “something of a ruin”. Modern languages are full of relic forms that we can
use to reconstruct features of vocabulary (sometimes even sounds and grammar)
from earlier times. The following words are historically compounds. Identify the
elements from which each word was formed.
(a) werewolf (b) cobweb (c) gossip (d) hatred (e) midriff (f) garlic
(g) lukewarm (h) daisy (i) window (j) tenterhooks
4 Backformation
a The following words are the result of backformation. Investigate the ety-
mology of these words, and identify whether the backformations involve a
derivational or inflectional ending.
(i) burial (ii) to burgle (iii) greed (iv) grovel (v) edit (vi) pea (vii) couth
(viii) dishevel (ix) liaise (x) cherry
b Investigate the etymology of the following four words. Describe why the
process of word formation that is common to all of them is really the
opposite of backformation. Note: Give just one general statement here that
applies to all four words.
(i) bodice (ii) chintz (iii) quince (iv) news
c The English word incident is on a similar path of development, evident
in the non-standard plural form incidentses. Why do you think this is
happening?
5 Research project: word formation processes in English

As we described in this chapter, John Algeo’s 1991 study showed affixation and com-
pounding way out ahead, and blending and borrowing to be insignificant sources
for new words in English. Your task is to determine whether this is still the case.
Take a random sample of around 100 expressions; you could use the same source
as Algeo (the “Among the New Words” publications), or you could select another
means of collecting neologisms. If you have access to the OED online through your
library, you will find on the homepage under “What’s New” links to lists of new
words for different years. (Be careful to focus on brand new expressions, and not the
new senses that established expressions have sprouted.) An even better source is the
Oxford Dictionaries Online (http://oxforddictionaries.com/), since this dictionary is
much faster to include the quirks and fads of the vernacular.
Your report should be in connected prose, not bullet points, and should be
approximately 1,000 words in length. A good idea is to provide an appendix to
include your findings (rather than have them in the body of the text).
NOTES
1 Interestingly, the Google corporation went through several lawsuits to protect its
name from becoming a verb (http://arstechnica.com/uncategorized/2006/07/7198-2/),
and finally established that even though google can be used as a verb, it is still not a
generic (and hence unprotected) term but remains a trademark: http://www.forbes.
com/sites/ericgoldman/2014/09/15/google-successfully-defends-its-most-valuable-asset-in-
court/#48cfd57a3f05
2 However, there is considerable variation in the representation of laughter across cultures
and languages; see http://forum.wordreference.com/threads/laughter-haha-hehe.419591/.
3
Changes to the semantics
INTRODUCTION
In Chapter 2 we saw that the lexicon is always changing – words disappear, while
new ones are being created (or borrowed) all the time. Meanings too are in a con-
stant state of flux. More than any other aspect of language, they are linked to the
life and culture of speakers, and this can take expressions on extraordinary jour-
neys. Attempts have been made to find patterns, even regular semantic principles
of change, but words and their meanings are strangely volatile beasts and many defy
classification.
Once it was possible to starve from overeating, and you could eat pears from
an apple tree. Lewd men could be models of chastity, girls could be boys, and so
could harlots and wenches. It might surprise you to learn that grammar and glam-
our are historically the same word, as are cretin and Christian. The word brave once
implied cowardice (an earlier meaning preserved in bravado). Enthusiasm was a
term of mild abuse, while crafty and cunning were terms of high praise (so were daft
and silly). Changes can be so striking that expressions can even come to mean the
opposite of what they once meant (and we’re not talking here about the deliberately
ironic use of words like wicked, vicious, sick, rancid, filthy and putrid to describe
things that are exceptionally good). The original meaning of fast ‘not moving at all,
fixed firmly’ (as in it stuck fast) sprouted the newer meaning ‘moving quickly’ (as in
she runs fast). Our current understanding of black probably goes back to an earlier
meaning ‘shining, white’ (making it historically related to words like bleach, blanch
and even bald). In Australia and New Zealand if you barrack for a team you give
them positive support and encouragement, but in the United Kingdom you verbally
abuse and jeer at them.
In their lifetime expressions can go through a number of different changes, which
can end up taking them far from their original sense. Witness the various stages
that an ordinary word like buxom has gone through (from Old English būgan ‘to
bend’): ‘flexible, compliant’ > ‘obedient, meek’ > ‘obliging, kind, affable’ > ‘attractive,
plump’ > ‘big breasted’. Many processes will interact to bring about such shifts, and
a number of forces we go on to discuss would have been at work here, including
euphemism and even sound symbolism (think of the number of related words that
begin with <b> – such as breasts, bosom, boobs, bouncers, bulbs, balloons, bazoomas
and so on).
But first we need to come up with an adequate understanding of meaning, and
this is not easy. As Geoffrey Leech on page 1 of his book Semantics (1981) points
out, “the word ‘meaning’ and its corresponding verb ‘to mean’ are among the most
Changes to the semantics 53
eminently discussable terms in the English language”. Entire books have been pub-
lished on the meanings of meaning. In this chapter we adopt a fairly simplistic view
whereby meaning is determined by the set of contexts in which a word occurs;
hence semantic change will be any change in that set of contexts. So how does this
occur?
Meaning shifts aren’t usually abrupt – they happen gradually, via polysemy (‘mul-
tiple meaning’) and meaning associations. As the simple word glass shows, different
meanings and different nuances of meaning can coexist for long periods of time,
with new ones always ready to hop on board.
(3.1) a It was made of glass [= hard brittle substance]

b She drank the wine from the glass [= drinking-vessel]
c He looked in the glass and didn’t like what he saw [= mirror, looking glass]
d The image was formed by the glass [= lens]
e Even without a glass we could make out the ship [= telescope, spy glass]
f He saw the glass was rising [= barometer]
g You holding any glass [= high-purity methamphetamine]
h That pipe is total glass [= surfing term for the inside of a perfect break]
Over many years Old English glæs (first attested in the 9th century) has been
sprouting new meanings and associations; we’ve only touched on them here. Last
century also saw a new slangy verbal use (to glass ‘to strike someone with a (bro-
ken) glass or bottle’), and more recently sci fi has hatched a new verb meaning ‘to
use high powered weaponry for mass destruction’ (as in planet glassing). Clearly,
much of our understanding of this word will be determined by context – the lin-
guistic, cultural and social settings all contributing to the intended interpretation.
Words are not like math symbols with a fixed and constant designation. At any
one time, they can hold a multitude of different meanings. Together with all the
associated baggage that arises from our personalities and prejudices, these slip and
slide around over time as language evolves and adapts.
“Tolerable friends”
The meanings we carry around in our heads seem to us so natural and inborn.
Yet, as Nick Enfield (2015) argues, since we aren’t telepathic and can never
know for sure what goes on inside other people’s heads, we can only hypothe-
size what they mean by the words they use. True, we refine these assumptions
along the way (not consciously of course) by being exposed to varied contexts
and uses, but our word meanings remain hypotheses.
It can happen that speakers in a community hold different ideas of what
words mean. Drawing on the idea of “false friends” in second language learn-
ing (expressions in two languages that resemble each other but are wrongly
assumed have the same meaning; for example German Fabrik means ‘factory’,
not ‘cloth’), Enfield describes words as “tolerable friends” at best; in other

words, speakers can have different semantic representations, but this mis-
match never becomes apparent and doesn’t cause problems. For example,
most people aren’t aware of the very different meanings that exist for the word
instep. Many dictionaries have only the early meaning ‘the upper surface of
the human foot between the toes and the ankle’ (Oxford English Dictionary
definition). Some (usually older) speakers have this meaning, but most now
understand instep as the underneath part of that section of the foot, and oth-
ers take it to be the entire middle section, the top and underneath part (one of
us recalls well a discussion some years back by the Macquarie Dictionary edi-
tors on exactly this point – and the thorny issue of when to include, or even
privilege, a new sense). As Enfield writes, less frequently used words (such as
instep) are more prone to this kind of variation, and because the contexts of
use aren’t varied, speakers aren’t given the same opportunity to revise their
hypotheses.
3.1 CONSEQUENCES OF SEMANTIC CHANGE
Semantic changes have generally been classified into various types. The classifi-
cation falls along two axes. First, there are three ways that words can change in
sense (dictionary meaning): meanings can broaden, narrow or shift. Second, there
are two ways that words can change in connotation (meaning association): they
can deteriorate or elevate. As you might expect from the contradictory nature of
these labels, the classifications have no actual explanatory value but are simply
convenient ways to describe such changes. Here we’ve given examples only from
English, but exercises at the end of the chapter give many more from a broad range
of languages.
3.1.1 Broadening
As the name suggests, words can expand their contexts and come to mean more
than they did before: Old French ariver ‘to land, bring to shore’ (< Latin ad ‘to’ +
rīpa ‘shore’) was the source of English arrive ‘reach’ and modern French arriver
‘come, reach, arrive’, and boucher (the word for the person who originally sold
goat’s meat) gave us the butcher. In the 1700s, grog referred to ‘diluted rum’, but it
extended to include ‘any spirits with water’ and from the 1800s to include ‘strong
drink in general’ (and in Australian English it has gone one step further to mean
‘any kind of alcohol’). Interestingly, German Grog (hot tea with rum) retained
something of its original meaning. Broadening commonly occurs when proper
names lose their capital letter and enter the general lexicon as ordinary words: in
Chapter 2 we described how the Earls of Cardigan and Sandwich were the inspira-
tion for cardigans and sandwiches.
3.1.2 Narrowing
More usually, the contexts of words reduce, and so they come to mean less than they
did before: a hound was ‘any sort of dog’, an apple ‘any fruit off a tree’, a girl ‘a young
person of either sex’ (compare youth, which is currently restricting its application
to ‘young male’), liquor ‘beverage of any kind’ (modern drink is heading in the same
direction). Borrowed words often undergo changes of this kind. In Japanese, sake
has a very broad meaning that includes ‘alcoholic beverage’ generally; in appropri-
ate contexts, however, it can be used to denote specifically rice wine, and in English
it has now narrowed to mean only that particular Japanese beverage.
3.1.3 Shift
Time can also totally alter the contexts in which words appear. The term grog men-
tioned earlier was originally the nickname Old Grog given to British Admiral Ver-
non on account of his grogram cloak (grogram being a type of cloth); in 1740 the
admiral ordered rum to be diluted with water to stop sailors intoxicating them-
selves, and the sailors dubbed the insipid beverage grog. Semantic shifts often open
these nice windows onto cultures and attitudes in the past. Another illustration (also
nautical) is the word noise, which has shifted its meaning from ‘sea-sickness’ to the
current-day ‘intrusive sound’. These days, we don’t tend to think of the sound of sea-
sickness, but as the wordsmith Eric Partridge (1961) pointed out in brilliant detail,
it must have been incredible. Here you have to conjure up the unfortunate sailors
in a tiny wooden vessel caught in the middle of a storm in the times of Ancient
Greece. Imagine the din – the ship creaking, the wind howling, the waves battering
the sides and, of course, the groans and moans of those poor wretches on board. The
racket and commotion would have been considerable. Now the transition from ‘sea-
sickness’ to ‘din, confusion’ and finally to ‘intrusive sound’ becomes quite plausible.
3.1.4 Changing values – amelioration and deterioration

All sorts of social and expressive information attach to meaning, and this is conno-
tation. Think of the different connotations surrounding the terms girl, woman, lady,
chick, broad, madam, tart, bird and sheila. All could potentially be used of the same
female person, but they have very different semantic nuances. Connotations arise
from our experiences, beliefs and prejudices about the contexts in which words and
expressions are used. Consequently, they vary between speech communities, even
individuals within speech communities, and they can also alter radically over time.
With time words can acquire both favourable connotations (= amelioration) and
pejorative connotations (= deterioration).
Amelioration can occur in two ways. First, unpleasant overtones can gradually
erode away. Intensifying expressions, like terribly, awfully, frightfully and horri-
bly, are particularly prone to this. These have lost their original strength and now
mean little more than ‘very’; witness “contradictory” usages like awfully good and
enormously small. Second, words can take on favourable overtones; they acquire a
kind of “semantic halo” (to use novelist C. S. Lewis’ term). For example, the current
egalitarian ideology has meant the elevation of terms like democracy and politician
that previously were quite negative.
But halos come and go (and sometimes they slip a little, as in the case of politician).
In fact, it is more usual for words to take on negative overtones rather than favour-
able ones. Perhaps we are all inherently pessimistic. Certainly, we’re far more likely to
look for the worst in things. We scold and disapprove far more than we applaud and
admire. Bad news is always more interesting than good news. All this is reflected in
the way words are so quick to take on negative meanings and associations. Weather
refers to the condition of the atmosphere, and it can be bad or good; yet is it often
used negatively (a broadcaster might report “We’re in for some weather”, meaning
‘adverse conditions’). You can see the same process at work with the cautionary label
that is slapped on television programs: May contain language (most programs do, of
course, but what’s understood here is bad language). The same process has shifted
the meaning of temper. It was originally neutral, referring simply to an emotional
condition or predisposition, but these days it has narrowed to mean a state of anger;
in other words, temper has come to mean ‘ill temper’. Economics has Gresham’s Law:
“Bad money drives out good”. Sociology has Knight’s Law: “Bad talk drives out good”.
Linguistics has its own Law of Semantic Change: “Bad connotations drive out good”.
Occasionally words climb out of the semantic abyss and take on positive associa-
tions once more. Consider the word attitude. Originally a technical term meaning
‘posture’, it shifted to mean ‘mental state/mode of thinking’ (that which is implied
by the physical position). It then deteriorated – to have (an) attitude implied ‘a bad
attitude’ or ‘an attitude problem’, something to be corrected by parents or teachers.
But, of course, how you understand this expression will depend on your view of
the world – neutral or bad concepts can become enviable qualities or sought-after
things. These days, attitude is becoming positive again. Its combination of insolence
and arrogance (with a bit of aggression thrown in) is touted as desirable, at least by
some. Many groups and societies now seek to describe themselves as being “with
attitude”. Interestingly, tude ‘a bad attitude towards someone’ seems to have picked
up the tab (see http://de.urbandictionary.com/define.php?term=tude).
3.1.5 Chain reaction changes

Sometimes a semantic change affecting a word can trigger changes to other words that
are in some way semantically related or belong to the same semantic field. For example,
jaw used to mean ‘cheek’, and cheek used to mean ‘jaw’, and as you saw in Chapter 1,
hound ‘generic term for canine’ and dog ‘specific canine’ swapped places. The following
shift affecting terms for ‘food’ presents a kind of “playing musical chairs” scenario:
meat ‘all solid food’ > meat ‘flesh of animals’

flesh ‘animal tissue’ > flesh ‘tissue of humans’
foda ‘animal fodder’ > food ‘all forms of solid nourishment’ (and so replaces meat)
A spectacular example of this kind of chain reaction change is provided by his-

torical linguist Raimo Anttila (1989: 146–7); it involves Latin legal terminology.
The vertical line here indicates the older meanings, and the angled line shows the
shift to the newer ones:
‘trust’ ‘damage’ ‘guilt’ ‘negligence’ ‘chance’ MEANING
damnum noxia culpa casus fortuna FORM
As you will see in Chapter 4, changes like these are rather like the sound shifts
that affect whole vowel and consonant systems. In both cases the systems are left
more or less intact; at least the same distinctions are maintained. As Anttila’s dia-
gram nicely illustrates, only the form-meaning pairings here have changed.
3.2 WHY WORDS CHANGE THEIR MEANINGS
There is a range of different factors that can bring about semantic shifts. We will
concentrate here on four broad areas.
3.2.1 Socio-cultural factors

Particular situations arising from the relationship between speakers and their world
are common triggers for change. In the discussion below, we look at some of the
most significant of these.
Social change
Shifts in societal mores or attitudes typically have semantic correlatives. The bud-
ding self-awareness of the modern western individual is driving a number of lexical
changes. Consider the meaning shifts that are currently taking place in words such
as celebrity (‘a solemn (religious) ceremony’ > ‘a popular public figure’), charisma
(‘divinely conferred power’ > ‘popular aura’), personality (‘pertaining to the per-
son’ > ‘celebrity’), cult (‘system of religious worship’ > ‘popular fashion involving
devotion to a person or thing’), image (‘representation of person/thing’ > ‘cultivated
favourable public reputation’) and model (‘representation of person/thing’ > ‘ideal
representation’ > ‘sexual clothes-horse’).
Language of special groups

When ordinary language swipes words from specialist languages the effect is usu-
ally semantic broadening. The original narrow specification is lost, and the term
expands its meaning. Epicentre comes from geology (where it refers to the true
centre of a disturbance) – in ordinary language it is acquiring a more general sense

of simply ‘centre’. The expression home in ‘to proceed towards, to focus attention on’
is aeronautical jargon of the early 1920s (although inspired originally by homing
pigeons). Often the changes involve a shift of contexts. Quantum leap has come a
long way from its technical origins (the increase in energy needed to make atoms
jump from one state to another). When we describe someone as having gone bal-
listic (in other words, having exploded with rage), this gets the technical concept of
ballistics quite wrong (when missiles go ballistic they don’t explode; they actually
coast).
Subreption
This is the process of external change – objects, ideas and institutions alter over
time, but the names for them remain. It’s as if the expressions outgrow their original
meanings (“old words in a new world”). These days we drink from glasses made of
plastic and through straws made of paper or plastic. Light was once by definition the
medium of visual perception, and the idea of invisible light was therefore nonsensi-
cal (a bit like a triangle with four sides or a round square); these days we have ultra-
violet and infrared light, and the meaning of light has shifted accordingly. Despite
the considerable and rapid developments in telephonics, we continue to dial num-
bers on our push-button phone, and we also hang up – though of course hanging
up doesn’t have quite the same satisfaction as slamming down the receiver of a
desk phone. Time will tell whether these words disappear or survive with shifted
meanings.
3.2.2 Psychological factors

There are all sorts of aspects of our mental make-up that we could consider here – we
procrastinate, we exaggerate, we are suspicious of intelligence, we are pessimistic, we
often underestimate the truth in certain circumstances and are given to understate-
ment. All of these traits have repercussions for the meanings of words.
Prejudice
The nasty habit we have of making preconceived (usually unfavourable) judgments
about people or things is responsible for many changes in meaning. Slut in Chau-
cer’s time referred simply to ‘a woman of untidy habits’, but in a short period of time
it had shifted to ‘a woman of a low or loose character’. Society clearly places very
different values on male and female sexuality, and this is reflected in such changes.
Untidiness gets linked to a woman’s sexual mores – if she’s messy, if she can’t keep
a tidy house, then she’s clearly someone of “loose” character. Words like slut, slag,
slattern, hussy, whore and more recently floozie, tramp and bimbo tell similar stories.
In fact, there are hundreds of words that refer to women in a sexually derogatory
way. Words for males are much more stable over time, and the result is a consider-
ably smaller list of negative terms. They also don’t have quite the same disapproving
sense of sexual promiscuity: even ostensibly pejorative terms like philanderer, flirt,
rake, lady-killer, gallant, gigolo, stud and womanizer refer at the same time to virility
and sexual prowess. They connote a Don Juan or a Casanova and have associations
of gallantry (compare earlier squire of dames and roué).
Emotion
Humans are natural born exaggerators, and hyperbole is a major driving force
behind semantic change. Nowhere is this more obvious than in the constant
renewal of boosting expressions mentioned earlier – something isn’t just good but
awfully good or terribly good. Inevitably, such dramatic words wear out over time
and become mundane. Alternative expressions then have to be found. Most have
strong, even quite gruesome beginnings. Originally, when something was quali-
fied with awfully, it would have communicated awe ‘divinely inspired terror, mixed
with veneration’. This made it a very effective intensifier, but time bleached it of this
energy and force, and by the early 1800s it meant simply ‘extremely’. Exaggerated
use is responsible for the bleaching of words in many other areas of the lexicon; e.g.
scallywag, rogue and rascal were once highly offensive expressions, and ratbag, once
an appalling insult, now pales in comparison with its modern relatives scumbag,
slimebag and dirtbag.
She was literally climbing the wall

Extravagance is what is behind the shift in meaning of the word literally. In its
core sense, literally still means ‘actually, without exaggeration’. The word sig-
nals that the phrase that follows is meant to be interpreted in its literal sense;
in other words, its real or non-figurative meaning. If someone were to say,
she was literally climbing the wall, we would understand that this person was
actually going up a wall – no exaggeration. But these days a more likely inter-
pretation is that the person was extremely anxious, almost frantic – and was
figuratively or metaphorically climbing the wall (in other words, not climbing
the wall at all!).
Speakers are always on the lookout for ways of embroidering expression,
and this kind of linguistic overkill has always been around and is well attested
across languages. A word like literally is tailor-made as an intensifier, and in
fact English speakers have been using it in this exaggerated way since at least
the mid 1800s. One example given in the Oxford English Dictionary reads:
“For the last four years, I literally coined money”. Presumably, this person
wasn’t actually manufacturing currency.
This change has meant that literally has become its own opposite. At the
same time, it means both ‘actually, not figuratively’ and also ‘not actually, figu-
ratively’. So it joins words like fast and barrack, mentioned earlier. These are
known as “contronyms”, or more colloquially “janus-words” (words with two
faces). Opposite meanings can coexist quite peacefully, but only if they occur
in such different contexts that there’s no chance of a misunderstanding. Blunt

instruments are dull and not sharp, for instance, but blunt comments are
pointed and very sharp. When you dust something, it frees it from dust – that
is, unless you’re dusting for fingerprints or dusting a cake with icing sugar. In
the case of literally, it is less clear that the opposite meanings can occupy the
same space in a peaceful way. If we were to tell you that she literally fainted
on the floor, you couldn’t be sure whether or not she’d actually fainted. In this
case, we suspect that the older meaning of literally is doomed (compare Ger-
man buchstäblich, which has shown the same transition).
There is also a kind of exaggeration via understatement. For example, in certain

contexts people routinely underestimate the truth, and this can drive change. Take
numerical expressions. Words for ‘pair’ or ‘couple’ frequently shift to something
meaning ‘a few’. Current changes to a couple give us a clue. Someone announcing
they’d had a couple of drinks would generally mean more than two, and for many
people now a couple means ‘a few’ in most contexts. There are plenty of other situ-
ations too where people resort to understatement to avoid speaking unrestrainedly
about certain topics. Since classical times, nasty verbs equivalent to die and kill have
been avoided; instead, dead people curl up, they go to sleep or go on a journey –
or they are described as simply having lived. Hence these verbs in many modern
languages derive from verbs with much milder meanings: kill originally meant ‘to
strike, hit’, and there is evidence to suggest that starve (which originally meant ‘to
die’) also began as a euphemism with the meaning of ‘to grow stiff ’. There are many
things in life that speakers and writers would rather not evoke too vividly, and this
takes us naturally to the next motivation.
Taboo and euphemism

When faced with the tricky problem of how to talk about things that for one reason
or another we would prefer not to speak of, we create euphemisms – sweet sound-
ing, or at least inoffensive, alternative expressions. But as Chapter 2 described, there
are internal forces at work that ensure that euphemisms are doomed to be short-
lived. Many simply wear out – like slang, insults and swearing, euphemism can
also lose its power. But what usually brings down euphemistic expressions are the
processes of narrowing and deterioration, as the taboo associations reassert them-
selves and undermine the euphemistic magic of the word. The next generation of
speakers grows up learning it as the direct term, or worse. Clear examples are words
for sexual activity like ejaculation, copulation, seduction, orgasm, intercourse, rape
and erection. All were once quite general terms, but as soon as they were pressed
into euphemistic service they quickly narrowed to the sexual sense alone. In 1666
Samuel Pepys wrote in one of his diary entries: “I could not but with hearty thanks
to Almighty God ejaculate my thanks to him”. Old novels abound in breathless
ejaculations. But these days, all senses of ejaculation, including the ‘short emotional
utterance’, have been overwhelmed by the sense of ‘discharging sperm’. Once a word
has acquired taboo or risqué senses, these come to dominate and eventually kill off
all other senses. Seduce meant ‘to lead astray (originally a soldier or subject)’; inter-
fere referred quite generally to a physical collision; liaison was a cooking term for
the thickening of sauces. In these terms, the sexual senses now rule.
3.2.3 Linguistic factors

Factors that drive semantic change can also come from elsewhere in the lexicon – in
various ways, words can be influenced by the meanings of other words, or groups of
words, that they hang out with or are related to in some way.
Contamination
Meanings of expressions often shift because of associations picked up from other
words that are connected in sound and/or meaning. Fortuitous is in the process of
shifting its meaning from ‘by chance’ to ‘fortunate’, clearly on account of its proxim-
ity to fortunate. In the dictionary flagrant is still defined as ‘blatant, brazen’; yet, for
many people it now describes ‘the way flowers smell’ (perhaps they see it as a blend
of fl[ower] and [fr]agrant). In some cases the meaning changes to such an extent
that the word actually comes to be its own opposite. Consider hoi polloi. Originally
Greek for ‘the many’, the word was borrowed into English to mean ‘the masses’. But
many English speakers report that they have the opposite meaning for this word,
namely ‘aristocratic persons, the high-born’. This shift has come about through asso-
ciation with expressions like high, haughty and hoity toity. Curiously, hoity toity has
itself undergone a similar shift. It derives from the reduplication of the now obso-
lete verb hoit ‘to behave boisterously’. Hoity toity originally described ‘riotous behav-
iour’, nothing like the meaning we have today. It then shifted to ‘snooty’ because
people associated it with haughty. And now it is influencing hoi polloi in precisely
the same way.
Sounds aren’t supposed to have meanings

In Chapter 2 we described how phonesthemes are “meaningful” sound
sequences that occur either at the beginning of words (tw-, gl- fl-) or in their
rhymes (-ash, -ump, -itter). Although they symbolize a certain meaning,
it’s often difficult to pin down exactly what it is. For example, the following
words (mainly verbs) show a characteristic -ump rhyme and seem to denote
‘heavy fall’: bump, dump, crump, clump, flump, jump, plump (verb), slump,
stump, thump. Others (mainly nouns) with the same rhyme denote some sort
of ‘round protuberance’: bump, chump, clump, dump, hump, lump, mump(s),
plump (adjective), rump, stump. Phonesthemes like these often encourage
similar sounding words to shift their meaning (they encourage words to shift
their sounds, too, but that is another story).
Take twit, now meaning ‘fool’ (the meaning appeared in the 1930s). The
word derives from a verb twit ‘to taunt’. So how do you get from the taunting
meaning to the current-day meaning? True, someone who twits (is given to
taunting banter) would be a twitter or a twit, and this kind of behaviour is
perhaps also a little idiotic. So perhaps the earlier idea of ‘someone prone to
twitting’ then slides to ‘annoying person’ or ‘fool’. But there are often many
factors driving a meaning shift, and this particular one would have had con-
siderable encouragement from a clash with similar sounding words.
In the early 1900s, there appeared a couple of other words beginning with
tw- and having similar meanings. There was the twerp ‘a silly, insignificant
person’ (inspired perhaps by art critic and odd-ball T. W. Earp, once described
by J.R.R. Tolkien as “the original twerp”), and twat, a word that originally
referred to ‘vagina’ but in the early 1900s had been recruited, like many other
bawdy body parts, to become a term of abuse. Now, when you get a couple
of salient words that happen to be similar in form and in meaning, they can
motivate meaning shifts elsewhere. It is as if speakers start to interpret certain
clusters of sounds as meaningful. In this case, the tw- of twerp and twat acts
as an additional trigger for the meaning shift in twit. And when the meaning
changes, twit joins the gang of other tw- words to describe someone who’s a
bit of a nincompoop.
Some shifts start life as something called a “malapropism”, where speakers mix
up words because of similarities in pronunciation. An inappropriate word is used
in place of another that resembles it in sound or spelling (there might be a vague
similarity in meaning). The term “malapropism” comes from Mrs Malaprop, a char-
acter in Sheridan’s 18th century play The Rivals. The term as such is formed from
French mal ‘ill’ + a ‘to’ + propos ‘purpose’. Mrs Malaprop had a chronic problem
with words and was famous for coming out with outlandish examples like “as head-
strong as an allegory on the banks of the Nile” (there usually wasn’t much inner
logic to Mrs Malaprop’s malapropisms!). A famous political example occurred in
1974 when John Ehrlichman, assistant to the US president at the time, described
John Dean, special adviser to the President, as “a piranha”. As one Washington
newspaper editorial later noted, was he really suggesting that Dean was a particu-
larly voracious South American freshwater fish, or did he mean a pariah, ‘a person
rejected or despised’?
Such mix-ups are commonplace in everyday language: mitigate for militate, prod-
igy for protégé, prostate for prostrate and economically depraved for economically
deprived. They might be slips of the tongue, or they might be driven by linguistic
snobbery – the desire to use a more impressive sounding word such as enormity
instead of the more mundane enormousness. The result can be a significant shift
away from the earlier meaning. Currently it looks as if the older meaning of enor-
mity ‘wickedness’ is well and truly on the way out.
It’s here that we are crossing over into the territory of the newly created eggcorn,
itself an eggcorn for acorn.1 The term was coined by linguists Mark Liberman and
Geoff Pullum, because they felt that examples like eggcorn weren’t handled by any of
the already existing categories of language error. They’re idiosyncratic slips too, but
they involve homophones or near-homophones, and, most importantly, they have
a logic to them (which Mrs Malaprop’s mistakes don’t really) – coleslaw becomes
coldslaw (it is a type of cold salad). The student who produced rope learning in place
of rote learning probably had in mind expressions such as learn the ropes (more
meaningful for him than the rarely used rote ‘mechanical manner’).
Slips of the ear/brain can endure and bring about change. When eggcorns spread
their wings and whole speech communities start using them, we move into the
territory of folk or popular etymology. The line between these is admittedly rather
blurred – think of folk etymology as involving triumphant eggcorns. Old English
shamefast ‘modest, shy’ (lit. ‘restrained by shame’) was misinterpreted as shame-
faced, and the meaning subsequently shifted to ‘embarrassed’ (something you reg-
ister on your face). A more recent example is be on tenterhooks. This doesn’t make
a lot of sense nowadays, because tenter, as the rack used to stretch woven cloth,
doesn’t exist anymore. So some speakers have altered the word to tenderhooks.
Remodelling tenterhooks to tenderhooks replaces an obscure word with something
similar but more transparent, with the additional bonus that tender contributes a
sense of ‘fragility’ and ‘sensitivity’, as might come of being in a state of painful anxi-
ety. It will be interesting to see whether this eggcorn also endures.
Ellipsis
American English fall ‘autumn’ derives from the old and longer phrase fall of the leaf.
It is a poetic but rather cumbersome way to describe the season; so not surprisingly
speakers abandoned the last few words, and the whole meaning of the phrase then
got dumped into the remaining word fall, which itself came to stand for the season.
Similarly, the word rifle was originally shorthand for rifled gun. This was a gun with
a special “rifled” groove inside the barrel. There are many such examples. Private
soldier was reduced to private; alarm clock to alarm; life sentence to life; underground
railway to underground. Adjectives can also turn into nouns when they find them-
selves left behind in this way; for example, a daily paper becomes simply a daily,
and a weekly magazine becomes a weekly. A more recent example involves bilateral
agreement – a reciprocal arrangement between two political parties – it’s frequently
described as simply a bilateral.
These truncations can involve significant meaning shifts when expressions involv-
ing rhyming slang are shortened. Take, for example, rabbit (on), which derives from
the longer rhyming slang version rabbit and pork (= talk). The meaning derives
from the unstated word that rhymes with the last part of the phrase; for example,
rabbit and pork equals talk. But time erodes the expression to rabbit, and the origi-
nal significance disappears because the rhyme is clipped. The sense then transfers
to the first part of the phrase, and the result is a meaning shift, in this case a striking
one. The verb to rabbit on now means ‘to talk incessantly’. Other examples include
brahms ‘drunk’ (< Brahms and Liszt = pissed), and berk ‘stupid person’ (Berkley
Hunt = cunt; the current meaning of berk is mild compared with the meaning of the
insult cunt, probably because berk is not widely recognized to be an end-clipping
of Berkley Hunt).
Metaphor
Metaphor (from Greek metaphorā ‘transference’) pervades language and without
doubt is one of the most significant forces behind vocabulary change. When people
use metaphor, they refer to one domain by using language expressions that are nor-
mally associated with some other domain. There is a transfer of meaning from one
given context to another. This explanation of metaphor has strayed little from Greek
philosopher Aristotle’s original account. As he put it back in the 4th century BCE,
“[. . .] a good metaphor implies an intuitive perception of the similarity in dissimi-
lars” (Poetics, Chapter 22, page 2335; translated by I. Bywater). So it involves a kind
of analogy, encompassing the comparison of two items where speakers see some
sort of connection.
Aristotle was probably the first to point to the ubiquity of metaphor. As soon as
we open our mouths (or put pen to paper) we produce metaphors. But most of them
are conventionalized – they are automatic. People talk generally about playing the
game, lifting the game, giving the game away, levelling the playing field, moving the
goal posts and so on, but, as with general expressions like aim and goal or even tack-
ling a problem, the figurative sense has become the conventional meaning. Even with
metaphors based on the specialist terminology of individual sports (like cricket,
baseball, football and boxing), people are unaware they are extending a sporting
term (people on a good wicket could also be described as having a fair innings, but
others not so fortunate would be on a sticky wicket ‘facing a difficult situation’ – they
might be caught out or stumped, if they are ‘non-plussed’ or ‘defeated’). When the
expressions are learned by a new generation as the common everyday term, all the
colour and emotional qualities fade. If the new meaning prevails and the metaphor
dies out, the alteration is complete and a meaning shift has taken place.
Expressions to do with seeing, hearing and touching often develop into mental
state verbs (understanding, supposing etc.); for example I see and I hear you for
‘I understand (you)’ or He finally grasped it for ‘He finally understood’. In this
last example you can see the jump from one domain (physical – taking hold) to
another domain (mental – comprehending). In these cases we are no longer con-
scious of the metaphorical links. Time has pushed them below the level of con-
sciousness. Often the imagery is well and truly buried, as in the verb comprehend.
It too comes from something that means ‘to grasp, seize’, but this is a metaphor
from long ago, and one that has been borrowed from Latin. Much of what we talk
about is in terms of something else. Scratch the surface of many expressions, and
you will find a dried-out metaphor of this nature.
The last example illustrates an important type of metaphor – synaesthesia or
association of senses (from Greek syn ‘together’ + aisthesis ‘perception’). It can
involve transfers of all kinds – from sight and touch to intellect (as in the case of
the mental verbs just given), from sight to sound, from touch to sound, from taste
to colour and so on. When hot ‘having a high degree of heat’ took on the additional
meaning ‘spicy’, it shifted from the area of touch to taste. Other words that changed
in this way are cold, dry, smooth and coarse. The word faint has shifted from colour
to sound – ‘pale, lacking clearness’ sprouted the later meaning ‘barely perceptible’.
Dimension words like big, deep, thin and high transfer to both colour and sound.
And they can transfer to taste too. Just look at current wine terminology that draws
on images like big, full, deep, even, thick, flat and small. This is not only the stuff of
great literature and poetry – it’s also the stuff of our everyday language.
A close relative (perhaps even a subtype) of metaphor is metonymy, or change
by association (from Greek meta ‘trans’ + onoma ‘name’). A word is extended to
refer to something that is closely associated with the meaning of that word. Take
the broadening examples we looked at earlier. Pavlovas (< Russian ballet dancer
Anna Pavlova) and sandwiches (< the Earl of Sandwich) get their names either from
the inventors or from people associated with the items in some way; champagne,
burgundy, cheddar, cologne, denim and so on are all goods named after their place
of origin. In Restaurantese, metonymy is commonplace (The chicken wrap wants a
glass of white wine), and it is a favourite strategy for creating linguistic disguises (or
euphemisms); spend a penny for ‘urinate’ (from the days when toilets cost a penny
to access); go to bed and sleep with for ‘copulate’.
An important kind of metonymy is synecdoche, or the relationship of part-for-
whole (from Greek sunekdokhé ‘inclusion’); something is named on the basis of its
most prominent or salient part; e.g. hand for ‘employed worker’, silver for ‘cutlery’
(originally made of silver), tit for ‘breast’ (denoting the most tabooed part of the
breast – the nipple). Synecdoche is often exploited in abusive language, as in calling
someone a prick, and can be used for humorous effect in nicknames (often with
an ironic twist to them): Shorty (for someone really tall), Curly (for someone with
dead-straight hair) and Blue or Bluey (for a redhead).
Examples of metonymy and synecdoche differ from the earlier examples of meta-
phor in that they don’t evoke new associations but simply exploit those already
existing between words that share a close logical relationship. But clearly all three
are very close. You might think of metonymy and synecdoche as involving changes
within the same semantic domain (involving “whole-for-part” and “part-for-
whole” substitutions), and metaphor as involving changes across different seman-
tic domains (though of course when speakers extend familiar concepts from their
experiences to new purposes via metaphor, they are making associations).
“The penultimate super marathon”

Often change involves a number of factors. Take penultimate; there is not
much call for this word in everyday language, and this makes it vulnerable
to change. Many younger speakers are now using it as an emphatic form
of ultimate. Ultimate itself cannot be graded (very ultimate?). So, how does
someone convey the idea that something really is matchless, the out-of-this-
world greatest? It seems penultimate is being recruited for this purpose.
This is an interesting development because, if the change does take hold,
it’s going to turn the orthodox meaning of penultimate on its head. In origin,
penultimate comes from the Latin paene, meaning ‘almost’, plus ultimatus
‘last’, and penultimate means literally ‘almost last’. This new colloquial usage
now takes it beyond last – to refer to something that is beyond all others. “It’s
the penultimate super marathon” was heard on ABC (the Australian Broad-
casting Commission) radio in September 2015. The person wanted to get
across that this was the best of these ultra distance races, even though many
of the listeners probably understood it to mean the second last event.
Clearly exaggeration is playing a role here, and perhaps assisting this shift
are also thoughts of pinnacles ‘high points’ and penthouses ‘luxurious apart-
ments on top of high-rise buildings’ (so clang association). It might even be
that the addition of a prefix simply provides extra weight. Compare what is
happening to penultimate with the current use of the technical term epicentre
(‘the place on the earth’s surface where an earthquake begins’). As described
earlier, many now use it instead of ordinary old centre, especially if they want
to convey the sense that important things are happening, as in He is at the
epicentre of a drug ring.
Enfield (2015) describes how changes might start off as alterations in the
behaviour of individual speakers, but if a linguistic concept is useful, it will
be “aired, and shared,” and result in a change in community-wide convention.
Penultimate and epicentre are definitely words to watch.
3.3 REGULARITY IN SEMANTIC CHANGE – A MORE EXPLANATORY ACCOUNT
A number of linguists have attempted a more general (cognitive) approach that lays
the foundation of a general theory of semantic evolution with sets of “laws” outlin-
ing the most usual directions of change. One of the most influential linguists in this
regard has been Elizabeth Closs Traugott (see Traugott 1982, 1985, 2003). While
many had pointed out that words typically shifted their meanings from concrete to
abstract, Traugott sought a more explanatory account. In various works she outlines
examples of “lawfulness in semantic change” (an area usually thought to be rather
lawless). On the basis of these, she claimed conclusions could be made about the
nature of the mind.
Her approach was therefore cognitive and contrasted with the socio-historical (or
philological) approach that was developed in the 19th century. Because the focus of
this traditional approach was on individual changes, little or no sense of regularity
could emerge, making it difficult to establish whether there were any unidirectional
forces at work. Traugott sought to focus on domains associated with conceptual
structure and mental life in order to establish patterns of change. On the basis of
this work, there has emerged a number of general claims about semantic change.
We will touch on some of these here (but see Traugott and Dasher 2002 for a full
account).
Space as a source of abstract terms

[. . .] thoughts and sayings are, like sentences, conceived as objects in space that
can be put in a row, stood on, or stood under, and transmitted. Since they can be
put in a row, they are conceived as objects arranged on axes aligned with ourselves:
particularly the front-back and the up-down axis.
(Traugott 1985: 163)
There are all sorts of regularities in the directions words shift their meaning: the
development of grammatical categories from body part terms which are used to
express spatial concepts (head, back, face and stomach come to stand for spatial rela-
tionships such as ‘on’, ‘back’, ‘front’ and ‘in’); future temporal expressions like will
often derive from words for ‘wishing’; some connective words derive from temporal
expressions (e.g. when, while); mental state verbs derive from vocalizations (recall
‘remember’) and even visual perception (see ‘understand’). All show the character-
istic shift from less to more abstract that is typical of semantic change. Here we will
concentrate purely on source domains involving space.
• Spatial meaning to temporal meaning

The prepositions before and after started out in life as spatial words mean-
ing ‘in front of ’ and ‘in back of ’; the verbs of movement go and come have
acquired temporal meanings (e.g. I am going to sing). The spaces used seem to
be restricted to axes that run front and back (as in before and after) or up and
down (as in burn up, slow up, slow down etc.).
• Spatial meaning to grammatical relation
Grammatical relations refer to the functional relationships between constitu-
ents of a sentence; they hold between predicates (usually verbs) and associated
arguments (noun phrases). These relations can involve concrete spatial activi-
ties like going from one place to another, but they can also involve less direct
ones, like giving something to someone. English to has the meaning ‘towards’
but also expresses the indirect object relation (in They gave it to Alex it is as if
Alex were a place, not a person); similarly, spatial of ‘from, out of ’ can express
possession and belonging (the cover of the book); spatial by ‘close to’ expresses
agency in the passive construction (in Kate was given a book by Alex, the idea of
proximity is preserved in the fact that the agent (Alex) is thought of as closely
involved in the event).
• Spatial meaning to connective
Connectives are used to express cohesion between clauses. For example, but
derives from butan ‘on the outside of ’; notwithstanding ‘nevertheless’ has a more
obvious spatial origin in not + against + stand; also transparent are connective
expressions such as hereupon, therefore etc. As described in the quotation at the
start of this section, it is as if speakers line up sentences and clauses just as they
would line up objects in the real world.
• Spatial meaning to mental verb
In many languages, spatial activities give rise to mental verbs, such as those
of understanding, supposing and deducing. Understand is transparent (from
a verb ‘to step under’); suppose and deduce come from original Latin construc-
tions with literal meanings ‘under + put’ and ‘down + lead’.
• Spatial meaning to speech act verb
Speech act verbs include those like assert, promise, command etc. that (given
the right circumstances) not only name the action being performed but also are
equivalent to that action (in other words, the act is accomplished by uttering the
words, as in I assert that Fritz stole the sock). There have been some 275 speech
act verbs identified for Modern English, and about 75% (including borrowings)
originate in spatial terms: suggest (< Latin sub + gerere ‘under carry’) and insist
(< Latin in + stare ‘stand upon’).
Other regularities
Traugott talks about three functional components in language to explain general
tendencies in semantic change (1985, modified from Halliday and Hasan 1976). The
first (the propositional component) refers to the resources languages have for talk-
ing about something. The second (the textual component) refers to the resources
languages have for creating cohesion. Finally, the personal (or expressive) compo-
nent includes the resources languages have for expressing speakers’ attitudes or feel-
ings about the event or situation that is the focus of the conversation. Over time
we see a transition from propositional to textual to personal; i.e. as Traugott has
described (p. 165), a shift from meanings that are “more verifiable in the world” (and
therefore concrete) towards meanings relating to “the internal world of personal
point of view, inference and belief ” (and therefore more abstract).
Hence, a broad tendency is for meanings to become increasingly anchored in the
speakers’ worlds, more and more inclusive of their point of view. This is also called
subjectification, and we touched on this idea in Chapter 1. Many of the changes we
have already discussed involve the increasing expression of speaker perspectives.
“Reading between the lines”

So much goes on when words are used, with context and background provid-
ing all sorts of clues that go beyond core meanings. Addressees seek to extract
all the interpretations possible from any message, often imputing meanings
that may not match original intentions. So a word might start off having a
particular meaning, but new ones are lurking in the wings, ready to hop on
at any time. Take the simple conjunction since as an illustration. Its earliest
recorded meaning is ‘from the time that’ as in Life has become dull since Fred
left. But we “read between the lines”, looking for enriched meanings that are
somehow relevant to the context of discourse. For some people, a tempo-

ral meaning implicates some sort of causal link between these two clauses
(why else would someone say such a thing?) – for them, Fred’s leaving has
something to do with life being dull. Implications can become so reliable
and widespread that they then get conventionalized and become part of the
actual meaning; since now has both temporal (‘from the time that’) and causal
(‘because’) meanings, and the latter is on the way to prevailing. Compare the
conjunction after, whose temporal meaning in some contexts can have a
causal interpretation. (One of us recalls being surprised when some years
back half the Structure of English class described after as a causal connector
(rather than the more expected time connector) in the sentence Life became
dull after they left – a good example of what Enfield has dubbed the “gap
between mind and community”).
Clearly there are regularities involved here, but we still don’t know exactly when
or even if a given change will occur. Typically there are a number of factors involved,
and those deriving from the socio-historical setting often take words on surprising
journeys. As we will see time and time again in this book, language change is never
one-dimensional, and there are always networks of different intersecting pressures
that influence languages at different times: psychological (the mental make-up of
speakers), physiological (the production of language), systemic (the linguistic sys-
tem with interacting components), social and political (the speech community and
the individual, the socio-political environment), external (contact and borrowing)
and so on. There are also human wildcard factors to consider; the cultural preoc-
cupations of speakers can be powerful triggers for dramatic, and often unexpected
changes, and not just in lexical change. We are not suggesting that changes are ran-
dom, like whims of fashion, but they are complex, with contingent internal and
external factors (systemic and speaker oriented) and snowballing effects – and per-
haps even an element of chance.
SUMMARY
This chapter introduced you to a number of types of semantic change: broaden-

ing refers to the expansion of the contexts in which a word can appear; narrowing
refers to a reduction in the contexts in which a word can appear; shift entails the
total alteration of the contexts. Connotations arise through our beliefs, prejudices
and experiences; words can elevate and deteriorate. We also identified a number
of forces for change: specialist languages can borrow from ordinary language,
and terms from specialist languages can enter ordinary language; subreption is
the process of external change – “old words in a new world”; social change will
always have semantic correlatives; prejudice is a powerful trigger for meaning
change, as are psychological factors like emotion and taboo; words can pick up
associations from other words (e.g. in folk etymology and ellipsis); metaphori-
cal change involves the transfer of meaning from one item to another which is
somehow associated.
We also identified a number of strong tendencies in semantic change, many
involving cognitive processes, such as subjectification – where meanings take on
more personal (e.g. moral) viewpoints. We can also predict that if a language has
temporal terms, markers of grammatical relations, connectives, or mental or speech
act verbs, then some of them will be derived from spatial terms. But never lose sight
of the many factors that can coalesce to drive a word along a particular semantic
path, and in a particular direction.
SOURCES AND FURTHER READING
Of the many introductions to semantics out there, the ones we recommend are
Leech 1981; Allan 1986, 2001; Aitchison 2003; Cruse 2004; Loebner 2013; and on
semantic change the various Traugott publications already mentioned (especially
Traugott and Dasher 2002) and Enfield 2015. Euphemism is a topic you can fol-
low up on in Allan and Burridge 1991, 2006, and for metaphorical expressions we
recommend Lakoff and Johnson 1981, as well the timeless account offered by the
4th century Greek philosopher Aristotle (see his various descriptions in Poetics and
Rhetoric; Aristotle 1984). For sound symbolism see LaPolla 1994. You can follow up
contronymy in English in Karaman 2008; Lutzeier’s 2007 impressive three-volume
dictionary of contronyms in German shows just how surprisingly widespread the
phenomenon is.
EXERCISES
1 Part 1: identifying meaning shifts

The following list shows the earlier and later meanings of words from a number
of languages. Note that the earlier meaning is given in brackets, and the arrow-
head indicates the change. State what type of meaning change(s) they illustrate: i.e.
broadening (B), narrowing (N), deterioration (D), shift (S) or amelioration (A).
a a couple ‘a few’ (< ‘two, a pair’)
b ake-sita (Japanese) ‘the following morning’ (< ‘dawning time’)
c albern (German) ‘silly’ (< ‘simple’ < ‘kind, gentle’)
d asita (Japanese) ‘tomorrow’ (< ‘the following morning’)
e bereaved ‘deprived by death’ (< ‘robbed’)
f billig (German) ‘cheap’ (< ‘just, adequate’)
g bird (< brid ‘little bird/chick’)
h boor ‘crude fellow’ (< ‘farmer’)
i callous ‘insensitive’ (< callus ‘hard skin’; compare thick-skinned)
j chap ‘fellow’ (< ‘customer’)
k coast ‘sea shore’ (< ‘side’)

l crust ‘job’ (< ‘outer part of the bread’)
m curly ‘difficult (of questions, problems)’ (< ‘disposed in curls or ringlets’)
n daft ‘crazy, foolish’ (< ‘gentle, mild’)
o Dirne (German) ‘prostitute’ (< ‘young woman’)
p discard ‘reject’ (< ‘throw out a card’)
q famikon (Japanese) ‘any video game console’ (< ‘the name for Nin-
tendo Entertainment System’)
r flibbertigibbet ‘mischievous person’ (< ‘name of a devil’)
s fond ‘affectionate’ (< ‘foolish’)
t fowl ‘barnyard fowl’ (< ‘bird’)
u Friedhof (German) ‘graveyard’ (< ‘enclosed plot of land’)
v frock ‘loose-fitting outer garment’ (< ‘monk’s loose-fitting habit’); frock
‘woman’s dress’ (< ‘loose-fitting outer garment’)
w fruition ‘completion’ (< ‘enjoyment’)
x garble ‘to mix up’ (< ‘to sort out’, originally sifting out of spices)
y gleichen (Pennsylvania German) ‘to like, be fond of’ (< ‘to be like, resemble’)
z go ‘move, travel’ (< ‘walk, travel by foot’)
aa grog ‘alcoholic beverage’ (< ‘rum and water’; compare English grog ‘rum
and water’ < ‘nickname for Admiral Vernon (who wore a grogram cloak)’)
bb hayaku (Japanese) ‘immediately’ (< ‘rapidly’)
cc inquisition ‘persecution’ (< ‘investigation’)
dd joker ‘chap, fellow’ (< ‘one who jokes’)
ee hell (German) ‘clever’ (< ‘bright’; compare English bright)
ff laafe (Pennsylvania German) ‘walk’ (< ‘run’)
gg Marschall (German) ‘high military officer’ (< ‘stable hand in charge of
the horses’)
hh Maul (Pennsylvania German) ‘mouth’ (< ‘mouth (of animals)’)
ii mischievous ‘playfully annoying’ (< ‘disastrous’)
ll nice ‘kind, considerate’ (< ‘foolish, senseless’)
mm notorious ‘widely and unfavourably known’ (< ‘widely known’)
nn obnoxious ‘very objectionable’ (< ‘liable to injury’)
oo omoitatsu (Japanese) ‘resolve, plan to do’ (< omou ‘think’ + tatsu ‘stand up’)
pp pas (French) ‘not’ (< Latin passus ‘step’)
qq rort (originally Antipodean English) ‘act of fraud’ (< ‘jolly good time’)
rr saoul (French) ‘drunk’ (< ‘seized, possessed by anything’)
ss schmacke (Pennsylvania German) ‘to smell’ (< ‘to taste’)
tt sehr (German) ‘very’ (< ‘painful’; compare English sore)
uu shamefaced ‘embarrassed’ (< ‘fixed in shame’)
vv shiawase (Japanese) ‘happiness’ (< ‘a chance course of events’)
ww silly ‘stupid, daft’ (< ‘happy, blessed’)
xx smirk ‘simper’ (< ‘smile’)
yy sugoi (Japanese) ‘wonderful’ (< ‘terrible’)
zz Tier (German) ‘animal’ (< ‘wild animal’) (compare English deer ‘spe-
cific ruminant quadruped with deciduous antlers’ < ‘wild animal’)
aaa to wax (of the moon) ‘to grow in size’ (< ‘to grow’)
bbb trouver (French) ‘find’ (< Latin turbāre ‘to take/find fish’)
ccc trover ‘legal: the act of finding and assuming possession of any per-
sonal property’ (< French trouver ‘to find’)
ddd Trubel (Pennsylvania German) ‘trouble, misfortune’ (< ‘hurly burly,
confusion’)
eee urbane ‘sophisticated, elegant’ (< Latin urbanus ‘townsman’)
fff Versammlinghaus (Pennsylvania German) ‘church’ (< ‘meeting house’)
ggg villain ‘evil person’ (< ‘inhabitant of a farm’)
hhh Wand (German) ‘wall’ (< ‘weave’; note that walls were once woven
from branches)
iii wench ‘servant girl’ and earlier ‘wanton woman’ (< ‘young woman’)
jjj white ant ‘to subvert or undermine from within’ (< ‘termite’)
kkk yakuza (Japanese) ‘someone involved in organized crime’ (< ‘a
scoundrel’)
lll yellow ‘cowardly’ (< ‘golden colour’)
mmm zonjiru (Japanese) ‘think’; ‘know’ (< zon ‘put’ + suru ‘do’)
Part 2: triggers for change

Having named the changes in the above list, suggest why these words might have
shifted in this way. Also indicate those that show an increase in subjective meaning
(i.e. indication of the speaker’s attitudes and beliefs).
2 Finding your own examples of change

Now find one example yourself (from any language) for each of these changes –
broadening (B), narrowing (N), deterioration (D), shift (S) or amelioration (A).
(There are many websites devoted to new additions, and these can involve old
words that have acquired new meanings; for example, the Oxford English Diction-
ary (OED) updates four times a year and publishes lists of the new words each
year. (You will probably have access to the OED online through your library, and
on their homepage under “What’s New” you can find links to lists of new additions
for different years.)
3 Relic meanings preserved in fossils

As described in Chapter 2, modern languages are full of relic forms that can be used
to reconstruct features of vocabulary and meanings of earlier times. The following
expressions (or parts of them) preserve older meanings that have dropped out of
use (the relevant parts have been italicized); some have preserved words that are no
longer used.
a sweetmeat
b Holy Ghost
c fishwife
d bank teller
e short shrift
f beg the question
g neck of the woods
h hem and haw

i hoist by his own petard
j raring to go
k the quick and the dead
l the exception which proves the rule
m more haste, less speed
n not a whit
o one fell swoop
p to and fro
i Give the etymology of each italicized “fossil”.
ii Label and explain any meaning changes that have taken place.
iii Some speakers have now recast expressions like short shrift and let ball
to short shift and net ball; many have shifted the meaning of to beg the
question from ‘to assume the point that is being raised in the question’
to ‘to raise the question’. Why do you think this is?
4 Lexical and semantic change

The following is a 10th century leechdom (‘remedy’) for abdominal pain. Under-
neath the original Anglo-Saxon (or Old English) text is a literal word-by-word
translation, and a loose translation follows. (Recall that the consonant symbol
<þ> thorn was from the futhorc or runic alphabet and was later replaced by
French inspired <th>; the vowel symbol <æ> (called “ash”) came from the Latin
script; the symbol 7 stands for ‘and’.)
Wiþ wambe wærce 7 rysel wærce þær þu geseo tord

against womb wark and belly pain when thou seest turd
wifel on eorþan up weorpan ymbfo hine mid twam handum mid

Weevil on earth up throw catch him with two hands with
his geweorpe wafa mid þinum handum swiþe 7 cweþ þriwa.

his throwing wave with thy hands strongly and say thrice.
Remedium facio ad ventris dolorem.

remedium facio ad ventris dolorem.
Weorp þonne ofer bæc þone wifel on wege beheald þæt þu ne locige æfter.
throw then over back the beetle on way take-care that you not look after.
þonne monnes wambe wærce oþþe rysel ymbfoc mid þinum handum þa wambe
the person’s womb pain or belly grasp with thy hands the womb
him biþ sona sel. XII monaþ þu meaht swa don after þam wifel
him is at-once well. 12 months thou have-power so to-do after the beetle
‘For stomach ache and pain in the belly (fat); when you see a dung beetle
in the earth throwing up (dung), catch him with your two hands along
with his casting up (i.e. dung balls), wave him strongly with your hands,
and say three times, “Remedium facio ad ventris dolorem”. Then throw the
beetle over your back; take care you don’t look backwards. When a per-
son’s stomach or belly (fat) is in pain, grasp the stomach with your hands,
it will soon be well with (the person); for twelve months after the beetle
(event) you shall have power so to do.’
Using some phonological imagination and a good etymological dictionary (such

as the Oxford English Dictionary online) give a brief history of the bolded words
(around one paragraph). As part of your account, include the following:
a Identify any meaning shifts that have taken place, and identify the type of
meaning change (i.e. broadening, narrowing etc.).
b Say whether the word is archaic (old-fashioned and therefore stylistically
restricted) or obsolete (no longer in use).
c Some words may no longer exist in ordinary language but are preserved
in specialist contexts or fixed expressions (e.g. riff ‘belly’ survives in mid-
riff; the older plural brethren survives in religious contexts; see Chapter 6).
Identify any modern expressions or contexts that have fossilized these
words and/or their meanings.
5 Research essay
Linguists have long noted that specific semantic nuances continue to “glimmer
through” modern forms, sometimes despite considerable change to both the form
and meaning of a language expression. Using the quotation below as the basis of
your discussion, write an essay (approx. 1,500 words) on the nature of semantic
change, reviewing the idea that “old ideas” persist over time. Your bibliography
should have at least six references.
A word never – well, hardly ever – shakes off its etymology and its for-
mation. In spite of all changes in and extensions of and additions to its
meanings, and indeed rather pervading and governing these, there will
still persist the old idea.
(Austin 1961: 149f.)
NOTE
1 And just to confuse things, acorn is itself a successful eggcorn – the chance similarity
of Old English æcern to oak and corn caused various remodellings including oakehorn,
akecorn and of course modern acorn.
4
Changes in sound structure
INTRODUCTION
If we somehow had access to a time machine and could transport ourselves back to
the Middle Ages, we might hear first hand the language of the great medieval poet
Geoffrey Chaucer (~1345–1400). Other differences aside, the accents would sound
at once tantalizingly familiar and bizarrely foreign. To illustrate, here’s another
extract from the prologue of The Wife of Bath’s Tale (from Chaucer’s Canterbury
Tales). We’ve given a loose translation below and the phonetics alongside, so that
you can appreciate the differences. (Note that Chaucer’s poetry is generally written
in a loose iambic pentameter, a common metrical form for English poetry; it con-
sists of five (that’s the penta bit) pairs of alternating unstressed and stressed beats.
If you want to remember the rhythm of iambic metre think of the pulse of “iambic
feet are firm and flat”.)1
But, lord crist! whan that it remembreth me bʊt lɔ:rd krɪ:st hwan ðat ɪt
rəmɛmrəθ me:
Vpon my yowthe, and on my Iolitee, ʊpɔn mɪ: yu:θ and ɔn mɪ:
dʒɔlɪte:
It tikleth me aboute myn herte roote. ɪt tɪkləθ me: abu:t mɪ:n hɛrtə
ro:tə
Vnto this day it dooth myn herte boote ʊnto ðɪs dæɪ ɪt do:θ mɪ:n hɛrtə
bo:tə
That I haue had my world as in my tyme. ðat ɪ hav had mɪ: wʊrld as ɪn
mɪ: tɪ:mə
But Age, allas! That al wole enuenyme, ðat a:dʒ allas ðat al wɔl
ɛnvənɪ:mə
Hath me biraft my beautee and my pith. haθ me: bɪraft mɪ: bɛʊte: and
mɪ: pɪθ
Lat go, fare wel! The deuel go therwith! lat gɔ: fa:r wɛl ðə dɛvəl gɔ:
ðɛ:rwɪθ
The flour is goon, ther is namoore to telle; ðə flu:r ɪs gɔ:n ðe:r ɪs namɔ:r to:
tɛllə
The bren, as I best kan, now moste I selle; ðə brɛn as ɪ bɛst kan nu: mo:st ɪ
sɛllə
But yet to be right myrie wol I fonde. bʊt jɛt to: be: rɪçt mɪrɪ wɔl ɪ
fɔ:ndə
Now wol I tellen of my fourthe housbonde. nu: wɔl ɪ tɛllən ɔf mɪ: fɔʊrθ
huzbɔ:ndə
‘But, Lord Christ! When I recall my youth and jollity, it tickles me to the
bottom of my heart. And to this day it does my heart good to think that
in my time I’ve had my fling. But age, alas, that poisons everything, has
deprived me of my beauty and spirit. Let it go then, goodbye! The devil
take it! The flour’s all gone; there is no more to say. Now I must sell the
bran as best I can. But all the same I mean to have my fun. And now I’ll tell
about my fourth husband.’
Here are a few differences to note:
1 All the letters are pronounced (so no silent letters as in Modern English).
2 Vowel sounds are very different, and there are fewer diphthongs (e.g. /u:/ in
[flu:r] ‘flour/flower’).
3 Certain consonants no longer exist (e.g. /ç/ in [rɪçt] ‘right’).
4 There are some unusual clusters of consonants (e.g. /hw/ in [hwan] ‘when’).
5 There is evidence of long (or geminate) consonants (e.g. /ll/ in [tɛllə] ‘tell’).
If we were to travel back even further in time to a performance of the great epic
poem Beowulf (composed around the 8th century), the language wouldn’t even
sound remotely English, though some might recognize it as being Germanic. Try
for yourself: https://www.youtube.com/watch?v=0zorjJzrrvA.
In this chapter we explore some of the major types of sound changes, includ-
ing those that have transformed the shape of English so radically over the centu-
ries. Phonology is probably the most highly developed area in the field of historical
linguistics. For one thing, the study of sound variation and change has a particu-
larly long and scholarly history – a fall-out of the 19th century preoccupation with
sounds and the oldest stages of languages. It is also the case that we know a lot about
speech organs and the mechanics of speech production; so we have a good idea
about what are and what are not likely sound changes.
4.1 PHONOLOGICAL PROCESSES
When we speak we don’t just speak in single words, but in groups of words. How-
ever, there are no convenient gaps between these spoken words, equivalent to the
little white spaces between written words – one word simply merges into the next.
This seamlessness of speech can dramatically change the shape of words.
For example, here’s the phrase phonetic transcription as it might sound in normal
speech. Some symbols here are probably unfamiliar to you, but just focus on the stops
/k, t, p/, which are pronounced very differently depending on their position in the word.
[fəˈn̥ɛɾɪkotʃɹæ̃nˈskr̥ɪpʃn̩]
Said in isolation, English stops are followed with a little puff of air, known as
aspiration (so if you lit a match and said the name Kim, you’d likely blow it out).
Changes in sound structure 77
But for the production of the stops in this phrase, this puff of air is suppressed
(you’d be uphill blowing out any matches here). At the end of the word phonetic,
the /k/ is almost swallowed; this is an unreleased stop (the organs of speech aren’t
opened again after the sound is produced). In the middle of the word phonetic /t/
is pronounced like a hastily articulated /d/; this is known as flapping (the tip of
the tongue rapidly strikes the alveolar ridge), and it has the symbol /ɾ/. Interesting
things are currently happening to /t/ when it precedes /r/ as at the start of words like
transcription. It sounds a bit like the <ch> in chap.
In this section we consider in detail some of the most common processes affect-
ing the evolution of sounds. They fall into three main types: loss of sounds, addition
of new sounds and modifications to already existing sounds. These adjustments are
going on all the time as we speak, and sometimes they stay just that – fast speech
phenomena. But, as we explore here, they might also lead to significant changes to
the sound system of a language.
The vast majority of changes, as you will discover, are reductive in some way,
and people often complain about them, describing speakers as “careless” or “lazy” –
usually unaware that many of their own cherished expressions are themselves prod-
ucts of the same erosion (e.g. English lady and lord are fragments of Old English
hlāfdige ‘loaf kneader’ and hlāfwēard ‘loaf warden’). These sorts of reductions are an
inherent characteristic of ordinary casual (and usually fairly rapid) speech – we all
make them and would sound quite peculiar if we didn’t.
There is a theory that speech production varies along a continuum that ranges
from distinct or clear speech to less distinct or less clear speech. According to this
theory, speakers will make just as much effort to speak clearly as is required by their
audience in order to understand what is being said (Lindblom 1983, 1990). So we
are looking at a kind of tug of war going on in the evolution of languages between
the clear speech necessary for intelligibility (hyperarticulation) and the need for
physiological economy (hypoarticulation, sometimes dubbed “Zipf ’s principle of
least effort” or “economy of gesture”). You could imagine that if it weren’t for the
need to communicate effectively, the reduction of speech gestures might well reach
the point where sounds would be unrecognizable.
The standard practice in linguistics is to use an arrow “—>” to indicate pro-
cesses underway, and to use an arrowhead (or shaftless arrow) “>” to represent
historical processes (i.e. changes that have taken place). In both cases the direc-
tion of the arrow(head) represents the direction of the process or change; for
example, tree /tri:/ —> /tʃɹi/ shows what sometimes happens to the word tree in
casual conversation; tree (< Old English trēow) shows what has happened to the
word tree over time. In this way we can highlight the distinction between process
(synchronic variation) and transformation (diachronic change).
4.1.1 Old sounds drop out – loss

When we speak we take short cuts, often omitting unstressed vowels and drop-
ping consonants: words like enough and escape are pronounced as ’nough [nʌf]
and ’scape [skeɪp] (with the initial unstressed vowel omitted), vulnerable often
loses its /l/ [vʌnrəbəl], and Wednesday usually gets two syllables /wenzdeɪ/. In
English place names and personal names, the losses can be extreme: Salisbury /
sɔlzbri/, Leicester /lɛstə/, Cholmondely /tʃʌmli/, St John /sɪndʒən/, Featherstone-
haugh /fænʃɔ/, Woolfhardisworthy /wʊlzi/. Small wonder linguist Ron Langacker
once described language as “a gigantic expression-compacting machine”!
When reduced forms come to replace the full forms, even in formal or careful
speech, they leave a lasting impression that might even be reflected in the spelling;
in German, nouns that end with -el or -er regularly lose the vowel when a suffix is
added (e.g. Himmel ‘heaven’ + -isch ‘ish’ > himmlisch). In English, we witness more
and more instances of ’nuff, gonna and wanna even in written texts.
Sounds and their spelling in English

When printing arrived in the late 1400s, English spelling started to set-
tle down, though it took a few centuries to standardize. At the time when
it started to regularize, the language experienced a number of significant
changes. Many sounds changed their shape, and many disappeared alto-
gether. As a result, English now has thousands of words spelled much as they
were pronounced in Chaucer’s time. Most vexing are of course the silent let-
ters in words such as write, right, sword, know and gnat – the bane of English
language learners today.
So whether or not spelling displays a change depends a lot on when it takes
place. For example, English /w/ disappeared before /l/ sometime during the late
medieval period. The word lisp comes from Old English wlispian, but by Chau-
cer’s time it was spelled lisp. Around the same period, English also lost /h/ from
a number of other clusters (e.g. loud < hlud, ring < hring and neck < hnecca).
Unlike a later change that saw /w/ disappear before /r/ (as in write), all these
changes were early enough to be reflected in the spelling of words.
As you can probably imagine, reductive changes occur more easily when
people have no notion of the written word. If spelling is there to act as a
reminder, pronunciation is less apt to change – the visual image of a word
like government probably means we are less likely to pronounce it gubment.
Writing acts a little like an anchor, slowing down the normal process of sound
change, in some cases even reversing a change (e.g. often lost its /t/ during
the 17th century, but it has made a comeback now through the pressure of
spelling – [ɒftən]).
In the case of consonant loss, the disappearance is sometimes balanced by a

lengthening of the neighbouring vowel (unless it’s already long, in which case it
doesn’t change). The process is called “compensatory lengthening” – the time taken
to pronounce the word ends up being about the same. Like the Cheshire Cat’s grin
(the creation of Lewis Carroll in Alice’s Adventures in Wonderland), it is as if the
consonant lingers on in the quality of the adjacent vowel.
If you compare the English and German examples in the next table, you can see
that English lost the original nasal before voiceless fricatives, and this then trig-
gered a lengthening of the adjacent vowel (signalled here by a macron in the Old
English). German preserved the original nasal-fricative cluster of the parent lan-
guage (Proto-Germanic).
Table 4.1
Compensatory lengthening in English
Proto-Germanic Old English Modern German Translation

*fimf, *fimfi f ı-f, f ı-ftig fünf, fünfzig five, fifty
*munþ- mu-þ Mund mouth
*gans- go-s Gans goose
*kunþ- cu-þ kund known (couth)
Nasal loss before voiceless fricatives is a common change and occurs in languages
as diverse as the Romance languages (Italian, French, Spanish), Ojibwa, Old Irish,
Swahili and Yao (see Ohala and Busà 1995); as in the Germanic examples, the loss
is accompanied by vowel lengthening (and often nasalization of the vowel as well).
Vowel deletion is such a common change that it has spawned a number of
impressive sounding labels to describe the loss of vowels from different parts of
the word (some linguists have extended these labels to include also loss of conso-
nants from these positions, which also seems sensible to us). To help you remem-
ber these names, we’ve applied the process to the labels themselves (these are in
brackets):
• Apocope (‘apocop’) involves the disappearance of the final vowel. Revisit The
Wife of Bath’s Tale (or /ta:lə/) and notice how the final silent e’s in many of
the Modern English words (such as tale and time) were pronounced; in other
words, /ta:lə/ > /ta:l/ > /teɪl/; /tɪ:mə/ > /tɪ:m/ > /taɪm/.
• Syncope (‘syncpe’) is used for the loss of medial vowels. These sorts of losses are
significant because they can result in new consonant clusters; e.g. monks shows
the syncope of two vowels: Old English munecas > monkes > monks /mʌŋks/.
• Aphesis (‘phesis’) is the loss of an unstressed initial vowel, as in opossum > pos-
sum and esquirrel > squirrel.
Change that involves the deletion of entire syllables is haplology (‘haplogy’);

for example interpretative > interpretive; Englalond > England. As here, it typically
involves consecutive syllables that are similar sounding. Both German and Eng-
lish show regular haplology when the addition of an affix creates successive similar
sounding syllables; e.g. Zauberin ‘sorceress’ < *Zauber-er-in (compare English sor-
ceress < *sorcer-er-ess). Back in the 14th century Chaucer might well have written
humblely with three syllables, but nowadays we pronounce (and also write) this as
humbly with just two syllables; in the same way, nobly, simply, idly and gently have
all been cropped a syllable.
4.1.2 New sounds appear

The label “epenthesis” can refer to any process in which a vowel is added to a word
(though it is occasionally used more generally for consonants too). The addition
of sounds to the front of a word has its own special label, “prothesis”; for example,
speakers of French and Spanish inserted a vowel at the start of initial sC-clusters,
as in Spanish escuela ‘school’, estado ‘state’ and French école, état (< Latin schola,
status); the equivalent English words you can see were borrowed straight from Latin
(though we also have estate from French).
Addition is a less common process than loss, and the motivation is quite dif-
ferent. Some languages show a preferred C(onsonant) V(owel) syllable structure,
so vowels might be inserted to break up consonant clusters (especially awkward
ones) to produce this optimal arrangement of sounds, e.g. non-standard pronun-
ciations fillum /fɪləm/ for film, athelete /æθəlit/ for athlete and umbarella /ʌmbərɛlʌ/
for umbrella. Japanese took the word Christmas from English but requires CV syl-
lables. Christmas, however, is CCVCCVC (the <t> isn’t pronounced). So to fit this
word into the language, Japanese inserted a few extra vowels: kurisumasu – which
is perfectly CV CV CV CV CV. The process is very common in creole languages,
e.g. Australian Kriol silip (< sleep) and balijiman (< policeman).
Extra consonants can appear too, and this has a different label – “excrescence” –
to distinguish it from vowel insertion. This motivation here is a mistiming, with
extra sounds being inserted as the vocal organs move from one sound to another. To
see how this change might happen, take a non-standard pronunciation somepthing
/sʌmpθɪŋ/ for something. When you articulate the nasal /m/, your soft palate is low-
ered (so the air can escape through your nose). If you then raise the soft palate too
soon, you will automatically insert a little /p/ and the word comes out as somepthing
(it is voiceless because of the following /θ/).
We associate examples such as this one with children, but precisely the same
thing happened to empty (< Old English ǣmtig). Table 4.2 shows the fate of some
Latin words in English and German, as well as some Germanic forms both lan-
guages share. You’ll notice only the English forms show the extra consonant (and
note too how the inserted consonant agrees in place of articulation with the nasal
and with the voice quality of the following sound).
Table 4.2
Medial consonant excrescence in English
Latin Modern English Modern German

numerus number Nummer
camera chamber Kammer
Old English Modern English Modern German
þunor thunder Donner
þȳmel thimble —
ganra gander (compare gannet) Ganre (earlier form)
Wrong timing can also cause consonants to appear at the end of words. If the
tongue goes back to rest position against the teeth ridge before articulation is com-
plete, the result will be an extra /t/; it is as if speakers want to provide the ends of
words with a more satisfying conclusion (and you can see this clearly in colloquial
pronunciations nope and yep). German shows many examples of this; the following
table is just a handful (and this time English is the less adventurous language).
Table 4.3
Final consonant excrescence in German
Old High German Modern German Modern English

ahsa Axt ax(e)
habuh Habicht hawk
sac Sekt ‘bubbly wine’ sack ‘white wine’
selp selbst self
This process is behind the shift from early English agains, middes and betwihs to
against, (a)midst and betwixt (also the dialectal form varmint alongside vermin).
Enhancing word endings with extra sounds in this way was once more common in
English, but typically the phonetic afterthoughts didn’t survive: wonst, twyst, neist,
acrost, margint and sinst > once, twice, nice, across, margin and since.
4.1.3 Old sounds are modified – assimilation, dissimilation, metathesis

The phrase phonetic transcription [fəˈn̥ɛɾɪkotʃɹæ̃nˈskr̥ɪpʃn̩] showed the extent to
which sounds can be modified in normal speech. More usually, they change their
shape to become more like their neighbours (assimilation), but occasionally they
change to become less alike (dissimilation). Certain combinations can cause sounds
to shift about and appear in other positions in the word (metathesis). We will also
look at lenition (or weakening), a shift that sees the strength of sounds diminish
(e.g. consonants become more vowel-like). All these processes are ever present in
ordinary connected speech and are reflected in the synchronic rules of languages;
they are also common types of sound change.
Assimilation
Sounds will change to be more like the sounds they hang around with – this is
assimilation. The process is extremely common and is responsible for the majority
of sound changes. In the word sink, for example, /n/ becomes velar /ŋ/ to be more
like the following velar /k/. Reducing the distance between two sounds in this way
means a smoother transition and less articulatory effort (but, as described earlier,
economy of gesture is constrained by the need for intelligibility).
Consonants vary along three parameters: place of articulation, manner of articu-
lation and state of the vocal cords (i.e. voiced or voiceless), and they can assimi-
late according to one or more of these parameters. In the sink example, the nasal
Table 4.4
Total assimilation in Italian
Latin Italian Assimilation

factum fatto ‘fact’ /kt/ > /tt/
doctor dottore ‘doctor’ /kt/ > /tt/
septem sette ‘seven’ /pt/ > /tt/
acceptus accetto ‘acceptable’ /pt/ > /tt/
somnus sonno ‘sleep’ /mn/ > /nn/
domna donna ‘woman’ /mn/ > /nn/
subsulto sussulto ‘leap’ /bs/ > /ss/
absens assente ‘absent’ /bs/ > /ss/
has assimilated in place of articulation. It is even possible for sounds to assimilate

along all parameters (= total assimilation). Italian has clear examples where original
Latin clusters have totally assimilated to become long (or geminate) consonants
(see Table 4.4).
In all the examples so far, sound segments have assimilated to a following sound;
this is called anticipatory assimilation, and it is the more usual direction. However,
sounds can assimilate to preceding sounds (this is known as lag). The following Ice-
landic changes are similar to the Italian except that they work in the opposite direc-
tion: finna ‘find’ < *findan; munn ‘mouth’ < *munþ-; gull ‘gold’ < *gulþ. (Compare
Proto-Germanic *wulno > *wullo > Old English wull / Old High German wolla ‘wool’.)
One type of assimilation is so common that it has its own name: palatalization. It
most commonly affects consonants appearing before high front vowels and the palatal
approximant /j/ (which is also produced by raising the tongue to the palate directly). It is
what triggers pronunciations like /mɪʃə/ and /fɪtʃə/ in high-speed articulation of miss you
and fit you, because speakers draw the /s/ of miss and the /t/ of fit towards the /j/ of you.
But it also leads to long-term changes, such as the current pronunciation of words like
nation, sure, measure, pleasure and so on. The spelling often gives a clue; e.g. the ti in
nation shows the original /i/ which triggered the pronunciation /ʃ/. It is this same process
that gave rise to the palatal phoneme /tʃ/ in chin (< Old English cinn /kɪnn/).
Russian has spectacular examples of palatalization; as Table 4.5 shows, certain
consonants become palatal when they occur before front vowels like [e] and [i]; the
palatal sounds are bolded here, and the palatal quality is indicated by the raised j.
Table 4.5
Palatalization in Russian
Basic forms Palatalized forms

stol ‘table’ (nominative) stol je ‘table’ (locative)
vkus ‘taste’ (noun) vkus jen ‘tasty’ (adjective)
dar ‘gift’ (noun) dar jit ' ‘s/he gives’ (verb)
dom ‘house’ dom jiʃko ‘cottage’
bomba ‘bomb’ (noun) bomb jit ‘s/he bombs’ (verb)
All the examples of assimilation so far have involved immediately adjacent sounds,
but it is also possible for non-contiguous sounds to assimilate. For example, Old
French had the verb cercher /sertʃer/ ‘search, look for’ (from which English borrowed
search); the expected modern form is /serʃe/, but instead we find chercher /ʃerʃe/. Dis-
tance assimilation between consonants is rare, but between vowels is more usual. It is
generally called vowel harmony but specifically umlaut when the change is anticipa-
tory, i.e. when vowels assimilate to those in a following syllable.
Umlaut is found in many languages including the Germanic languages, Irish and
Romanian (see Hock 1991: 66–68 for examples). English has only relics of the pro-
cess in the form of irregular plural nouns such as mouse-mice, tooth-teeth, man-men
and so on. They are all that remains of an earlier pronunciation rule (known as
i-mutation, already mentioned in Chapter 1) whereby the stem vowel of the word
harmonized with the vowel of the -i plural ending (see Table 4.6).
As the table shows, when the ending (which triggered the change) eventually
dropped off, we were left with a strange looking alternation that through later
changes eventually became the modern mouse-mice. Umlaut is responsible for
many unusual vowel alternations in Modern English; e.g. relic forms like old-elder-
eldest show the mutation that was triggered by the lost endings *-iro and *-isto.
Similarly, the vowels in the second member of pairs like full-fill, food-feed, lie-lay,
fall-fell and drink-drench are the result of assimilation with a palatal semi-vowel /j/
in the now lost causative suffix -jan (you can find more of these pairs in Emerson’s
1921 lavish account of vowel mutation in early English).
In Modern German the umlaut pattern is much more entrenched than in English, to
the extent that the vowel change brought about by umlaut has its own marker – two dots
above the vowel. Vowels written as a, o, u and au become ä, ö, ü and äu (see Table 4.7).
While umlaut is still alive in German, it is no longer the case that all high front
vowels produce umlauted vowels, e.g. Amt ‘office’ —> amtlich ‘official’ (*ämtlich);
umlauted vowels also show up in a wider range of environments (the trigger vowels
aren’t restricted to high front vowels), e.g. the -er plural in Mann ‘man’ —> Männer.
Umlaut illustrates one of the great puzzles for historical linguists – how (and why)
do close siblings end up diverging in this way? In this case, why did umlaut bite the
dust in English but not German?
Table 4.6
i-mutation in Germanic
Singular Plural
Proto-Germanic *mu-s- *mu-siz
Pre-Old English assimilation *mu-s /mu:s/ *mu-si /mu:si/
*mu-s /mu:s/ *my-si /my:si/ /y/ = high fronted rounded vowel
Old English mu-s /mu:s/ my-s /my:s/
Middle English mus /mu:s/ mis /mi:s/
Modern English mouse /maus/ mice /maɪs/
Modern German Maus /maus/ Mäuse /moɪsə/
Table 4.7
Umlaut in German
Basic forms Umlauted forms

Mann /man/ ‘man’ männlich /mɛnliç/ ‘manly’
Stunde /ʃtundə/ ‘hour’ stündlich /ʃtyndliç/ ‘hourly’
gut /gu:t/ ‘good’ gütig /gy:tiç/ ‘kind’
Not /no:t/ ‘need’ nötig /nø:tiç/ ‘necessary’
Gott /got/ ‘god’ Göttin /gø:tin/ ‘goddess’
Hund /hunt/ ‘dog’ Hündin /hyndin/ ‘bitch’
Orangutangs and pengwings

Some of you might have come across references to Benedict Cumberbatch’s
voice-over for the BBC documentary South Pacific, in particular the part
where he repeatedly pronounces the word penguin incorrectly. It comes out
as something like pengwing:
So why are these woodlands so attractive to pengwings? A fresh water

stream through the forest makes a handy highway for a parent peng-
wing heading home from a fishing trip [. . .]
You can listen to the clip here: https://www.youtube.com/watch?v=-GnLD

JAgrws.
We are thrilled by this example because it illustrates this rather rare pho-
nological process of long-distance assimilation. Ordinary assimilation is
commonplace, but it’s much more interesting when non-contiguous sounds
harmonize, especially consonants. You might find examples in the speech of
young children (dog occasionally becomes gog for example), and in random
slips of the tongue in adult speech, such as one of the authors’ recent fear-
some phenomenom when she meant to say fearsome phenomenon (of course
it didn’t help that there’s a double whammy of m’s here). Where you are more
likely to get long-distance consonant harmony is where syllables are similar
and the resulting form is almost a reduplication. Smorgasbord is sometimes
heard as smorgasborg, and orangutang is now probably more usual than origi-
nal orangutan. And now there’s pengwing.
But these remain sporadic changes that haven’t yet made any long-term
systematic impression on the language. Probably none of us is tempted by
wingspang for wingspan or hangmang for hangman. The written word does
slow down these sorts of changes, but orangutang does seem to have stuck –
it’s been recorded since the word first entered the language in the 1600s (orig-
inally Malay orang ‘person’ + utan ‘forest’). In fact, the example is repeated
over and over in linguistics textbooks because long-distance consonant har-

mony is so rare!
Interestingly, there’s evidence that pengwings have been around for a bit too.
You’ll find an entry in Urban Dictionary dating from early 2007 (http://www.
urbandictionary.com/define.php?term=pengwing&utm_source=search-
action). But will it endure? Well, it seems we like to turn words into rhymes
(smorgasborgs and orangutangs), and in the case of pengwing there’s the
added motivation that the remodelled worded nicely captures those rather
rudimentary little wings so much a feature of the penguin – pengwing is apt.
And then of course nothing beats celebrity – whether it’s the promotion
of products or linguistic forms. The actor, whose beautiful and authoritative
voice journalist Caitlin Moran once likened to “a jaguar hiding in a cello”, says
pengwing (The Times, August 5, 2010). What’s more, none of those penguin
experts working on the documentary ever challenged him! And in the special
way the internet can make things happen, Cumberbatch’s gaffe has now a
worldwide profile. There are even T-shirts to be had.
Things are looking good for pengwings.
Lenition
Lenition is the name given to the general weakening of sounds. Via this process,
segments involve a reduced use of articulators, and they might be shortened – the
end point of lenition is loss. Such changes occur when sounds find themselves in
weakened prosodic positions. As we discussed in Chapter 1, the history of English
shows the gradual erosion (or lenition) of grammatical endings (the stress shift in
early Germanic to the initial syllable had left these endings vulnerable and prone to
reductive processes). Two main changes took place:
1 /a, o, u, e/ > /ə/

The vowels reduced to the indeterminate weak vowel, and originally distinct
endings like -an, -en, -on, -um became unclear; compare the pronunciation of
Belgian and Belgium today.
2 /m, n/ > ø finally
Final nasals dropped off.
To illustrate, consider the stān ‘stone’ paradigm (based on Bynon 1977), shown
in Table 4.8.
As you can see, sound changes operate without any regard for grammar. The
inflectional endings were seriously disturbed by these changes; the eventual loss of
the final schwa meant that now only two forms survived: stone /stoʊn/ and stones
/stoʊnz/.
Table 4.8
Changes to the ‘stone’ paradigm
Old English Forms after changes Modern English

Nominative/accusative sta-n sta-n stone
Dative sta-ne sta-ne
Genitive sta-nes sta-nes stones (possessive)
Nominative/accusative sta-nus sta-nes stones

Dative sta-num sta-ne
Genitive sta-na sta-ne
Consonants can be described as stronger or weaker according to a number of

different scales; they are particularly prone to lenition, especially when they are
flanked by vowels and find themselves becoming more vowel-like. So lenition is
often the fall-out of the very same assimilatory processes we’ve just discussed.
Table 4.9 shows some fairly typical trajectories of more > less articulatory effort
(adapted from McColl Millar 2015: 51).
Table 4.9
Some typical paths for lenition
Stronger > weaker Articulatory effect of weakening

1 Geminate C > simple C less obstruction of airflow, reduced duration
2 Stop > fricative > approximant less obstruction of airflow, economy of gesture
3 Stop > liquid > approximant less obstruction of airflow, economy of gesture
4 Oral stop > glottal stop economy of gesture
5 Non-nasal > nasal less obstruction of airflow through nasal cavity
6 Voiceless > voiced easier transition in the proximity of vowels
German offers another clear example of lenition. At the end of words speakers
devoice stops and fricatives (think of this as a kind of assimilatory change with
the voiced consonants assimilating to the voicelessness of the following silence), as
shown in Table 4.10. We can date the devoicing to the medieval period, because of
Middle High German spellings such as rat ‘wheel’ (versus rades ‘of the wheel’) and
tac ‘day’ (versus tages ‘of the day’).
Table 4.10
Final consonant devoicing in German
Old High German Modern German Translation

tag /tag/ Tag /ta:k/ ‘day’
rad /rad/ Rad /ra:t/ ‘wheel’
hund /hund/ Hund /hunt/ ‘dog’
bad /bad/ Bad /ba:t/ ‘bath’
Weakening of unstressed grammatical words

The following is a letter that one of us received from five-year-old Isaac
Broadhead:
My name is Isaac and I am 5 years old. I am learning to read and

think there is something a bit funny about the English language. Can
you tell me why words like ‘is’ and ‘as’ are spelt with an ‘s’ and not a
‘z’ – like they sound? Also, why the word ‘of ’ is spelt with an ‘f ’ and not
the ‘v’ that it sounds like.
Isaac’s complaint about the behaviour of is, as and of offers another exam-
ple of lenition – and we could also throw into the mix other functional words
such as was, has and his. These short grammatical words were all once pro-
nounced with the voiceless sounds still preserved in the spelling (hence
Isaac’s frustration), but via a weakening process typical of unstressed words,
they became voiced (sometime during the 1500s). Interestingly, the original
pronunciation of of is preserved in the related preposition off – both were
simply variant pronunciations (compare with /wɪθ/ and /wɪð/). We might
point out that if and us also used to have voiced variants (and still do in dia-
lects of northern England), but the pronunciation ended up falling in line
with the spelling.
Dissimilation
Dissimilation, as the name suggests, describes how sounds change to become
less like other sounds in the vicinity. The reason behind dissimilation is not clear,
though some linguists attribute it to the fact that humans can find repetition of the
same muscular activity problematic. You might have experienced the difficulty of
repeating the same action over and over again (e.g. whether it’s stuffing envelopes,
playing the same note over again or repeating a sound in a tongue twister like “In
me, many an enemy anemone enema”, which has too many nasals) – suddenly you
find yourself doing something weird, perhaps substituting a totally different action.
Similarly, in a sequence of repeated sounds, we can avoid the difficulty by replacing
one of the sounds (which is of course what often happens in tongue twisters).
Dissimilation is behind the non-standard pronunciation /ɛksɛtərə/ for etcetera
/ɛtsɛtərə/ (with added reinforcement coming from other words beginning with ex-
/ɛks/). The Modern German word for ‘six’ sechs is pronounced in the standard as /zeks/;
the first of the two fricatives /xs/ has dissimilated to a stop /k/. The native dialect of one
of the authors, however, has retained the original fricatives: /sexs/. Nasals and liquids
(like /l/ and /r/) are more prone to this process than other sounds. For example, some
English speakers dissimilate the second nasal in chimney —> chimley. The Russian word
Table 4.11
Examples of dissimilation in English and German
Original word Modern English Modern German

randon (Old French) random —
rainson (Old French) ransom —
cardamo-mum (Latin) cardamon (non-standard) Kardamom
peregrine (Latin) pilgrim Pilger
(compare peregrine ‘falcon’)
purpure (Latin) purple purpur
purpure (heraldry term)
marmor (Latin) marble Marmor
tartoffel ‘potato’ (Old High German) — Kartoffel
for ‘February’ fevral´ /f ʲevrˈalʲ/ (< Latin Februārius) shows dissimilation of the second
/r/; something similar happens to English February (the first /r/ changes to /j/ as in
Febyuary).
As examples in Table 4.11 show, dissimilatory changes can involve preceding or
following segments (though again anticipation is more common than lag). Here we
have bolded the consonants that have dissimilated to highlight the rather unsys-
tematic nature of the process.
In English the second nasal in a sequence was changed to a nasal in a different
place of articulation. In a sequence of /r . . . r/, English (and occasionally German)
changed one /r/ to an /l/. German Kartoffel shows a more unusual dissimilation
involving stops (the first /t/ > /k/).
All of the above changes are sporadic (even successions of nasals and rhotics
won’t regularly prompt dissimilation in this way). Only occasionally do you find
examples of regular dissimilation. Recall our discussion of Grassmann’s Law in
Chapter 1. Hermann Grassmann discovered that in a sequence of aspirated conso-
nants, the first one lost the aspiration. So in the reduplicated forms for the perfect
tense in Greek, the consonant was de-aspirated if the initial consonant was aspi-
rated: the original form thi-thē-mi ‘I put’ (with a reduplicated prefix) became in
Greek tí-thē-mi. While energy expenditure is difficult to measure, there is specula-
tion that the motivation for this law is economy of effort – the fact that aspirated
consonants require considerable respiratory exertion means that successive aspira-
tion will be difficult and therefore pruned back.
Metathesis
These days the pronunciation anenome is more common than the historically accu-
rate anemone. The process that has switched the nasals around in this way is called
metathesis (or sometimes jokingly methatesis). The cause is some sort of mistim-
ing or miscoding, and it is common where difficult combinations of sounds are
involved. English examples frequently involve those notoriously unstable sonorous
consonants that have loomed large in many of the processes so far – liquids, rhotics
and nasals; e.g. renumeration for remuneration, aminal for animal and emeny in
place of enemy.
Sometimes the switching leads to arrangements of sounds that better fit into pat-
terns of existing words (compare anenome / an enemy; renumeration / numeral;
aminal / criminal). But switching sounds doesn’t always ensure more straightfor-
ward pronunciations – not too many words seem to rhyme with emeny for instance
(though for some speakers perhaps hegemony, and maybe even lemony). And some-
times both forms happily coexist: German has both the metathesized and origi-
nal forms of the proper name Birgit/Brigitte (compare English Bridget); both crud
‘dirt’ and metathesized curd ‘coagulated milk’ continue in English, as do task and tax
(originally variant pronunciations of the same word).
This last example shows other typical shifters, namely s-clusters like /sk/ and /sp/:
asterix often appears instead of asterisk; pasghetti instead of spaghetti. These forms
can remain childhood errors, or one-off slips of the tongue, but when a number
of speakers keep making the same sorts of adjustments, tongue slips can endure.
Table 4.12 gives some examples in English and German where the slips (over-
whelmingly r-metathesis) have survived and become permanent fixtures in one or
both of the languages.
Table 4.12
Metathesis in English and German
Proto-Germanic Modern English Modern German

*þurih through durch
*horso- horse (compare walrus ‘whale horse’) Ross
*þridja- third (compare three) Dritte
*brennan/*brannjan burn brennen
*wops- wasp Wespe
*aiskon ask heischen (shows palatalization)
aks (non-standard)
Metathesis is typically sporadic and not productive; however, occasionally the

addition of morphemes can trigger a regular reordering of the segments. Hanu-
noo (Philippines) shows a regular switching of segments in numerals, as shown in
Table 4.13.
Table 4.13
Metathesis in Hanunoo
Number Number + ‘times’

ʔusa ‘one’ kasʔa ‘once’
duwa ‘two’ kadwa ‘twice, two times’
tulu ‘three’ katlu ‘thrice, three times’
ʔupat ‘four’ kapʔat ‘four times’
lima ‘five’ kalima ‘five times’
ʔunum ‘six’ kanʔum ‘six times’
pitu ‘seven’ kapitu ‘seven times’
In the first example, the addition of the ka- prefix ‘times’ causes the /u/ to delete;
the glottal stop then swaps places with the following consonant, as in ka+ʔsa —>
kasʔa (but note that this does not involve other consonants (e.g. ka + tlu —> katlu).
4.2 PHONETIC VERSUS PHONEMIC CHANGE
We can view sound change from two perspectives: the phonetic (the physical
aspects of sounds) and the phonemic (the sound system). Phonetics gives us a way
to describe the phones of languages such as English and German and how they
differ; so it deals with unprocessed, raw sounds – or phones. The phonemic per-
spective considers the structure of sounds and how they function within a system;
it identifies the meaningful sound units (phonemes or distinctive sounds) of a lan-
guage and their variant pronunciations (or allophones).
4.2.1 Phonetic change – changes in accent

Phonetic changes involve adjustments to the phonetic realization of a phoneme, or
to the distribution of the allophones of a phoneme, but cause no disruption in the
number of phonemes or in the relationships between them.
The history of English shows a number of these phonetic changes. We have exam-
ples of alterations to the environments where allophones of a phoneme occur and
examples where even the phonetic form of a phoneme has altered. These are simply
pronunciation changes; they have no impact on the overall sound system.
To illustrate, the oral and nasal stops /t, d, n/, which are now pronounced as
alveolar sounds, were thought to have been dental stops; however, from around
the time of Chaucer (about 1400), the articulation of these sounds started to move
progressively further back in the mouth (see Anttila 1989: Chapter 4). Although
the shape of the sounds has changed acoustically, this hasn’t done anything to
change the phonemic system of English; /t, d/ still contrast with the other stops
/p, b/ and /k, g/, and they distinguish the same words just as they did in Chaucer’s
time. This is pure phonetic change and can be represented as follows:
[t̪ , d̪ , n̪] > [t, d, n]
The pronunciation of /r/ offers another example. The approximant that appears in
many varieties of English is believed to have originated in a flap or trill, as in Scots
English today (trills are sequences of rapid stops, with the tip of the tongue vibrat-
ing against the roof of the mouth; a flap is a single trill). Again this is just a simple
change in pronunciation; nothing has altered in terms of the overall system.
These examples involve the phonetic shape of a phoneme in all environments
in which it appears, but changes can also occur in restricted environments. We
know from spelling and from other sources (e.g. historical descriptions and mod-
ern dialects) that at one time speakers of British English pronounced /r/ wherever
it appeared in a word. Ben Jonson, writing in 1640, said about r: “It’s the dog’s letter,
and hurreth in the sound; the tongue striking the inner palate, with a trembling
about the teeth. It is sounded firm in the beginning of the words, and more liquid in
the middle and ends; as in rarer, viper” (Jonson 1640: 773). Sometime between 1600
and 1700, the pronunciation changed, and /r/ vanished except in positions where it
appeared before vowels, e.g. her /hɜ/ sausage versus her /hɜr/ appetite. The distinc-
tion between r-less (or non-rhotic) and r-full (or rhotic) varieties is now one of the
great divides in English dialectology.
R-less dialects of English still have a phoneme /r/, but the rule for its pronuncia-
tion has changed:
/r/ > [r] before vowels

ø everywhere else
This simply says: pronounce /r/ before vowels, but delete it everywhere else.
These sorts of phonetic changes are going on all the time, but we don’t always
know much about them, because they aren’t usually reflected in writing systems,
especially if spelling is regularized.
4.2.2 Phonemic change – changes that alter the sound system

We now turn our attention to the types of changes that lead to some sort of altera-
tion in the phonemic system of a language. They fall into two broad types: mergers
and splits.
Merger
When two (or more) separate phonemes combine to become a single phoneme,
this is known as a merger. The front rounded vowel /y:/ in Old English ended
up unrounding and merging with /i:/. This is a complete merger; i.e. the change
affects /y:/ and /i:/ in all environments. We can represent complete merging as in
Figure 4.1.
This simple diagram is a little misleading in that it suggests the original two
sounds have merged into a different sound, whereas the resulting phoneme can be
one of the original merging phonemes; i.e. either /x/ or /y/ (as the /y:/-/i:/ merger
shows).
Much more usual is a partial merger. This means that the sound change is condi-
tioned; i.e. the phonemes merge only in certain environments and are kept distinct
in others. We can represent it as in Figure 4.2.
Figure 4.1
Complete merger
/x/
/z/
/y/
Figure 4.2
Partial merger
/x/ /x /
/y/ /y/ (in certain environments)
In the Englishes around the world, there is a lot of vowel merging going on, espe-
cially in the environment before /l/. Currently, many younger speakers in New Zea-
land and Australia are neutralizing /e/ and /æ/ before laterals; for these speakers,
shell and shall are indistinguishable. There is a parallel phenomenon for /o/ and
/oʊ/, as well as /ʊ/ and /u/, so these speakers don’t distinguish doll and dole or fool
and full. In New Zealand this merging is extending to /ɪ/ and /ʊ/, also before later-
als, so that pill and pull are homophones (meaning that for some New Zealanders
the /ɪ/ and /u/ vowels may also be indistinguishable, as in pill and pool).
Split
Primary splits occur when allophones of a phoneme split off from that phoneme
and merge with a different phoneme. So words that originally contained the same
phoneme end up with different ones. In Old High German there was a fricative
allophone of /t/ which merged (partially) with /s/ between vowels (e.g. Old High
German ezzan > Modern German essen). So what we have is a re-alignment in the
system of phonological contrasts (as in Figure 4.2 for partial merger).
A similar example is something called Latin rhotacism. Pre-Latin had phonemes
/s/ and /r/; a change caused /s/ to become /z/ between vowels, and later this shifted to
/r/ (a prime example of lenition): flos ‘flower’ but flores ‘flowers’ (formerly floses). In
other words, /s/ split into /r/ between vowels (thus merging with existing /r/) and /s/
everywhere else; no new phonemes results, but the distribution of /s/ and /r/ changed.
With a primary split the overall inventory of phonemes doesn’t change; there is
simply a reorganization of the phonemes (more occurrences of a particular pho-
neme, but no new phoneme). More interesting are secondary splits where a pho-
neme splits, but there’s no merging going – the split causes the introduction of a
new phoneme. (As with a lot of this terminology, don’t ponder too much on the
labels “primary” and “secondary” – they are not entirely clear.) This can be repre-
sented as in Figure 4.3.
Figure 4.3
Primary split (split + merger in Old High German)
/t/ /t/
/s/ /s / (between vowels)

Figure 4.4
Secondary split
/x/
/x/
/y/
What happens here is that there is a change in the conditioning features of allo-
phones. To illustrate, we take the evolution of a new nasal consonant in English and
German. Both languages currently have three nasal phonemes:
Bilabial Dental Velar

m n ŋ
On the basis of evidence from dialect variation, spelling, rhyme, early descrip-
tions and our knowledge of sound change, we know the velar phoneme began life
as a pronunciation variant of /n/, occurring only before other velar sounds (most of
us pronounce pancake as /pæŋkeik/). In other words, /ŋ/ was simply an allophone
of /n/. So an English word like sing was pronounced /sɪŋg/. The pronunciation rule
was something like this:
Early English and German pronunciation of /n/

/n/ —> [ŋ] before velars {k, g}
[n] everywhere else
How, then, did the velar nasal acquire its new status as a phoneme? Sometime
during the history of both languages, stops fell off the end of nasal+stop clusters. So
the pronunciation rule was:
/b, d, g/ —> ø after nasals (in final position)
It is this rule that accounts for the silent <b> in lamb and climb, and the modern
pronunciation of sing and long with final /ŋ/, not /ŋg/ (i.e. /sɪŋg/ > /sɪŋ/). English
regional pronunciations of sing as /sɪŋg/ are relics of this earlier pronunciation, as
are German dialect pronunciations of lang ‘long’ and jung ‘young’ with a final /k/
(compare relic English pronunciations nothink and everythink).
Once /g/ disappeared from words like sing, the velar nasal (now in word-final
position) came to be in contrastive distribution with the other nasals in this posi-
tion – some, sun and sung now make a minimal group for English. The meaning
contrast between these words depends entirely on the distinction between the
nasals. As a relatively new phoneme, /ŋ/ shows a fairly restricted distribution; it
can only appear at the end of syllables, either by itself (as ring, singer) or before
velar stops (as in finger, sink). But borrowings like ngara ‘New Zealand lizard’ (from
Māori) are helping it to spread its wings.
As a final illustration, return to an example we gave in Chapter 1. While Old
English did have voiced fricatives [v, ð, z] they weren’t independent phonemes.
Speakers produced them automatically whenever the fricatives were flanked by
vowels (e.g. if /f/ found itself between two vowels, it assimilated to the voicing on
either side and became [v]). So in Old English there was a pronunciation rule that
said:
/f/ —> [v] between vowels

[f] everywhere else
Recall that French borrowings like veal placed the voiced fricative into initial
position, so that it contrasted with /f/ (feel versus veal). The loss of final vowels
in words like save also had the effect of introducing the voiced fricative into final
position. And so in this way /f/ split into two phonemes, /f/ and /v/, as shown in
Figure 4.5.
Figure 4.5
Secondary split in early English
/f/
/f/
/v/
The separate status of /z/ and /ð/ as phonemes also came about because of
changes occurring elsewhere in English. Modern pairs of words like grass/graze
and bath/bathe show voiced and voiceless alternating final consonants. The final e
in the spelling preserves the original trigger for voicing these fricatives. With the
loss of these vowels, however, the fricatives suddenly found themselves word finally,
contrasting with the voiceless set, e.g. grass /gras/ versus graze /greiz/; bath /baθ/
versus bathe /beið/. And so it was that English acquired its series of voiced fricative
phonemes /v, ð, z/ to match the voiceless series /f, θ, s/.
So you can see how the phonological processes that are going on all the time in
ordinary speech can have a significant impact on the sound system. Fast speech
phenomena like assimilation and loss can lead to the demise of phonemes but also
the creation of new ones.
4.3 ON EXCEPTIONS IN SOUND CHANGE
As we will discuss more fully in Chapter 9, an important principle of historical

linguistics is the orderliness of sound change; in other words, the idea that changes
such as the ones we’ve been seeing in this chapter are not capricious but regular. You
might find this puzzling. Some of those examples of dissimilation and metathesis
were anything but regular – and for the most part these sorts of changes are irregu-
lar or sporadic. However, the majority of changes are not of this nature.
The absolute regularity of sound change, or what came to be known as the “regularity
hypothesis”, was a major tenet of the Neogrammarian school of linguists (or Junggram-
matiker ‘Young Grammarians’, to use their German nickname) in the late 19th century
and remains a cornerstone of the discipline. These linguists even described them as sound
laws (or Lautgesetze). Accordingly, the sound system of any language, as it developed
through time, was subject to the operation of sound laws, which were understood to be
absolutely regular in their operation, except where words deviate from their “proper”
patterns under the influences of non-phonetic factors. Below are some of these non-
phonetic processes that interfere with the otherwise regular outcome of sound change:
1 Analogical change (by meaning)

Speakers perceive a similarity between two forms on the basis of meaning
(no formal similarity) and then force the forms to be phonetically alike (it’s
rather like extreme long-distance assimilation). For example, femelle changed
to female on analogy with male; grine changed to groin on analogy with loin.
Numerals are particularly prone to this sort of assimilation; for example, Latin
novem ‘nine’ should be noven but was remodelled on analogy with decem ‘ten’.
2 Analogical change (by form)
Analogical change can operate without any similarities in meaning, but purely
on the basis of partial similarities in form; e.g. the reassignment of boundar-
ies, as where the n of indefinite articles or pronouns is wrongly interpreted as
belonging to the following noun. The word nickname comes from ekename (lit.
‘also-name’ or ‘additional name’). Since, in speech, there’s no difference between
an ekename and a nekename, the n shifted across and attached to the following
word (compare colloquial a nother from an other).
3 Folk etymology
Folk etymology (as discussed in Chapter 3) is rather similar; this is where
speakers misunderstand a word and render it into a more familiar looking
form. Often the remodelling is semantically justified; e.g. bridegroom < bryd-
guma ‘espoused man’ (compare German Braütigam) on analogy with groom.
4 Spelling pronunciation
Spelling pronunciation occurs where speakers change the pronunciation of words
to fit more closely with the spelling; the effect is the reversal of the sound change.
Often usually receives a spelling pronunciation rather than the t-less pronunciation
typical of the 18th and 19th centuries (compare soften and glisten). The English
words sure and sugar show the palatalization you would predict from their history.
Earlier, suit, consume and sue were similarly pronounced, but the non-palatalized
pronunciations have been restored because of the influence of spelling.
5 Onomatopoeia
All languages have words that imitate sounds. Vocabulary of this sort often resists
otherwise regular sound change. For example, Middle English pipen (sound made
by little bird chicks) should have become Modern English pipe /paip/ (as did the
tube people use to smoke), but it remained peep (to better echo the chirping sound).
6 Taboo
Irregularities can occur when speakers distort the pronunciation of words for rea-
sons of taboo. For example, the vowel in ass ‘donkey’ didn’t change to /a/, as it did in
words like grass, because it would have clashed with the bawdy body part arse /a:s/.
These counter-agents to regular sound change (such as the association processes

driving analogical reformulations) operate without threatening the validity of
sound laws as purely mechanical physiological processes. But there is still another
aspect to this whole notion of the exceptionless sound laws that we need to explore
here. Take the example given earlier of the Law of Latin Rhotacism:
/s/ > /r/ between vowels
It is a classic example of lenition; examples include forms like est/erit (present and
future forms for ‘be’).
Sound change is supposed to be exceptionless, but there are a number of famous
exceptions to the law – Latin words that show /s/ between vowels. So why didn’t the
following words change to fall in line with the others?
rosa ‘rose’
causa ‘cause’
miser ‘miserable’
soror ‘sister’
The label “law” is misleading. A sound change is not so much a law (like the law
of gravity) but more like a historical event – something that occurs “at a certain time
in a certain language under certain conditions”. Edward Sturtevant in 1942 offered
a rather interesting analogy here. He likened the Law of Latin Rhotacism to a ficti-
tious historical event that he dubbed the Law of Waterloo:
All Prussians six feet tall were killed in the Battle of Waterloo
And he then proceeded to use this analogy to explain the apparent exceptions to Latin
rhotacism. As he demonstrated, these do not fall under the conditions of the “law”.
Like historical events, sound changes have a natural term, and Latin rhotacism
was done and dusted by the 4th century BCE. After this time, a number of words
like rose ‘rose’ were borrowed into the language (e.g. from Greek), and though they
had intervocalic /s/, they weren’t affected by rhotacism – it was too late. To use Stur-
tevant’s analogy, they were Prussians naturalized after the battle.
Apparent exceptions like causa ‘cause’ originally had geminate /ss/ (Old Latin
caussa), and because geminate consonants are stronger, they could resist the weak-
ening to /r/. Only later did they weaken to /s/. To use Sturtevant’s analogy, they were
Prussians not yet born at the time of the battle.
Slightly more complicated are the words like miser ‘miserable’ (and also caesaries
‘hair’). They didn’t undergo rhotacism because the following /r/ blocked the change
from happening (recall that speakers generally don’t like too many liquids in the
one word). To use Sturtevant’s analogy, they were Prussians under six feet tall and
therefore not subject to the Law of Waterloo.
This leaves us with soror ‘sister’ – why not sosor as predicted by the previous
group of words? As you might have guessed, this time it’s the presence of the initial
/s/ that allowed the change to sneak through. These were Prussians who were six
feet tall, but their posture was stooped!
Like many analogies, you don’t want to push this one too far. Nonetheless, it
nicely illustrates two important aspects to sound change. One is that a sound
change takes place at a particular time in the history of a language, and only then.
So it’s like an event in history (or perhaps like an epidemic disease that works its
way through a community and then peters out, as we will see in Chapter 7). Second,
sound changes aren’t whimsical but regularly conditioned, and we can usually find
principles behind groups of apparent exceptions. Words that enter the language
later won’t undergo a previous change; other sound changes can obscure the full
story; sound changes can even be thwarted by competing changes. And of course
there are other factors we haven’t been able to consider here involving the progress
of phonological changes, both socially and geographically (see Chapter 7).
4.4 WHY DO SOUNDS CHANGE?
Over time sounds will vanish, brand new ones will appear on the scene, and old
ones will change their shape. Why does this happen? The motives that have been
proposed over the years to account for consonant and vowel changes are many and
varied.
Early theories linked sound change with changes in the anatomical structure
of the organs of articulation; e.g. the Australian accent has been attributed to a
national nose inflammation (through excessive amounts of pollen or hay), bad
dentistry and even excessive alcohol consumption early on in the colony’s history.
Ethnic character was another popular theory; e.g. a national inferiority complex, a
free-wheeling and adventurous spirit and an outlaw heritage have all been put for-
ward at some time to account for the Australian accent. Early linguists also argued
that geographical and climatic conditions influence sound systems, especially the
notion that somehow harsh climates produced harsh sounds (harsh sounds were
never clearly defined, though “guttural” sounds were often implied). There is no
shred of evidence for any these theories.
Change is never one-dimensional. At any one time, languages are being influ-
enced by a network of intersecting pressures that work to bring about change:
physiological (the production of language), systemic (the linguistic system with
interacting components), psychological (the mental make-up of speakers), social
and political (the speech community and the individual, the socio-political envi-
ronment) and external (contact and borrowing). Here we touch on just some of
these factors.
4.4.1 Simplicity
All languages show sounds dropping out, merging or assimilating over time, and these
sorts of changes do involve a lessening in muscular effort. They’ve been described as
economy of effort, following the line of least resistance, human laziness, sloth – but the
end result is also a more efficient, more streamlined production. So yes, many changes
do result in simplification. However, it isn’t this straightforward. English speakers
stopped pronouncing the /k/ in knee [ni:] presumably because it was easier to say it
this way – so why didn’t German speakers do the same in Knie [kni:] ‘knee’? Simplic-
ity also can’t account for the fact that many changes result in greater complexity; e.g.
the English consonants /ð/ and /θ/ (rather rare sounds) evolved from simple stops.
A change might also simplify one part of the language but introduce complications
elsewhere. You saw earlier how many of the irregularities in Modern English grammar
(such as was-were, foot-feet and wife-wives) were the fall-out of regular sound changes
that occurred. In short, when sounds drop out or assimilate, this can contribute to
greater complexity overall. Simplicity is a factor, but it can’t be the only one.
4.4.2 Contact induced change

We’ve seen that when languages borrow words from outside, this can influence the
sound system. Also important are substratum effects. Think of the Celtic-English
contact situation, where the original inhabitants (Celtic speaking) adopted the
language of the newcomers (English speaking). In such situations there is usu-
ally a period of bilingualism when the locals speak the new language but with
some interference from their primary language. If some of these elements from
the primary language are transmitted to later generations of speakers of the pre-
vailing language, this is described as a substratum of that language. Typically the
substratum modifies the phonology of the adopted language, and we look more
closely at examples in Chapter 8.
4.4.3 Structural pressure

Many sound changes occur that bring about greater symmetry in the vowel and
consonant systems of a language. Consider the evolution of English consonants.
Below we have reproduced a rough consonant inventory for the three major stages
of English: Old, Middle and Modern. Compare the three systems and take note of
the bolded new phonemes at each stage.
Old English Middle English Modern English

p t tʃ k p t tʃ k p t tʃ k
b d dʒ g b d dʒ g b d dʒ g
m n m n m n ŋ
f θ s ʃ x h f θ s ʃ x h f θ s ʃ h
v ð z v ð z ʒ
w l r j w l r j w l r j
Each change has had the effect of bringing about greater symmetry within the
system. Languages tend to line up their consonants in pairs – voiceless sounds are
usually matched with voiced ones, and oral stops with nasal stops. While we can’t
go as far as saying that all sound change brings about a more balanced sound sys-
tem, many changes in English and other languages have either helped to fill gaps or
eliminated them somehow.
Good examples of structural pressure driving change are what are called chain
shifts. It is well attested that a change to one vowel can affect the system overall. One
vowel moves, and this sets off a kind of chain reaction, with neighbouring vowels
all shifting in solidarity (presumably driven by the requirement for intelligibility).
A famous example is what Danish linguist Otto Jespersen dubbed “The Great Eng-
lish Vowel Shift”: sometime between Chaucer and Shakespeare, the long vowels /i:/,
/e:/, /ɛ:/, /a:/, /u:/, /o:/ and /ɔ:/ came to be articulated with a higher tongue position;
/i:/ and /u:/ became diphthongs (these high vowels can’t rise any further without
becoming consonants) (see Figure 4.6). Further changes then took place. In Shake-
speare’s time the modern diphthongs in mice and house were more like the modern
Canadian /əi/ and /əʊ/; /ɛ:/ and /o:/ later diphthongized to /eɪ/ and /oʊ/, and some-
time during the 18th century the original mid vowel /e:/ (in meat) changed to /i:/.2
Middle English Chaucer Shakespeare Modern English

/i:/ [mi:s] [məis] [mais] ‘mice’
/e:/ [me:t] [mi:t] [mi:t] ‘meet’
/ɛ:/ [mɛ:t] [me:t] [mi:t] ‘meat’
/a:/ [na:mə] [nɛ:m] [neɪm] ‘name’
/u:/ [hu:s] [həʊs] [haʊs] ‘house’
/o:/ [go:s] [gu:s] [gu:s] ‘goose’
/ɔ:/ [stɔ:nə] [sto:n] [stoʊn] ‘stone’
Because it happened so early, it’s difficult to determine which was the rogue vowel
here. One possibility is a drag (or pull) chain. One vowel moves, leaving a gap, and
this has the effect of pulling another vowel in to fill the gap etc. Another possibility
Figure 4.6
The Great English Vowel Shift
i: ‫׀‬ u:
e: o:
ε: ɔ:
a:
is a push chain. One vowel invades the territory of another vowel, pushing it into
the territory of yet another etc. Most evidence (by way of rhymes, spellings and
borrowings) points to a pull chain, with /i:/ as the possible initial trigger (but the
mystery is still, of course, why did this vowel move in the first place?).
Southern hemisphere varieties of English are currently going through short front
vowel raising relative to many British and American dialects (the second Great
Vowel Shift; see Bauer 1992). For New Zealand English, historical evidence suggests
a “push chain” effect; the already raised /æ/ vowel (in words like trap) of the colo-
nial input in the 19th century then went on to influence the /ɛ/ and /ɪ/ vowels (in
dress and kit). Australian English was going the same route; however, the past few
decades have seen an about-face with vowels now moving in the opposite direction
to the New Zealand change (in other words, they are lowering, not rising). Regional
chauvinism is a major incentive for people to start highlighting their distinctiveness
linguistically, and perhaps the strong rivalry between the two countries, combined
with the substantial physical distance, is one reason for this unexpected reversal of
sound change. This then leads us nicely on to social factors as motivators for change.
4.4.4 Social change

Early linguists noted that social upheaval (epidemics, civil wars, revolutions, eco-
nomic crises) appeared to go hand-in-hand with periods of dramatic linguistic
change. For example, the Black Death coincided with a number of major changes
that took place in English around that time. This was an epidemic that brought
death on a scale comparable with a nuclear holocaust, and was made all the worse
by a number of subsequent epidemics. Such devastation triggered mass emotional
disorders and would have torn social networks apart. As we’ll see more in Chap-
ter 7, it is precisely at such tumultuous times that variants are able to take off, spread
and eventually embed themselves as long-term changes in the language system. It
is as if changes are always lurking in the wings, but it is moments of social mayhem
that enable them to catch on and flourish.
With the massive social changes that are accompanying the current technological
revolution, you might contemplate how the internet, in particular social media, is
now changing language. On the internet people converse with each other in virtual
communities just as they do face-to-face in real communities – except of course
they are writing, not speaking. They are doing this to convey information, but never
lose sight of the second important function of language – to define the gang. For
members of virtual speech communities to express their identity (both personal and
shared with other members of the group), they have to break away from the written
standard and do their own thing. This is the ideal set-up for variation and change.
4.4.5 Frequency factors

An important motivating factor for sound change is the frequency with which we use
words and phrases. Indeed, we’ve already seen how aspects of our linguistic behav-
iour are shaped by repetition (e.g. frequency leads to the reduction of meaning). In
particular, repetition has a profound effect on pronunciation. Individual words and

strings of words that are common typically shorten over time (e.g. daisy began life
as ‘day’s eye’). Reduction is most obvious in formulaic expressions like greetings and
leave-takings; the ritualized string God be with you is ground down to bye. Pairs of
words used frequently are also more likely to reduce; in normal rapid speech don’t
you is pronounced as donchew [doʊntʃu] or even doncha [doʊntʃə].
There are a number of reasons repetition leads to the reduction of forms. One is
that frequent or everyday words will often crop up in casual settings where short
cuts and reductions are more tolerated – a lot can be taken for granted when you’re
chatting with someone you know well in a familiar environment. Words of high
frequency are also easier to understand, so listeners won’t require the same amount
of phonetic detail. Reduction is also the result of the automation that follows from
the repetition of articulatory movements. Linguists have shown that when gestures
of articulation become automatic this triggers a reduction in the timing and mag-
nitude of these gestures. The speed of execution automatically increases, the transi-
tion between sounds becomes more fluent, and the movements might even overlap.
This is undoubtedly the motivation behind the extreme reductions we saw earlier
with place names like Worcestershire (in 2015 the popular social networking site
reddit ran a survey to uncover the most difficult words to pronounce – Worcester-
shire was at the top!).
SUMMARY
In this chapter we reviewed the most significant phonological processes; these pro-
cesses change the speech sounds we produce as we speak, but they can also have
consequences for the actual sound systems of languages. Despite the fact that the
majority are reductive, the imperative of intelligibility means that speech gestures
don’t reduce to the point where sounds become unrecognizable. As phonetician
Joseph O’Connor once put it, “[l]anguage does what it has to for efficiency and gets
away with what it can” (1973: 251). We also touched on some of the motivating
factors behind sound change. Typically involved is a complex network of differ-
ent social, psychological and linguistic pressures – as well as external or foreign
influences. Language is not governed by deterministic law-like processes, but it is
not frenzied or anarchic either – order emerges from this complex interaction of
factors.
FURTHER READING
There are some wonderful accounts of sound change in some of the general his-
torical linguistics books; in particular, we recommend Aitchison (2013), Campbell
(2004), Crowley and Bowern (2010), Hock and Joseph (1996), McColl Millar (2015)
and (for English) Bauer (1994); on the Neogrammarians and the 19th century con-
tribution to the study of sound change, see Burridge (2013). Labov (1994) writes
extensively on vowel shifts in varieties of English and also cross-linguistically; and

on the pull chains and push chains see Martinet (1955). On the duelling forces of
economy of gesture and the imperative of comprehensibility, see Lindblom (1983,
1990) and Zipf (1949). Lesley and James Milroy pioneered the work on model-
ling language contact between individuals; for their ideas about social networks
and change see L. Milroy (1987) and J. Milroy (1992, 1993). On the motives gener-
ally that lie behind change see various chapters in Hickey (2003). Joan Bybee has
researched the many different aspects of our linguistic behaviour that are shaped by
repetition (see Bybee 2003).
EXERCISES
1 Phonological processes – synchronic variation in Modern English

Explain and name the phonological processes that bring about the pronunciation
of the following words. Try to think of at least one other example that illustrates the
same process (in any language).
a The non-standard pronunciation of family as fambly, youngster as younkster

and chimbley for chimney.
b The pronunciation /fɛbri/ for February and /tɛmpri/ for temporary.
c The pronunciation of seven as /sɛbm/ and eleven as /əlɛbm/.
d The non-standard pronunciation of the word diphthong /dɪfθɒŋ/ as /dɪpθɒŋ/.
e A TV viewer’s complaint that the newsreader pronounced Italy as idilly.
f The pronunciation of assume and presume as /aʃum/ and /prəʒum/.
g The normal pronunciation of swoon and twice is /sw̥ uːn/ and /tw̥ aɪs/ (the
empty circle under /w/ indicates the devoicing of the approximant.
h Using appropriate phonetic symbols and descriptions, trace the series of
changes that reduces (in fast speech) the pronunciation of Did you have a
good weekend to /dʒævəgʊdwigɛn/. Name each of the processes involved.
i The personal names Dwight and Dwayne are often pronounced with a vowel
between [d] and [w].
j The pronunciation of hospital as hostipal.
2 Phonological processes – diachronic change in English

Explain and name the phonological processes that are behind the changes in the
following words. Try to think of at least one other example that illustrates the same
process (in any language).
a Proto-Germanic *emna- > ev(e)n and *heman- > heaven in English.

b The Italian colonella ‘little column’ > Modern English colonel /kɜrnəl/ or
/kɜnəl/.
c Kriol is an English based creole spoken in northern parts of Australia.
Explain (with labels for the processes) how the English word afternoon
/aftənun/ has become Kriol apinum.
d Kriol [ʃarap] < English shut up; [palɪɟɪman] < English policeman (/ɟ/ = voiced
palatal stop).
e The transitions Old English hlāfdige > lafdi > lady and Old English hlāfwēard
> hlāford > laverd > lord.
f Old English Sunnendæg > Sunday.
g Proto-Germanic *samftō > sōfte > soft.
h Anglaland ‘the Land of the Angles’ > England.
i Old English eahtatyne [axtətin] and eahtatig [axtəti] > eighteen and eighty.
j Old English mommele > mumble.
3 Phonological processes – languages other than English

Explain and name the phonological processes that have triggered the pronunciation
of the following words.
a Proto-Germanic *emna- > eben; *heman- > Himmel in German.

b Czech /pta:k/ ‘bird’ > /pta:tʃek/ ‘little bird’ (-ek = the diminutive suffix).
c Swedish: /æ:dəl/ ‘noble’ versus /æ:dla/ ‘noble (definite)’; /vakker/ ‘pretty’
versus /vakkra/ ‘pretty (definite)’.
d German: /buntə/ ‘colourful’ (attributive) and /bunt/ ‘colourful’ (predicative)
versus /bundə/ ‘league’ (dative) and /bunt/ ‘league’ (nominative).
e The pronunciation changes in Turkish when the accusative suffix is added:
/dip/ ‘bottom’ versus /dibi/ ‘bottom (accusative)’; /ʃɛrit/ ‘tape’ versus /ʃɛridi/
‘tape (accusative)’; /grup/ ‘group’ versus /grubu/ ‘group (accusative)’.
f German Superintendent > Superintend ‘Superintendent’; Bewundererin >
Bewunderin ‘admirer (feminine)’.
g In Fore /pone/ ‘banana type’ becomes /to bone/ or /to βone/ ‘another
banana type’; /kune/ ‘netbag’ becomes /to gune/ or /to ɣune/ ‘another net-
bag’ (explain the alternative pronunciation).
h The pronunciation of Yoruba verbal prefixes: /ba/ ‘hide’ versus /mba/ ‘is
hiding’; /fɔ/ ‘break’ versus /mfɔ/ ‘is breaking’; /kɔ/ ‘write’ versus /ŋkɔ/ ‘is
writing’; /lɔ/ ‘go’ versus /nlɔ/ ‘is going’; /wa/ ‘come’ versus /ŋwa/ ‘is coming’.
i French: /pətit ami/ ‘little friend’ and /gʁoz ami/ ‘big friend’ versus /pəti
gaʁso~/ ‘little boy’ and /gʁo gaʁso ~/ ‘big boy’.
j Pre-Old Irish */magl/ > Old Irish /ma:l/ ‘prince’; Pre-Old Irish */kenetl/ >
/kene:l/ ‘gender’.
4 Historical development of sounds in English

Below are four different versions of the Lord’s Prayer taken from different periods
in the history of English. Examine each version and answer the questions that fol-
low (the answer for each question requires about a paragraph).
Old English text, ca. 1000

Fæder ure þu þe eart on heofonum
si þin name gehalgod
tobecume þin rice
gewurþe þin willa on eorþan swa swa on heofonum

urne gedæghwamlican hlaf syle us to dæg
and forgyf us ure gyltas
swa swa we forgyfað urum gyltendum
and ne gelæd þu us on costnunge
ac alys us of yfele. Soþlice
Middle English text, ca. 1400

Oure fadir that art in heuenes,
halewid be thi name;
thi kyngdoom come to;
be thi wille don in erthe as in heuene;
yeue to vs this dai oure breed ouer other substaunce;
and foryeue to vs oure dettis, as we foryeuen to oure dettouris;
and lede vs not in to temptacioun,
but delyuer vs fro yuel. Amen.
Early Modern English text, ca. 1611

Our father which art in heauen
hallowed be thy name.
Thy kingdome come. Thy will
be done, in earth, as it is in heauen.
Giue us this day our daily bread.
And forgiue us our debts,
as we forgiue our debters.
And lead us not into temptation,
but deliuer us from euill. Amen
Modern English text

Our Father, who art in heaven,
May your name be kept holy.
May your kingdom come into being.
May your will be followed on earth, just as it is in heaven.
Give us this day our food for the day.
And forgive us our offenses,
Just as we forgive those who have offended us.
And do not bring us to the test.
But free us from evil. Amen
a Take five words from the prayers above that illustrate Langacker’s descrip-
tion of language as “a gigantic expression-compacting machine”. Name and
describe the changes that have taken place.
b Give two Old English consonant clusters (not double (or geminate) con-
sonants) that are no longer in the modern language or are disappearing.
Illustrate with examples from the prayer.
c Using words from relevant versions of the prayer to illustrate, describe the
changes that affected the velar nasal consonant in the Middle English/Early
Modern English period.
d Using words from relevant versions of the prayer to illustrate, describe the
changes that affected fricative consonants in the Middle English period.
e Describe the system for the use of the letters <u> and <v> in the Middle
English prayer. (There are enough clues in the prayer, but this might also
require some research on your part.)
f Take a word from the prayer and track its changes from Old to Modern
English; in a general statement, describe how the pronunciation of this
word has changed.
5 Research essay
Write an essay (approx. 1,000 words) that explores the following quotation from
Ronald W. Langacker (1977: 106). Your bibliography should have at least five
references.
It would not be entirely inappropriate to regard languages in their dia-

chronic aspects as gigantic expression-compacting machines.
NOTES
1 We have rendered the sound spelled <e> here as a schwa when in final position, but bear
in mind that at this time these vowels were on the way out (think of Modern English tale
/teɪl/, which would have been pronounced /ta:lə/). Poets like Chaucer could play with
this feature to accommodate matters of rhythm and metre.
2 We’ve had to simplify the changes considerably here, and there are a number of dialectal
complications and competing shifts that account for anomalies in the modern language
(e.g. words like great and steak didn’t go on to shift to /i/).
5
Changes in word structure
INTRODUCTION
Morphology is concerned with word structure and word formation. On the one
hand, it deals with inflectional morphology, i.e. the creation of word forms, such
as book (singular) versus book+s (plural). Books is not a new word, and it would
not get its own entry in the dictionary. Rather, it is just a new form of the word (or
lexeme) book. On the other hand, morphology also looks at how lexemes are built
through processes such as affixation (e.g. possible and im+possible), compounding
(e.g. wall + paper = wallpaper), clipping (refrigerator > fridge) and so on. Both of
these domains of morphology are affected by linguistic change.
Inflections and inflectional systems develop over time, and so do word formation
patterns. For example, in Chapter 1 we mentioned the loss of inflectional morphology
in English between 1000 and 1500 CE. In Chapter 2 the focus was on morphology as
it relates to word formation (the creation of lexemes). In this chapter we now shift
the focus once more onto the grammatical life of words (the creation of word forms).
Over the last 150 years or so, inflectional morphology has received considerable
attention in historical linguistics and language change theory, and most of this chap-
ter will deal with this. In particular, we will be concerned with three major aspects
of morphological change: reanalysis (the reinterpretation of structure), analogy (or
attraction to structure) and morphosyntactic typology (the shift of languages from
one grammatical type to another). We will conclude with some explanations for
morphological changes, mostly based on psycholinguistics and cognitive psychol-
ogy, and the usual summary, exercises and further reading.
5.1 REANALYSIS AND ACTUALIZATION – REINTERPRETATION OF STRUCTURE
Language is communication between speakers and hearers. Speakers have certain

thoughts in their heads and encode these into words and sentences. These words are
transmitted through air, on paper, electronically or visually to the “hearers”. Hear-
ers in turn have to perceive these words and sentences and decode them in their
heads. Figure 5.1 shows a simplified visualization of this process. Needless to say,
the whole act of communication is a lot more complicated and involves different
contexts for sender and receiver, and also “noise” (i.e. additional material not part of
the transmitted information) inside the speaker’s head (other stuff you’re thinking
about while speaking, for example), inside the hearer’s head (other stuff the hearer
is thinking about while listening to you) and in the channel (e.g. background noise
or ads that get posted next to your email message).
Changes in word structure 107
Figure 5.1
Communication process (simplified)
ENCODING DECODING
IDEA MESSAGE
DECODED
Internal Internal and

Noises Semantic Noises
External Noises
As we mentioned in Chapters 1 and 3, this encoding and decoding do not always

happen perfectly. The thoughts in the speaker’s and the hearer’s heads are not iden-
tical. This applies both to the message itself (you are not a mind reader and don’t
know exactly what the person you are talking to you has in mind) and to the form
of the message. Part of decoding the message is to (re-)construct the original gram-
matical form. One very simple example from syntax would be the sentence John
kissed Mary. The speaker encodes this as John [subject] kissed [verb, past tense],
Mary [object]. The job of the hearer is to understand the sentence by decoding (or
parsing) this as efficiently as possible. And in the ideal case, the hearer parses this
as John [subject] kissed [verb, past tense], Mary [object] and arrives at roughly the
same interpretation. But this is not always the case.
Here’s one example of what can happen when speakers and hearers are not on
the same page with their language structures. It comes from word formation. The
English word apron comes from French naperon ‘small table cloth’ and was bor-
rowed in the 14th century as a napron (compare the related napkin/nappy). Appar-
ently, in the 15th or 16th century, this structure led to confusion, and we find both
a napron (old) and an apron (new). It seems as though some speakers thought that
the word is apron, preceded by the indefinite article an! So, owing to this ambigu-
ous sound structure, speakers and hearers could not be sure if the head noun was
actually napron (which requires a single vowel a as the indefinite article) or apron
(which requires the vowel-consonant sequence an as the indefinite article). Even-
tually, they seem to have settled for the new form, an apron, the word that we use
today. The same happened to adder (and a number of other words, such as umpire).
Adder started its life as Old English nadder ‘snake, serpent, viper’ (Modern German
still has the cognate Natter). Between the 14th and 16th century, we see a nadder
gradually turning into an adder. As you saw briefly in Chapter 4, Modern English
nickname shows the opposite process: Middle English (about 1300 CE) had an eke
name ‘an additional name’ (eke comes from Old English eaca ‘an increase’). In the
15th century, this turned into a nekename. Note that this involves not one but two
“misinterpretations”! On the one hand, the consonant of the indefinite article was
seen as part of the head noun; on the other hand, the independent word eke was
reinterpreted as part of the head noun name, i.e. ekename instead of eke name.
This process is called reanalysis (or metanalysis, or rebracketing – unfortunately,
there are many different names), and it is one of the most fundamental mechanisms in
linguistic change. It has been successfully applied in the explanation of morphological
and syntactic change, and in grammaticalization (see Chapter 6). Roughly, reanalysis
can be defined as a process that changes the actual (underlying) linguistic structure
without necessarily affecting the visible or audible surface manifestation of that struc-
ture. So it presupposes ambiguity in linguistic structures, or at least some uncertainty
about the underlying structure of a given utterance. Strictly speaking, reanalysis itself
is not necessarily visible in the output. When speakers/hearers reanalysed a nadder as
an adder, the sound structure [ənædər] remained more or less the same. But reanaly-
sis is usually followed by what we call actualization, i.e. the gradual surfacing of the
new reanalysed structure. This, then, is the point where the new structure becomes
visible and it is clear that something must have changed underneath (e.g. when saying
two adders, without the n, or writing <an adder>). This usually goes in tandem with a
third aspect, sometimes referred to as “extension”: the new and reanalysed structure
spreads to other contexts and establishes itself as a new, more generalized pattern. So,
the whole process usually involves three individual steps:
REANALYSIS  ACTUALIZATION  EXTENSION
The Polynesian passive

A very famous example of this process can be found in the development of
the passive in Polynesian languages. Consider the Māori forms in Table 5.1,
which we have roughly glossed as ‘active’ and ‘passive’.
Table 5.1
Ma-ori passive formation
Active Passive
mataku ‘to fear’ matakuria ‘to be feared’
fau ‘to tie’ faufia ‘to be tied’
neke ‘to move’ nekehia ‘to be moved’
hopu ‘to catch’ hopukia ‘to be caught’
inu ‘to drink’ inumia ‘to be drunk’
tohu ‘to point out’ tohungia ‘to be pointed out’
awhi ‘to embrace’ awhitia ‘to be embraced’
When you look at this table, you would probably say that the Māori passive
is quite complex because it’s formed by a range of different suffixes (-ria, -fia,
-hia, -kia, -mia, -ngia, -tia). This is also how Māori speakers view this aspect
of their language (and you could compare it to the various suffixes for English
plural forms, such as hat-hats, child-children, ox-oxen, foot-feet and so on). In
fact, this complexity has arisen because of reanalysis.
Originally the passive suffix was -ia, and all of the active forms in Table 5.1
ended in consonants. But speakers stopped pronouncing the final conso-
nants (which speakers are wont to do) – so matakur became mataku, and fauf
became fau and so on. What happened then was that speakers reanalysed
the structure of the passive forms by shifting the morpheme boundary: so
matakur-ia > mataku-ria; fauf-ia > fau-fia and so on (compare the plight
of English nadders, naprons and nicknames in the previous discussion). This
ended up making things a whole heap more complicated. Nowadays, when
they form passives, speakers have to choose from many different (and totally
unpredictable) passive endings (whereas once it was simply -ia). This is a
good example of how a simple sound change (final consonant deletion) can
make things a lot more complicated for the grammar (something we touched
on in Chapter 4 and return to in a moment).
The usual pattern in language change is that another change steps in to help
clean up the mess. What appears to be happening now in Māori is that the -tia
suffix (the most common ending) is spreading its wings. It’s become a kind of
default ending, so that when verbs are coined or borrowed they will take -tia. This
suffix will probably oust the others eventually, and regularity will be restored. It’s
the process that has helped popularize -tia that we now turn our attention to.
5.2 ANALOGY – ATTRACTION TO STRUCTURE
Analogy is – arguably – one of the most powerful cognitive capabilities available.

Douglas R. Hofstadter, author of the famous and award winning Gödel, Escher, Bach
(1979), eloquently described it as “the very blue that fills the whole sky of cognition”
(“Analogy as the Core of Cognition” is the title of Hofstadter’s 2006 talk; this can
be accessed at https://www.youtube.com/watch?v=n8m7lFQ3njk). Raimo Anttila,
a well-known historical morphologist, was just as poetic when he characterized it
as “the warp and woof of cognition” (Anttila 2003).
Analogy is commonly understood as a very simple mathematical equation:
a : b = c : d (a is to b as c is to d)
Or, in more everyday language terms: foot is to toe as hand is to finger:
foot : toe = hand : finger

Given analogical thinking, we can now solve these equations with ease:
3 : 4 = 9 : x where x = 12

shoe : foot = x : hand where x = glove
This is commonly understood as proportional, four-term analogy. It is interesting to

note now that Hofstadter in particular points out that this is certainly one (small) aspect
of analogy, but that, for him, analogy is a much more general and powerful process.
In his 2006 talk he defines analogy making as “the perception of a common essence
between two things” (ibid.; compare our discussion of metaphor in Chapter 3) where
“the perception of a common essence” is not something metaphysical and mysterious
but takes place in one’s current frame of mind, and “things” refers to “mental things”
or mental representations. This broader view allows us to gain a more encompassing
and psychologically plausible view of analogies. Instead of just focusing on tiny, math-
ematical (proportional, four-term) analogies, analogical thinking is seen as something
much broader and can encompass frames, scripts, stories and so much more. In that
sense, analogies don’t just exist out there (as many people think) but analogy making is
a truly remarkable cognitive capacity, a dynamic, fast paced, fleeting process.
We’ve already seen some examples of analogy at work in word formation (Chap-
ter 2) and in phonology (Chapter 4) – but what does it have to do with the more
grammatical life of words? In this chapter we will see that analogy making is one of
the most powerful capacities of speakers and hearers when they deal with the mor-
phology of their languages. Arguably, analogy making drives many of the changes
we see. This has been shown very impressively in child language acquisition.
As early as 1958, the psychologist Jean Berko Gleason designed her famous “wug
test”. She presented several children with the drawing of a little animal (like the one
we show in Figure 5.2). And she said something like: “Look, this is a wug [wʌg]”
(note that the word did not exist in the English language and the children had never
heard it before). Then she showed them another picture, this time with two animals
of the same type. And she told them something like “There are two of them now.
There are two ______”. And the children replied, without hesitation: “wugs” [wʌgz].
How did they do that? They did not know the word, and nobody ever told them
about the rules of English plural inflection. The answer is simple. They perceived
similarities between what they had already experienced and what they now encoun-
tered, and they used analogical thinking to come up with the answer! So, for exam-
ple, they might have perceived a similarity between wug and bug, and thus abduced
(a kind of backwards reasoning, see Chapter 1):
bug : bugs = wug : x where x = wugs
Note that this is not necessarily the only answer. The “correct” plural of wugs could
have been something else, for example wag, as in one wug, two wag (since the word
does not exist we can make up any plural form we like!). But this alternative plural
form is something they could not have guessed. Even these small children already
used the power of analogical thinking to explore the morphology of their native
Figure 5.2
The “wug” test (based on Gleason 1958)
This is one wug.
Now there are two of them.

There are two
language. Also, note that they did more than just add “plural -s”. The plural mor-
pheme in English has a fairly large number of allomorphs, or concrete realizations
(in other words, different pronunciations). Only consider the difference you hear
in cats [kæt-s], dogs [dɒg-z] and foxes [fɒks-ɪz]. So the regular, phonologically con-
ditioned forms include [ɪz] after stem final sibilants (hissing sounds) such as [s, z,
ʃ, ʒ], voiced [z] after voiced sounds, and [s] after voiceless sounds. And there are a
number of more or less irregular, lexically conditioned patterns such as mouse-mice,
child-children, sheep-sheep and so on that you can’t predict but simply have to learn.
The patterns that get extended in analogy are usually the live, or productive, ones. So
out of this complex of rules, children who produced wugs [wʌgz] picked the “right”
form; i.e. they analysed the input wug in such a way (with a word-final voiced con-
sonant) that they arrived at a “correct” analogy with similar words. In other words,
they created [wʌg-z] and not [wʌg-s] or [wʌg-ɪz]. Quite an accomplishment, con-
sidering that nobody ever tells you about these things when you’re three years old!
Analogy can make paradigms and forms simpler in that it reduces the number of
allomorphs (or forms) for a given paradigm. Recall from Chapter 1 that paradigms
are sets of different word forms (such as am, is, are, was, were for the verb to be).
Here’s just one example: English used to have a large number of irregular verbs that
have three different forms for the present tense, the past tense and the past participle.
Some of them still exist in present-day English, like see, saw, seen. These contrast with
the regular verbs that only have two forms for present tense, past tense and past par-
ticiple: the simple present (kiss), past tense (kiss-ed) and past participle (also kiss-ed).
Analogy now turns irregular verbs into regular ones. Examples include strive, strove,
striven, which is now for many speakers strive, strived, strived; help, holp, holpen, which
is now help, helped, helped; and climb, clomb, clomben, which is now climb, climbed,
climbed. And there even used to be laugh, low, laught (or laughen), which is now laugh,
laughed, laughed. All in all, we estimate that out of ca. 300 irregular verbs in Old Eng-
lish, about 150 underwent change and became regular. Table 5.2 summarizes some of
these developments (but note that this is also simplifying the whole story a bit, as there
may be additional analogies, sound changes and other change processes involved).
Table 5.2
Analogy and irregular verbs in the history of English
Modern Earlier
strived strove-striven
stepped stope-stapen
climbed clomb-clomben
laughed low-laught (laughen)
crept crope-cropen
helped holp-holpen
yielded yold-yolden etc.
Old French offers another illuminating example (based on McColl Millar 2015:
101). In Old French, the verb aimer ‘love’ had two different stem forms: one with
ai- (in the singular and third person plural) and one with a- (in the first and second
person plural). Analogical levelling has now led to a regularization and levelling of
the paradigm, so that Modern French only has stem forms with ai-, even in the first
and second person plural.
Old French Modern French

1 sg aim 1 pl amons aimons
2 sg aimes 2 pl amez aimez
3 sg aimet 3 pl aiment aiment
You may be tempted to think now that analogy always leads to regularity and
transparency in the morphological system by eliminating all those irregular and
unpredictable forms. In many cases, this is true. But there are some counterex-
amples where analogy is the source of irregularity and allomorphy in a given
paradigm. Consider the past tense of dive. Is it dived (regular) or dove (irregular)?
If you are from the United States, you probably opt for dove. If you are from the
United Kingdom, you probably go for dived. As a matter of fact, the traditional,
old past tense of dive is dived. The Oxford English Dictionary has its first entry
of dove from 1855, whereas dived goes back to the Middle Ages. So irregular
dove was probably created in the 19th century, maybe in analogy to other irregu-
lar (and well-established) patterns like drive-drove or ride-rode. Similarly, arrive
has a non-standard past tense arrove, and squeeze sometimes has squoze, though
these obviously never made it into the major varieties of English.
As we have said before, analogy is one of the strongest forces in linguistic change,
and perhaps even in human cognition. Nevertheless, it should also have become clear
that analogical changes are somewhat random in the sense that we never know when
analogies will be created and which forms will be affected. Why was holp eliminated in
favour of helped, but not bought in favour of *buyed (admittedly buyed does exist, but
it is comparatively rare)? Why is it climbed today but not *drived? On the other hand,
though we never know exactly when and where analogy will strike, we know that ana-
logical levelling usually leads to greater regularity in the linguistic system (for some
exceptions, see above). Paradigms are usually a lot simpler after analogical levelling.
With sound change it is exactly the opposite. Sound change is mostly highly regular
and predictable (see Chapter 4), but it often causes irregularities in the linguistic sys-
tem. In Chapter 4 we touched on the famous sound change of i-umlaut in the pre-Old
English period, the vowel harmony rule that caused the regular plural forms for mouse,
goose, tooth and many more to change into what we know today as irregular plural
forms: mice, geese, teeth. What happened was that all these words had a highly regular
plural form in -iz.
Sg Pl
mu-s *mu-s-iz (note that * here means that the form does not exist in
go-s *go-s-iz any documents, but had to be reconstructed)
to-θ *to-θ-iz
Now the high front vowel [i] of the plural in the second syllable changed the quality of
the vowel in the stem. These vowels were fronted (an example of distance assimilation).
Sg Pl
mūs *mȳs-iz
gōs *gø̄s-iz
tōθ *tø̄θ-iz
In a next step, the final syllables were eroded and lost, and the front rounded
vowels were unrounded.
Sgold/new Plold Plnew

mūs mȳs > mīs
gōs gø̄s > gēs
tōθ tø̄θ > tēθ
Finally, all these underwent a massive sound change (part of the Great Vowel
Shift, as we outlined in Chapter 4), i.e. a raising of all long vowels.
Sgold Sgnew Plold Plnew

mūs > maʊs mī s > maɪs
gōs > gu:s gēs > gi:z
tōθ > tu:θ tēθ > ti:θ
And what we get eventually are two word forms in the paradigms which appear
to be highly irregular (mouse-mice, goose-geese, tooth-teeth and so on). But what
caused these irregularities were highly regular and perfectly predictable sound
changes. This phenomenon is known as “Sturtevant’s paradox”:
Analogy is irregular but causes regularity. Sound changes are regular but cause
irregularity.
As an important mechanism in linguistic change, analogy has received a lot of

attention in the linguistic literature. Scholars have offered numerous accounts of
how it works and, more importantly, how it can be constrained. In other words,
they have tried to find out when it happens and which direction it usually takes.
Jerzy Kuryłowicz, a well-known Polish historical linguist of the 20th century, inten-
sively studied analogy as a factor in linguistic change and came up with six rules,
or “laws”, that help us constrain and understand analogical changes. Kuryłowicz
used the metaphor of rain, gutters, sewers and pipes to illustrate his understanding
of analogy: We cannot actually predict when it will rain. The rain, like analogical
change, is not a necessity. It happens, or it doesn’t, but when it rains, the water can
take predetermined paths defined by gutters, sewers and pipes, and we can pretty
much say what will happen when it does rain. These gutters, sewers and pipes in
some sense are Kuryłowicz’s “laws”. We’ll look at each of these six rules in turn.
Kuryłowicz No. I: a complex morphological structure (a so-called bipartite marker)

tends to replace simpler (isofunctional) morphological markers
We can generally distinguish between simpler morphological structures that only
have, for example, one grammatical suffix (e.g. dog – dog+s) and more complex
grammatical structures that have, for example, both a grammatical suffix and
a change in pronunciation (e.g. German ‘dream’ Traum [traum] (sg) – Träume
[trɔɪmə] (pl), which shows both a change of its vowel (au [au] > äu [ɔɪ]) and a
suffixation with -e [ə]). As a tendency in analogical change, simpler structures are
usually replaced by more complex structures. One textbook example for this comes
from German plural formation:
Sg Pl
Old High German gast gest-i ‘guest’
boum boum-a ‘tree’
Modern High German Gast Gäst-e ‘guest’
Baum Bäum-e ‘trees’
In German, the plural of gast developed more or less predictably: the word-final
vowel [i] led to a raising of the mid-word vowel [a] to [ɛ], represented in Modern
German by the umlaut spelling <ä>. The word-final vowel [i] was then weakened
to a so-called schwa [ə], a very common process in most languages. The unsur-
prising result was the bipartite (double) marker for the plural with a change of
the mid-word vowel and a suffix. What is surprising is the development of Baum-
Bäume. The Old High German boum-a (with a single plural marker -a) should have
developed regularly into *Baum-e (again, with a single marker) since the word-final
-a would not have influenced the mid-word vowel here. But what we actually find is
a change of this mid-word vowel into äu [ɔɪ] in analogy to other plurals like Gäste
and a weakening of -a to schwa [ə]. The result is a complex, bipartite marker that we
did not expect – and an example for Kuryłowicz’s tendency No. I.
However, there are also a number of counterexamples which illustrate that this
is really only a tendency, and nothing like a law (such as the law of gravity). One
example where a complex marker is replaced by a simpler one is the plural of English
brother. According to the Oxford English Dictionary, before the end of the 16th cen-
tury broðor, broþeren, breþeren and the like were the only available plural forms (with
one strange exception, a text called Laʒamon’s Brut, a Middle English poem written
around 1200, which also has broþeres). After 1600 brothers became the regular form
(and brethren took over another function; see Kuryłowicz’s Law No. IV below). Here,
obviously, a complex marker with a vowel change and suffix was replaced by a simple,
regular marker in analogy to other regular markers – a clear counterexample to Law
No. I. Another counterexample can be seen in past tense forms like stope, which got
replaced by stepped (see above). One can argue that stope is not morphologically more
complex than stepped, which means that one simple marker is replaced by another
simple marker, and not by something more complex as predicted by Law No. I.
Does this mean that the “law” is not valid? Not quite. It doesn’t appear to be a law
in the strict sense of the word (a fact that applies universally). The law of gravity,
for example, does not allow for exceptions and counterexamples. As a tendency,
however, Kuryłowicz’s “law” is revealing, and even though it does not capture all
changes, it helps to explain at least some of them. And it is interesting to think about
why some changes don’t follow our predictions! We can search for opposing forces,
disturbances of any kind, and we also need to look at the other “laws” and how they
interact with each other.
Historical linguist Hans Heinrich Hock (1991) has also suggested ignoring the
letter of this law but accepting its spirit: “forms which are more ‘clearly’ or ‘overtly’
marked tend to be preferred in analogical change” (p. 212). As a broader concept,
“overt marking” captures more cases than either “bipartite” or “complex”.
Kuryłowicz No. II: analogical changes are from a “basic” form

to a “derived” form of the word; the relationship between the
two is a consequence of their “spheres of usage”
Law No. II sounds very complicated at first, but it actually claims two very simple
things. First, the two forms must be related as basic and derived in order to be
susceptible to analogy. So the past tense of dive can be built on analogy to drive, i.e.
drive:drove, and thus dive:dove. Drive and drove are basic and derived forms of the
same word, and so are dive and the new dove. This usually does not happen with
unrelated pairs of words, for example “drive (verb) relates to hive (noun) as drove
(verb) relates to X”, where X is *hove (noun). The latter does not exist, and it is not
the plural of hive, of course. In other words, drive and hive do not form a pair of
basic and derived forms and therefore cannot give rise to other analogous pairs.
The second part of Law No. II is slightly more complicated, partly because
Kuryłowicz used the somewhat mysterious term “spheres of usage” here. But again
the idea is simple: patterns which are more common and which are used more often
in creating new words or word forms are more likely to be used as models in analogy
than others. In other words, creating a plural in English with the suffix -s is a lot more
common than any other pattern. Therefore, brethren was ousted by brothers, which
was created on analogy to this very common pattern. It is very unusual to find the
opposite; e.g. a word like sister with its plural sisters is very unlikely to have its plural
replaced by something like *sistren1 or the like, simply because the pattern for the
latter is a lot more infrequent and unusual. Of course, the early example of dove is
exactly of this type, and would appear to be working in the opposite direction. But
here we can say that this was influenced by a whole group of verbs in -ive, including
drive-drove, strive-strove and thrive-throve (which is where this “sphere of usage” pro-
viso comes in – the influence of this bunch of verbs was strong enough to pull dive in
their direction).
But just like Law No. I, Law No. II also has its problems and exceptions. What, for
example, is the direction of the levelling of the verb ‘choose’ in English and German?
‘choose’ Old English Old High German

Infinitive ce- ozan kiosan
1st person sg past tense ce- as ko-s
1st person pl past tense curon kurun
Past participle (ge)coren (gi)koran
Modern English chose the infinitive as its model for the analogy and dropped
all other forms in favour of it, so that cēas and curon became chose, and (ge)coren
became chosen. Modern German picked the plural forms as models and now has
küren (basic), kor (1st person singular past tense), koren (1st person plural past
tense) and gekoren as the past participle. In other words, kiosan and kōs were lev-
elled out in favour of the other forms. This actually means that in German derived
forms served as models, not basic forms, as Law No. II would lead us to expect.
Other real problems for this law are cases of backformation, where what is affected
is the basic, not the derived term of the proportion. For example, Modern English
pea is the result of a backformation from an original singular pease (as in the “pease
porridge hot” nursery rhyme). When -s became the standard plural marker, some
nouns already ending in a final consonant [-z] were mistaken for plurals. This is
certainly analogy at work, but it is in the opposite direction from the usual; it’s the
creation of a simple form on analogy to cases where a complex form and a simple
form exist together. Again linguists have suggested a useful adaption of this second
law which would allow for these backformations without throwing out the law. For
analogy to take place, the two forms making up the first part of a proportion must
be related by a productive morphological process. Take the past tense forms of the
English verbs in Table 5.3.
Table 5.3
Basic and past forms of [ai:t] verbs in English
Basic Past
write wrote
fight fought
cite cited
Only cited represents a productive process of past tense formation – and it shows
the one pattern that will extend into new situations. We can understand the lan-
guage learner who says I writed or I fighted, but not the one who comes out with I
wrought, I cought or I fote, I cote as potential past tense forms of write, fight! These
made-up past tense patterns do not have the capacity to extend by analogy.
And there is yet another problem for this law, and this has to do with the effects
of frequency. While we expect the more common, more productive pattern to be
dominant, in American English regular (and common) dived was ousted by the less
common dove on the basis of verbs in -ive like strive-strove, drive-drove and thrive-
throve, as mentioned above. So, every now and then (and for some complex and
maybe unknown reasons) we see changes in the opposite direction, with rare forms
or patterns winning out.
Celebrity endorsement
For some time now English has been regularizing peculiar plurals like leaf-
leaves. The plural of cliff is no longer cleves but cliffs, for example. And while
leafs hasn’t got there yet, forms like silver leafs (a kind of white poplar) and the
Canadian Maple Leafs (the ice hockey team) show that the regular plural has
snuck (sneaked) in the back door. One noun, however, has gone against this
path of change, and that is dwarf. It might sound older, but dwarves is actually
a new form. In Old English the word was dweorg [dweorx] (so ending in the
same consonant as Scottish loch). Over time, [x] either dropped out or shifted
to [f], as it did in words like trough, laugh and also dwarf (in this case the spell-
ing reflects the shift). This change happened long after we’d lost the voicing
rule that turned [f] into [v] between vowels. Hence, the plural of dwarf was
always dwarfs, pure and simple. Tolkien chose dwarves, however, even though
as a philologist he knew this was historically wrong. He has a note to this effect
in the beginning of The Hobbit, where he writes, “In English the only correct
plural of dwarf is dwarfs, and the adjective dwarfish. In this story dwarves and
dwarvish are used.” These forms have an antiquated ring to them, and this must
surely be the reason that he chose dwarves to describe the ancient people in his
tales (even though in Lord of the Rings Tolkien offers a different explanation).
The popularity of Tolkien’s writing will probably mean that archaic forms like
elves, elvish, wolves and, indeed, the late arrival dwarves will not be regularized
to elfs, elfish, wolfs and dwarfs but will remain as an earmark of the fantastical.
Kuryłowicz No. III: if there are words which are put together (as stem and affix),
and if this is transparent, these will serve as a model for other, non-transparent
combinations of stem and affix
McColl Millar (2015: 105–6) cites an example from Basque to illustrate Kuryłowicz’s
third law. In Basque we find the locative case ending -n to mark words that some-
how have to do with place. Examples include heme-n ‘here’ and etxea-n ‘at home’.
The question word non/nun ‘where’ consists of the stem no/nu- plus the locative
ending -n. This can combine to form an indefinite structure nonbait/nunbait ‘some-
where’. This, however, is unusual because it no longer ends in -n. It seems that in
some western varieties of Basque nunbait changed into nunbaiten, so that the loca-
tive ending -n is reattached and transparency is restored.
This law has received only little attention in the literature. One reason for this
could be that these processes are comparatively rare and examples are hard to find.
And while it can account for those rare cases where more basic forms are affected
by change, there are plenty of counterexamples or at least phenomena which cannot
be clearly identified with respect to this law. So, for example, how do we explain that
German and English moved in completely different directions with their analogical
changes on the forms of cēozan/kiosan (see above)?
Kuryłowicz No. IV: when a form is subject to analogy, the new (analogical) form takes
over the main function or meaning of that word, and the old form can still be used in
a secondary function
In contrast to Law No. III, this law has been widely discussed, and there are numer-
ous examples to illustrate how it works. Take the plural of the word brother, for
example. As we mentioned before (see Law No. I), up until the 16th century, the
plural form(s) for brother were broðor, broþeren, breþeren and many, many more, but
only very rarely something like broþeres. The (regular) plural form brothers began
to appear in the late 16th century. While Shakespeare, who wrote his plays around
1600, still used brothers and brethren as synonymous, the former quickly took over
later in the 1600s. The older form brethren, however, did not disappear. It came to
be used as a special plural with a particular (secondary) function “in reference to
spiritual, ecclesiastical, or professional relationship” (OED, s.v. brother).
Kuryłowicz No. V: if a language has “the choice” between losing a more important
grammatical distinction and a more marginal one, it will give up the marginal one
Consider the following forms and changes in Latin and early and later Iberian
Romance.
Latin ‘bread’ Sg Pl
Nominative (“subject”) panis pane-s
Accusative (“object”) panem pane-s
Early Iberian ‘bread’ Sg Pl

Nominative (“subject”) panes panes
Accusative (“object”) pane panes
Later Iberian ‘bread’ Sg Pl

Nominative (“subject”) pane panes
Accusative (“object”) pane panes
In the development of early Iberian the special singular nominative form panis
underwent some sound changes and turned into panes. It thus became indistin-
guishable from the plural forms; the accusative singular form lost its word-final
consonant and turned into pane. Later Iberian rectified this situation, and the
nominative singular changed into pane through analogy with the accusative form.
In terms of a four-way analogy we could say panes (nom, pl) : panes (acc, pl) as
X : pane (acc, sg) where X = pane. This is unexpected considering the laws we have
discussed so far: why would the accusative be more basic? This becomes clear when
we consider the alternative. If pane had been changed to panes, the whole paradigm,
or group of words, would have become indistinguishable. But the change came at
a price. When panes (acc, sg) changed to pane it became indistinguishable with
regard to case (so whether it’s available as subject or object). What we gain is a re-
established number distinction, and we can’t confuse the nominative singular with
any of the plural forms, as was the case in early Iberian. Now it can be argued, con-
sidering Law No. V, that the number (singular-plural) distinction is more impor-
tant or basic than the case distinction. If this is true, then the law would lead us to
expect that the case distinction (nom-acc) is given up in the singular in order to
(re-)establish the more important number distinction.
Law No. V can count as fairly well established and uncontroversial. The only
real problem is that we do not have a definition of what counts as more basic. Is
it the number distinction on verbs (singular or plural) or the person distinction
(first, second, third)? These are questions that can be debated and discussed empiri-
cally, so that we “only” need to check what we observe more often in the languages
around the world.
Kuryłowicz No. VI: Analogy can work across language (or dialect) boundaries; for ex-
ample, a native word may be subject to analogy on the basis of a non-native word,
especially if that non-native word comes from a prestigious other language or variety
McColl Millar (2015: 106) cites another nice Basque example, this time from deri-
vational morphology, to illustrate Law No. VI. He states that Basque has a highly
productive suffix -tasun, which makes abstract nouns, so bakar ‘lonely’ turns into
bakartasun ‘loneliness, solitude’ and eder ‘beautiful’ turns into edertasun ‘beauty’.
But Basque adopted a large number of Spanish words with Spanish suffixes like
-dad and -dura. And Spanish is often regarded as the more prestigious language
of the two. As a consequence, some speakers began to produce some new, mixed
formations like bakardade ‘solitude’ and ederdura ‘beauty’, with a mixture of Basque
stem and Spanish suffix. Just like Law No. IV, Law No. VI is also a widely accepted
tendency in analogical change, though examples are not that frequent.
In sum, we can say that Kuryłowicz’s six laws offer an interesting and controver-
sial attempt at constraining and explaining certain patterns in analogical change.
As we have said before, these could be the gutters, pipes and drains that channel the
water after the rain. But predicting when and where it will rain remains difficult, if
not impossible. Some of these “laws” (like Nos. I, II, IV and V) are less controversial
and more “powerful” than others (with lots of examples and fewer counterexam-
ples), while Laws No. III and VI are more problematic. It is probably fair to say that
as long as we see all of these as tendencies (stronger or weaker) rather than actual
laws of nature, these patterns uncovered by Kuryłowicz can be very illuminating
and instructive.
Mańczak’s tendencies
Kuryłowicz was not the only one who pondered on the laws and regularities of
analogical changes. The Polish linguist Witold Mańczak was another prominent
20th century scholar who worked on the problem. He formulated nine hypoth-
eses, or “tendencies”, in response to Kuryłowicz’s “laws”. For reasons of space,
we can only list them here without any discussion. We strongly encourage you
to compare these to the “laws” developed by Kuryłowicz. Can you see any cor-
respondences between Mańczak’s tendencies and Kuryłowicz’s laws? Where do
we find contradictions? Think about possible examples for these tendencies.
1 Longer words are usually reshaped on the model of shorter ones, except
in inflectional paradigms.
2 Changes in word stems (e.g. vowel changes in the stem of word) are usu-
ally lost rather than introduced.
3 Longer inflections are often remade on the model of shorter ones, unless
these are a “zero” affix (no marking) and a strong, clear marker.
4 Zero endings (no marking) are usually replaced by some visible marker.
5 Endings with only one syllable are usually replaced by endings with more
than one.
6 The form of the indicative (factual) mood is usually the basis for reshap-
ing the other moods (e.g. the imperative (commands, prohibitions) and
subjunctive (unreal thoughts wishes, hopes)).
7 The form of the present tense is usually the basis for reshaping the other
tenses (past, future).
8 If a geographical noun and a common noun differ only in their inflection (but
are otherwise similar), the geographical (local) cases will be the traditional,
archaic ones, and the non-geographical (non-local) ones the innovations.
9 If a geographical noun in a paradigm is changed owing to analogy, the
starting point for that change usually lies in the local cases.
5.3 TYPOLOGY – CHANGE IN MORPHOLOGICAL TYPE
How do languages get their job(s) done? Remember, one of their jobs is to com-
municate messages between speaker and hearer, like “who did what to whom when
and how” (as in Andrea tickled Helena this morning with a feather). Andrea in this
case would be described as the subject, tickle as the predicate (in the past tense),
Helena as the object, this morning as a temporal adverbial and with a feather as an
adverbial of manner. How do languages make clear which of these is the subject,
and which is the object, or the adverbial? Subject, object, predicate and adverbial
are also called syntactic functions.
English mostly does that by word order. First comes the subject, then the verb,
then the object. The girl kissed the boy reports of a female human kissing a male
human. If you turn this around you get The boy kissed the girl, and you get a male
human kissing a female. If you were a native speaker of Latin, that problem would
not exist. In Latin, ‘boy’ is puer, ‘to kiss’ is bāsiāre and ‘girl’ is puella. However, Latin
has many grammatical word forms which change depending on what you want to
say. So if the girl should be the subject (that does the kissing) she takes the so-called
nominative inflectional marker (simple -a in this case: puella). The boy would then
be the object (and gets kissed) and so needs the so-called accusative inflectional
marker -um: puerum. And to make matters even worse, the verb needs to have the
right tense, person and number marking, in this case -vi-t, so bāsiāvit. And the
resulting sentence is puella bāsiāvit puerum (note that Latin did not use articles
such as the in the same way as English does). If you want the boy to kiss the girl, you
need to change all that, and puer needs to be in the nominative (no change), and
puella needs to be in the accusative with -am: puellam. And so you get puer bāsiāvit
puellam. We can summarize this as in Table 5.4.
Word order in English has to be relatively fixed, simply because you cannot dis-
tinguish between subject and object when you just look at the forms. Latin mor-
phology may be a bit more complicated, but it allows for much more flexible word
order. So in Latin it makes no difference in meaning if we turn the words around:
The girl kissed the boy = puella bāsiāvit puerum

= bāsiāvit puella puerum
= bāsiāvit puerum puella
= puerum puella bāsiāvit
= puella puerum bāsiāvit
= puerum bāsiāvit puella
Note that the Romans actually did like some orders better than others. In this
case they would have preferred puella puerum bāsiāvit, with S(ubject), O(object),
V(erb). But all the other orders were also possible and can be found here and there
in Latin texts. In English there is also some flexibility (the boy the girl kissed . . . is
Table 5.4
Latin inflection and word order
Subject (nominative) Object (accusative) Verb

‘boy’ boy / puer boy / puerum kissed /
bā siā vit
‘girl’ girl / puella girl /puellam kissed /
bā siā vit
The girl kissed the boy – puella bā siā vit puerum
The boy kissed the girl – puer bā siā vit puellam
possible), but saying that these options are equally possible and that English speak-
ers only “prefer” SVO would be stretching it a bit.
To cut a long story short: a brief comparison of languages like English and Latin
shows that these two languages do their job in different ways. English employs word
order to tell you who did what to whom, while Latin prefers inflections. Both of
them are equally functional; i.e. neither of them is in any way better than the other –
they only use different means; in other words, they are built in different ways.
This fact was recognized a long time ago, in 18th and 19th century philology and
philosophy, by eminent German scholars such as August Wilhelm Schlegel and Wil-
helm von Humboldt. Schlegel and Humboldt suggested that we could divide lan-
guages into different categories, depending on how they did their job. The actual
distinctions they proposed are a little bit complicated, but roughly we can say that
Latin is a great example of what can be called a fusional language, i.e. a language
that signals syntactic functions (subject, predicate, object etc.) and relations (what
belongs to what in a sentence) by complex inflectional (grammatical) word forms.
Our example of puella puerum bāsiāvit illustrated the idea of syntactic functions,
and example (5.1) below shows the idea of syntactic relations. Note that, unlike Eng-
lish, Latin does not require that words which belong together also stand together.
So they can be separated by other words. What you need to do is check for the
inflections. These are marked in the second line of the example (the so-called gloss)
by a little dot plus some abbreviation. So, “.acc” stands for accusative, “.nom” for
nominative and so on (you will find a list of abbreviations at the beginning of this
book). This allows you to read and understand foreign languages even if you don’t
speak them.
(5.1) Parvam fɪ-lia mea casam no-n amat.

Little.ACC daughter.NOM my.NOM cottage.ACC not loves.
‘My daughter does not love the little cottage’
Note that parvam ‘little’ and casam ‘cottage’ share the same inflection, accusative,
which normally marks objects of all kinds. Similarly, both filia and mea are both marked
for nominative (which is often used for subjects). We know that a verb like amat (infin-
itive amāre ‘love’) requires a subject (the one who loves) and an object (somebody or
something that is loved). And when we put all of this together, we see that mea filia
(‘my daughter’) can only belong together and act as the subject, while parvam casam
‘(the) little cottage’ also belongs together and forms the object, even though parvam
and casam are separated by other words. What would change if we told you that mea is
the nominative form but meam is the corresponding accusative form, as in (5.2) below?
(5.2) Parvam fɪ-lia meam casam no-n amat.

Little.ACC daughter.NOM my.ACC cottage.NOM not loves.
The translation is very different. It is no longer ‘my daughter’, because filia is still in
the nominative, but meam is not – it’s accusative now, so they can’t belong together.
Instead, meam unites with the other accusative elements in the sentence, parvam
and casam, leading to ‘my little cottage’. So the new translation is ‘(The) daughter
does not love my little cottage’. This is what inflections and fusional languages can
do for you. The important point to remember is that with many inflections, the
individual morpheme carries more than one function; the different functions are
“fused” into the morpheme. The inflection -at on the verb amāre ‘love’ tells you that
this is present tense, indicative, active, third person and singular: that’s five func-
tions in one! Inflectional morphemes in fusional languages can have a number of
(different) functions; sometimes it’s five or more, and sometimes it’s only one. Latin,
Armenian, French, Russian and many other languages show that kind of structure.
A second type of language that can be distinguished from fusional languages is
the so-called isolating language. Good examples of this type are Mandarin Chinese
and Vietnamese. The following sentences from Chinese (adapted from Language
Files 165) should illustrate the idea:
(5.3) [w�ͻmən tán gā ngtçín]

We play piano
‘We are playing the piano’
(5.4) [w�ͻmən tán gā ngtçín lə]

We play piano past
‘We played the piano’
(5.5) [tā dǎ w�ͻmən]

s/he beat we
‘S/he beat us’
As you can easily see here, every single “word” or morpheme has its own func-
tion or meaning (e.g. there is no past tense form for the verb play but rather the
invariant verb plus an independent past tense marker). Also note that the pronoun
[wɔ̌mən] ‘we’ does not change when it is used as the object, as in (5.5). In English,
you have to change we to us for that. In Chinese, it is just the word order that tells
you if ‘s/he hit us’ or ‘we hit him/her’. In English, you can count on word order and
different pronoun forms for that (but note, as we mentioned earlier, that English
speakers generally also have to rely on word order). And Latin has all these com-
plex inflections for that job. So, in some sense, Chinese is a little bit like English in
its dependence on word order, but it is even more extreme in this respect in that
it hardly marks any grammatical features at all. But it is also not 100% pure in this
typological categorization as an isolating language. The formerly independent plu-
ral “word” [mən], for example, seems to have turned into a morphological plural
“marker”, like an inflection. We will return to this below.
A third common type of language is the so-called agglutinating language. Exam-
ples include Finnish, Nahuatl and Swahili, among many others. The set of examples
in (5.6) below illustrates such an agglutinating language, in this case Swahili (from
Language Files 167).
(5.6) [ni-na-soma] I-present-read ‘I am reading’

[u-na-soma] you-present-read ‘you are reading’
[a-na-soma] s/he-present-read ‘s/he is reading’
[ni-li-soma] I-past-read ‘I was reading’
[u-li-soma] you-past-read ‘you were reading’
[a-li-soma] s/he-past-read ‘s/he was reading’
[ni-ta-soma] I-future-read ‘I will read’
[u-ta-soma] you-future-read ‘you will read’
Agglutinating languages appear to be morphologically rather straightforward.

Every (invariant) morpheme has only one meaning, and they are all attached
(“glued”) rather loosely to their stem. In the case of Swahili, there are invariant
pronoun forms (ni = ‘I’, u = ‘you’, a = ‘s/he’) and invariant forms for the tenses
(na = present, li = past and ta = future). And all you need to do is glue them together!
And we are willing to bet that you can already speak a bit of Swahili now: how do
you say ‘s/he will read’ in Swahili?
Note that both fusional languages and agglutinating languages have inflections. In
fusional languages, inflections tend to have more than one function (which means that
words can be rather short), whereas in agglutinating languages inflections typically only
have a single function, so that they tend to pile up and can give you pretty long words.
If we compare the Swahili example ‘s/he read’ with, say, German (which is fusional) you
will see the difference: German las ‘read (past)’ includes (1) the idea of reading is taking
place; (2) the event takes place in the past, (3) the subject is third person; and (4) the
subject is singular. In a prototypical agglutinating language, each of these four meanings
would have clearly identifiable parts, but in las we can’t draw clear boundaries in this
way. Moreover, change any of these meanings, and the verb takes a different form (e.g.
‘s/he reads’ in German is liest). A good description of fused (unsegmentable) words like
las is “portmanteau morph” (here you have to picture a bag with multiple compartments).
There is one last type that we need to look at briefly – the so-called polysynthetic
language. The term “synthetic” describes those languages whose words usually con-
tain more than one morpheme (so it includes both inflectional and agglutinating
languages). So polysynthetic languages belong to the synthetic type, but they are
rather extreme examples. Yupik, an indigenous language in Siberia and Alaska,
presents one such example (5.7) (Mithun 2001: 38):
(5.7) kaipiallrulliniuk
kai -pia -llru -llini -u -k
be.hungry -really -PAST -apparently -INDICATIVE -they.two
‘The two of them were apparently really hungry’
In (5.7) you can see one single word – kaipiallrulliniuk – expressing one single
sentence in English, ‘The two of them were apparently really hungry’. This is obvi-
ously an extremely complex structure and can be taken to represent the other end
of the spectrum, with isolating languages at the simpler end. Again, this is not to
say that one of the two languages is less elegant, efficient or perfect. All existing
languages by definition fulfil their function(s), and they all do their jobs, albeit in
different ways. In a polysynthetic language like Yupik, many different morphemes
(and what other languages would treat as independent words) are integrated into
one single word. This single word then is a whole sentence. The results can be very
opaque and impressive for the non-native speaker.
A final word of caution seems in order before we turn to questions of linguis-
tic change. So far, we have given the impression that there are ideal languages out
there that clearly belong to one type or the other. This is not so. There are very
few, if any, languages out there that clearly and only belong to the fusional, (poly)
synthetic, isolating or agglutinating type. Most languages fit a type to a matter of
degree – Swahili is not a purely agglutinating language, just as Mandarin Chinese is
not purely isolating, and neither Latin nor German is purely fusional. For all these
languages we find exceptions. So one could argue, for example, that the invariant
pronoun ni ‘I’ in Swahili marks not only first person but also singular (the first
person plural pronoun is wa). This would mean that ni has not one but at least two
functions. We saw that Mandarin Chinese showed traces of inflections, and some
words in Latin never change. So, in a nutshell, this is a question of degree, rather
than of “yes” or “no”. Languages tend to belong to one type or another. And they
seem to be in constant flux, as we will see in the next section.
How do these different morphosyntactic language types relate to linguistic
change? Schlegel, Schleicher, Humboldt and many others saw some evolutionary
development behind these different types. They assumed that isolating languages
(e.g. Chinese) gradually evolve into agglutinating languages and then into fusional
languages (which were sometimes, erroneously, also seen as superior to the other
types). The whole idea of a change in language typology was refined by the hypothe-
sis that this kind of change could be cyclical, i.e. proceeding from isolating to agglu-
tinating to fusional and back to isolating, starting all over again. This is visualized
in Figure 5.3.
As we will see more in Chapter 6, a striking theme running through the story
of English is the unrelenting erosion of grammatical endings and the shift to a
more isolating structure. Functions carried out by the -er and -est on adjectives are
being taken over by more and most (e.g. more beautiful has evicted *beautifuller);
possessive -s, past -ed and plural -s are also endangered. Some of the more recent
Figure 5.3
From fusional to isolating to agglutinating and back again
Fusional Isolating
Agglutinating
changes in Chinese appear to be shifting it in the direction of agglutination. Tra-

ditionally, the first person pronoun was wŏ, no matter whether plural or singular.
But now we seem to find that men has turned from an individual word into a suffix,
so that we see wŏ ‘I’ versus wŏmən ‘we’ (see examples (5.3) to (5.5)). Similarly, the
formerly independent lə (the past tense marker) seems to have turned into a
completed-action suffix: qù ‘go’ versus qùlə ‘went’. Scholars of Chinese have also
noticed the acquisition of agglutinating features in the area of word formation.
Banfi and Arcodia (2008: 202) give the example of the noun forming suffix -shēng/-
sheng from the abbreviation of a compound (e.g. īshēng ‘physician’; compare Eng-
lish -gate from Watergate). Similar changes in word formation and compounding
lead us to believe that Chinese is gradually turning into an agglutinating language
(though it’s still a long way before it looks like Turkish or Swahili).
Dixon (1997) also proposes a cyclical development of language from fusional to
isolating to agglutinative and back to fusional. In his discussion, Dixon uses the
elegant metaphor of a clock to illustrate this. Fusional languages could be placed
at 12:00, isolating languages at 04:00 and agglutinating languages at 08:00. Thus,
it becomes immediately clear that when languages move between these different
times (and types), they show a mix of different structures and rarely can be found
in their pure forms. Dixon claims that Early Chinese, for example, could have been
at 03:00 as it was mostly isolating but still showed some fusional elements. Classical
Chinese would then be at 04:00 and Modern Chinese (mostly isolating but on its
way to becoming agglutinating; see above) around 05:00. Classical Egyptian, Dixon
says, is one language that has gone full circle: from fusional through isolating to
agglutinating and back again in about 3,000 years.
But how do languages move from one type to another? When they move from
fusional to isolating, languages develop simpler morphology; i.e. they lose mor-
phological (inflectional) information and develop independent and autonomous
elements that now encapsulate the different grammatical pieces of information that
were formerly coded in inflectional morphemes. During the move from isolating to
agglutinating these independent elements are fused again, i.e. glued onto the lexical
material in the sentence. Note, however, that each of the elements still carries its
particular piece of information independently. When languages finally move from

agglutinating to fusional, we see the merger of different and independent grammat-
ical markers into one single complex inflectional affix. There are many reasons for
these apparently dramatic changes: language contact, erosion through performance,
and expressivity, to name but a few (we discussed many of these in Chapters 1 and
4). Nevertheless, it is important not to forget that changes such as these do not have
to occur (some languages are stable over a long period of time) and that, if they do,
this can be a very long-term process that can easily stretch out over several millen-
nia. There are some counterexamples, such as English, which seems to have moved
from fusional to almost isolating in less than 1,000 years. But this seems to be a rare
case of rapid development when we look at the broad picture, and undoubtedly
language contact had a role to play here (inflections, as you will see in Chapter 8, are
generally the casualties in any contact situation).
5.4 WHY – EXPLAINING MORPHOLOGICAL CHANGES
In this final section we need to briefly talk about the reasons for morphologi-
cal change. Why do inflectional systems change? Why do languages shift in their
typology?
As we have just mentioned, one of the most important reasons, probably, does
not lie in the morphology of language(s) but in phonology. The vast majority of
languages code grammatical meaning in suffixes, i.e. in little markers that follow
the stem, as we have seen above in English and Latin. (The World Atlas of Linguistic
Structures Online list 406 languages that use predominantly suffixes in their inflec-
tional morphology versus only 58 that are predominantly prefixing; Dryer 2013.)
The problem is that the ending of a word is very susceptible to erosion, i.e. the loss
of phonetic material (speakers often don’t pronounce the endings of words very
clearly, as we have seen in Chapter 4). For example, in informal spoken German,
the two words hast Du [hast du] ‘have you’ are often fused and pronounced as hasse
[hasə]. In informal spoken English, we find expressions such as gonna and wanna.
We saw in Chapter 4 how with time phonetic erosion eliminates a lot of word-
final sounds, and this is where grammatical meaning is usually encoded. If a lan-
guage goes down that path long enough, it might lose its means to signal syntactic
functions and relationships through grammatical inflections. As a consequence, it
makes a lot of sense to find another means to do the job, for example word order.
And this is what we see when a language like Old English (strongly synthetic) turns
into Modern English (which is a lot more isolating). We return to this topic in
Chapter 6 when we focus more on syntactic change.
A second reason for morphological change of the kind discussed here is also of a
very general nature. This has to do with frequency and exposition. It’s a fairly well
established fact in cognitive psychology that human cognition is very susceptible
to how often it is exposed to something. To put this more simply: it makes a huge
difference for your cognitive processing if you encounter something only once or
a million times. This is also true and important for language. We can generally
distinguish between high-frequency and low-frequency items. There is no clear

definition for when some linguistic item is low or high frequency, but we can trust
our intuitions in this matter. Just look at what could be the ten most frequent words
in English:
1 the
2 be
3 to
4 of
5 and
6 a
7 in
8 that
9 have
10 I
It is immediately clear that these are fairly short, monosyllabic words; that they
come from the Germanic word stock; and that they are grammatical rather than
lexical.
Compare these to some words that we would classify as low-frequency items: gal-
actophorous ‘bearing or carrying milk’, mulct ‘fine, penalty’ and truckle ‘a small barrel
shaped cheese’. The likelihood that you have heard these words before (or will ever see
them again) is comparatively small. But what does that have to do with morphology?
On the one hand, high-frequency items are more often affected by phonological
erosion, and we touched on this in Chapter 4. This is, of course, very plausible. Only
think about your favourite teddy bear, the one you touched and played with several
times a day – he’d be the one with the threadbare fur. Similarly, the words that you
use very often will be more likely to erode phonetically. One nice example could be
and. You can use the full form [ænd], but in many cases speakers opt for shorter
forms, such as [æn], [ən] or even just [n] (as in Guns N’ Roses). But even though
these high-frequency forms tend to erode a little bit faster, they are less likely to be
affected by other factors, such as analogy or language contact. This is due to the fact
that high-frequency items are more deeply entrenched in the mind (again, you are
less likely to forget your favourite teddy bear or confuse it with another one). One
example is the plural of child: children. This is clearly a high-frequency item, and yet
there are practically no examples of regular childs in present-day English. So we can
expect phonetic erosion of children but not analogical change and regularization.
Low-frequency items, on the other hand, are usually not subject to erosion; i.e.
they don’t wear out that much. Rather, these are more often affected by analogical
changes! In Chapter 1 we asked you what the plural of fungus is. No doubt, this should
be a low-frequency item for most speakers (unless you are a biologist specializing in
mycology). This means that we don’t readily know the answer to questions like these
and therefore tend to reach for some analogical pattern: one bus, two buses = one fun-
gus, two funguses. The actual plural is fungi – but this is not very deeply entrenched for
most of us. So even if we have heard that before, we may have forgotten about it. You
could have built another analogy with words like syllabus-syllabi of course, but this
would require that you know these words and use them at least occasionally.
Note that this problem is not just about nouns. We could also ask for the past
tense of verbs like thrive or plead (throve or thrived? pleaded or pled?), and again, if
you don’t remember the past tense forms for these rather infrequent verbs, you look
for analogical patterns.
In sum, while high-frequency items are more affected by phonological erosion,
they are less likely to be subject to analogical change. Low-frequency items are less
likely to wear out but are rather subject to analogical change.
Tokens versus types

What we have just discussed for high- and low-frequency items also applies
to patterns, of course. In this case we need to distinguish types and tokens in
language. Tokens are quite simply the actual number of items that you find in
a given utterance. Types, on the other hand, only count each form once and
so help us to identify the underlying patterns. Consider the famous example
(5.8) below.
(5.8) A rose is a rose is a rose is a rose.

(Gertrude Stein)
How many types and tokens do we find here? There are 11 words here, and
we count all of them as tokens; so the number of tokens is simply 11. But how
many types? How many different words? We see a (4 times), rose (4 times)
and is (3 times). Each of them is counted only once, so that we find 3 different
words, or types. Now consider example (5.9):
(5.9) Johnny sings in the choir, Mary plays in a band, and their father
draws excellent pictures.
Again, there is the question of how many tokens and how many types we
find in (5.9). If we just look at the words, the answer is very simple: 21 tokens
and 20 types (only in occurs twice). But if we ask for morphological pat-
terns, matters are somewhat different. How many verbs do we have in (5.9)?
Three: sings, plays, draws. And how many different types of verb inflection?
Only one, the famous, regular third person -s. But we encounter this type of
inflectional morphology with three different tokens (sing-s, play-s, draw-s).
This gives us a broad database and a good chance to recognize the pattern (or
type). Now compare (5.10) and (5.11):
(5.10) Johnny plays the guitar, Mary plays the piano and their father
plays tennis.
(5.11) Johnny sings in the choir, Mary can play the piano, and their
father won a gold medal in figure skating.
At first sight, (5.10) and (5.11) seem very similar, but as databases they are
actually very different. In (5.10) we have 13 tokens, in (5.11) we find 20. Look
at the verbs again. In (5.10) we find plays, plays and plays: 3 tokens which are
exactly the same. They also exemplify the pattern of third person -s, of course,
but they are not as “impressive” as the 3 different tokens in (5.9). In (5.11) we
find sings, can and won: 3 different tokens but also 3 different types! We have
third person -s on sing-s, no visible inflection on can, and a past tense form
(without -s or even -ed) in won: 3 different patterns, i.e. 3 different types make
it very hard to discern a pattern here.
Let’s try to connect these findings with our ideas on frequency effects, out-
lined above. The more often we see a pattern (a type instantiated with differ-
ent tokens), the more likely we are to recognize this as a pattern, and the more
likely we are to use it as a pattern when we build our words. In other words, a
high type frequency (or better: many tokens with the same type) fosters ana-
logical thinking. A high token frequency as such helps to establish the token
but not necessarily the type, or pattern.
SUMMARY
In this chapter we discussed some principles and mechanisms in changes in inflec-

tional morphology. We considered reanalysis and actualization as major mecha-
nisms of change when hearers encounter certain linguistic structures and try to
decode them. The idea behind reanalysis is that grammars in the minds of speak-
ers and hearers do not always match perfectly, and that “misunderstandings” in
the widest sense can lead to change, which can then be seen only in actualization
(when the outputs differ). We then considered analogy as a major force in lan-
guage and cognition. After a discussion of the mechanism itself and of analogical
thinking more generally, we elaborated on ways to constrain this idea. We followed
Kuryłowicz with his idea that analogical change could be like rain: you never know
when and where it will happen, but there are pipes, drains and gutters which can
help you predict where the water might flow (though this is also not infallible, as we
saw). After our critical discussion of Kuryłowicz’s six “laws” of analogy, we finally
turned to language typology and how different languages can use different means
to express the same ideas. We discussed the venerable distinction between isolating,
agglutinating and fusional languages and how languages can move from one type to
another. In the last section of this chapter we talked about some of the reasons for
morphological change, and we particularly highlighted language economy as one of
the factors, next to the equally important frequency effects.
Generally, we can say that morphology seems to be one of the most complex
and interesting problems in linguistics and linguistic change, not least because it
plays such an important role for many languages and is intricately connected to
almost every other linguistic level, from phonology through syntax and semantics
to pragmatics.
FURTHER READING
Dixon (1997) introduces you to some ideas on linguistic typology, typological

developments and the large-scale developments we find not just in languages but
also in language. Hock (1991), especially Chapters 9 and 10, is perhaps not an easy
read for the beginner but provides a very interesting and comprehensive intro-
duction to analogy, and a fair treatment of both Kuryłowicz and Mańczak. Bybee
(2010) is an extremely smart, interesting and readable up-to-date introduction to
frequency effects in language and language change – another must-read. Greenberg
(1960) calculated the ratios of the number of morphemes over the number of words
for different languages and is a useful reference for morphological typology; see
also Lyons (1968) and Haspelmath and Sims (2010). We briefly mentioned gram-
maticalization here (i.e. the development of grammar out of lexical items) and take
it up in more detail in the next chapter. If you’re interested in following it up now,
we recommend Hopper and Traugott (2003), which was one of the first textbooks
ever to discuss the omnipresent and (to this day) highly controversial process.
EXERCISES
1 Morphological typology
Where on Dixon’s clock would you place Modern English? Give your reasons for
your assessment.
2 Morphological change in English

Trace the history of elder versus older in English and discuss your findings. (Hint:
Check out the history of the words in the OED.)
3 Morphological change in German

In Modern German, there is some confusion about the “proper” past tense of some
verbs, while others show no such variability. This is illustrated in the following table.
backen buk? backte? ‘bake’

hacken — hackte ‘chop’
weben wob? webte? ‘weave’
leben — lebte ‘live’
preisen pries? preiste? ‘praise’
weisen wies — ‘show/point’

gleichen glich? gleichte? ‘resemble’
weichen wich — ‘give way’
fragen frug? fragte? ‘ask’
tragen trug — ‘carry’
Can you guess what the older and the newer forms of the variable verbs are? Where
does this variability come from, and does it follow our predictions? The irregular
past tense frug seems to be pretty rare today (in contrast to glich). Can you speculate
why this is the case?
4 Analogy
Analogy can affect a number of different linguistic phenomena — phonological,
morphological, syntactic, semantic and even orthographic patterns can be general-
ized in this way. With examples drawn from any part of the language, briefly discuss
why it is that analogy is very often described as some kind of tidying up or pattern
neatening process.
5 Research project: degrees of morphological complexity

It is possible to calculate the degree to which a language is isolating by taking a
sample of continuous text and working out the number of morphemes over the
number of words – the lower the ratio, the more isolating the language. A ratio of
1.0 would be an ideal isolating language, and agglutinating and fusional languages
would show a higher morpheme-to-word ratio. Modern English at 1.68 is much
more isolating than Old English at 2.12 or Sanskrit at 2.59. The ratio for polysyn-
thetic Greenlandic is a whopping 3.72 (Greenberg 1960 calculated these ratios; see
also Haspelmath and Sims 2010: 5–6).
Your task here is to take English and one other language and calculate how much
morphology they have; in other words, the degree to which they are isolating.
a Select comparable texts across the languages, and take around 300 words of
running text.
b Calculate the ratio of morphemes to words.
c Compare your findings and describe how the two languages differ. Plot
them along a continuum of isolation.
d The figures are average ratios over running text, but languages are often rel-
atively isolating with respect to certain classes of words and agglutinating/
inflecting with respect to other classes of words. Hypothesize how different
your figures might be if you could calculate the ratios over all the words in
the language, counting each word only once.
e Can you describe the effects of style on this exercise? Would you expect dif-
ferent results if you had selected, say, a piece of more formal writing?
f With examples, describe how English has characteristics of all three mor-
phological types (isolating, agglutinating and inflecting). Give examples
to suggest how it is losing its morphology and becoming more isolating.

(Some of you might be brave enough to take a sample of Old English to
calculate the ratio of isolation.)
NOTE
1 In fact, sistren was around in Middle English but was eventually ousted by sisters; there
was no sign of it by the late 16th century.
6
Changes in sentence structure
INTRODUCTION
For a long time, historical linguistics concentrated mostly on phonological and

morphological changes (as we discussed in Chapters 4 and 5). Historical grammar,
more particularly syntax and syntactic theory, was a relatively neglected area. How-
ever, from the late 1960s onwards, together with the growth of influential general
theories of syntax have also come studies in the historical principles and processes
governing syntactic change. Indeed, research in this area has become extensive and
now encompasses a wide range of very different frameworks, including Chomskyan
linguistics and “universal grammar” (e.g. Lightfoot 1998; Crisma and Longobardi
2009), typology (Miller 2010; Ledgeway 2012), construction grammar (Traugott
and Trousdale 2014; Barðdal et al. 2015) and grammaticalization theory (Fischer
et al. 2000; Hopper and Traugott 2003). This chapter focuses on two theoreti-
cal perspectives of grammatical change: namely the typological perspective and
grammaticalization.
Because morphology and syntax interact so closely, some of our discussion here
draws on concepts already introduced in Chapter 5. Change often involves both
these aspects of language, and the blended label “morphosyntax” is a useful cover
term that captures this relationship. In fact, grammatical change can also include
change at the phonological and semantic levels. Recall our account of typological
shifts: fusional > isolating > agglutinating > fusional and so on. Here we see the
interplay of morphosyntactic as well as phonological and even semantic processes.
To move from fusional to isolating, languages develop simpler morphology; they
lose morphological information (inflections). To move from isolating to agglutinat-
ing, such independent elements must now be fused again. And from agglutinating
to fusional, we see the merging of independent grammatical markers into one single
complex (inflectional) affix. All this happens via the sorts of reductive phonological
processes we examined in Chapter 4, working together with semantics and mor-
phosyntax. And we will see other examples of this interaction later in this chapter.
Doing historical syntax is hard

It is worth pointing out that much of the early neglect of historical syntax had to do
with the practical difficulties of carrying it out, and mostly this arose from the lack
of adequate documentation of earlier stages of languages. As Labov once famously
wrote, “[h]istorical linguistics may be characterized as the art of making the best
use of bad data” (1982: 20). Even for well-recorded languages (and they are in the
Changes in sentence structure 135
minority), empirical evidence can be lost or damaged, the scripts illegible and the
orthography difficult. Such deficiencies are especially problematic for a grammati-
cal investigation because we are faced with hundreds of different constructions and
categories – the chances of accidental gaps and discontinuities in the data are very
real. In contrast, sounds can be listed, and because of the finiteness of phonological
inventories, descriptions can be gained of lost phonological systems sometimes on
the basis of even a very limited corpus (though of course recovering spoken styles
of the past from textual records comes with its own host of difficulties).
Added to the problems is the difficulty of making grammaticality judgments
without recourse to native speaker intuitions, especially when you come across
constructions that are of low frequency. Consider the following actual example of
a subordinate clause (the so-called linking relative) from modern conversational
English.
(6.1)
I’m taking them to Kangaroo Ground, which hopefully they won’t have too
much culture shock over there.
These linking relatives are rarely encountered in spoken corpora (collections of

spoken and/or written texts), but researchers can evaluate sentences and grammati-
cal constructions that occur rarely (or perhaps not at all) in their text collections
by asking native speakers for grammaticality judgments, for example via surveys.
Of course, where a construction occurs sporadically in early texts and then gains
currency later on, we can assume that this represents an innovation in the language
(and probably reflects the conservatism of the written language with respect to the
spread of the change). However, the chance of an early “performance error” is also a
very real one. At best we might assign it marginal grammatical status.
Equally, it is impossible to say with any confidence that a certain construction
type is ungrammatical (that it does not exist) just because it isn’t present in the data.
We call this the lack of negative evidence. If you didn’t find a linking relative in your
corpus of Modern English it would be a gap in the data. Here is another example.
These days we can make an offer for something by using a very much reduced ques-
tion. Instead of saying, “Would you like some coffee?,” we can just say, “Coffee?”
with a rising intonation. When you check the major corpora of English (try it for
yourself at http://corpus.byu.edu/corpora.asp), you will find that this mostly occurs
with food, drinks and drugs. Does this mean that we can’t use a reduced question to
offer anything else? Absolutely not. It is perfectly natural to say, “Handout?” when
offering a handout, or “Towel?” when offering a towel. So if we were to claim, on
the basis of our corpus findings, that this is not possible, we’d be very wrong. True,
a structure that cannot be found in one billion words of text may well be ungram-
matical – but it may well be quite grammatical too. So corpora can only give you
positive evidence; with negative evidence (or a low degree of likelihood) we need to
be much more careful.
This is of course precisely the problem we have in historical linguistics. All the
data we have is necessarily corpus data. We can’t go back in time and ask native
speakers if some structure is acceptable or not. Even worse, most historical corpora
are fairly small and limited in scope. While modern corpora of English, for exam-
ple, easily go beyond half a billion words, many historical corpora hover around
one to two million words of text. So it becomes even more difficult to say what sen-
tence structures could have been ungrammatical at a certain point in time. Again,
we have to rely on statistical likelihood, and because we need so much more data to
do historical grammar, there is always a very real chance of gaps.
Another wrinkle is that grammatical structures are governed by both linguistic
and extra-linguistic factors, including the vagaries of style. Think of the array of
syntactic patterns you might expect in different text types such as personal letters,
biblical sermons, drama, legal documents and so on. So to gain an accurate picture
of the relative chronology of any changes, investigations of grammar ideally draw
on text material that is as homogeneous as possible over the time span selected
(comparable genres, authors and even thematic material). Stylistic matters do not
pose the same problems for phonology. As Romaine (1982) discusses, phonemes
are largely independent of factors like style. The occurrence of phonemes is gov-
erned by language alone, so that a relatively small sample, regardless of stylistic
considerations, will yield typical results for any phonemic investigation. Compare
the occurrence of word order patterns, which is governed by both linguistic and
extra-linguistic factors like style (i.e. involving positive choice). The fact is, the more
language conditioned a feature is (independent of extra-linguistic factors such as
stylistic factors), the more random its frequency is likely to be and therefore the
smaller the sample needed. A study of the history of word order patterns will require
a much larger sample than the historical study of any sound system. Unfortunately,
for most languages this is unattainable, and the conclusions we draw regarding their
early syntax are all the more uncertain for it.
Gothic provides the earliest continuous text in a Germanic language and offers
a good illustration of this. The surviving evidence of Gothic (thanks to Wulfila
the Visigoth) exists in fragments of a Bible translation from the 4th century. From
this, much definitive work has been carried out on the phonological system of
the language, but little can be said about its grammatical system, in particular its
syntax – all the more because the text is a close translation of the Greek. When a
new manuscript page was discovered on 23 April 1971, it sent waves of excitement
through the world of historical linguistics. But while it shed additional light on
the inflectional morphology and provided some new lexical items, it could add
nothing to the knowledge of Gothic syntax.
Standardization
When languages change this creates fuzziness – a murky time when the new
expressions and constructions fall into a kind of linguistic no man’s land.
They’re not clearly “misuse” but for some speakers not yet “use” either. If
asked, many native speakers of Modern English would find linking relative
clauses peculiar, and they probably would reject sentences like (6.1). The dif-
ficulties are compounded because of standardization – a process that forces
languages into tidy classificatory systems with clear boundaries as to what is
and is not acceptable.
And this standardization process causes additional difficulties for histori-
cal linguists. Imagine you’re conducting a long-term study of English syntax
from Old English to modern times. You would try to minimize the interfer-
ence of stylistic considerations by choosing texts with little stylization and lit-
erary ambition, and keeping the texts as uniform as possible over the selected
time span. These texts would show some striking changes over this period.
One is the shift from flexible word order to grammatically controlled Subject-
Verb-Object word order (which we discuss below), and another is the marked
increase in syntactic complexity (lots of subordinate clauses). But how can
you be sure that these changes aren’t simply the symptom of an emerging
autonomous prose style; in other words, the fall-out of the transition from the
organizational principles of unplanned discourse (characteristic of speech)
to the more elaborated code typical of planned discourse (characteristic of
writing)?
Writing conventions (even ones that are just emerging) will always restrict
what we can observe. In the case of English, the emergence of the standard
language and the effects of a growing literary tradition, together with increas-
ing literacy, have meant that written texts that once would have provided
clues about early forms of speech are no longer as revealing.
In short, there is always greater uncertainty about conclusions made for syntax
than for phonology or morphology. No finite corpus of utterances, no matter how
large, will provide material for a full description of the syntax of a language at some
earlier stage of its development.
6.1 CHANGE IN WORD ORDER
Shifts in the ordering of sentence constituents (or what is usually referred to as word
order change) offer an example of a systematic and far-reaching type of change that
influences the behaviour of whole classes of words and affects the fundamental syn-
tactic organization of a language. The history of English is a good illustration of this
kind of significant transformation.
Old English showed considerable flexibility in the ordering of its basic constitu-
ents. In fact, all logically possible arrangements of subjects, verbs, objects and other
trappings appeared during that time. But this doesn’t mean that word order was
free. Old English represents a time of expressive word order, where the placement
of elements was controlled by information structure and other contextual consid-
erations (e.g. given versus new information, what the speaker considers important,
what s/he wants to emphasize and what s/he assumes the audience already knows).
By contrast, English today demonstrates what can be termed grammatical word
order. As described in Chapter 5, the fixed ordering of items functions syntactically
to indicate clause types and the grammatical relations within them (such as subject
and object); in a sentence like the queen loved the king, it is the constituent ordering
that tells us who is the lover and who is the lovee.
The following sentences show the possible word order variation documented for
Anglo-Saxon times. To make things straightforward, we have considered only the
ordering of the subject (=S), the main verb (=V) and everything else (=X), and we
have highlighted the verb strings in each example. All examples are taken from a
collection of 10th century Old English leechdoms ‘medicines’, covering everything
from hangover cures and advice for the removal of unwanted hair to remedies
against monstrous nocturnal visitors and the bites of mad dogs.1 Such medical
and medico-magical texts are useful for an investigation of syntax. Being techni-
cal prose with little in the way of literary ambition, they provide some of our best
evidence of a living, breathing speech community at that time.
Main clauses showed predominantly verb-second order (i.e. SVX and XVS).
(6.2)
and he ut bræcƥ wurmsig bloð
and he.NOM out breaks purulent.ACC blood.ACC
‘He coughs up infected blood’
(6.3)
ƥanne wite ƥu gewyslice
then know you.NOM certainly
‘Then you know for certain’
Although Modern English word order is fixed SVX order, there are relics of ear-
lier XVS (as in 6.3). Today, subject-verb inversion has a lively (even mock dramatic)
effect, which speakers and writers still love to exploit. Here’s an actual example:
Out will come beef dusted with Japanese pepper, fingers of salmon with dill
(6.4)
sauce and all that rocket in olive oil.
(Grand Bouffe, Good Weekend Age Magazine 16/2/96)
A number of other minor orders appeared in Old English; these include verb-
third (XSV), verb-initial (VSX) (which was a more expressive order often found in
vivid prose) and verb-final or near-to-verb-final order (SXV(X)).
(6.5)
ƥone mon ƥu meaht gelacnian æltæwlice
that.ACC man.ACC you.NOM might cure completely
‘That man you might cure completely’
(6.6)
gewitaƥ he sona aweg
go they NOM soon away
‘They (= the warts) will soon disappear’
(6.7)
ƥus ƥu hit scealt agitan
thus you.NOM it.ACC shall understand
‘Thus you will understand it’
(6.8)
yfelum deaƥe he his lif geændaƥ
evil.DAT death.DAT he.NOM his.ACC life.ACC ends
‘He shall end his life by an evil death’
Basically, verb-final order had a linking function and was common in coordinated
clauses in sequence (recall, the symbol 7 stands for ‘and’; it was one of the Tironian nota
‘signs’, from the shorthand conventions developed by Cicero’s scribe Marcus Tiro).
(6.9)
[. . .] 7 he eaþelic nan þing forswoligon ne mæg
and he.NOM easily no thing swallow NEG may
‘[. . .] and he is not able to swallow anything easily’
Subordinate clauses showed dominant verb-final (SXV) or near-to-verb-final

order. The order here is the consequence of the same cohesive function as in linked
main clauses (as in 6.9); in other words, verb-final signalled clauses that were bound
contextually to other clauses.
(6.10)
Gif men his wamb sar sy [. . .]
if person.DAT his.NOM belly.NOM sore be
‘If a person’s belly [lit. ‘to a person his belly’] is sore [. . .]’
It was also possible to find subordinate clauses with something closer to the mod-
ern ordering of the subject and the main verb in sequence (SVX).
(6.11)
Gif him biƥ ælfsogaƥa [. . .]
if him.DAT be elfsucked
‘If he [lit. ‘him’] is elf possessed [. . .]’
As an interesting aside, the subject in both (6.10) and (6.11) shows dative mark-
ing, so not the expected marking of subjects (which was nominative). In Old
English you could indicate that entities weren’t wilfully involved in an event by
placing them in the dative case, instead of the expected nominative case. Relics of
this construction are examples like me thinks.
Old English differed from Modern English also with respect to the ordering
within complex verb units; in other words, where there was a main verb and one
or more non-finite verb forms. In the following example, you can see that the main
verb, meaht, is in second position, and the rest of the verb unit, gebetan, comes later
in the clause. This construction is sometimes called the “verbal brace” because parts
of the verb form a kind of brace or bracket around the other clause constituents:
(6.12)
ne meaht hwaeƥere aeltaewlice gelacnian
not might however completely cure
‘Yet [you] cannot cure [it] completely’
All of these verb-final features of Old English are the fall-out of earlier times
when verb-final ordering was more widespread. Evidence from early Germanic dia-
lects strongly suggests that the neutral, unmarked order for both main and subordi-
nate clauses in the parent language, Proto-Germanic, was verb-final.
Verb-final ordering has left only a handful of remnants in Modern English. There
are some fixed expressions such as With this ring I thee wed, as well as compounds
like typeset, bloodshed, leasehold, roll call and woodcut that show fossilized OV
order (with someone setting the type, shedding blood and so on; we’ve bolded what
would have been the verb here); by comparison, compounds such as pickpocket and
pastime with VO order are much less usual.
Clearly English has undergone a major shift in word order. As the modern
translations here show, word order is now rigidly SVX. So why abandon this ear-
lier flexibility in favour of a system of grammatically rigid ordering? As touched
on briefly in Chapter 5, this change has traditionally been attributed to the loss
of morphological markers (the case system) – something is obviously going to
have to replace a collapsing inflectional system if basic grammatical functions are
to be signalled reliably, and fixed word order is one such remedy. But it is not a
straightforward matter of one system disappearing and then the other stepping
in to fill its shoes.
A more likely scenario is this: the gradual erosion of morphological markings
(through phonological processes) begins to take place, creating a need for some
form of grammatical change, if the language is to remain an effective vehicle of
communication. Increasing reliance on word order then renders these endings
redundant, and this hastens their demise. So intricately linked are the processes
that it probably doesn’t make much sense arguing for cause or symptom here. But
we emphasize, as we have on other occasions, that language change is never one-
dimensional, and the explanation for it isn’t either. There are undoubtedly many dif-
ferent interlocking factors involved here, which would also account for the different
chronologies we find in the Germanic languages. Behind this shift to SVX word
order would have been a network of different intersecting pressures: psychological,
physiological, systemic, social, political, external (contact and borrowing) and so on.
The complexity of changes of this nature is something we return to in Chapter 10.
6.1.1 Grammatical change is gradual

For a long time, scholars believed that grammar is a black and white, binary matter.
Either some change took place, or it didn’t. But we find more and more evidence
that grammatical change can also be gradual, i.e. that it can start with a few words
and constructions, then gain momentum, so that more words and constructions are
affected, until it runs out of steam.
There are close linguistic relatives of English that preserve some of the earlier
verb-final tendencies of Proto-Germanic. Modern Dutch and German, for exam-
ple, have SVO (verb-second) word order, but conservative SOV order remains in
these languages as a marker of subordinate clauses. Here’s an example from Dutch:
(6.13)
a Hij zegt dat hij een auto koopt
he say that he a car buys
‘He says that he buys a car’
Main clauses have also retained the “verbal brace” – only finite verbs have moved
to second position in the clause, while all non-finite verbs remain in final (or near-
to-final) position, forming a kind of bracket or brace around all other constituents
(shown here by the bolding):
b Ik heb een auto gekocht

I have a car bought
‘I have bought a car’
We would expect verb placement to eventually fall in line and SVO order to gen-
eralize to all clause types. In both these modern languages, certain material, princi-
pally adverbials (or what might be termed “afterthought material”), can appear after
the verb in subordinate clauses, and to the right of the end brace in main clauses,
and we give examples of this in Chapter 8. From what we know about the nature of
syntactic change, these kinds of violations of verb-final order are like the “thin end
of the wedge”, undermining and eventually destroying the OV remnants in these
languages. This is what Naro and Lemle once referred to as “sneaky diffusion”. As
they nicely describe it:
Syntactic change tends to sneak through a language, manifesting itself

most frequently under those circumstances in which it is least salient or
noticeable. This sneaky diffusion occurs along several distinct dimensions
simultaneously until eventually the whole language is entrapped.
(1976: 237)
In Chapter 7 we see examples of this kind of “sneaky diffusion” in phonological

change – a handful of words are affected at first (usually the frequent ones), and
the change then gathers momentum before it peters out. Grammatical changes are
subject to the same gradual dispersal.
“Sneaky diffusion” in English

The erosion of grammatical endings is a striking theme running through the
story of English. Of the rich inflectional system of Old English, there remains
only a tiny group of eight survivors. Two are the -er and -est endings that cre-
ate the comparative and superlative forms of adjectives. Typical adjectives are
gradable, which means they take part in the three-term system – something
is tasty, tastier or tastiest.
However, not all adjectives can take these endings, and the group that
does is becoming smaller. Phrases with more and most such as more tasty
and most tasty are pushing out forms like tastier and tastiest. Rather than
a clean takeover, however, we find the usual variability associated with lan-
guage change. But it is not random. Typically, adjectives of one syllable retain
the endings (e.g. bigger/biggest). Adjectives of three or more syllables have
lost the fight completely and require more and most (*beautifuller and *beau-
tifullest sound awful). Adjectives of two syllables are in flux – some take
the endings, some don’t, and some go both ways, and not all speakers agree
(compare simpler/simplest versus more simple/most simple). Adjectives with
-ly (e.g. costly) are at the forefront of the change, whereas adjectives with -y
(e.g. tasty) are lagging. But there is a lot of variation (see Bauer 1994: Chap-
ter 3; Peters 2000). Shakespeare’s horrider and certainer sound odd to us now.
But what about wickeder versus more wicked, or quieter versus more quiet –
they seem somewhere in between.
In Shakespeare’s time double constructions were commonplace – forms
like more larger and most unkindest where speakers had a bob each way. And
though they are still heard today, by the end of the 17th century they had
become linguistic outlaws; perceived as tautological, they were condemned
by the rules being laid down for the standard language.
It is conceivable all inflections will disappear with time, replaced by free-
standing (or analytic) structures (such as more and most here). We’ve seen
that languages do change typology according to a kind of cycle. Fusional
languages become less fusional and tend towards a more isolating structure;
isolating languages tend to move towards agglutinating languages, which in
turn tend to move towards the fusional type. However, it’s difficult to know
for sure what will happen. We can take note of what we imagine to be changes
underway in the language, but we can never be certain they’ll run their full
course. Predicting linguistic change can be one of the trickiest (most tricky?)
tasks confronting historical linguists.
6.2 TYPOLOGY AND WORD ORDER CHANGE
We’ve already seen that because all languages of the world show a range of similari-
ties and differences in their structure, it is possible to set up language types, that is,
groups of languages classified according to the features which they have in common
and which in turn differentiate them from other languages. These are typological
classifications, and they have been enormously important for historical linguists
because they help to define parameters for potential change. Chapter 5 looked at
groupings based on morphological structure (isolating, agglutinating, fusional).
Here we consider another central typological classification built on the basic (or
unmarked) ordering of the major constituents S(ubject), V(erb) and O(bject).
Of all the six mathematically possible word order combinations (SOV, SVO, VOS,
VSO, OVS and OSV), the majority of languages fall into one of three: SVO, SOV or
VSO. Of these, SVO and SOV are the most frequent, and VSO the least frequent.
The three other logically possible orders, VOS, OSV and OVS, are unusual – it is
rare to find the object preceding the subject. A factor here is likely to be the greater
saliency of the subject as the topic of a sentence, which places it more naturally in
initial position (it’s more important and interesting to know WHO did WHAT to
WHOM than WHOM was something done to by somebody . . .). This would also
explain the strong tendency towards SVO as an alternative order in VSO languages.
Thanks to the World Atlas of Language Structures Online (Dryer 2013), we can give
you an idea of the relative frequency of these six orders here. The figures provided in
Table 6.1 are based on declarative main clauses in which both the subject and object
are nouns (not pronouns), as in The boy read the book.
Typological classifications may or may not coincide with genetic classifications that
are based on shared inherited features (see Chapter 9). Languages belonging to one
genetic family can show vastly different word order typologies. Welsh, for example, is a
distant relative of English, but true to its Celtic origins has a basic word order pattern of
VSO. German and English are close relatives, but we’ve just seen that they differ typo-
logically. German has split word order: SVO in main clauses and SOV in subordinate
clauses. Russian is also related to English, but it has flexible word order – while SVO
Table 6.1
World Atlas of Language Structures Online frequencies for Subject, Object and Verb
Order of Subject, Object and Verb Representation

Subject-Object-Verb (SOV) 565
Subject-Verb-Object (SVO) 488
Verb-Subject-Object (VSO) 95
Verb-Object-Subject (VOS) 25
Object-Verb-Subject (OVS) 11
Object-Subject-Verb (OSV) 4
Lacking a dominant word order 189
Total: 1,377
is more usual, all possible permutations of S, V and O are grammatical. On the other
hand, we can find typological homogeneity between genetically diverse languages.
English shares basic SVO ordering with a number of genetically related languages like
Norwegian, French and Greek, but also many unrelated languages like Malay, Thai and
Mali. This is something we come back to in greater detail in Chapter 9.
Language change is what produces these mismatches between typological clas-
sifications and genetic classifications. A language’s basic type is just as prone to the
process of change as any other aspect of the language (and a good illustration is
the fact that the three dominant patterns, SVO, SOV and VSO, are all represented in the
languages that have descended from Indo-European). A decline of SOV order in
the world’s languages is due to the fact that many languages of Continental Europe
and Africa have undergone the same shift from SOV to SVO type (see, for example,
Hyman 1975 for Niger-Congo). Hence, typological similarities between genetically
unrelated languages can be because the languages have moved in the same direction.
Such changes can be the result of “forced change” (or language change through
contact; see Chapter 8). A striking example of this is the shift in Amharic from basic
VO to OV type as a result of contact with Cushitic. Unfortunately, it is often difficult
to distinguish between this type of change and a case of parallel but independently
motivated change. For example, what role did contact with SVO French play in the
increasing VO character of early English? Perhaps all we can say is that contact with
French simply accelerated changes that were already well underway.
6.2.1 The contribution of Joseph Greenberg and others

It was during the 1960s that the American linguist Joseph Greenberg wrote his
famous essay “Some Universals of Grammar with Particular Reference to the Order
of Meaningful Elements” (1963). This approach, and those it inspired, used typol-
ogy and typological universals to account for existing syntactic patterns and their
historical development.
Greenberg examined selected grammatical features (for the most part word order
patterns) from 30 languages of the world. He divided these into the three basic types
mentioned earlier – SVO, SOV and VSO. Data in his research showed correlations
between certain word order patterns and other grammatical properties. From his
findings, he listed 45 implicational universals of the type – “If a language A has fea-
ture F1, then (with more than chance frequency) it also has feature F2.” For example,
his Universals 3 and 4 read:
Universal 3: Languages with dominant VSO order are always prepositional.
Universal 4: With overwhelmingly greater than chance frequency, languages
with normal SOV order are postpositional.
Implicational statements of this sort seek to establish constant relationships

between different grammatical features of language. They represent near-universals,
tendencies based on those properties which emerge from the data as being the statis-
tically predominant ones, and against which patterns of change may be studied.
After Greenberg’s work appeared, there were a number of proposals that sought
to account for these word order correlations by appealing to the existence of a basic
ordering of elements to which all languages naturally conform. The work of two
linguists in particular exemplifies this position, namely Winfred P. Lehmann and
Theo Vennemann. Both linguists reformulated Greenberg’s original classification
into only two types: OV and VO. On the basis of his findings, they made one overall
generalization that correlated the position of all sentential elements to the relative
ordering of the verb and its direct object. Those languages conforming to either OV
or VO type in their word order patterns were therefore described as typologically
“consistent” (or “harmonious”) and those that did not as typologically “inconsis-
tent”. A consistent VO language, for example, would show patterns in which the
items in the A column in Table 6.2 precede those in the B column (consistent OV
languages would naturally show the reverse of these patterns).
According to this schema, then, English is an inconsistent VO language by
showing some noun phrase patterns harmonious with OV type (except for rela-
tive clauses and prepositional genitives that follow their head nouns and therefore
conform). Modern Romance languages like French and Italian are more consis-
tently VO by placing most modifiers after the noun: French un livre vert (lit. ‘a
book green’) versus English a green book (note that there is some variation in the
Romance languages, as we go on to discuss).
Both Vennemann and Lehmann offered structural principles (“the principle
of natural serialization” and “the structural principle of language”, respectively)
to account for the correlations they drew from Greenberg’s data. Roughly, these
principles work this way: constituents in the A and B lists have the same status as,
respectively, heads and modifiers (or dependents) of traditional and more recent
linguistic theory (for example, in the cross-categories central to X-bar theory).
What we have here, then, is a level of “pure type”. Pure OV and VO languages
represent a consistent implementation of the modifier-head relationship. This then
Table 6.2
Word order correlations (based on Vennemann 1974 and Lehmann 1973)
A B
Verb Object
auxiliary main verb
modal main verb
verb adverb
preposition noun
noun demonstrative
noun genitive
noun numeral
noun relative clause
noun descriptive adjective
comparative adjective standard of comparison
has historical application – the pressure to conform to this basic ordering is viewed
as strong enough to initiate change. Accordingly, structural changes are explained
as being goal driven (teleological). The ordering relationship between the syntactic
patterns in both the verb phrase and the noun phrase and the fact that they exhibit
parallel developments are therefore explained by both linguists in terms of direct
analogy; that is, the modifier stands in precisely the same relationship to its noun
as the complement or object stands to the main verb (Modifier : Noun :: Object :
Verb). As we go on to discuss, some linguists now give a processing explanation
for the harmony – the reduced cognitive load that comes with consistent ordering.
According to both linguists, only a change in verb position (brought about through
language contact, for example, or the breakdown of a case system) could trigger a
typological shift. Their account allowed for a considerable delay between this initial
change and later “harmonic” changes in other phrase types, to account for inconsis-
tent languages like English. The shift in English from OV to VO is complete, but the
noun phrase ordering remains principally that of Modifier-Noun. The appearance of a
prepositional possessive with Noun-Genitive order (e.g. cover of the book) that is push-
ing out the older inflectional possessive (e.g. the book’s cover) is therefore a predictable
move towards greater typological harmony. Similar is the development of Noun-
Adjective ordering in French – the handful of aberrant common usage (so highly vis-
ible) adjectives (like grand ‘large’ and petit ‘small’) that usually appear before the noun
can be explained as remnants of the OV type that characterized early Romance.
More recently, researchers have been greatly developing both the field of typo-
logical universals and their dynamic application. While these developments take
us way beyond the scope of this book, we should say something briefly about later
attempts to refine and further clarify the original implicational universals and also
find a rationale for them.
Early problems were not so much with the typological approach itself, but rather
with the misuse of it. Greenberg’s initial statements were extremely cautious, based
only on the data from his sample of 30 languages. For example, he found only one
verb phrase pattern implicationally related to a noun phrase pattern, namely the
correlation between VSO order and Noun-Adjective order. Subsequent research
(e.g. Dryer 2011) has confirmed the absence of any connection between the order
of object and verb and the order of adjective and noun. While there does exist a
strong correlation between Verb-Object order and adpositions (prepositions versus
postpositions), there is nothing to justify the move to correlate the ordering of all
sentence constituents to the head-modifier ordering as just described. Moreover,
Greenberg’s initial statements were largely unidirectional; e.g. OV order is a good
predictor of a case system (Greenberg’s Universal 41), but a case system doesn’t
predict OV order, nor does the lack of a case system imply VO order. There is no
justification to reformulate all correlations as bidirectional (as in Table 6.2). Finally,
there is no evidence to suggest that only verbs can motivate change. Many languages
show changes in the noun phrase to have occurred prior to any changes in the verb
phrase; for example, Germanic shifted to prepositions while it still had OV syntax,
as did Latin (see Lockwood 1968: Chapter 7 on the development of prepositions in
German; Miller 1975 on the shift from post- to prepositions in Latin).
Problematic cases
Word order typologies assume the viability of categories like subject, verb,
object, noun and adjective as basic linguistic entities of all languages. They
also assume the viability of basic word ordering in natural language. Such
basic orders do not exclude the possibility of other word orders. English
speakers can deviate from basic SVO order if they want to highlight a par-
ticular part of the message; for example, fronting elements is a common strat-
egy for emphasis, as in Salted licorice I adore! (OSV). Such variation on basic
ordering relates to factors to do with information flow (the communicative
function of sentences) and affects more particularly the positioning of the
verb and its arguments and not the ordering within other phrasal categories
like noun phrases and prepositional phrases.
But sometimes it’s not easy to decide on what is basic. We’ve seen that the
English possessive construction allows two possible orders: Noun-Genitive
and Genitive-Noun (the cover of the book versus the book’s cover). Which
of the two is more basic? True, we can use frequency, in terms of both lan-
guage use and grammar; markedness (for instance, a marked construction
may have a specific meaning associated with it); and also historical con-
siderations to help decide the question of basicness. But, even so, there are
times when these criteria conflict, and consequently no order emerges as
being truly basic.
Dryer (1997) also points out that the six-way word order typology (SVO,
SOV etc.) represents a clause type that is really rather rare in ordinary spoken
language, and he suggests that a more useful typology would be one based
on the relative position of the object and verb (OV or VO) and of the subject
and verb (SV or VS). This would give a four-way classification: OV/VS, OV/
SV, VO/VS and VO/SV.
But there are also many languages that just don’t seem to fit into this
schema at all – those with flexible or free word order, for example. For most
Australian languages (such as Yulparija and Dyirbal), it is theoretically
possible for words (even words belonging to the same notional phrase) to
appear in any order with no grammatical significance. While an analysis of
discourse may reveal word order preferences based on thematic organiza-
tion (considerations of topic and focus, for example), it is still not possible
in these languages to distinguish a basic ordering. Just as problematic are
those languages whose functional categories S and O don’t work in the same
way as in other languages; for example, ergative languages where the sub-
ject of an intransitive verb behaves like the object of a transitive verb. There
are also those that show ellipsis of these categories; for example, many lan-
guages, from diverse families, don’t require explicit subjects. Work is now
being done to accommodate these sorts of differences (see, for example,
Croft 2001).
Greenberg’s initial sample of 30 languages was small and also geographically

and genetically biased. Later research has been doing much to expand the sample;
the World Atlas of Language Structures (WALS) is a massive database of structural
properties of language, including different word order features and the correlations
among these features. Linguists such as Hawkins (2002, 2004) and Dryer (1992,
2011) have also attempted a more explanatory account of these universals, examin-
ing them from a performance perspective. The claim is that dominant word orders
and harmony arise from innate processing mechanisms, the forces involved in the
production and comprehension of language. Finally, since an important trigger
for languages to change their word order is contact with other languages, much
work has been done of late to understand the nature of contact induced change
(see Chapter 8), and therefore also the forces that are at work to shift a consistent
language into an inconsistent state.
We have spent some time on this approach because it was a real turning point in
the study of diachronic syntax. Its synchronic framework in the form of these word
order universals offered linguists a means of linking together a number of changes,
at the same time giving them meaning by showing them to be part of some overall
long-term trend, that is, towards consistency of type. In short, it allowed a theory of
syntactic change to emerge. There is no doubt that certain word order patterns “go
together”, and languages do seem to motivate changes favouring these patterns – even
the greatest critics of the approach don’t dispute this.
However, the intersecting pressures that work to drive languages in particular
directions (psychological, physiological, linguistic, social and cultural) operate
against regular and utterly predictable changes, and these so-called universals will
only ever be statements of likelihood or tendency, holding X% of the time. Moreover,
there will always be exceptional cases precisely because languages find themselves
in the middle of a shift. As many linguists have said to their students, “shift happens”.
So schemas for word order changes aren’t following prescribed courses determined
by absolute universals or rigid “laws”, but there are preferred pathways of change.
Typological universals point to these preferred pathways, and they give us a good
idea of what constitutes likely and unlikely changes. Think again of Kuryłowicz’s
“gutters of change” (discussed in Chapter 5).
6.3 CREATING GRAMMAR
Most speakers are very aware of change happening to the vocabulary of their lan-
guage, and many also notice changes to pronunciation. However, few realize that
languages are also constantly renewing their grammar. So where do grammatical
elements come from? Germanic languages like English and German developed
their own class of verbs indicating past tense with some sort of dental suffix (Eng-
lish played and German spielte) – where did this suffix come from? And what about
other tense and aspect markers, articles, conjunctions, pronouns and so on?
Where we know the history of grammatical bits and pieces, they have grown
out of ordinary words – everyday lexical items (those concrete items that relate
to different aspects of human experience). In the case of English -ed and German
-te, these suffixes emerged from a verb meaning something like ‘do’. So played was
originally akin to play + did. The creation of grammar is something called gram-
maticalization, and linguists studying it examine the lexical sources that give rise to
grammatical forms and constructions, as well as the changes that take place as the
items become more grammatical. And, as we will be seeing, these changes involve
a parcel of closely synchronized processes happening at the phonological, semantic
and grammatical levels.
There are two major mechanisms involved in grammaticalization – reanalysis in
the first place and extension (or analogy) in the second place. As you saw in Chap-
ters 2 and 5, both are important concepts in the explanation of lexical and morpho-
logical change. In this chapter our focus is on syntax, but the changes often involve
alterations to word structure as well. It isn’t easy (or useful) to separate morphology
and syntax, especially when it comes to the creation of grammar.
Recall that reanalysis is the process whereby a form comes to be analysed in a
different way. It involves the development of new structures out of old structures
and can have far-reaching implications for the grammar of a language – it can even
be the trigger for the rise of a new grammatical category. Colloquial French offers
a clear illustration of the process at work. And you’ll see now just how closely mor-
phology and syntax act together.
Similar to English, French generally forms many questions by reversing the pro-
noun subject and the verb. The following is an example:
(6.14)
Statement: Il vient. ‘He is coming’ [usually [ivjɛ̃]]
Question: Vient il? ‘Is he coming?’ [usually [vjɛt̃ i]]
A basic rule of French pronunciation is that final consonants are deleted unless
there is a following vowel. So, as in this example, sounds suggested by the spelling
are silent – the subject pronoun il ‘he/it’ is phonetically reduced to [i], and the -t
verb ending of vient is unpronounced in the statement [ivjɛ]̃ ‘He is coming’. How-
ever, when the word order is reversed for a question, the rules of French pronuncia-
tion (liaison) result in the phonetic re-appearance of the [t] in [vjɛt̃ i] ‘Is he coming?’.
It seems that in the minds of many French speakers, this stream of sound [vjɛt̃ i]
has been reanalysed, and the sequence -ti- has been interpreted as a marker indicat-
ing that a question is being asked (compare the English speaker who misinterprets
ice cream as I scream).
(6.15) Je savais ti?

I knew question particle
‘Did I know?’
Via reanalysis, we have here the creation of a totally new grammatical marker. Its
immediate ancestor is a -t ending of certain verbs and i (the reduced form of the
third person pronoun) – there’s been a boundary shift, and these two forms have
squished together to form the new marker for yes-no questions (for all persons). It
will be interesting to watch the progress of this question marker (as yet it remains
regional and has made no inroads into the standard language).
Only reanalysis can lead to the creation of new grammatical structures. Yet anal-
ogy is also hard at work in changes in this part of the language. It involves the gener-
alization of types of linguistic structure. As we saw in Chapter 5, it essentially takes
a new pattern (like the question particle ti) and spreads it around. Moreover, being
overt, analogy provides the evidence, for both speakers and linguists, that a change
has taken place (as in the reanalysis example we’ve just seen for French).
The development of a new infinitive clause marker in Pennsylvania German
offers a paradigm example of these dual forces of reanalysis and analogy at work.
As the following reconstructed chain of events shows, many different structural
aspects can be affected in the process – here constituency, hierarchical structure
and category membership. Consider the following example (6.16):
(6.16)
[Es is ungweenlich [fer de John]] [harti Bicher zu lese]
It is unusual for the John hard books to read
‘It’s unusual for John to read difficult books’
The original clause boundary would have occurred after the fer phrase with fer de
John ‘for (the) John’ forming a prepositional phrase that was part of the main clause,
and the complement clause signalled by the infinitival marker zu. The close logical
relationship existing here between the complement of the preposition and semantic
subject of lese ‘read’, John, comes to be expressed by reanalysing fer as a conjunction
(or complementizer) heading a new subordinate clause. The boundary shifts, and
the prepositional complement is reanalysed as the subject of the infinitive. Once
this stage has been reached, the fer-clause can be moved to appear in sentence ini-
tial position as in (6.16’).
(6.16’) Fer de John harti Bicher zu lese is ungweenlich

‘For John to read difficult books is unusual’
As the prepositional force of fer weakens and it expands into more and more con-
structions, it shifts from the status of a prepositional-like element to a complementizer-
like element, eventually taking over from zu. With further generalization, the fer-clause
can extend to appear after adjectives, nouns and verbs that would not otherwise take the
preposition fer. The (understood) subject of the infinitive verb is no longer restricted to
an original prepositional complement.
(6.17) Ich hab net’s haez fer saage er sol net

I have not.the heart for say he should not
‘I haven’t the heart to say he shouldn’t’
(6.18) Ich bin ledig fer so lang waarte

I am tired for so long wait
‘I am tired of waiting so long’
It is precisely this same development that has taken place to give rise to the Eng-
lish complementizer construction for . . . to, as well as the German infinitival con-
struction um . . . zu ‘in order to’. In both these cases the prepositions (for and um
‘around’) + NP originally belonged to the main clause but were later reanalysed as
part of the infinitival construction. Consider the following Middle English example
from Chaucer (from Harris and Campbell 1995: 62):
(6.19) [It is bet for me] [to sleen my self than ben defouled thus]
‘It is better for me to slay myself than to be violated thus’
The pronoun me was syntactically part of the constituent for me, but at the
same time it functioned as the logical subject of the infinitive to sleen ‘to slay’.
With time the boundary shifted, and the prepositional complement was reana-
lysed as the subject of the infinitive. So it’s a boundary shift, such as we saw
with the French and Pennsylvania German examples (and the phonetic equiva-
lent would be the change from mine Ed > Ned; an ekename > a nickname, as
described in Chapter 5). The reanalysis is now evident in the Modern English
version, where the whole clause can be fronted:
(6.19’) [For me to slay myself] [is better than to be violated thus]
6.3.1 Case study – the evolution of not in English

A good example to illustrate grammaticalization involves developments that have
taken place to produce the English negator not, and its journey from Old English
ne-a-wiht ‘not-ever-anything’. This is a change that involves both the morphology of
the negator and also its placement syntactically. In fact, as is typical of these changes
that give rise to new constructions, there is a parcel of closely interrelated processes
occurring at the grammatical, semantic and phonological levels.
Old English negated by placing a negative particle ne before the finite verb, but
often this was supported by one or more additional negative words, and we saw this
in the earlier example (6.9), repeated here.
(6.9) [. . .] 7 he eaþelic nan þing forswoligon mæg

and he.NOM easily no thing swallow may
‘[. . .] and he is not able to swallow anything easily’
By Middle English multiple negation was the norm, with two (and often more)
negators present.
(6.20) þere nys noþynge þat so sone smyteþ with grevaunce

there not-is nothing that so soon smites with grievance
þe hed as wyne
the head as wine
‘There’s nothing that damages the head as quickly as wine’
During this period, original ne-a-wiht ‘not-ever-anything’ became the favoured

reinforcer, and together with preverbal ne emerged as the normal expression of
negation. This is sometimes called “embracing negation” because these two negators
snugly wrapped themselves around the verb: hyt ne swellyþ noght ‘it not swells not’.
Eventually preverbal ne disappeared, and single postverbal noght/not established
itself as the preferred negator (of course multiple negation remains robust in non-
standard varieties today: I don’t know nothing).
(6.21) þe rede cale suffers noght þe wonde hele

the red kale suffers not the wound to-heal
‘The red kale doesn’t allow the wound to heal’
If we jump to the 16th century things are looking more familiar.
(6.22)
I would not that you should observe a certaine houre, either for dinners or suppers.
One and the same order of diet doth not promiscuously agree with all men.
Drinke not above four times.
At this time there are still some different negative patterns, as the third exam-
ple shows. In Modern English we need the support of do here; i.e. Do not drink.
Although auxiliary do was available to speakers at this time, it wasn’t until much
later that it became obligatory in negation (if no other auxiliary was present). The
origin of this do is disputed, and it won’t concern us here – suffice to say that it is
another example of grammaticalization at work, its source being the lexical verb do
‘to make, perform’. The rise of “dummy” do has been a significant development in
English negation (and it's something we investigate in Chapter 7).
To summarize, we can view the remodelling of English negation as an orderly
progression through three clearly identifiable (but overlapping) stages:
(Stage 1) preverbal ne > (Stage 2) embracing ne-not > (Stage 3) postverbal not
In what has become a classic account of negation, Otto Jespersen (1917) sug-
gested that the changes for English were cyclical in nature, and since then most
linguists have referred to this kind of negator renewal as “Jespersen’s Cycle”. Cross-
linguistic studies show that the change is extensively attested and in a wide range
of languages well beyond Germanic (and of course is potentially more widespread
given that poorly documented examples of single negation could have evolved via
precisely such a pattern of erosion and creation). Jespersen’s Cycle of negation is a

paradigm case of grammaticalization.
We can describe the different types of intertwined changes that have taken place
as follows. All are predictable changes associated with the making of grammar.
• Semantic changes
With grammaticalization words lose their more concrete lexical meanings and take
on much more general or abstract meanings. And the more general they become
in their meaning, the more frequently they’re used. The full expression ne-a-wiht
shifted from the literal meaning of ‘not-ever-anything’ to a general expression of
negation. You can compare some of the extra reinforcing expressions we find in
Modern English negation: I don’t care a bit, which shows the start of this kind of
generalization, as bit derives from the lexical item bite (and would originally have
strengthened verbs of eating).
• Phonological reduction
When words become more grammatical they typically become unstressed; con-
sonants and vowels reduce. Put quite simply, the words shorten. Bit with its short
vowel already shows signs of this kind of reduction. In the case of not, what was
formerly a three morpheme item (ne+a+wiht) is now an unsegmentable unit
not and -n’t in normal speech. In most rapid speech, in fact, not even disappears
entirely. Like the “Cheshire Cat’s grin” only the shadow remains in the form of
nasalization, for example, can’t be done [k~a:bidʌn]! So grammatical morphemes
can reduce to such an extent that all of their segmental phonemes disappear, a
process Jim Matisoff has appropriately dubbed “cheshirization” (see Hopper and
Traugott 2003: 157).
In Chapter 4 we saw how repetition leads to the reduction of form; since gram-
matical words are used more often than lexical words, it follows that frequency of
use will cause them to shorten. Note how all the grammatical elements of English
are little. They are also the less informative parts of a message, and for that reason
too they don’t need to be long (there exists in language an iconic relationship
between informativeness and size, the theory being that this correlation comes
about through the pressure of communicative efficiency; compare Zipf ’s law).
• Structural changes
Grammatical words can’t be stressed, they can’t be modified and we don’t move
them around for special focus as we do lexical words. As they become fixed in posi-
tion and as they reduce in size, they become increasingly dependent on surround-
ing material, eventually even fusing with it. The development not > -n’t nicely shows
this transition. But note that there is a continuum from free-standing to bound
morphology, and somewhere between an independent word and an affix lies a class
of morphemes known as clitics (based on the Greek verb meaning ‘to lean’). In
speech, clitics “lean on” neighbouring words (just like -n’t leans on verbs) – not able
to stand on their own but not yet an affix.
content item > grammatical word > clitic > inflectional affix
Jespersen’s Cycle of negation

French also offers a clear-cut example of Jespersen’s Cycle of negation. Here’s
a reproduction of Jespersen’s original 1917 schema (with English alongside
for comparison):
French English
(1) jeo ne dis ic ne secge
(2) je ne dis pas I ne seye not
(3) je dis pas I say not
(4) I do not / don’t say
The following summary of the changes spells out the dual forces of reanaly-
sis and analogy at work:
1 Negation is by preverbal ne.

2 Verbs are optionally strengthened with postverbal nominals (such as pas
‘step’, mie ‘crumb’, goutte ‘drop’).
3 The nominal pas ‘step’ (originally confined to motion verbs) is reanalysed
as a negator (reanalysis).
4 Pas is extended to verbs having nothing to do with movement (analogy).
5 Pas is reanalysed as the obligatory concomitant of ne (ne-pas).
6 Pas is now the obligatory negator (ne has been reanalysed as optional).
We only know that reanalysis has taken place (stage 3) because of stage 4
(analogical extension). This then makes the reanalysis at stage 6 possible.
6.3.2 Where is English negation heading now?

From original ne-a-wiht ‘not-ever-anything’, clitic n’t is an impressive illustration
of our love of compacting, reducing and obliterating boundaries. In fact, recently
there have been some forensic linguistic cases triggered by a barely vocalized nega-
tor. One of the most famous is detailed by the forensic linguist Roger Shuy, who
had to testify that the defendant actually said I wouldn’t take the bribe, rather than
I would take the bribe, as was being claimed (see Hitt 2012, who discusses other
examples of forensic linguistics at work).
Linguists Paul Hopper and Elizabeth Traugott (2003), who work in the area of
grammaticalization, emphasize the perpetual tug of war between, on the one hand,
creativity and expressiveness (i.e. speakers’ introduction of new and innovative
ways of saying things) and, on the other hand, routinization (i.e. speakers’ tendency
to reduce the speech signal – compacting, reducing, obliterating boundaries – as in
Old English ne-a-wiht ‘not-ever-anything’ > noght > not > -n’t). Given the signifi-
cance of negation, Modern English -n’t is clearly ripe for renewal.
The sources for new negative markers are relatively restricted. Willis et al. (2013)
outline three main options. Most commonly, negators arise from nominal minimiz-
ers, or what are sometimes termed “accusatives of smallest measure” such as we’ve just
seen with pas ‘step’. Also widespread are inherently negative pronouns and adverbs
such as nothing, nowhere, never, at all and so on. A third option is similar to the first,
except the strengthening marker is some sort of clause-final resumptive negator, such
as the simple interjection no (e.g. I didn’t do it, no) or an ordinary sentential negator
(e.g. I didn’t do it, no way), which is then reanalysed as part of the clause.
Change is typically marked by rivalry between variant forms, and in Modern
English there are a number of contenders for the job. One is the adverb never (lit.
‘not ever’), which in Englishes around the world is used as the general expression
of negation: I never ate it ‘I didn’t eat it’. Another candidate involves a group of
concrete nouns used to express minimal quantities when combined with negation.
They are typically emphatic, for example I didn’t drink a drop; He didn’t get a spot
of praise; I couldn’t walk a step; I didn’t eat a crumb; It didn’t hurt a bit; I don’t give a
damn. We could include here the highly colloquial ones, such as bugger all and jack
all, and even the flourishing of expressions that occur after I don’t give, like a rat’s
(arse), a toss or a fig. All of these are reminiscent of the early days of not, and any
one could potentially evolve into a negative marker (though the taboo quality of the
colourful ones probably means they are unlikely to ever gain the currency needed
to start the grammaticalization ball rolling).
There is nothing intrinsic about their meaning that predicts their development
into negators, but it is rather the way they combine with negators in discourse that
determines this particular development. Any one could start expanding its con-
texts. A likely contender is bit. It is already showing signs of semantic and phono-
logical change and has started to generalize into other contexts, no longer being
confined to verbs of eating, as in examples like It didn’t hurt a bit (compare a crumb,
which can only be used with verbs of eating: *It didn’t hurt a crumb).
Post-sentential not
Another interesting development in modern informal spoken English (since
the 1990s) is a form of post-sentential negation as in “What a totally amazing,
excellent discovery . . . NOT!!” (a quote from the movie Wayne’s World, which
helped to popularize the expression). These sorts of colloquial constructions
often provide the basis for real change, but in this case the positioning of not
is an obstacle. Negative markers typically precede the elements they negate,
with preverbal placement as the norm for standard clausal negation. There
appears to be something natural about the preverbal placement of the nega-
tive particle with evidence from second language learning and child language
to support this (Dahl 1979: 96–97). Such constraints on the positioning of
negators suggest that an innovation like post-sentential negation is never
likely to convert to a change.
The reason for negation to come early is probably obvious. It seems a
good idea to delay such a crucial part of the message as the negative ele-
ment . . . NOT! And of course this is exactly what stranded not plays on. It’s
a brilliant example of what is called a “garden-path sentence” – one that lures
the readers into a false interpretation. So while post-sentential negation has
been around for more than 20 years (surprisingly long for a piece of teenage
lingo of this kind), constraints on the positioning of negation suggest it is not
a serious contender for general negation but will remain playfully ironic –
though its recent take-up by adults has seen it disappear as a fashion item of
teenspeak (we will talk some more about the role of teenagers in the propaga-
tion of linguistic change in Chapter 7).
SUMMARY
We opened the chapter by looking at some of the difficulties facing linguists inter-
ested in studying grammatical change, difficulties that go some way to accounting
for the relative neglect of historical syntax. But times have changed, and a flourish-
ing of general theories of syntax has brought with it a flourishing of insights into the
historical principles and processes governing sentence structure.
One area where considerable advances have been made is in word order change,
and we used the history of English to illustrate the kind of significant transforma-
tions these changes bring to the overall structural organization of a language.
Our initial focus was on typological and universalist approaches, in particular
the area of word order typology and implicational universals. This approach has
provided the necessary basis that hitherto had not existed for the description and
evaluation of syntactic change in language and has encouraged many more studies
in this field. This schema offers a means of linking together a number of changes,
at the same time giving them meaning by showing them to be part of some overall
long-term trend; that is, towards consistency of type.
We then shifted our attention to the mechanisms driving grammaticalization,
namely the twin forces of reanalysis and analogy. Reanalysis modifies actual
underlying representations and brings about significant rule changes. It is the
development of new out of old structures and is an important mechanism for the
creation of grammar. Analogy does not bring about rule change but involves the
generalization of types of linguistic structures and only has the effect of modifying
the surface (rule generalization). But by being overt, it provides the evidence that
grammaticalization has taken place (i.e. new grammar has been created). Gram-
maticalization doesn’t occur without either of these mechanisms.
We started this chapter rather bleakly, but we hope we have now left you optimis-
tic about future research in historical syntax. Much like Tolkien’s cauldron, its story
continues to bubble away – continually being enriched and improved by ongoing
research that we’ve only been able to touch upon here (though see Chapter 10). The
development of strong synchronic theories of syntax has meant big advances in our
understanding of the whys and wherefores of historical syntax – the mechanisms of
the changes and their causes. Moreover, with progress in technology it is now much
easier to log and classify changes that have occurred and that are in progress. Vast
improvements in corpus design are providing massive digitized collections of texts
(from early and more modern periods) that are annotated and searchable, and his-
torical evidence better suited to the study of language change is now more readily
available. Despite the bleak beginning, we have every reason to be optimistic about
future research in this area.
SOURCES AND FURTHER READING
For a cross-linguistic account of syntactic change across a wide variety of languages

see Harris and Campbell (1995). It’s in this book that you will find details of the
development of the English complementizer for-to; and for German and Pennsyl-
vania German see Börjars and Burridge (2011).
Our discussion on word order typology and change was drawn from the follow-
ing books: Comrie (1989); Croft (1990); Greenberg (1963); Hawkins (2002, 2004);
Mallinson and Blake (1981); Vennemann (1974); and Lehmann (1973).
Grammaticalization is now a hot topic, and there are some excellent accounts
around: see Heine, Claudi and Hünnemeyer (1991); Hopper and Traugott (2003);
and the papers in the journal Language Sciences 23 (2001), a special issue devoted
entirely to this topic. Bybee (2003) looks specifically at the role of frequency. And
for those specifically interested in the diachronic aspects of negation, we recom-
mend Willis et al. (2013) and Van der Auwera (2010).
EXERCISES
1 Word order
Revisit Table 6.2 and investigate the word order patterns in Japanese, Turkish and
Spanish and decide how consistently these languages fall into one typological
category.
2 Short-answer questions – English and other languages
a The English based creoles Bislama and Tok Pisin (spoken in the Pacific)
have evolved a new verbal suffix, illustrated below:
dring ‘to be drinking’ dringim ‘to drink (something)’

marit ‘to be married’ maritim ‘to marry (someone)’
giaman ‘to be lying’ giamanim ‘to deceive (someone)’ (< early slang gam-
mon ‘humbug, rubbish’)
i Describe the form of this suffix and its function. Hypothesize how it
developed.
ii This suffix appears on new verbs entering the language (e.g. Bislama
imel-im ‘email (someone)’). However, there is a group of common
usage verbs, such as save ‘know’ which regularly appear without any
suffix. How might you explain this aberrant behaviour?
b Some linguists argue that grammatical change is goal driven; in other words,
languages develop grammatical categories (such as future marking) because
they need them. Outline one reason why this cannot be the case. Give an
example to support your answer.
c Examine the following examples, and explain why (iii) is ungrammatical.
i Fred’s going to university
ii Fred’s going to/gonna go to university
iii *Fred’s gonna university.
d In Kriol (spoken in northern Australia), gotta has a variety of functions.
Two are illustrated in the following examples:
If you kick him here you gotta kill ’im. ‘If you kick it there, you will kill it’
One big plate gotta three little candle ‘A big plate with three little candles’
Explain the grammatical functions of gotta in these examples and name the
change that is illustrated here.
e Briefly explain linguist T. Givón’s catchphrase “Today’s morphology is yes-
terday’s syntax” (1971: 413). Support your discussion with examples.
f How might structural pressure cause a language to change? Give an example
of change that has been caused by structural pressure.
g Examine the following examples from the Niger-Congo language Ewe
(taken from Heine et al. 1991), and describe the developments that have
occurred to the word for the body part ‘back’:
é-pé megbé fá
3SG-POSS back be cold ‘His back is cold’
é le xo á megbé
3SG Is house DEF behind ‘He is at the back of the house’
é no megbé
3SG stay behind ‘He stays back’
é kú le é-megbé
3SG die be 3SG-behind ‘He died after him’
é tsí megbé
3SG remain behind ‘He is backward/mentally retarded’
3 Grammatical change in Pacific Creoles – Tok Pisin

In creoles such as Tok Pisin, verbs are typically unmarked for tense. Depending on
the context, Mi kam can mean ‘I am coming’, ‘I came’ or ‘I will come’. If necessary,
speakers can add extra sentence modifiers such as bin or bai to indicate either past
or future time: Mi bin kam ‘I came’ and Bai mi kam ‘I will come’. Increasingly, it is
becoming normal to include this sort of information, even when it is clear from
context, and Tok Pisin now has a range of temporal, aspectual and modality par-
ticles (usually preverbal). Some of these are listed below.
bai/baimbai (future)
bin (past)
laik (proximal future)
klosap (inceptive – specifies the beginning of the action)
pinis (perfect)
save (habitual)
stap (continuous)
inap (ability)
ken (permission)
mas (necessity)
a Briefly describe what is happening in the language. Your answer should

include phonological, semantic and grammatical details. What sort of over-
riding change is illustrated here?
b Give an example of this type of change that has occurred/is occurring in
English or in any other languages you are familiar with.
4 Historical development of English morphosyntax

Below are four different versions of the Lord’s Prayer taken from different periods
in the history of English. Examine each version and answer the questions that fol-
low (the answer for each question requires about a paragraph).
Old English text, ca. 1000

Fæder ure þu þe eart on heofonum
si þin name gehalgod
tobecume þin rice

gewurþe þin willa on eorþan swa swa on heofonum
urne gedæghwamlican hlaf syle us to dæg
and forgyf us ure gyltas
swa swa we forgyfað urum gyltendum
and ne gelæd þu us on costnunge
ac alys us of yfele. Soþlice
Middle English text, ca. 1400

Oure fadir that art in heuenes,
halewid be thi name;
thi kyngdoom come to;
be thi wille don in erthe as in heuene;
yeue to vs this dai oure breed ouer other substaunce;
and foryeue to vs oure dettis, as we foryeuen to oure dettouris;
and lede vs not in to temptacioun,
but delyuer vs fro yuel. Amen.
Early Modern English text, ca. 1611

Our father which art in heauen
hallowed be thy name.
Thy kingdome come. Thy will
be done, in earth, as it is in heauen.
Giue us this day our daily bread.
And forgiue us our debts,
as we forgiue our debters.
And lead us not into temptation,
but deliuer us from euill. Amen
Modern English text

Our Father, who is in heaven,
May your name be kept holy.
May your kingdom come into being.
May your will be followed on earth, just as it is in heaven.
Give us this day our food for the day.
And forgive us our offenses,
Just as we forgive those who have offended us.
And do not bring us to the test.
But free us from evil. Amen
a Examine the morphology in the Old English version and give an example
from the nouns in the prayer to illustrate the inflectional richness of the
language. In a general way explain what these inflections are doing (i.e.
you don’t need to give the specifics of the inflections).
b Generalize about overall trends that were underway in the language from
the Old to the Modern English period with respect to these inflections.
Support your answer using the example of the verbs from the prayers.
c Compare the Old, Middle, Early Modern and Modern versions and (using
examples from the prayer) briefly find examples to support the changes
in negation described in this chapter. Include how negation as repre-
sented in the contemporary version differs from negation in current-day
conversation.
d Compare the Middle English version with the Early Modern English ver-
sion. Is the Middle English prayer more like the Old English version or the
Early New English version in its grammar? Give two reasons (with exam-
ples from the prayers) to support your answer. (*Don’t include negation.)
5 Research essay – is “grammaticalization theory” really a theory?

Argue the extent to which recent scholarship on grammaticalization corresponds to
“The Emperor’s New Clothes” or to a “Great Leap Forward” (Campbell and Janda
2001: 93). Critics of grammaticalization (see the papers in a special issue of Lan-
guage Sciences 23, 2001) argue that it is simply a handy cover term for a variety of
linguistic changes (such as “semantic bleaching” or desemanticization, extension,
phonological erosion and reanalysis) that can apply independently of one another
and in all sorts of situations; in other words, grammaticalization has no indepen-
dent standing and certainly no explanatory power. Others (such as Heine 2003)
argue in favour of its theoretical status because they see it as offering an explanatory
account of how and why grammatical forms arise and develop. Assess the funda-
mental aspects of grammaticalization and determine whether or not there is some-
thing that corresponds to “a theory of grammaticalization”.
NOTE
1 The leechdoms were collected and edited by Oswald Cockayne in 1865; all three vol-
umes are now digitized and available online: https://archive.org/details/leechdom
swortcun01cock.
7
The spread of change
INTRODUCTION
How does something new spread through society, from one person to the next?
A closer look at the outbreak and spread of diseases may be somewhat unpleasant,
but it is very instructive. The Bubonic Plague (aka the Black Death) struck Europe
in the 14th century. Responsible for this horrible, deadly disease was probably the
bacterium Yersinia pestis, transmitted by flea bites. The bacterium originated in Asia
and reached European seaports in the 14th century (note that by then the plague had
already wiped out large parts of the populations of China, India, Syria, Mesopotamia
and Armenia). The story goes that in 1347 the trading city of Caffa in Crimea was
under siege by the Mongol Army. After it became clear that the city was almost unde-
featable, the besiegers decided to catapult the corpses of people who had died from
the plague into the city. In the face of this, traders from southern Italy immediately
fled by ship – and, unknowingly, took the plague with them to Sicily, from where it
spread in no time all over Europe, leading to an appalling death toll of 75–200 mil-
lion people across Europe and Asia within a few years. In some macabre sense, we
can call this a very successful disease. Compare this to the current outbreak of Ebola
in parts of Africa. Again, this is a horrible, deadly disease, caused by the Ebola virus
and at least as infectious as the plague. But while the number of fatalities in the
affected areas is still tragically high (there are about 12,000 reported fatalities so far),
it does not reach the magnitude of the plague, and the disease has not yet spread
widely beyond these regions. There is even a good chance that it will be controlled in
the very near future. So, in some sense, we can call this a failed disease.
These examples are about the spread, or, better, the diffusion of innovations. At
some point in time, something new came up and it caught on; in other words, it
spread through society (or, in the case of fails, obviously did not spread). We can see
exactly the same phenomenon in linguistic change. A given innovation (for exam-
ple a new word like chick lit ‘literature for female readers’) may or may not catch
on. If it does, we can see it spread through society, from speaker to speaker – like
some flu outbreak. So linguistic forms can spread through society from speaker to
speaker, like germs. But when, why and how does a given innovation spread? Why
do some fail? Have you ever heard the word crapulous? It comes from Latin crapula
‘intoxication’, and in the 1500s it meant something like ‘hung over’ or ‘feeling sick
because of excessive eating or drinking’ – a helpful word, especially if you actually
feel crapulous, but it did not catch on and thus died out (note that our present-
day word crap has nothing to do with crapulous. Crap probably comes from Latin
crappa ‘chaff ’ and the Middle English cropper/crop.)
The spread of change 163
At the same time, linguistic innovations may also spread through the linguistic
system itself. If we want to stay within the disease frame for a second, we might say
that a contagious rash not only spreads from one person to the next but may also
spread across your body. The latter could be compared to spread in the linguistic
system.
Imagine somebody introduces a change to the sound system: a new sound might
be added, an old sound might be lost, or perhaps the context of use of a given sound
changes. As described in Chapter 4, there is good reason to believe that Old English
was rhotic, i.e. that speakers of Old English pronounced words like bēor ‘beer’ and
sar ‘sore’ with an [r] after the vowel: [be:ɔr] and [sa:r]. Obviously, many varieties of
present-day English (at least in England) have lost this [r] after vowels (but kept it
elsewhere). Only consider Received Pronunciation (RP) in England beer [bɪə] and
sore [sɔ:], but run [rʌn] and break [breɪk]. There are many words in the English lexi-
con that have <r> in their spelling after a vowel. What happened to them? Did they
all lose postvocalic [r] at the same time? Or did some of them go first, and others
follow gradually? What we often observe in such cases is called lexical diffusion, or
the gradual spread of change through the lexicon of a language.
In the following sections, we will take a careful look at the diffusion of changes
both in society and in the linguistic system. Let’s begin with the latter.
7.1 DIFFUSION WITHIN THE LINGUISTIC SYSTEM
How do changes spread within the linguistic system? Before we try to answer such
a seemingly simple question, we need to distinguish between the different linguistic
levels of phonology, morphology and syntax. The patterns of diffusion might be
different for the different levels.
Let us begin by looking at phonological change. How does a change in the sound
system affect the vocabulary of a language? There seem to be two possibilities: it
could either be lexically abrupt and affect all items that can be affected simultane-
ously, or it could be lexically gradual and affect only some items at a time. The
former can be exemplified when we look at one of the most widely discussed sound
changes in historical linguistics: Grimm’s Law, or the First Germanic Consonant
Shift. This shift, named after the 19th century scholar Jakob Grimm (the elder of
the two Grimm brothers, who became famous for their fairy tales), took place when
Proto-Germanic1 developed out of Proto-Indo-European (PIE), probably more
than 2,500 years ago. Essentially, this shift turned voiceless stops into voiceless fric-
atives, voiced stops into voiceless stops, and voiced aspirated stops into voiced stops
or fricatives. The results are the differences we see today between the Romance
languages (Italian, Portuguese, Spanish, Latin, Sanskrit etc., which mostly still have
unshifted PIE consonants) and the Germanic languages (German, English, Dutch,
Danish etc., which mostly have new, shifted consonants). This is summarized in
Table 7.1.
The Neogrammarians (a group of scholars of language and linguistic change in
the 19th century) claimed that Grimm’s Law affected all words simultaneously.
Table 7.1
Grimm’s Law (First Germanic Consonant Shift)
Proto-Indo-European (PIE) Proto-Germanic English cognates

/p/ *pisk, Lat. piscis /f/ fish
*peter, Lat. pater father
/t/ *treyes, Lat. tres /ð, θ/ three
*tu, Lat. tu thou (Middle and Early Modern English)
/k/ *korn, Lat. cornus /h/ horn
*kemtom, Lat. centum hundred
/b/ *d hewbu /p/ deep

/d/ *dva, Lat. duo /t/ two
Skr. *dant tooth
/g/ *gno-, Lat. gnoscer e /k/ know (pronounced with a /k/ in Middle English)
/bh/ Skr. *bharati /b/ s/he bears

Skr. *bhra-ter brother
/dh/ Skr. *dhwar /d/ door
/gh/ Skr. *gha-ns /g/ goose (German Gans )
Skr. *ghost guest
Wherever there was a /p, t, k, b, d, g, bh, dh, gh/ these got shifted to /f, ð, θ, h, p, t, k,
b, d, g/, respectively. So for the Neogrammarians sound change is lexically abrupt.
Note that we do not know this for sure; in other words, this is only a hypothesis,
and there is little concrete evidence for this lexically abrupt shift. But assuming the
regularity of sound changes such as Grimm’s Law proved to be very important for
the reconstruction of undocumented historical language stages and languages (see
Chapters 4 and 9).
The Neogrammarian rule that sound change is lexically abrupt remained almost
unchallenged for several decades. From about the middle of the 20th century, how-
ever, scholars began to discover phenomena which cast some doubt on this idea.
Their studies quite often go back to one of the early critics of this principle, Hugo
Schuchardt. “Rarely-used words drag behind; very frequently used ones hurry
ahead. Exceptions to the sound laws are formed in both groups” (Schuchardt 1885:
58). Jean Aitchison asks us to consider the following brief text:
A cursory glance at the newspapers suggests that adultery is on the

increase in this century. If you think slavery has been abolished, go and
look at the factory at the end of the road. Every mother will tell you that
nursery schools are a mixed blessing.
(Aitchison 2013: 88)
The text contains a number of words that can have variable pronunciations: cur-
sory, adultery, century, slavery, factory, every and nursery. These words can either
have a schwa [ə] in their middle [sleɪvərɪ] or drop it [sleɪvrɪ]. Would you drop the
schwa in all cases? Or in none of them? Betty Phillips, an expert on frequency effects
in language, tested this in her class at Indiana State University. These are her results.
Table 7.2
Variable pronunciations of -Vry words
Word Celex* Schwa-less Schwa-ful

every 9,788 22 (85%) 4 (15%)
factory 1,064 18 (70%) 8 (30%)
nursery 580 19 (73%) 7 (27%)
slavery 134 10 (40%) 16 (60%)
adultery 107 15 (58%) 11 (42%)
cursory 32 9 (35%) 17 (65%)
(* = a general database with raw word frequencies)
There seems to be some correlation between raw word frequency (in CELEX, for
example) and the amount of schwa-deletion. The more frequent the word generally,
the more often schwa is lost, and vice versa. The very frequent word every is affected
85% of the time [ɛvrɪ], while the very infrequent cursory is only affected 35% of the
time [kərsərə]. What this amounts to is that changes can also happen one item at a
time, or, in other words, they may be lexically gradual! What we see in these con-
texts is the famous S-curve of diffusion, shown in Figure 7.1.
The graph in Figure 7.1 roughly looks like the letter S. It shows how innovations
begin to spread slowly at the beginning (just a few words are affected), then begin
to catch on, with rapid changes in many words, until they finally peter out, so that
most of the words that could be affected, or even all of them, show the change. This
curve is very famous in a number of disciplines such as physics, chemistry, medi-
cine, politics and economics. Most sorts of innovations, from germs to technology
and fashion, actually show this kind of diffusion.
Figure 7.1
The S-curve of diffusion
Percentage
Time
Figure 7.2
Percentage of -s vs. -th in the third person present tense (based on Nevalainen and Raumolin-Brunberg
2003: 220)
100
90
80
70
60
50
40
30
20
10
0
1410–1459 1460–1499 1500–1539 1540–1579 1580–1619 1620–1659 1660–1681
Changes in morphology can happen in very similar ways. Only consider the
development of “third person singular s” in Early Modern English, e.g. sing-s ver-
sus sing-eth, which we already mentioned in Chapter 1. This is summarized in
Figure 7.2.
Figure 7.2 shows a remarkable similarity to the S-curve presented in Figure 7.1. So,
in general, it seems as though the overall development is lexically gradual and follows
the S-curve of diffusion, with a slow start (until about 1579), followed by a rapid spread
from 1579 to about 1659, when the curve begins to peter out. Ogura and Wang (1996)
zoom in on this development and try to identify some of the generalizations underly-
ing this pattern of diffusion. Are there any words that change earlier on? Are there
others that are more conservative and lag behind? What Ogura and Wang find is that
first we need to distinguish between verbs that end in sibilants (“hissing sounds” such
as [s, z, ʃ, ʒ] etc.) and those that don’t. Words that end in sibilants tend to be slower in
adopting the -s suffix, mainly for phonotactic reasons – perhaps it is somehow harder
or awkward to pronounce kiss-es rather than kiss-eth (just as we don’t use friendlily as
the adverb of friendly – the repeated sound pattern feels weird).2 Within the group of
words that do not end in a sibilant, high-frequency verbs (such as make or come) tend
to change earlier than low-frequency verbs (such as endeavour or prevail). However,
once infrequent verbs start to change, they are much quicker to complete the change
than the frequent verbs, so that we find a time (around 1640–1710) when infrequent
verbs have already completed the process and some frequent verbs still show some
forms with -th. Similarly, the two most frequent verbs, have and do, begin to change
early but show variability between -s and -th forms for the longest time. This goes
to show that morphological changes can also spread through the lexicon, not unlike
phonological changes, and that the changes do not necessarily affect random words
but may be constrained by factors such as frequency, phonotactics and the like.
Figure 7.3
The development of periphrastic do, 1400–1700 (based on Ellegård 1953)
100
90
80
70
60
50
40
30
20
10
0
1400 1500 1600 1700
Negative Questions Affirmative Questions

Negative Declaratives Affirmative Declaratives
Last but not least let us look at the diffusion of syntactic changes, taking again an
example from Early Modern English. In this case, we will look at the introduction
of periphrastic do, i.e. the (new) use of do in questions and in negative sentences. In
a seminal study, Ellegård (1953) showed that this innovation did not affect all syn-
tactic environments at once. Periphrastic do was first introduced and spread most
quickly in negative questions (e.g. Don’t you like cheesecake?), followed by affirma-
tive questions (Do you like cheesecake?). Negative declarative sentences (e.g. I don’t
like cheesecake) were considerably slower. At around 1600, about 75% of all affir-
mative and negative questions were formed with do, but only 35% of all negative
declaratives showed the new pattern. These, however, quickly caught on during the
next 50 years or so. Affirmative declaratives (e.g. I do like cheesecake) never really
caught on. The whole development is summarized in Figure 7.3.
So, syntactic changes can also gradually spread through the linguistic system. Pho-
nological and morphological changes can gradually spread through the lexicon, i.e.
from word to word, and syntactic changes can gradually spread through various syn-
tactic contexts. Needless to say, not all changes spread through the linguistic system.
The sporadic phonological changes we looked at in Chapter 4, such as metathesis (e.g.
aks – ask), usually do not spread through the lexicon, and neither do some word forma-
tion patterns such as abbreviations or blends. Similarly, some syntactic changes seem
to occur in fairly isolated contexts, or they affect the whole syntactic system at once.
7.2 SPREAD IN SOCIAL STRUCTURES – THE SPEECH COMMUNITY AND THE

INDIVIDUAL
Let us now look at how innovations can spread through the linguistic community.
As we have pointed out before, a linguistic innovation that does not spread is exactly
that: an individual innovation, but not change. The investigation of the diffusion of
linguistic features in geographical and social space has to reach out and include
ideas, methods and findings from other disciplines such as sociology, (social) geog-
raphy, social psychology, migration studies and urban studies, among others.
7.2.1 Wavy, gravity, cascady . . .

One very old, but still popular model of linguistic change describes the diffusion of
new features as concentric waves, spreading out from a centre of origin and becom-
ing progressively weaker the further they move away from the centre.
One textbook example for this is the so-called Rhenish Fan, a very interesting
patterning of linguistic features in the Rhineland area around Cologne and Düs-
seldorf in the North-West of Germany. Pre-Old High German (the German lan-
guage before ca. 750 CE) had three different plosives, */p/, */t/, and */k/, in words
such as [dorp] ‘village’, [apəl] ‘apple’, [dat] ‘that’, [ik] ‘I’ and [maken] ‘make’. During
the High German Consonant Shift (between ca. 200 and 500 CE), High German
started replacing these with /f/, /pf/, /s/, /ç/ and /x/, respectively: [dorf] ‘village’,
[apfəl] ‘apple’, [das] ‘that’, [iç] ‘I’ and [maxen] ‘make’. This replacement started in
the South of Germany and spread northwards. For the largest part of Germany,
in the mid and easterly regions, there is a dividing line today (a so-called isogloss)
that runs roughly from Düsseldorf to Berlin which distinguishes between the High
German pronunciation in the South [dorf, apfəl, das, iç, maxen] and the Low Ger-
man pronunciation in the North [dorp, apəl, dat, ik, maken]. In the West, however,
in the Rhineland area, the story was a lot more complicated. Not all forms shifted
at the same point in space. So there is one isogloss that starts far in the south that
distinguishes between [apəl] (north) and [apfəl] (south); another one around Trier,
a little further north, that divides between [das] and [dat]; next comes a line near
Aachen and Bonn a little further north that distinguishes between [maken] and
[maxen], [dorp] and [dorf|. Last but not least, even further north, we find a fourth
line to divide the area between [iç] and [ik]. The whole situation is summarized in
Map 7.1.
In the case of the Rhenish Fan, the wave began somewhere in the South, and in
extending northwards it progressively lost its strength, so that not all features were
spread in all regions.
As mentioned above, the idea of a wave-like pattern of diffusion is fairly old and
goes back at least to Schmidt (1872) or even Schuchardt (1868 and 1870). While some
people saw this as a counter-movement to the very influential family tree model (see
Chapter 9), most of the discussions today now emphasize the compatibility of the two
models. The long-term development of languages comprises both of these comple-
mentary aspects, so that languages can both develop independently (in a tree-like
fashion) and in interaction with each other (a pattern with several overlapping waves).
In 1974 the British sociolinguist Peter Trudgill suggested some important modi-
fications to the wave model of diffusion. He called this the gravity model of dif-
fusion and argued that the spread of changes must be intricately tied up with
communication between speakers. The more speakers interact, the more likely it is
Map 7.1
The Rhenish Fan: 1 Dutch (West Low Franconian), 2 Limburgian (East Low Franconian), 3 Ripuarian
Franconian, 4 and 5 Mosel Franconian, 6 Rhenish Franconian (adapted from Hans Erren 2010
after Georg Wenker 1877, https://en.wikipedia.org/wiki/High_German_consonant_shift#/media/
File:Rheinischer_faecher.png CC BY-SA 3.0)
Iss
el
wi makt
1 (5)
ik
ich Uerdingen
2
Benrath
maken
machen
Bad Honnef
Dorp
Bad Hönningen
Dorf
4 op
auf St. Goar
dat 6
das
Speyer
appel
apfel
that changes will be passed on from one speaker to the next (this idea actually goes
back to Bloomfield (1933), who discusses the density of communication as the basis
for the wave model). Trudgill now introduced the idea that geographical distance
and population size are two measurable predictors for the spread of innovations.
Linguistic changes are promoted and spread through the more densely populated
centres (centres of gravity, if you like). What we then see is that changes do not
radiate uniformly, wave-like, from one place or origin, but rather jump from one
centre to the next, before and while they spread into the areas around these centres.
Trudgill was able to show this with the two English cities of London and Norwich,
and the countryside surrounding these places. One current change in England is
(or maybe was, this study is now more than 40 years old!) the dropping of initial h
in words like hotel, ham, heel and so on. This change had its origin in the vernacular
varieties of London. From there it first seems to have spread to Norwich, the next
biggest population centre in East Anglia, about 150 kilometres (or 100 miles) to
the North-East. From there we see gradual, perhaps wave-like diffusion into the
surrounding countryside. If the original wave model had been at work, we would
have expected the strongest changes around London, the centre of origin, and then
a gradual petering out while the innovation spread across the countryside, with
considerable less intensity in Norwich. This, however, is not the case. The centres
of gravity, London and Norwich, show the most widely advanced changes, while
the countryside between these places is lagging behind. The reason for this is all
too plausible: on the one hand, we see more interaction between speakers in the
bigger cities, the centres of gravity, and thus a greater likelihood for change. On
the other hand, there is a considerable number of speakers commuting to London
from Norwich every day. The London innovations can travel with the commuters
and can thus spread in Norwich without necessarily affecting the places in between
the two cities.
More linguistic diffusion

American sociolinguist William Labov gives another example of linguistic
diffusion that illustrates this principle (http://www.ling.upenn.edu/~wlabov/
Papers/PCM.html). Note that Labov refers to this as the cascade effect, rather
than the gravity effect, but essentially it is the same phenomenon. Labov
studied the spread of food terminology in the 1950s, 1960s and beyond in
Pittsburgh and Philadelphia (two large cities in Pennsylvania, on the East
Coast of the United States). First, he noted that the term hoagie became gen-
eralized in Philadelphia shortly after the Second World War as the term for a
particular large sandwich (today usually referred to as a sub in many places).
He simply checked entries in the Yellow Pages’ restaurant section in Philadel-
phia and found the distribution shown in Table 7.3.
Table 7.3 clearly demonstrates that hoag(ie) was well established by the
1950s in Philadelphia food terminology. It can be found in the city and to
some extent in the surrounding areas even on today’s dialect maps. For some
Table 7.3
Lexical stem of hoagie in Philadelphia telephone listings
hogg- hoog hog- hoag- submarine

1945 3 1 1
1950 10 2 1
1955 4 6 47 1
Table 7.4
Diffusion of hoagie to Pittsburgh from 1961 to 1966
1961–2 1962–3 1964 1965 1966

Grab’n Go Pizza H * S S S
Pat’s Pizzeria # # H H *
Village Pizza H H H H
Campiti’s Don Pizzeria # # H *
Luigi’s Pizza * * H
Frank & Betty’s Pizza Shop H H H
Aspinwall Pizza/Augie’s Hoagies H H
Mac Tony’s Nationwide H H
Luara’s Restaurant & Pizza H
Brookline Pizza H
Arudy’s Pizza/Famous Hoagie H
Total listings in Restaurant section 68 75 75 91 93
H = hoagie; S = sub(marine). * = listing, no ad; # = ad, no mention of H or S.
curious reason, hoagies in Philadelphia became closely associated with pizza

restaurants, at least after the 1960s, and listings containing the word hoagie
were then mostly found in the pizza section of the Yellow Pages.
Interestingly, hoagie is also a very widespread term in Pittsburgh. So Labov
then compared the Pittsburgh Yellow Pages. Table 7.4 shows what he found.
The term hoagie seems to have established itself in Pittsburgh’s restaurant
scene from 1962 onwards. Village Pizza was the first and most consistent user
of the term, and many shops such as Frank and Betty’s Pizza Shop, or Mac
Tony’s Nationwide, followed suit. It can be speculated that Village Pizza must
have played a very prominent and influential role in the city. Eventually, the
term hoagie got established as a hallmark of Pittsburghese – but we do not
see it diffused between Philadelphia and Pittsburgh as predicted by the wave
model. Rather, the term seems to have jumped from one big city to the next
(even though there is a considerable distance between the two of about 300
miles (500 kilometres)).
What we see today are both patterns. On the one hand, there is gradual, wave-like
diffusion around certain points of origin, though maybe slower and less pervasive
than originally thought. At the same time, gravity or cascade effects are operative.
These lead to the jumps that we can observe in many other cases.
7.2.2 Social factors

There is a huge list of social factors that have been investigated in connection with
the origins and diffusion of linguistic changes: social class, age, gender, ethnicity,
income, education, faith, sexual orientation and many, many more. We won’t be
able to discuss all of these here. Let’s concentrate on the most widely (and some-
times even hotly) debated factors: social class, age and gender.
Social class
Social class is a very complex sociological construct, and it would be hard to do
full justice to the concept in only a few paragraphs. (Levine (2006), Breen and
Rottmann (2014) and Grusky (2014) provide interesting and helpful sociological
surveys. One of the key problems is that class and cultural context are intricately
intertwined so that universal principles which apply to all cultures and societies are
hard to identify.) Suffice it to say, for present purposes, that the class or social strati-
fication system in most “western” societies and cultures (Great Britain, America,
Canada, France, Germany etc.) is immensely complex and in constant flux. Clas-
sifications that were helpful in the 1960s and 1970s (when modern sociolinguistics
was born and began to use the concept of class widely) do not apply anymore, and
certain historical generalizations are no longer valid.
In the 1950s, for example, class membership could partly be identified by asking
informants about their family background: working class parents usually meant
working class children, middle class parents brought up middle class children etc.
Working class usually meant basic education, while the middle class could attend
better schools and perhaps even go to university. This has changed in many societ-
ies. There are some changes that make straightforward historical equations (like
“lower class parents equal lower class children”) difficult, if not impossible today. As
an example: Labov in his famous department store study in New York City in the
1960s simply assumed that people who worked at Saks on Fifth Avenue represented
a higher social class than sales clerks at Macy’s, who in turn were higher up the lad-
der than those at Klein’s – a low-price department store (Labov 1972).
The bottom line of this discussion is that whenever we read about social class, or
perhaps even use the term, we need to be very careful what it means, and we need
to be aware of the context in which it was or is used.
Having said that, what are some of the factors that traditionally make up class
membership (you don’t get a membership card at birth or graduation, so we need
to establish this independently)? As mentioned before, class membership is usually
associated with education, occupation, income, family background, place of liv-
ing, type of housing, frequency and type of travel, and the like. Most studies in the
past distinguish between lower, middle and upper classes, and various subdivisions
within these groups. The following two diagrams present two studies that correlate
linguistic variation with social class and style.
Figure 7.4 presents another study by Trudgill in Norwich (1974). The variable
investigated is the pronunciation of words like class, last and dance. These could
either be pronounced with a back vowel like [da:ns] (imagine a very British, Hugh
Grant-like pronunciation) or with a front vowel like [dæ:ns] (imagine a more
American pronunciation). The latter is associated with lower class speakers. In this
study (see Figure 7.4) speakers could score points for their pronunciation of (a:).
Figure 7.4
(a:) in Norwich (based on Trudgill 1974)
200
180
160
140
Index of raised (a:)
120
100
80
60
40
20
0
Casual Careful Reading Formal
LWC MWC UWC LMC MMC
The more often they pronounced this like [æ:], the higher their score. Trudgill tested
this in four different styles: casual speech, careful speech, reading style and formal
(word list) style, and with five independently established social groups: lower work-
ing class (LWC), middle working class (MWC), upper working class (UWC), lower
middle class (LMC) and middle middle class (MMC). Figure 7.4 shows you the
results.
Predictably, the higher the social class, the more back [ɑ:] we find; the lower
the class, the more fronted [æ:] speakers use. Interestingly, however, this hardly
changes with style. So, even if speakers are made aware of the fact that their lan-
guage is being recorded and tested, they do not shift towards the more overtly pres-
tigious, standard pronunciation. They still produce front [æ:] with a remarkable
frequency. We can call this stable variation; i.e. language variation in the commu-
nity does not seem to be going anywhere. The lower social classes seem to be happy
with the fronted [æ:], while the higher social classes keep their back [ɑ:]. If there
were any ongoing changes, we would expect something to happen when we shift
styles, because speakers then perhaps want to use the more prestigious form (or
sound more local and use the fronted form). But we see nothing of that sort.
Compare this to the situation in Figure 7.5. This is taken from Labov’s study of (r)
in New York City in words like car, floor, beer or burn, i.e. in the syllable coda. The
(r) less pronunciation ([ka:], [flɔ:], [bi:ə], [bɜ:n]) is associated with the lower social
classes; the higher the score here, the more speakers use (r) in these cases. We find
three styles (casual, reading and formal) and six social classes: upper middle class
(UMC), MMC, LMC, UWC, MWC and LWC.
Figure 7.5
(r) in New York City (based on Labov 1972)
90
80
70
60
50
(r) in %
40
30
20
10
0
Casual Reading Formal
UMC MMC LMC UWC MWC LWC
The results of this study are very different. On the one hand, we can generally say
that the higher social classes tend to use more (r). This fits in with our prediction.
But in contrast to Trudgill’s Norwich study, we find that all classes change their lin-
guistic behaviour radically when they shift styles. The more formal it gets, the more
(r) is being used in all social classes. So, speakers seem to be very aware of the value
of (r) and consequently use it more often when it really counts. Most importantly,
however, we can see that the LMC speakers seem to overtake the UMC in their for-
mal style. How is that possible, and what does it mean? Let us assume that the UMC
is like a linguistic role model for the LMC. Sociologically speaking, members of the
LMC can be characterized as socially mobile aspirers, as people who have just left
behind the working classes and who have now set their eyes on a better life. And one
way to distinguish themselves from the working classes is by adopting a linguistic
marker that shows their social superiority. The MMC feel no such need for such an
extreme move since their distance to the lower classes is already wide enough. The
LMC as social aspirers, however, seem to overdo it. They use (r) more often in cer-
tain (conscious) styles than do their role models, the UMC. We thus get the striking
crossover pattern in Figure 7.5. But what does this mean for linguistic change? Here
we do not find stable variation, but a highly dynamic situation. A linguistic variant
that was typical for a certain social class, or style, spreads and diffuses into the lan-
guage of other social classes. In this case we find a pattern that is called change from
above, since the change (and pressure) originates from higher social classes using
overtly prestigious forms. And this is one important pattern for the diffusion of
linguistic changes, especially when varieties of a language move towards the overtly
prestigious standard.
Note that the reverse is also possible. We then call this change from below. This is
what happens when changes originate in informal language use, or in lower social
classes, and diffuse “up the social ladder” or into more formal styles. Montreal
French, for instance, saw the rapid replacement of generic on ‘one’ by tu ‘you’ (as
in one can never be sure versus you can never be sure). Laberge and Sankoff (1980)
and later Thibault (1991) found that this change was part of a massive restructuring
from below which first replaced former generic nous ‘we’ with on and then with tu.
Speakers are usually aware of changes from above (since these are often discussed
or even prescribed in the education system and the media), while changes from
below often fly below the radar and do not get noticed by speakers. Labov (2006:
203) even said that, in hindsight, it might have been better if “change from below”
had been called “change from within” and “change from above” had been called
“change from without” to show that the former usually happens within the linguis-
tic system, without reaching the level of social awareness, while the latter is imposed
on the linguistic system owing to outside factors (such as prestige and language
contact) and is thus mostly visible to speakers. Note, however, that the dividing lines
between these two sometimes get blurred, and that changes from below can also
reach the level of consciousness and social awareness. In those cases speakers and
the linguistic community are faced with the dilemma of having to decide whether
they are willing to accept the “new” variant into the mainstream (standard) system,
or not. If they do, it may even become subject to a change from above mechanism.
We might be witnessing such a change in present-day (North American) English.
Many speakers subconsciously shift their pronunciation of [s] in words like street,
strong, store and restaurant to [ʃ], i.e. shtreet, sthrong, shtore and reshtaurant. Very
few people are actually aware of this pronunciation. At the same time, we are wit-
nessing not only a spread of the new pronunciation (both as lexical diffusion and
in the speech community), but also a gradually increasing awareness. If you search
for “shtreet” and “shtrong” on the internet, you will find a number of examples and
discussions. It is hard to predict whether the palatalized [ʃ] variants will ultimately
win and will become the prestigious standard pronunciation, but this is certainly
not unlikely.
Age
Another interesting factor to look at regarding the diffusion of linguistic innova-
tions is age, not in the least because many innovations take place during childhood,
the teenage years and early adulthood. Only think about the creative potential of
youth language with new words like shan ‘unfair’ (UK), peng ‘looking good’ (UK),
mupload ‘mobile upload’ (US), fungry ‘fucking hungry’ (US) or deso ‘designated
driver’ (nominated as word of the year for 2015 and the fall-out of a drinking cul-
ture in a law abiding society) (AUS). But do these words actually spread? Do other
(adult) speakers adopt them? Or do teenagers at least keep using them through-
out their lifetime? In many cases, they don’t. There is a variety of language that
is specific for teenagers. In the public this is often dubbed “teenage slang”, but in
linguistics we would rather refer to it as an agelect, i.e. a “dialect” depending on
age. And just like teenagers, middle-age adults often have an agelect, and so do
senior citizens (although the latter two are less noticeable than the special language
of teenagers). If such an agelect (or some of its features) is really only specific for a
certain period in the life of speakers and does not spread beyond that, we call this
age grading. Examples for this would be the use of “teenage slang words” such as
the ones we just discussed. These mostly drop out of use after the age of 20 or 25,
and they never get adopted by adult speakers. Similarly, teenagers show a tendency
to use more non-standard forms, such as ain’t or multiple negation. This might not
be exclusively limited to teenagers, but we usually see a peak in use between the
ages of about 12 and 20. A third and last example comes from Canadian English.
Many Canadian children grow up with US television and, accordingly, pronounce
the letter z as one would in the United States [zi:] (this is also the pronunciation you
would hear in the various Sesame Street alphabet songs). But once they grow up,
most of them seem to drop this usage, align with the adults in Canada and instead
use the Canadian pronunciation [zɛd].
Let’s suppose age grading did not happen. In other words, whatever you said in
your youth would stay with you forever. What would be the consequence? The obvi-
ous consequence would be change. If there is some innovation in youth culture and
this does not go away when teenagers turn into adults, we see actual linguistic change
in the community. This could be called “generational change”. So one generation
introduces and keeps something that the previous generations did not have. Imag-
ine there is a young generation of Canadians who decide to keep the [zi:] instead of
changing to [zɛd]. In that case, there would be a high likelihood that Canada would
also become a [zi:] country eventually. Another example would be the use of the
word geil in Modern German. Originally, this meant ‘sexually aroused (animals in
particular)’ or ‘prurient’. In the 1970s and 1980s, teenagers and young adults began
using this to say ‘cool’, ‘attractive’, ‘highly desirable’. Expectedly, adults frowned upon
this (“wrong”) vulgar use of this (“dirty”) word. Today, almost 40 years later, geil is
quite a frequent word in informal spoken language and is used by teenagers and
adults alike. Some people still consider it vulgar, but it can even be heard on TV
and in pop songs, for example Die Fantastischen Vier’s “Zu geil für diese Welt” ‘Too
cool for this world’ (1993) or Deichkind’s “Leider geil” ‘Unfortunately cool’ (2012).
It appears in yellow press headlines (“So geil wird die WM” ‘That’s how cool the
championship is going to be’) and in advertisements (“Geiz ist geil” ‘Tight is right/
Being stingy is cool’). These would have been absolutely unthinkable 40 years ago.
One big problem in language change studies is to distinguish between age grad-
ing and linguistic change in apparent time. The latter rests on the assumption that
whatever we see in a younger generation today will be the community language of
the future. In other words, if teenagers show other linguistic structures than their
parents, this might indicate linguistic change. But what if these structures fade away
when the teens grow up? Looking back in history, this is fairly easy to distinguish
since we can trace actual changes (or successful innovations) and can identify the
individual trends or generations that started a particular change. In history we can
study change in real time and don’t have to make assumptions or predictions on the
basis of a synchronic state of affairs.
Apparent versus real time

But what about today? Synchronic studies, i.e. studies looking at changes in
apparent, not real time, are sometimes faced with the dilemma of having to
distinguish between teenage language use, and real change. Let’s look at one
example. In 1961 Labov studied the language of the people who lived on Mar-
tha’s Vineyard, a small island off the coast of Massachusetts. In particular,
he was interested in the pronunciation of the two diphthongs [ai] as in time
and [aʊ] as in house. His 69 informants pronounced these diphthongs either
more like [ai] and [au], as in General American, or more centralized like [əi]
and [əu]. In order to represent these differences, he calculated a centraliza-
tion index. The higher this index, the more speakers centralized their diph-
thongs to [əi] and [əu]. While there are many complex social factors at work
here, let us focus on age as a variable. Figure 7.6 represents Labov’s results,
stratified by age.
Something seems to be going on here. The group of 31–45-year-olds has
by far the highest centralization index; i.e. they are the most frequent users
of non-standard [əi] and [əu]. The younger generation does not follow this
trend (for very complex reasons that we cannot go into here), and the older
generations also show lower frequencies. How can that be explained? Labov
makes two suggestions. The older speakers could have had higher central-
ization values earlier in their lives (let’s say in the 1930s), which have then
progressively declined to the point that we see today. This would be the effect
of age grading, and we would suspect that the generation of 31–45-year-olds
in this study would show a similar decline and exhibit lower centralization
Figure 7.6
(ai) and (au) on Martha’s Vineyard, centralization index stratified by age (based on Labov 1972)
100
90
80
70
60
50 (ai)
(au)
40
30
20
10
0
14–30 31–45 46–60 61–75 75+
scores in the 1990s, i.e. 30 years after this study was carried out in the early
1960s. This would mean that this is not an example of linguistic change, but
rather of age related language use. Alternatively, what we see in the people
age 31–45 could be a sign of linguistic change. The older people maybe never
had high centralization scores, and what we see here is an innovation by the
46–60-year-olds, which caught on in the following generation (31–45). If this
innovation does not fade away, but maybe spreads further across the genera-
tions, we see actual change (in apparent time). How do we know? How can
we distinguish between these two explanations? There are two ways. On the
one hand, we could try to find earlier records from the 1930s to see if cen-
tralization was popular. If it was, we may suspect age grading, because these
speakers would now, in the 1960s, have lower scores. Luckily, such records
do exist. Between 1931 and 1944, 412 speakers were recorded and analysed
for the new Linguistic Atlas of New England. When Labov looked at these his-
torical documents he could find little to no centralization, which led him to
conclude that age grading in this case seems unlikely. Still this is no positive
proof for ongoing change, but it points in this direction.
Another way to provide positive proof for any changes is to go back to
Martha’s Vineyard several years later and to see what happened with these
two variables in the language of the Islanders. Pope et al. (2007) is one such
study that replicates Labov’s study as closely as possible, only about 40 years
later. Figure 7.7 presents the results of this new study, along with Labov’s
original findings.
The data in Figure 7.7 suggests that we do see changes in real time,
rather than age grading, although perhaps only for (au). The overall rate of
Figure 7.7
Martha’s Vineyard 40 years later (based on Pope et al. 2007)
1.8
1.6
1.4
1.2
0.8
0.6
0.4
0.2
0
<1887 1887–1901 1902–1916 1917–1931 1932–1948 1949–1961 1962–1976 1977–1988 1989–1998
year of birth
(ay) 1962 (aw) 1962 (ay) 2002 (aw) 2002

centralization seems to have increased over time, and even the youngest gener-
ation, which did not pick up the trend in the 1960s, now shows much higher
index scores. And while it is true that those who were 31–45 years old in
the 1960s (born 1917–31) are leading the change, we see an overall increase
across all generations, so that we might suspect that this is an actual ongo-
ing change in the direction of centralization. However, Figure 7.7 also shows
that the younger generations (especially after 1977) are increasing their use
of centralized [au] while gradually dropping the centralization of [ai]. The
reasons for this may be very complex and cannot be discussed here. The
fact itself is still interesting and shows that linguistic change means not only
the spread of new features, but also the loss of elements and structures over
time. By going back to the island and replicating the earlier study, Pope et al.
(2007) have turned an initial apparent-time study (hypothesizing based on
synchronic data) into a real-time study of linguistic change.
Table 7.5 nicely summarizes the main points so far.
Table 7.5
Five types of variation
Individual Community Synchronic pattern Example

1 Stability stable stable flat (a:) in Norwich
2 Age grading unstable stable monotonic slope with age [zɛd] in Canada
3 Generational change stable unstable monotonic slope with age geil in German
4 Communal change unstable unstable flat (r) in New York City
5 Lifespan change unstable unstable monotonic slope with age
(Adapted from Wagner 2012, adapted from Sankoff 2005)
It distinguishes between four different types of sociolinguistic variation: (1) sta-

bility – nothing happens (as with (a:) fronting in Norwich); (2) there is age grading,
as with z = [zɛd] in Canada, so that the individual is unstable and changes over
his or her lifetime, while the community stays the same; (3) there is generational
change, where the individual remains more or less stable, but the new generation
of speakers brings change; (4), communal change, where both individuals and their
community seem to change simultaneously; and (5) lifespan change, where indi-
viduals and communities exhibit more or less random effects and changes without
any apparent pattern (apart from the fact that the individual changes his or her
language at a later point in life).
Gender
The third and last factor we will look at is gender. But first we need to define what
gender is. For this, we need to distinguish between biology and culture. Biology
determines your physiological, bodily sex as most commonly male or female. Your
gender, however, is not biologically determined, but dynamically negotiated in a

specific culture. It comprises how you act, feel, dress, behave and so on as a member
of society. This can also be male, female or somewhere in between, and it can change
constantly (unlike your biological sex which ordinarily doesn’t change). So you can
have a biologically male sex, but behave in a very female (gender) way, or vice versa.
And this may be different from time to time, as you may play a different role for dif-
ferent people, or different people may perceive you in different ways. In the follow-
ing, we will not look at biological sex, but cultural gender. The way you talk is only
very minimally influenced by your physiology. Bodily differences between males
and females are probably only relevant for some acoustic features such as pitch,
voice and vowel quality, and a few others (if you are interested, Simpson 2009 gives
an excellent overview). But gender may play a key role in language variation and
change, as we will show in the following sections.
Gender has been investigated as an independent variable in sociolinguistics for
many decades now. And most research found that males and females speak in dif-
ferent ways (this is hardly surprising), and also that they consistently differ in the
way they relate to language norms (this is surprising). We can generally distinguish
between standard language forms, which carry overt prestige and which are propa-
gated through norms and prescription, for example at school, and non-standard
language forms, which often carry covert prestige for some people and which are
prohibited by language norms. One very simple example: in Standard English, we
expect only simple negation, as in I haven’t done anything. This is the Standard Eng-
lish structure that we all learn at school. Multiple negation, as in I haven’t done noth-
ing, is the non-standard structure, frowned upon and corrected by teachers. Simple
negation carries overt prestige; i.e. it is sanctioned and encouraged by mainstream
society. Multiple negation, however, while being sanctioned by mainstream society,
can still be cool and attractive for some (for example drop-out teenage boys). The
latter is called covert prestige.
On average, it seems, female speakers tend to use forms more often that carry
overt prestige. Figure 7.8 presents data from a study in Philadelphia carried out by
Labov. The variable is (ing), i.e. the pronunciation of kissing as [kɪsɪŋ] (standard) or
[kɪsɪn] (non-standard). The lower the score, the more standard forms [ɪŋ] can be
heard.
As we can see in Figure 7.8, two significant patterns emerge. First, the rate of non-
standard forms in casual speech is considerably higher for both genders and across
all social classes. Second, for most social classes, women show fewer non-standard
forms than men in the respective style. So, MWC women in casual style use fewer
non-standard forms than men do, and so on. The only exceptions are UWC women
in careful style, who show the same rate as their male counterparts in the same style.
But, statistically speaking, the difference between men and women is significant
across the board, and we can conclude that, at least for this variable, women tend to
show fewer non-standard forms than men do and, conversely, use more prestigious,
standard forms than men do. This seems to be a fairly robust pattern in many socio-
linguistic studies: women, on average, use more standard, prestigious forms than
men and avoid non-standard forms.
Figure 7.8
(ing) in Philadelphia (based on Labov 2001: 265)
100
90
80
70
60
% (in)
50 Men:casual
40 Men:careful
30 Women:casual
20 Women:careful
10
0
las
s ss las
s
as
s ss
C Cla C Cl Cla
ing kin
g ing dle id dle
ork Wo
r ork rM
id
rM
r W le r W e e
we idd pe ow Upp
Lo M Up L
What does this mean for linguistic change? In the previous discussion, we have
distinguished between change from above and change from below. Change from
above is usually initiated and propagated by prestigious social groups. Schools,
universities and the media play an important role here. So, generally, change from
above works towards establishing (or maintaining) standard forms. Change from
below, on the other hand, establishes or maintains non-standard forms and struc-
tures; it works from within the linguistic system, often goes unnoticed and leads
to greater linguistic variability. Looking back at what we have said before about
tendencies in female language use, it becomes clear that women seem to play a
vital role in changes from above. They tend to use more standard, overtly presti-
gious forms. And what is their role in change from below? This is a bit more tricky.
Figure 7.9 shows a study on the devoicing of [ʒ] to [ʃ] in Buenos Aires by gender
and age (the first person pronoun ‘I’ yo is pronounced either [ʒo] or [ʃo], and ‘call’
llamo is pronounced either [ʒamo] or [ʃamo]). The higher the score, the more new,
devoiced forms we find.
Here, in this change, women across all age groups are clearly the leaders. What is
interesting, however, is that this is a change from below (or within). It was not initi-
ated or propagated by powerful social groups, it leads away from the standard pro-
nunciation, there is little to no social awareness of this change, and there is hardly
any style shift involved when the frequencies are compared for more informal and
formal styles (we would expect fewer non-standard forms in formal styles if speak-
ers were aware of what they are doing). So what we see here is a change from below
with women as leaders. This is in obvious contradiction to what we discussed before,
namely that women are the leaders for changes from above. What is even worse: this
Figure 7.9
Devoicing of [ʒ] in Buenos Aires by gender and age (adapted from Wolf and Jiménez 1979)
90
80
70
60
50
[∫ ] in %
Male
40 Female
30
20
10
0
55+ 36–55 24–35 18 15 12 9
is not a single incident. Labov (2001) documents several such cases where women
are the leaders in change from above – and change from below. This leads him to the
formulation of a gender paradox in linguistic change: “Women conform more closely
than men to sociolinguistic norms that are overtly prescribed, but conform less than
men when they are not” (Labov 2001: 293). What does this mean? The first part of the
paradox is fairly clear: if there is an overt linguistic norm (e.g. not to use the double
negative) women will follow this rule, on average, more closely than men. The same
also holds true for linguistic variability that reaches the level of awareness or that car-
ries some stigmatization: speakers are aware of the [ɪŋ]/[ɪn] variability, and of the fact
that [ɪn] is a non-standard, less overtly prestigious form (even when there is no open
rule against it). In those cases, women will tend to move towards the standard, overtly
prestigious forms. But there are also changes which stay below the level of awareness
and which are not subject to overt, explicit bans and stigmatization, even though the
new forms do not belong to the overtly prestigious standard. In those cases, women
are also the leaders of change. It is only in those cases where people are aware of the
non-standard forms and the stigma they carry that men take the lead in change and
women use fewer of these non-standard forms than men.
7.2.3 Social networks

The study of social groups, or social classes, dominated sociolinguistic studies for
quite a while and led to some remarkable insights. However, in the 1970s and 1980s
there was some growing dissatisfaction with the sociological background to this.
After all, we lead very dynamic lives and a stable social class categorization does
not really do justice to this fact. Every day you interact with different people in very
different ways, and only part of this behaviour is reflected in stylistic differences.
Furthermore, we need to ask ourselves how changes can actually spread from one
class to the next. In the previous sections we suggested that one such mechanism
could be tendencies in certain social classes to use language variables in particular
ways owing to the desire to be upwardly mobile, to belong to a better social class.
But this does not explain all the different change patterns that we see.
For that reason, Lesley Milroy (and others) introduced the notion of social net-
works to sociolinguistic analyses (Milroy 1987; see Bergs 2005 for a comprehensive
overview). Instead of grouping speakers into different classes, depending on certain
social variables, social network analysis looks at the social connectedness of speak-
ers. The assumption is that it makes a huge difference whether you have few contacts
or many, and also whether you know these contacts only in one single function (for
example as boss and employee) or in many (as friend, neighbour and boss). The for-
mer would be looser and more shallow, while the latter would be a very intensive
connection. Note that all this is fairly independent of a social class point of view. One
example: picture the contrast between people living in a small village in the country,
and people living in a large urban area. How do their lives differ, on average? People in
the country would usually have smaller networks; i.e. they tend to know fewer people,
but they would know these people in a variety of functions. Your neighbour is also
your workmate and your friend, and she plays on the same soccer team. Moreover,
many of your contacts would know each other. The result is a very dense network.
And what about the city? Usually, people in the city have a lot of contacts. But these
contacts tend to be uniplex; i.e. you know people in only one or two functions (e.g. as
neighbour or as friend or as workmate). Moreover, your contacts probably don’t know
each other. The resulting network is very loose and wide. What are the consequences
of these differences? Villages and their inhabitants tend to be rather conservative, or
norm enforcing, in their overall behaviour. Cities and their inhabitants, by contrast,
tend to be rather innovative and open to new things.
So how do we measure or determine social network structures? In Milroy’s case,
she started by asking her informants very simple questions like: Where do you live?
Who are the people you frequently interact with? In what capacities do you know
these people? Where do you work? Do other people from the neighbourhood work
there as well? The result was a list of five more general criteria that helped to estab-
lish the quality of the network of a given speaker. These were:
• “membership of a high-density, territorially based cluster”

• “having substantial ties of kinship in the neighbourhood”
• “working in the same place as at least two others from the same area”
• “the same place of work as at least two others of the same sex from the same area”
• “voluntary association with workmates in leisure hours”
(Milroy 1987: 142)
Speakers could score up to five points on this network strength scale, where five
points meant a very dense network, zero or one point a very loose-knit one. A close-knit
network here would be norm-conservative in the sense that it is sealed off against out-
side influences, so that, for example, local non-standard forms can be maintained.
Central members of such a network are usually very resistant towards change. Mar-
ginal members, i.e. those with a low network strength score, may have more contacts
outside the network, and are more susceptible to outside influences. This can mean
either another network with different local norms, or even supralocal standards. In a
given network we can distinguish between at least three or four different roles. There
is the norm-conservative central member, the marginal member, the average member
and the peripheral member, or bridge. This is visualized in Figure 7.10.
A peripheral member such as (4) in Figure 7.10 can act as a bridge between two net-
works and can transport forms from one to the other. However, these new forms have
to catch on in the network. Central members are usually resistant towards change,
but if they encounter the new form with sufficient frequency, they may become early
adopters, i.e. speakers who quickly adopt the new form and propagate it in their net-
work. The reason for such behaviour is actually quite simple. Central members tend
not to introduce new forms, so as not to endanger their central position. But when
they realize that there might be a new, dominant form on the way, it is crucial to be
among the first to use it in order to stay ahead of the group. Still, in many cases, new
forms will not make it into the network and will be restricted in their use to marginal
members (who themselves don’t have enough power to bring about change).
A social network approach to linguistic change has a number of advantages. It
looks away from abstract social groups based on particular features (such as income
or education) and focuses on the individual in his or her particular social environ-
ment. In that way it allows us to trace changes and how they diffuse in social struc-
tures in much greater detail. The obvious drawback is, of course, that it requires a
Figure 7.10
The expanded “dots-and-lines model” (Bergs 2005: 29)
(5) °
(3) °
° ° ° °
(2) °(4) ° ° °
°
°(1)
(1) Marginal member (2) Focus/central member (3) ‘Average’ member

(4) Peripheral member/bridge (5) Cluster/Clique
solid line = ‘strong tie’ dotted line = ‘weak tie’
lot more fieldwork and data in order to identify speakers’ networks. The second
advantage of this approach is that networks can be seen as dynamic and constantly
evolving. By definition, you are a member of only one social class, and usually you
remain a member of that class. Obviously, this is a very static view of social inter-
action. But you can be member of several networks, and these networks continue
to evolve. So when you join a new university club, or a sports team, your network
immediately expands, just like when you start at a new work place. Similarly, exist-
ing networks disappear, for instance when you move to another city. All these events
have consequences for your networks and for your verbal behaviour. You meet new
people, you encounter new linguistic structures and expressions, and you bring
your own language to these networks!
In a nutshell, social network analysis is the ideal complement for large scale anal-
ysis based on social class, for example. Studies of social classes help us to see the big
picture and the major changes in a linguistic community, while social network anal-
ysis gives us the tools to look at changes in individual speakers and their networks.
7.2.4 Communities of practice

A third concept we can look at here is communities of practice. These are commonly
defined as aggregates of people in mutual engagement, with a joint enterprise and
a shared repertoire of practices. In other words, these are groups of people who
get together with a common goal and who share the same language use, or style of
clothing, or other signs of their belonging together. This does not mean that com-
munities of practice are typically peaceful, united groups. There can be disagree-
ment between members, and even fights. Sometimes members will drop out, and
others will join. Some examples of communities of practice are the burnouts, jocks
and nerds at high schools – or the trekkies (the fans of Star Trek), if you like. These
all have specific enterprises (hanging out and rebelling against school and society for
the burnouts; participating in school life and being sporty, competitive and active
for the jocks; and discovering scientific challenges for the nerds, for example). They
show mutual engagement (they meet in some way and interact), and they share cer-
tain repertoires (clothing, haircuts and music, but also language). In this way com-
munities of practice are not unlike social networks. In fact, social networks are the
structural backbones of communities of practice. But our look at them is different.
In social network analysis, we focus on structures and on the quantity of relations.
When studying communities of practice, we are more interested in the quality of
these relations, i.e. how a given individual or group interacts, and what the features
of these groups could be, especially from a linguistic point of view.
Communities of practice are obviously very interesting when it comes to the dif-
fusion of linguistic changes. In developing their repertoires, communities of practice
distinguish themselves from other groups in society. Members can share these rep-
ertoires, and they can develop them. New members often change their own language
in order to fit; sometimes new members can bring new forms into the group. Com-
munities of practice can allow for certain forms, or even encourage their use, just as
they can disapprove of certain linguistic expressions. In other words, communities
of practice can promote or stall change in the linguistic community at large. This
is even more interesting, as sometimes we find communities that explicitly discuss

these issues. Some of them even focus on language: the Verein deutsche Sprache e.V.
(www.vds-ev.de), for example, has the explicit goal of keeping the German language
free of Anglicisms, i.e. English loanwords. This is their joint enterprise. Their mutual
engagement happens in meetings, through a newsletter and online, and the mem-
bers of this group clearly share a common repertoire (they replace alleged Angli-
cisms with German words, e.g. Ruhmeshalle instead of Hall of Fame). Very similarly,
the NoCussing Club (www.nocussing.com) promotes the avoidance of swearing.
Again, the joint enterprise, with a mutual and shared linguistic repertoire (in this
case avoidance of swear words), is easy to discern. And the potential impact of such
communities of practice on language development is obvious. But other communi-
ties of practice which are not concerned about language as such can have similar
effects. Penelope Eckert showed in her 1999 study of the language of high school
students how communities of practice such as jocks versus burnouts can have sub-
stantial influence on the language use, especially pronunciation features, in these
groups. And it is important to remember at this point that communities of practice
can cut across boundaries of social class and gender in their composition. All it takes
is that you subscribe to the joint enterprise, engage with the others and share their
“code” or repertoire (provided, of course, that you are admitted to the group).
SUMMARY
In this chapter we had a detailed look at how language change can spread. We started
out by looking at diffusion in the linguistic system itself, and we distinguished
between abrupt changes that affect all items in question and gradual changes which,
for example, can exhibit lexical diffusion in an S-curve.
In the second part we looked at how linguistic changes can spread in the commu-
nity. We looked at models of geographical diffusion (the wave, gravity and cascade
models) and social diffusion. We focused on various social factors (class, age, gen-
der), social networks and communities of practice. All these offer different perspec-
tives on one and the same phenomenon, namely the fact that linguistic change seems
to creep gradually through society, like inventions and viruses. It is no surprise that
for many changes we find the same pattern in social diffusion that we found in
lexical diffusion: the S-curve. At first only a few speakers are affected, and then we
get a rapid spread in the middle part, with many speakers being affected, until the
whole change process peters out. Class, age, gender, social networks and communi-
ties of practice help us to capture and to explain this at various levels of granularity:
from the macro-level (class) to the micro-level (network), from quantitative aspects
(classes and networks) to questions of quality (communities of practice).
FURTHER READING
There is a wealth of literature now on innovation and diffusion in linguistic change.

Lesley Milroy (1987) introduces you to the details of social network analysis for
present-day English; Bergs (2005) utilizes the approach in the analysis of language
variation in Middle English. James Milroy (1992) offers a fascinating and influential
perspective on innovation, actuation and diffusion in society. Eckert (1999) pres-
ents a comprehensive account of youth culture and the language of jocks and burn-
outs from a communities of practice perspective. Tagliamonte (2012) gives a very
modern and readable introduction to the empirical study of language variation.
Denison (2003) introduces you to the S-curve in linguistic change and elsewhere.
Other classic must-reads are Trudgill (2011), a collection of his most influential
works, as well as the ground-breaking trilogy Principles of Linguistic Change by
Labov (1994, 2001, 2010).
EXERCISES
1 Schwa-deletion
As described in this chapter, English is currently losing [ə] from the middle of
words. For example, most English speakers now pronounce every as [ɛvri], not
[ɛvəri]. But because sound change is gradual, [ə] doesn’t disappear across the board.
Do you pronounce delivery as [dəlivri] or as [dəlivəri]? Read the following words
out loud and take note of where you delete [ə].
victory, adultery, century, every, cursory, delivery, hickory, elementary, factory,

nursery, slavery, forgery, family, burglary, boundary, scenery, desultory, library
2 Yod-dropping
First read the following words out loud and take note of whether or not your pro-
nunciation has [j] (yod) – for example [tjun] versus [tun].
assume, new, askew, crew, blue, superb, Susan, huge, Hugh, mute, pew, abuse,
argue, lewd, issue, beautiful, dew, tune, stew, suit, tutorial, enthusiasm, nude, cute,
tissue, clue, cue, rule, few, flute, overdue, fuel, presume, dune, duty (there are other
words you can add here).
• Now find 10 people – 5 of each sex.

• Take note of their age and educational background.
• Put the above words into a larger list with other words (so your infor-
mants don’t twig to what you are investigating and alter their pronunciation
accordingly!).
• Get each of these 10 people to read this list. Make note of the pronunciation of
these words.
• Determine where there is yod-deletion and where there isn’t.
• Can you explain the variation you find? (Hint: Look at the other sounds in the
vicinity.)
• Compare the pronunciations of these words in a dictionary (such as the Oxford
English Dictionary). What generalization can you make?
3 Short-answer questions
a Why do you think linguists traditionally regarded change as unobservable?
b Frequency of use has a profound effect on linguistic behaviour – frequent
everyday words are first to be affected in sound change but last to be affected
in grammar change. With examples, explain this paradox. (You might also
want to revisit the discussion in Chapters 4 and 5.)
c What is meant by lexical diffusion?
d Briefly describe the social modal of language change based on social net-
working theory; in particular, show how network analysis can contribute to
our understanding of language variation and the mechanisms of language
spread.
e Using examples, discuss what is meant by the following catchphrase: “Varia-
tion is the synchronic aspect of change – Change is the diachronic aspect of
variation.” How accurate a picture does this present of the spread of change?
4 Social networks and communities of practice

Try to analyse some of your own social networks and communities of practice.
What are the similarities and differences? What kind of role do you play in the net-
works? Can you identify bridges, early adopters and marginal members?
5 Research essay
The last couple of decades have seen some massive social changes in many parts of
the western world. We are witnessing a dramatic increase in unemployment rates
for young people, and some changes in attitudes towards (female) gender identities,
i.e. the development of a “girl power” movement and the association of female iden-
tities with formerly rather male attributes such as independence, power, strength
and toughness (some interesting discussion can be found in Rosalind Sibielski’s
2010 PhD thesis, “What Are Little (Empowered) Girls Made Of? The Discourse of
Girl Power in Contemporary U.S. Popular Culture”3).
Do you expect these developments to have long-term effects on the role of
women in linguistic change, as discussed above? If so, try to find some evidence for
the changing role of girls and women in linguistic change!
NOTES
1 As a brief reminder: The prefix Proto means that this language or dialect is only recon-
structed (see Chapter 9), so we don’t have direct evidence for it. Reconstructed forms are
usually marked with an asterisk “*”.
2 In Chapter 4 we discussed exceptions to the Law of Latin Rhotacism – words like miser
‘miserable’ where /s/ didn’t change to /r/ because the following /r/ blocked this from hap-
pening (speakers don’t like too many liquids in the one word).
3 https://etd.ohiolink.edu/rws_etd/document/get/bgsu1277091634/inline).
8
Languages in contact
INTRODUCTION
There are said to be around 200 countries in the world and approximately 7,000 lan-
guages that somehow have to fit into these countries (we’ll ignore here the difficulty
of identifying and classifying countries and languages; for that, see Chapter 9). So,
clearly, we’d expect a lot of contact going on between these languages, and we’d also
expect most children in the world to be growing up bi- or multilingual (which is in
fact the case).
In this chapter we focus on what happens to languages when they encounter one
another, especially on the interaction of the internal and external factors that go
together to shape the linguistic outcomes of the contact. These outcomes can be any-
thing from the incorporation of occasional vocabulary items to the drastic overhaul
of the fundamental grammatical organization of the language. They might entail the
happy coexistence of two (or even more) languages or the dominance of one language
over others. They might even result in the creation of a brand new hybrid language.
Languages don’t have to come into direct contact for linguistic features to cross
borders. Even overwhelmingly monolingual countries will not be unaffected by
outside influences. In places like Britain, Australia, Japan, France and Germany, for
example, children will usually become native speakers of only one language, but of
course there are quite large groups of people in these countries that make regular
use of other languages, and there is considerable linguistic diversity. So although
people may have little or no face-to-face contact with speakers of other languages,
foreign features (usually words and expressions) can be transferred by mass media
and via language teaching. The fact that English is currently the global lingua franca
(or common language) places it in an interesting position in this regard. Over many
years it has been making contributions to the lexical coffers of many languages;
some originally English words (like ok) are now so widespread as to be truly inter-
national. But this position also has its down side, and we find a number of people
who “blame” English for the decay of their language(s) – a claim which is untenable
from a linguistic perspective – or who even describe English as a “killer language”
that gradually eliminates other languages. There is some truth to the latter, but the
story is certainly more complex, as we will go on to describe.
There will always be an array of different factors that go to shape the types of changes
that arise from language contact. These factors can be linguistic, including how differ-
ent or similar the languages are (for example, it is easier for languages to incorporate
constructions that are comparable to ones they already have). The factors can also
be social and psychological, including the length and intensity of the contact, as well
as the power and prestige of the relationships involved. Also relevant is the fact that
different levels of language structure (e.g. phonology, prosody, grammar and lexicon)
are differentially sensitive to these factors, and this will have a bearing on the nature
and direction of the transfer of features, structures and rules from one language to
another. Before we consider these complexities, we need to look generally at the differ-
ent contact situations that can trigger such very different cross-linguistic influences.
8.1 TYPES OF CONTACT
To make our account of contact induced change more straightforward, imagine for
the moment a fictitious speech community whose language we will call Karmaic.
This group of speakers comes into contact with another group of speakers whose
language is totally different. We’ll call this language Wok. Ignoring for the moment
details to do with the degree of bilingualism, the nature of the contact, the size
of the linguistic groups and the power relationships holding between them (and
how the respective speech communities feel about each other), we can predict three
overriding kinds of contact scenarios:
1 Language maintenance – where speakers of Karmaic continue to use their

native language from generation to generation
2 Language shift – where speakers abandon Karmaic in favour of Wok
3 Language creation – where long-term language contact between the two lan-
guages gives rise to a new contact language (it might be a pidgin, a creole or a
mixed/intertwined language showing features of both Wok and Karmaic)
There are real contact situations that won’t fit neatly within these umbrella cat-
egories (language is always more complicated than textbooks make out). But by
imagining these three scenarios we can broadly explore the different patterns of
influence that potentially take place when language groups encounter one another.
8.1.1 Language maintenance

Contact with Wok over a long period of time will mean that some Wok features will
enter Karmaic. As described in Chapter 2, borrowing is the usual term given to this
process whereby a language takes and incorporates some linguistic element from
another language. Strictly speaking, of course, this isn’t borrowing (which suggests
items are simply on loan with the understanding they will be returned at some later
stage) – language loans are not handed back.
When the arrivals (in this scenario, speakers of Wok) are linguistically absorbed
into the indigenous population (here Karmaic), the influence of their language
is described as being either superstratum (if Wok speakers were a socially domi-
nant group) or adstratum influence (if Wok speakers were a socially equal group).
All parts of the language can end up being borrowed, even grammatical patterns,
though usually this requires long-term and widespread bilingualism between
borrowers and lenders. So we can imagine a kind of hierarchy of “borrowability”
Languages in contact 191
(though no absolute hierarchy can be proposed, since there are so many other inter-
fering factors having to do with type and nature of the contact and the compatibility
of the linguistic systems).
Lexical items are the most easily transferred, and these will be content morphemes
(especially nouns and verbs with real world meanings). Contact doesn’t even have to
be that extensive for this to occur. A bewildered Karmaic speaker just has to point at
something and look puzzled, and the Wok speaker will probably supply the missing
lexical item. Structural borrowing (grammatical morphemes, features of syntax) is
much rarer, however. Basically, the more grammatical the element, the less likely it
will be borrowed. Free-standing grammatical forms (such as conjunctions and prepo-
sitions) are borrowed more easily than bound forms (such as a past tense or plural
suffix), but if Karmaic borrows heavily enough from the lexicon of Wok, it might
also acquire some derivational and (more rarely) inflectional affixes along the way –
perhaps even some phonological features, although this is unusual (we discussed one
such case in Chapters 1 and 4 – the phonemicization of English /f/ and /v/ due to con-
tact with Norman French; however, there were also changes internal to the language
that supported this). Generally, the extent and nature of this kind of borrowing will
depend a lot on how intense the interaction is between the languages, and also on how
compatible the two systems are. It will be easier to borrow Wok constructions that are
similar to existing ones (or at least that don’t conflict with the basic structure of Kar-
maic). For instance, in rare cases where languages do take on inflectional affixes, they
are usually well matched (or they end up getting regularized); for example, the -s plu-
ral ending on English words in German (like Jeeps) coincides with a native -s ending.
There will also be crucial social and psychological factors (beyond simple questions of
prestige) that play a role in determining what kinds of linguistic features are involved,
and the extent of the borrowing. Changes can occur even when they encompass lin-
guistically divergent systems because of the social dynamics at play.
It is usually the case that borrowing involves some sort of gain – the borrowers
profit in some way from the transaction. The gain might be social. Karmaic speakers
might take on Wok words (and even pronunciations) for reasons of fashion or social
clout. Cross-linguistic influences (or contact induced changes or transfers, as they’re
sometimes called) might also be therapeutic. In the case of vocabulary, perhaps the
transferred material replaces an expression that has become obsolete or has lost its
expressive force. Perhaps the motivation is sheer necessity. It may simply be that a
borrowed word from Wok fills a gap in the Karmaic lexicon; so it might involve an
artefact or a concept known to Wok speakers but not to Karmaic speakers.
English in contact with French and Scandinavian

English has two examples in its history where very heavy lexical borrow-
ing took place. After the French speaking Normans conquered England in
1066, English was well and truly under the Norman French thumb. French
was the superstratum – it was the prestige, or educated, language (as also
was Latin) and English remained the language “of the street”. It eventually
prevailed and French was ousted, but French left its mark on the English
lexicon (see Serjeantson 1936; Stockwell and Minkova 2001). The French
controlled the state, the military and cultural and intellectual interests, and
French words flooded into these areas, sometimes as brand new additions,
sometimes ousting the English expressions. Around 10,000 words were
adopted – the majority between 1250 and 1400 – and some 75% of these
survive today. Had there been no French conquest, many native English
words that were ousted may well have survived (though it can be hard to
predict the shelf-life of lexical items); for example, medical practitioners
might still be called leeches instead of doctors, physicians or surgeons (words
that came into English via French). The influx of these French words also
created the system of stylistic levels that characterizes the modern lan-
guage. The native English words are the fundamental everyday vocabulary
(typically shorter, more concrete and stylistically more neutral), and they
support the lexical superstructure comprising those vocabulary items of
refinement and nuance that come to us from French; ask and rise are Eng-
lish, and question and mount are French (and interrogate and ascend are
Latin; see Hughes 1989).
Contact between English and Scandinavian languages was similar, but with
a different bunch of invaders and a very different outcome. At various stages
during the 9th and 11th centuries there were Scandinavian raids on Eng-
land. These culminated in 25 years of Danish rule, with large settlements of
Scandinavians, particularly in the northern and eastern parts of the country,
and considerable influence on native English. However, Viking Norse and
Old English were in an adstratum relationship. They existed side by side with
more or less equal prestige, and the borrowings reflected this. The expres-
sions English took in during this time are striking for their everyday nature.
They include basic verbs like die, get and give; basic nouns like husband, fellow
and egg; and even grammatical words like they, them and their. These are just
a handful of the hundreds of words that entered the language at this time.
Borrowings of this sort suggest that the languages must have been very close –
so close, in fact, that it is actually difficult to assess the true extent of the
Scandinavian contribution. There could have been many more borrowings,
but it is hard to know.
8.1.2 Language shift

We now look at a different situation, in which our Karmaic speakers give up their
native language in favour of Wok. Language shift involves changing patterns of lan-
guage use within a multilingual society, and it can be total or partial (e.g. Wok is
used for the majority of functions and Karmaic for special uses). We can identify
three broad categories of language shift:
1 Complete language shift. The entire community switches to another (usu-

ally dominant) language. In migrant communities the original language is
generally still spoken in the country of origin. In indigenous communities
this leads to the death of the original language.
2 Partial language shift: domain restriction. The dominant language of the
wider community encroaches on the domains in which the minority lan-
guage was formerly used. The minority language continues to be spoken in
one or two domains.
3 Partial language shift: population reduction. The proportion of people
who speak the minority language (whether as monolinguals or bilinguals)
is markedly reduced in relation to people who only speak the dominant
language.
If elements from the primary language (Karmaic) are transmitted to later gen-
erations of speakers of the prevailing language (Wok), it is typically pronunciation
that is affected, and possibly also grammatical patterns. This is different, then,
from borrowing into a maintained language, which, as we have just seen, involves
the incorporation of lexical aspects first and foremost and (in cases of close con-
tact) possibly also structural features. The sorts of contact induced changes in
situations of language shift are also not usually referred to as borrowing but rather
as substratum influences, and they involve the transfer of first language habits into
a second language (sometimes given the rather judgmental label of “improper
learning”).
The Celtic substratum

English didn’t originate in Britain, and when Germanic tribes arrived in the
fifth century they came into contact with various Celtic tribes. The majority
of Celts seem to have learned the language of their Germanic conquerors
thoroughly, and the impact of Celtic was slight – confined to place names like
Thames, Devon and London and a handful of lexical borrowings such as ban-
nock ‘scone-like cake’, down ‘hill’ (preserved only in place names) and pos-
sibly iron. This is very similar to the impact of indigenous languages on the
standard varieties of English in the United States and in Australia; it is largely
names of places, plants, trees, wildlife and cultural items that have been
incorporated (such as moccasin, opossum, skunk, tomahawk, wigwam, Mani-
towoc, Susquehannah, Texas, Utah and boomerang, corroborree, paddymelon,
kookaburra, budgerigar and about one third of Australia’s place names). The
spread of English throughout the British Isles led in later centuries to the shift
of Celtic languages to English in places like Wales, Scotland and Ireland. The
Celtic underlay that lingered gave rise to the very distinctive pronunciations
you still hear in the former Celtic regions.
As an aside, there are obvious practical advantages to knowing about substratum

influences in areas such as language teaching. What is often called “a foreign accent”
involves the transfer of speech habits into a second language. This is understand-
able – we naturally hear and perceive sounds and sound patterns through the filter
of our own language.
Say an English speaker is learning Russian. The Russian phoneme /r/ is pro-
nounced as a trill (with the tip of the tongue vibrating against the teeth ridge),
whereas in many varieties of English /r/ is a postalveolar continuant (with the
curled tongue tip approaching the area behind the teeth ridge). The English pro-
nunciation tends to carry over into Russian from English speakers. So “interfer-
ence” can involve the physical aspects of sound production, and it can also involve
trickier aspects of the sound system. English has voiced and voiceless (inter)dental
fricatives ([θ] and [ð]), but these sounds aren’t found in many languages (suggest-
ing there is something intrinsically difficult about them). Russian speakers learning
English substitute /d/ and /t/; French speakers substitute /s/ and /z/. What is inter-
esting here is the nature of the replacement sounds. French speakers could opt for
the stops, but they maintain the fricative feature (so the mode of articulation); Rus-
sian speakers could go the fricative route too, but they opt to maintain more of the
English place of articulation (and you’d have to look closely at the sound systems of
these two languages to know why speakers make different choices here).
Phonemic distinctions of the first language are often imposed on the second
language, and this contributes strongly to a perceived foreign accent. Language X
might lack a distinction that is present in language Y, and if language X speakers
impose their sound patterns on language Y they will underdifferentiate sounds.
Japanese, for example, has one liquid consonant, which can be realized as either a
lateral [l] or a rhotic [r]; English has two liquid phonemes, /l/ and /r/ (and there are
many minimal pairs, like light-right, climb-crime, arrive-alive etc., to support this). It
requires extra effort for speakers to perceive and produce sounds accurately if they
lack the contrast in their first language.
Restrictions on the arrangements or sequences of speech sounds can also pro-
vide stumbling blocks for foreign language learners, who might fail to acquire the
patterns of their new language or, more usually, end up transferring the rules of
their primary language into the new language. Take something as straightforward
as word-final position. In English, /h/ is the only consonant that can’t appear
word finally, but other languages are more restricted. In Chapter 4 we described
how German speakers devoice final stops and fricatives; many carry this rule over
to English (so trend is pronounced as [trɛnt]). Japanese prefers words that end
in vowels. In fact, the only final consonant that doesn’t present difficulties for a
Japanese speaking learner of English is final [ŋ] (as in bling); everywhere else, the
natural solution is to insert an extra vowel at the end, and English loanwords (or
gairaigo) reflect this (e.g. songu < ‘song’; kēki < ‘cake’; batto < ‘bat’). Spanish has
restrictions on final nasals, and Spanish speakers can have trouble identifying and
pronouncing nasals other than /n/ in final position (so long might be pronounced
as [lɒn], not [lɒŋ]).
8.1.3 The creation of new languages

We now move to the third kind of contact scenario, where the outcome of contact
between our two hypothetical languages is much more striking. Rather than Wok
or Karmaic simply changing as a result of contact, their interaction triggers the
birth of a brand new language (a pidgin or a creole) or perhaps a type of bilingual
mixed language, combining, say, the grammatical elements of Wok and the lexical
items of Karmaic.
Generally speaking, pidgins are a type of makeshift language that springs up
when speakers of different linguistic backgrounds come into contact and there is
an urgent need to talk. In the formation of a pidgin, there are always two (or more)
languages involved, although the pidgin takes one language, usually the socially
dominant one, as its point of origin for the lexicon (the lexifier language). This lan-
guage contributes most of the vocabulary, and significant features of the grammar
are likely to derive from other sources.
Traditionally, people thought of pidgins as simply debased forms of whatever
the dominant language was, and they used derogatory labels like “broken English”,
“monkey-talk”, “impoverished English” and even “gibberish”. This “me Tarzan –
you Jane” image of pidgins is totally inaccurate. Pidgins have their own grammars.
There might be a lot of variation (especially in the early stages of their creation), but
they are not simply defective varieties of another language. However, they do differ
from a natural language in a number of respects:
• Pidgins aren’t anybody’s first language but are used when speakers venture out-
side their usual speech community.
• Because they are used in limited contexts, pidgins lack the range of stylistic
variation characteristic of a normal language.
• Pidgins make do with reduced vocabularies and require far less complex and flex-
ible structures (English based pidgins, for example, are syntactically and morpho-
logically much simpler than the variety of English on which they are based).
Typically, a pidgin is short-lived. Speakers often abandon it in favour of the source

language, or if the contact ceases, the pidgin will simply die. However, if the situ-
ation stabilizes, and the contact continues, the makeshift pidgin metamorphoses
into a fully fledged and dynamic language. Because the language must now serve
its speakers in all kinds of settings and circumstances, the changes are typically
rapid, especially in vocabulary and grammar. For this reason, pidgins and creoles
have become central to the study of language change – and also subjects of intense
debate; as McWhorter (2003: 202) describes, “the interface between creole studies
and language change has been a tumultuous area since the 1990s [. . .] Concepts
often taken as assumptions ten years ago are now widely questioned, even the very
definition of creole itself.”
In theory, it is straightforward to say when a pidgin ends and a creole begins. As
soon as children in a community are brought up speaking the pidgin as their first
language, it becomes a creole by definition. Accordingly, a creole is simply a nativ-

ized pidgin. The linguistic reality, however, is another matter – linguistically it is
difficult to say where the boundary lies. Even before a pidgin becomes somebody’s
first language, it can develop an elaborate structure (close to that of a creole) if it
is used for a number of different purposes. For this reason, some linguists even
avoid the labels “pidgin” and “creole” and refer to these varieties straightforwardly
as “contact languages”.1
English based creole varieties cluster in three main areas – the Caribbean, West
Africa and the West Pacific. They include Bislama (Vanuatu), Krio (Sierra Leone),
Tok Pisin (Papua New Guinea), Gullah (spoken by African Americans in Florida
and the Sea Islands), Costa Rican Creole, Jamaican Creole and Kriol (northern parts
of Australia). All of these languages derive their vocabulary predominantly from
English, the superstrate language (or lexifier language), but they are grammatically
very different. Structurally they show parallels with their substrate languages. How-
ever, it is also true that creoles all round the world share features, most dramatically
in their grammars. For example, they are overwhelmingly isolating languages (so
they do not have much in the way of grammatical morphology). Here are some of
the characteristics that commonly appear in creoles, even those of very different
linguistic backgrounds:
1 Few bound forms (e.g. number words often used to indicate plurality; tense
and aspect indicated by context or by particles)
2 Fixed word order (often Subject-Verb-Object)
3 Negative concord (more than one negative element occurs in the sentence)
4 Use of reduplication
5 Possession expressed by juxtaposing the possessor and possessed noun
phrases, or by particles
6 No case distinctions for pronouns
7 Few adpositions (prepositions or postpositions)
8 Use of repetition for rhetorical effect
In the 1980s creolist Derek Bickerton put forward what he called “the language
bioprogam” (e.g. Bickerton 1981, 1986). Drawing on neurology, biology and a num-
ber of other fields, he argued that the structural similarities across different creoles
can’t be attributed to the substrate or superstrate languages, but rather are the fall-
out of genetically programmed innate faculties. It remains a highly controversial
idea, and you can find out more about both his theory and its critics in the readings
given at the end of this chapter.
We shouldn’t give the impression that all creole grammars are the same; there is
also considerable variation between them due to a number of factors: (1) the fall-
out of typological differences shown by their respective substrate languages (with
respect to pronominal forms, case systems, verb properties, negation and discourse
structures, among other features); (2) contact with local vernacular languages
(which may or may not be an additional first language of speakers; (3) ongoing
contact with the superstrate language (particularly for urban groups, a growing
force of influence). Typically, contact languages range along a continuum of vary-

ing features and abilities, from varieties (called “acrolects”) close to the superstrate
language (e.g. Standard English) in everything but accent through to varieties (the
“basilects”) that are not mutually intelligible with the standard. Situated between
these two polar extremes, there exists a range of varieties (or “mesolects”). Speakers
have command of a number of these linguistic types, and their use depends on the
situation and the audience.
Contact languages downunder

Both New Zealand and Australia offer situations where English came into
close contact with other languages. In the early 1880s, a “foreigner-talk” sys-
tem known as South Seas Jargon was used in various parts of the Pacific,
primarily between European whalers and indigenous crew members, some of
whom were Māori (jargon here refers to a kind of pre-pidgin hybrid language
and is different from the more usual understanding of the word jargon). In
New Zealand this variety developed into Māori Pidgin English, which was
used for early contact between Māoris and Pakehas (Whites). However, the
pidgin never stabilized enough to evolve further, and the New Zealand pidgin
was short-lived.
In Australia pidgins based on English appeared not long after the arrival of
the Europeans. The pidgin varieties became increasingly important for con-
tact, not only between Aboriginal people and English-speaking settlers, but
also as a lingua franca between speakers of different Aboriginal languages.
The disruption that followed the arrival of the Europeans forced the move-
ment of people to mission stations, pastoral properties and towns. Although
there had always been widespread multilingualism among speakers, this
wasn’t adequate to cover communicative needs in these new settlements,
where people of different linguistic backgrounds were thrown together and
where there was continued uneven interaction between Aboriginal people
and English speakers. Pidgins therefore fulfilled a communicative need. Out
of these, creoles evolved in the Kimberley Region, the Roper River area and
parts of North Queensland. These various English based creoles have much
in common, but they also show some regional differences too. These depend
on the Aboriginal languages represented in the community where the pidgin
originated and also influences from other pidgins and creoles brought into
Australia from the outside.
It is possible that many of the languages we know today started life as a con-
tact language – perhaps even English. In Chapter 1 we identified various stages in
the development of English: Old English/Anglo-Saxon (approximately 450–1100),
Middle English (approximately 1100–1500) and Early Modern English (approxi-
mately 1500–1800). As you saw, English experienced some rather radical changes
between the Old and Middle English periods, so radical in fact that the varieties
would not even have been mutually intelligible. A popular theory in the 1980s
attributed these drastic changes to the fact that medieval English was not a mod-
ernization of Old English but rather a creolized pidgin. They supported this idea
with evidence like the following:
1 The content of the lexicon – a considerable portion of the Middle English

lexicon was French.
2 Structure – the shift from synthetic structure (grammatical meaning con-
veyed by inflections) to analytic structure (grammatical meaning conveyed
by separate words and word order).
3 The socio-historical context – a small aristocratic group maintained their
French ways (e.g. speaking French), with the rest continuing to speak English.
While evidence like this is suggestive, the theory of a medieval creole is far-
fetched and not generally accepted. Linguists now do not believe that the language
of, say, the medieval poet Geoffrey Chaucer is a linguistic innovation, but that it
simply reflects the contact situation with French (and also Norse). The types of
changes that took place between the Old and Middle English periods are well docu-
mented for contact situations generally and do not challenge the idea of “unbroken
continuity” from Old through to Early Modern English. In short, the processes that
create creole languages are no different from universal patterns of language change
that we have been observing throughout this book (see Trotter 2012 for a readable
up-to-date assessment of Middle English creolization).
In a number of bilingual settings, the intertwining of languages has seen the cre-
ation of new mixed speech varieties that are not mutually intelligible with either
of the source languages being mixed. Like pidgins and creoles, these mixed lan-
guages present a challenge for historical linguistics because they don’t fit the stan-
dard genetic picture (see Chapter 9). True, all languages are intertwined with other
languages to some degree, but all these contact languages (and this includes pidgins
and creoles) are extreme cases because they don’t represent the daughter languages
of a single parent but are genetically related to two (or even more) languages. Mixed
languages are different from pidgins or creoles, however, because speakers are usu-
ally fluent (or at least start off being fluent) in both of the source languages (so
there’s no imperfect learning going on). They also take and adapt their lexical and
grammatical subsystems from each source language, and they don’t share gram-
matical features as creoles do. They also arise from very different socio-historical
settings (i.e. not the unequal contact existing in the settlement colonies that gives
rise to pidgins).
The code-switching behaviour of bilinguals is an important factor in the devel-
opment of these mixed languages. In bilingual situations where languages are
maintained and in constant close contact, it is common for speakers to move from
one language to the other in mid-speech; this can involve isolated words, phrases
and occasionally even entire clauses. Code-switching has been described as bilin-
guals “keeping a foot in both camps”, and is a normal and natural feature of the
conversations between speakers who know the same two (or more) languages (see
example 8.1 below). The fact that mixed languages arise in this way out of expres-
sive needs rather than communicative needs is another point of difference between
mixed languages and pidgin/creole languages.
Map 8.1
Map showing location of Gurindji and Warlpiri2
In Australia extensive language contact has seen new mixed languages created via
bilingual speakers’ code-switching between an English based creole and an Austra-
lian traditional language; these have then conventionalized into new autonomous
languages that are now acquired by children as a first language. Two new mixed
varieties are Gurindji Kriol and Light Warlpiri. Both have systemically mixed the
lexical and morphosyntactic systems of a traditional Australian language and an
Australian contact variety.
Gurindji Kriol is spoken by Gurindji people in the Victoria River District of
northern Australia (see Map 8.1). It emerged during the 1970s and combines the
lexicon and structure of Gurindji (the traditional Australian Aboriginal language
of the Gurindji people) and Kriol. In their account of the transition from code-
switching to a mixed language, McConvell and Meakins (2005: 18) give the follow-
ing example of the sort of early code-switching that gave rise to the mixed language
Gurindji Kriol; in this example the Gurindji words are in italics.
(8.1)
ail teik-im botum an go bek ngalking-ku kungulu-yawung
I’ll take-TRANS bottom and go back greedy-DAT blood-HAVING
‘I’ll take the bottom and go back. Bloody meat is for greedy people’
While nouns and main verbs can come from either Gurindji or Kriol, there is a
structural split between the noun phrase and verb phrase grammar, with Gurindji
contributing the noun structure (including case marking), and the verb structure
(including auxiliaries) coming from Kriol. Kriol also dominates the Gurindji Kriol
lexicon, contributing around 64% of verbs and 67% of nouns. The following exam-
ple from Gurindji Kriol shows the linguistic mix (the Gurindji-derived forms are
italicized, and Kriol forms are in plain font).
(8.2)
warlaku-ngku i bin katurl karu leg-ta
dog-ERGATIVE 3SG PAST bite child leg-LOCATIVE
‘The dog bit the child on the leg.’
(Meakins and O’Shannessy 2012: 218)
This example shows a VP structure (which includes the past tense marker bin)
derived from Kriol, with an NP structure from Gurindji, and this includes ergative
case marking -ngku (showing the agent) and locative case marking -ta (showing
place). As this sentence illustrates, Gurindji contributes the noun phrase structure,
but the actual nouns are derived from both Gurindji (e.g. karu ‘child’) and Kriol/
Aboriginal English (e.g. leg).
Light Warlpiri is spoken in Lajamanu, in the Northern Territory of Australia
(see Map 8.1). It draws most verbs and verbal morphology from Kriol, nouns from
Warlpiri and English, and nominal morphology from Warlpiri.
(8.3)
jarntu-ng i=m bin bait-im kurdu wirliya-ngka
dog-ERGATIVE 3SG=NFUT PAST bite-TRANS child leg-LOCATIVE
‘The dog bit the child on the leg.’
(Meakins and O’Shannessy 2012: 218)
We need to emphasize the complex nature of multilingualism that characterizes

these two indigenous communities. In addition to the traditional languages Warl-
piri and Gurindji, the linguistic repertoires of these speakers vary along a contin-
uum, and there is considerable variation (reminiscent of the creole continuum). At
one end of the continuum, the way of talking is close to the English many people in
rural Australia use. At the other end are the mixed languages, in which the structure
contains words and features of the different parent languages. In the middle of the
range are varieties of an English based creole. It is still early days – these hybrid
languages may well be stabilizing, or else they could be in the process of a more
dramatic shift towards Kriol.
While there are different types of mixed language, they all show this sort of ances-
tral/linguistic split, often between grammar and lexicon (with one language con-
tributing the grammar and the other the lexicon). Ma’a (spoken in Tanzania) is
historically Cushitic (of the Afroasiatic language family) with 50% Cushitic vocabu-
lary, but a structure that is largely Bantu (of the Niger-Congo language family).
Media Lengua is another example with a lexicon from Spanish and its grammar
derived from Quechua (a native South American language spoken primarily in the
Andes). This admixture of features is a point of difference between these contact
languages and pidgins/creoles – recall that creoles draw their vocabulary from one
language (the socially and economically dominant one) and, though their gram-
matical systems have a lot in common with the substrate languages, the source of
the structural features is not always identifiable.
You will find some differing accounts of what constitutes the class of mixed lan-
guages, and the socio-historical and linguistic characteristics vary and are com-
plicated; for example, not all mixed languages arise through fluent bilingualism
and the code-switching behaviour of skilled bilinguals. So, to see examples of other
intertwined varieties and more detailed accounts of these fascinating contact lan-
guages, we recommend you visit the references provided for further reading.
8.1.4 The linguistic Sprachbund – a special case of language convergence

Where languages occur in close proximity and there is extensive and stable bilin-
gualism, you can find structural features spreading throughout the area in a spec-
tacular fashion; the result is that the languages come to resemble each other. The
process is called convergence. In its most extreme form (and this is widely reported
around the world), this kind of mutual interference can even occur between lan-
guages that are unrelated. So languages that start life being genetically diverse can
end up being typologically similar. In these cases, the influence is two-way – features
are shared by the converging languages and so they become very similar in struc-
ture, though there is usually little effect on the respective lexicons of the languages
(speakers retain their own vocabulary).
Speech communities that share features in this way are referred to as linguis-
tic areas, and the features are called areal features. One of the most famous exam-
ples is the Balkan Convergence area, or Balkan Sprachbund, involving Romanian
(Romance); Bulgarian, Serbo-Croat3 and Macedonian (all three Slavic); Albanian;
and Greek. The following are some of the features that are shared by most, and
sometimes all, members of the group; in general, you can say there is an overall
tendency for grammatical functions to be marked by free-standing (or analytic)
structures (see Tomić 2006 for a full account):
• definite articles that follow rather than precede nouns (e.g. house the)
• analytic future constructions based on ‘want’ or ‘have’ auxiliaries
• prepositions in place of cases
• collapse of the dative and possessive cases (i.e. merging of form and function)
• double marking of animate objects
• analytic comparatives (compare English more lovely)
• loss of the infinitive
• numbers from ‘eleven’ to ‘nineteen’ share a similar structure (lit. ‘one on ten’ etc.)
The combination of these and other shared features (including lexical and pho-
nological features) is the result of multilingual contact over a long period of time.
The features do not come from a single Balkan source; they aren’t unique to the
Balkans. What we have here are four distinct language groups that have converged
into one language area (or Sprachbund).
8.2 CASE STUDY: PENNSYLVANIA GERMAN AND ENGLISH IN CONTACT
As we’ve stressed throughout this book, linguistic and social factors are closely
intertwined when languages change, and this is certainly the case when the changes
are triggered by contact. The assortment of linguistic factors includes the closeness
of the linguistic systems, their structural compatibility, and also naturalness (i.e.
whether a form or construction is intuitively plausible or more frequent across lan-
guages),4 Into this mix there are always crucial social and psychological aspects (the
intensity of the contact, prestige and attitude factors). Without precise information
on the socio-historical setting, we cannot predict when transfer will take place and
what the nature of it will be. Phonological, prosodic or grammatical features can
never be considered independently of the context in which they appear.
To illustrate just how complicated the factors can be in a contact situation, con-
sider different patterns of change that result from the interaction between Pennsyl-
vania German and English in North America. Pennsylvania German is the language
of conservative Anabaptist communities, the Amish and Mennonites (colloquially
the Fuhrleit or “horse and buggy people”), in Ontario (Canada) and Pennsylvania
(United States), and it has been in contact with English for nearly 400 years. Penn-
sylvania German is the language first learned at home, and English is generally
acquired at school (from the ages of five or six). The direction Pennsylvania German
to English, therefore, involves substratum influence – in other words, the transfer of
language habits into a second language (so-called “improper learning”). The other
direction, English to Pennsylvania German, represents the more restricted sense of
borrowing – the transfer of items to a group’s first language.
Speakers in this community are balanced bilinguals with both languages happily
coexisting, each within its own separate domain of use – a situation known as diglos-
sia.5 Pennsylvania German (in this schema the L(ow)-variety) is usually only spoken
and is the language of home and community. English (the H(igh)-variety) is read and
written and is only spoken when dealing with those outside the community. Where
bilingualism has the support of diglossia, the situation is usually stable (i.e. mainte-
nance). Switzerland has shown this kind of variability with Standard High German (H)
and Swiss German (L) for many decades, if not centuries. However, when the languages
are no longer compartmentalized in this way, it is much more likely that one language
will ultimately be used in all contexts and the other language will be lost (i.e. shift). It
seems this linguistic “compartmentalization” is necessary for languages to live together.
8.2.1 English to Pennsylvania German – superstratum influence

The number of English expressions coming into Pennsylvania German is enormous
and growing. Some are clearly needed borrowings for new objects and concepts,
but by no means all arise from need. There are many examples of everyday Eng-
lish words replacing original Pennsylvania German forms (e.g. remind, decide, sure,
busy, really, a lot). Prestige is often the reason for unnecessary loanwords, but is not
an obvious motivating force here, given the Anabaptist world-view – its emphasis
on demut ‘humility’ rules out any positive evaluation of status. However, the impe-
tus may derive from the important position of English as the written language.
Early lexical borrowings from English didn’t have much effect on the overall
phonological structure of Pennsylvania German. At that time speakers were not
bilingual, and the English borrowings simply assimilated to Pennsylvania German
phonological patterns. But with bilingualism and the increase in borrowings over
the years, their impact is now more obvious. Striking are changes involving the
sound written <r>. Until recently, the Pennsylvania German /r/ was pronounced as
a trill before a vowel, and elsewhere as a schwa. It is now usual to hear something
resembling the North American retroflex before consonants and word finally after
long vowels; for example, Darm ‘intestine’ is now frequently pronounced [dæɻm].
There are two aspects to the contact with English to explain this pronunciation.
One involves direct borrowings such as car. The other is an indirect influence from
English via literacy. Pennsylvania German is not a written language, but people
do read aloud (archaic) Luther German (which is the language of their Bible and
hymns). When they do, their pronunciation is strongly influenced by English; writ-
ten <r> is always pronounced, and as a retroflex rather than a trill (North American
English is r-full; i.e. /r/ is pronounced everywhere).
At first blush, grammatical influence also appears strong. For example, the lan-
guage has a new future construction derived from geh ‘to go’. This is an immediate
future, comparable to the English gonna.
(8.4) Ich hab geglaubt es geht ihm happene

I have though it goes to-him happen
‘I thought – it’s gonna happen to him’
However, the role of English is difficult to pin down here, as it is for most of the
putative structural transfers. Contact may have triggered the change, but since this
particular development (a ‘go’ future) is replicated in languages around the world,
English may simply be accelerating a change that was independently motivated.
The passive construction also shows a number of English influences:
(8.5) De Schtrump s/waar geschtoppt bei der Maem

the stocking is/was darned by the mother
‘The stocking is/was darned by mother’
Here you can see (1) the bei ‘by’ phrase is after the verb geschtoppt and not
before (the expected German word order); (2) the preposition bei has now replaced
original vun ‘of ’; (3) the auxiliary sei ‘to be’ has taken over from the auxiliary
warre/waerre ‘to become’. The language has also developed a ‘get’ passive similar to
that in English, as in the stocking got darned (in fact, Pennsylvania German kriege
‘to get, receive’ is moving along strikingly similar paths to get in English; see Bur-
ridge 2007).
There are many similar examples where Pennsylvania German appears to be con-
verging to English in its structure, and the closeness of the linguistic systems and
structural compatibility would be assisting here. But, in truth, most of the changes
probably have multiple origins with different internal and external factors hav-
ing a role to play. For example: (1) We have here two genetically related languages
that because of their common origin are traveling in similar directions; in other
words, the seeds for many of these changes would have been sown long ago in the
parent language. (2) The Pennsylvania German speakers may also independently
have sown the seeds, and since many of the developments illustrate extremely well-
trodden paths of grammatical change (e.g. the ‘go’ future, the ‘get’ passive), they are
likely products of the universal cognitive processes driving language use. (3) In
some cases English contact probably induced certain of these seeds to sprout, but
in other cases it might simply have stimulated the growth of seedlings that had
already emerged.
8.2.2 Pennsylvania German transfers into English – substratum influence

This time consider the other side of the contact coin, namely Pennsylvania Ger-
man influence on the English of these speakers. The least amount of interference
occurs at the lexical level; the few lexical transfers are confined to untranslatable
concepts such as Freindschaft (which inadequately translates into English as ‘fam-

ily’, ‘relations’). There is also little in the way of grammatical transfers. The most
obvious influence from Pennsylvania German occurs at the prosodic level, in this
case a distinctive rise-rise-fall pattern for a range of sentence types, including
questions and statements. There are also significant transfers of consonant and
vowel features such as [dʒ] —> [tʃ] and [ʌ] —> [ɔ] (e.g. just is pronounced as
[tʃɔst]).
To summarize both directions of influence, we can arrange the patterns of influ-
ence into two reverse hierarchies of increasing and decreasing strength respectively,
shown in Table 8.1.6
Table 8.1
Pennsylvania German-English contact – direction of influence
Influence E —> PG Influence PG —> E

Phonological weak strong
Grammatical strong weak
Lexical very strong very weak
Why is it that we find precisely these patterns of interference? Genetic or typo-

logical considerations are not much help here. Although they predict that the simi-
larity of the linguistic systems would facilitate changes, they cannot explain why
interference seems sensitive to the different levels of grammar and why, in particu-
lar, the reverse hierarchies exist when we compare the direction of influence.
Consider the child language experience and the order in which different linguis-
tic levels are learned (see McLaughlin 1984). Intonation and stress features are the
first to be acquired and are therefore the most deeply secured features. These and
segmental features (vowels and consonants) are acquired during a child’s first three
years. The acquisition of grammatical features continues until much later, during the
early school years, and vocabulary of course continues to be acquired throughout
one’s lifetime. It is not surprising, therefore, that the most deeply anchored speech
habits are the ones we find most readily transferred from a first to a second language
in a language shift scenario. Sometimes a distinctive intonation is all that remains
of a previously bilingual speech community. And to consider the other side of the
coin, it is also not surprising that these deeply embedded phonological features are
the ones that show the greatest resistance to influence in a maintained language.
Both types of contact, borrowing and substratum induced influence, can result in
grammatical changes – so why do we find grammatical transfers from English into
Pennsylvania German and not the other way around? During the early school years,
children have not yet stabilized their Pennsylvania German grammatical structures,
and they are vulnerable to English influences at this time. Socio-historical studies of
change have shown that linguistic innovations have a much greater chance of taking
hold in a language if they attract no attention and no resistance from speakers (see
the discussion in Nadkarni 1975). Pennsylvania German is not standardized and
has no written form; people are used to variation and are tolerant of it. Conversely,
there are many factors working against the transference of Pennsylvania German
features into their English. Significant here is the way English is acquired in the
formal school setting. There is considerable prescriptivism with a strong emphasis
on “proper English”, and outside of school the children’s experience of English is
within formal contexts only; i.e. in contact with outsiders and as a written language
(recall the diglossia of this speech community). There would be little in the way of
“imperfect learning”.
The special case of discourse particles

When it comes to those elements that have some kind of discourse regulating
function, Pennsylvania German draws heavily on English. There is a striking
number of English discourse markers that have been fully integrated into the
language. These include actual transfers such as well and loan translations
(or calques) such as ennigweg < anyway. These markers have been borrowed
along with the English functions: Pennsylvania German well appears before
surprising information, answers to questions and self-repairs; ennigweg sig-
nals a shift of topic (often to an earlier topic). Similarly, native markers have
adopted functions akin to the equivalent English discourse marker – weeschte
‘you know’ emphasizes common ground between the players.
(8.6) Er war nett weeschte arrig reich awwer er war well-to-do

he was not you-know very rich but he was well-to-do
‘He wasn’t you know rich but he was well-to-do’
In contrast, there are very few Pennsylvania German discourse/pragmatic

markers in the English of these speakers. So why draw on English in this
way? As in the case of the lexical borrowings, need is not the answer, and
neither is social prestige. Matras (1998) has argued that these kinds of bor-
rowings are highly automaticized and gesture-like. Rather than being social
triggers, they are cognitively motivated. He claims that bilingual speakers
will cut down to just one set of these items, and this set will always favour
those that belong to “the pragmatically dominant language”. The notion of a
pragmatically dominant contact language is an interesting one here. This is a
speech community that tries to maintain its isolation and limit its interaction
with the outside world, but the reality is that it has to do business with the
English speaking majority in order to survive. So this gives English consid-
erable clout, especially as the language of literacy – no small wonder some
people perceive English as a “killer language”. But we need to keep in mind
that, in contrast to a real killer, English never had any intention to “kill” any
other language.
8.3 LANGUAGE DEATH
As earlier described, one unhappy outcome of language contact is language loss,

where multilingual speakers of small, indigenous or minority languages abandon
them in favour of the socially dominant language of the wider community and/
or the nation-state. This situation is now usually referred to as language death. Of
course, some people describe languages like Latin and Old English as dead lan-
guages, but these languages didn’t die – they simply altered their structures over
time as a result of natural evolution. Alongside the Classical Latin of the late Roman
Empire (this was a kind of Standard Latin), there were vernacular (or colloquial)
varieties of Latin that were spoken in various parts of Europe. These colloquial Lat-
ins then evolved into modern-day languages such as French, Italian, Catalan and
Spanish. So Latin metamorphosed into these Romance languages (and you can find
a family tree for these languages in Chapter 9). In this section we are talking of the
complete disappearance of a language, something that can cause a good deal of
pain for people who find themselves unable to speak their ancestral languages –
and additionally painful because it is strongly associated with other experiences of
cultural trauma and community breakdown.
Around the world, languages are dying and at distressing rates – American Indian,
Celtic, Austronesian and Australian Aboriginal languages, to name just a few. Even
in those places where local languages appear healthy, the high status of bulldozer
languages such as English, Mandarin Chinese, Hindi and Indonesian suggests their
survival is precarious. Creole languages, such as Tok Pisin in Papua New Guinea and
Kriol in Australia, are also threatening many of the indigenous languages in those
regions; paradoxically, these creoles themselves are threatened by their superstrate
(or lexifier) languages (those that originally contributed most of the creole’s vocabu-
lary). Social pressures brought on by increased formal education and emphasis on
schooling can cause creoles to gradually move more and more in the direction of the
socially dominant language. For example, English based creoles like Tok Pisin and
Kriol might become more and more like Standard English. It is a kind of “language
suicide”, although its more official title is decreolization – through convergence and
borrowing the creole moves closer to the status variety of the language.
UNESCO (the United Nations Educational, Scientific and Cultural Organization)
has identified language survival as a high priority. According to some estimates,
the linguistic monopolies created by the expansion of English and other national
languages will see more than half of the world’s roughly 7,000 languages disappear
this century – some predictions are as grim as 90% (and others even grimmer):
“English is even being promoted by many governments in countries where English
is nobody’s first language, and nearly all nations promote their national language
at the expense of minority and migrant languages” (Bradley and Bradley 2002: xii).
To illustrate how dire the situation is for some language groups, consider the devas-
tating effect of English on Australian Aboriginal and Torres Strait Islander languages.
Before English speakers settled in Australia, the country was remarkably diverse lin-
guistically. Table 8.2 shows the approximate numbers of Aboriginal dialects, languages
and language families at the time of earliest European contact.
Table 8.2
Number of Aboriginal languages originally spoken in Australia
Language families 26–29

Languages 200–250
Dialects 500–700
Table 8.3
Comparison of the 2005 and 2014 National Indigenous Language Surveys
Language status 2005 2014 Change

Languages still spoken 145 120 25 languages
Severely or critically endangered 110 100 10 languages
Spoken by all age groups 18 13 5 languages
For Australian indigenous communities, language contact, with the accompany-

ing imposition of colonization, has brought about wholesale extinction of many
languages. As Table 8.3 shows, only around 120 of the original 200–250 languages
remain today, according to the National Indigenous Languages Survey published
in 2014. Even the remaining robust languages are under threat, despite vigorous
efforts being made to maintain them. Only 13 have more than 500 speakers, and five
languages have fallen out of this category in the last ten years. It has been estimated
that the number of surviving languages might decline by as much as 50% in the next
20–30 years, as the most critically endangered languages lose their last speakers.
Over and over again, we see this pattern of language spread leading to language
shift and then language loss. And consider what we lose when these languages dis-
appear. Crucial parts of peoples’ social and cultural identity vanish. For linguists,
important linguistic systems disappear forever – and who knows what else vanishes
with those dying words and constructions. Wisdom unique to these linguistic com-
munities becomes extinct with their languages – Evans (2010: xviii) describes it as
the “looming collapse of human ways of knowing”. This might involve knowledge
connected, say, with the environment and its flora and fauna – knowledge that could
well have scientific value for the development of new materials, drugs and foods.
There are important implications here for historical linguistics, since during
language loss significant changes take place. As speakers forget forms and con-
structions, they might lose them altogether or change them to something closer to
those in the language they are in the process of shifting towards. Similar changes
happen to the sound system, as speakers drop phonological contrasts and rules.
Consider again the Pennsylvania German speech community. We’ve only looked
at the religiously conservative groups (the so-called horse and buggy people),
where the language is thriving. For these speakers, language has a deeply religious
significance, and this works to safeguard its survival. Amongst the more modern
groups, however, language proficiency ranges from fully competent speakers to real
“semi-speakers” (speakers who have forgotten or have never learned much of the
language). But even the most proficient speakers aren’t teaching their own chil-
dren Pennsylvania German, and language death seems inevitable here. The dying
language of these groups shows particularly rapid and numerous changes. They
involve major reduction or “stripping away” of morphological complexity such as
case levelling, collapse of gender and number distinctions, and reduction of tenses –
in general, a movement towards greater isolating structure. Compared to the lan-
guage of fully competent speakers, the language of semi-speakers appears in a
greatly reduced form. Of course, as we’ve seen, these are the same sorts of processes
that affect robust languages, especially in a contact situation, but language death
and these sorts of grammatical developments are intimately connected, and the
changes are more rapid and more drastic.
8.4 LANGUAGE PLANNING AND POLICY
Language planning and policy is a large area within sociolinguistics that tackles
problems within speech communities. It includes the work linguists do to assist
communities to stop or reverse the decline of a language or even to revive one
already extinct with no native speakers. All over the world communities are cry-
ing out for support to help document and maintain their language in the face of
probable extinction, and many linguists are now involved in some way in fieldwork,
language documentation and language revival. Unlike the linguistic changes we’ve
been considering elsewhere in this book, we are now looking at changes that arise
out of deliberate intervention (by educators, policy makers and linguists) to alter
either the languages themselves or the ways in which their speakers use them. This
is one of the few areas where linguists cheerfully admit to being prescriptive.
There are different aspects of language planning – and all aspects can of course
be highly political:
1 Status planning deals with selection (of the language variety to be promoted
as the standard), as well as attitudes towards alternative varieties and the
political implications of these choices.
2 Corpus planning deals with the codification of the linguistic features of the
standard and requires the production of grammars, dictionaries, textbooks
of instruction and other prescriptive literature; for example, it involves the
development of orthographies (i.e. introducing or changing a writing sys-
tem), and this includes the alphabet, spelling, punctuation, accentuation
and capitalization.
Another aspect of corpus planning concerns modernization – the expan-
sion of the lexicon to incorporate terms for modern artefacts, activities and
concepts that are otherwise referred to by code-switches or borrowings from
outside sources; sometimes this practice is motivated by linguistic purism,
a desire to prevent the language from being swamped by lexical aliens (and
when communities seek to revive or revitalize traditional languages, mod-
ernization can be a hot topic).
3 Acquisition is part of the language planning process in which institutions

(usually a national, state or local government body) seek to influence which
language(s) will be used as the medium(s) for instruction (this can be from
primary schools through to universities); obviously, this process can alter
the status and level of prestige of a language, and might even reverse a situ-
ation of potential language shift.
Clearly, the decisions made around policies in these areas can go on to affect lan-
guage status. They can also have a major impact on language vitality, even determin-
ing which languages are nurtured and thrive, and which are neglected and decline.
8.5 THE DANGERS OF PURISM
Speakers don’t generally like their languages to change (see Chapters 1 and 10), and
in situations where languages are battling for their lives, speakers can be particularly
sensitive to change. In this situation, one very real danger is a tendency towards
purism; in particular, an unrealistic insistence that the current-day language reflect
the norms of the past, and that the endangered language remain uncontaminated
by outside elements.
Again this is where Pennsylvania German is instructive. As seen earlier, the language
of the religiously conservative groups shows considerable evidence of convergence to
English in its structure, and of course the incorporation of English loanwords is com-
monplace. But the absence of purist attitudes in this community is striking, and this
bodes well for the future of this language. Speakers constantly remark on the variation
and change they see in their language, often commenting wie Englisch as mir sin ‘how
English we are’. Despite an isolationist philosophy, a desire to be abgesandert vun die
Welt ‘apart from the world’, this intrusion of English into Pennsylvania German is never
criticized or judged harshly. In this case structural compromise is a sign of health, and
the absence of resistance will enhance the chances of the language’s survival.
This is not to belittle puristic attitudes – the desire to keep a language pure and
free of elements from dominant languages such as English is understandable, espe-
cially in a situation of potential language shift to the dominant language. Moreover,
linguistic purism does seem to be something that is deeply ingrained in the human
psyche. But puristic attitudes can be a real barrier to natural, healthy change, change
that all languages need if they are to remain viable and versatile tools for a society.
And in a language endangerment context, they can have the disastrous effect of
discouraging younger speakers, who feel they don’t speak an authentic form of the
language, for example the “proper” Māori or the “proper” Irish that older speakers
in the community are insisting on (again, see Chapter 10 on the myth of purism
and related effects). The result can be that younger people give up altogether (see
Dorian 1994). Linguistic straightjacketing hardly ever works, but it can have serious
repercussions for a language under threat.
Many communities have shown the improved wellbeing that results from efforts
to reconnect individuals and families to their culture and to relearn or at least
learn about their languages. But successful language revitalization does not mean
retaining the original structural complexity of the heritage language. This lesson is
an important one for language maintenance and revival programs – the language
revived in the present can never be the same as it was in the past.
SUMMARY
Contact induced change refers to the process whereby a language takes and incor-
porates some linguistic element from another language. This can involve any part
of the grammar, but the most readily transferred items are lexical (so content rather
than grammatical words). However, the extent of the influence will always be
affected by the nature of the contact and what can be rather unpredictable social
attitudes.
There are two different contact scenarios: (1) When we have a situation where
the original inhabitants adopt the language of the newcomers, there is usually a
period of bilingualism when they speak the new language but with some inter-
ference from their primary language. If some of these elements from the primary
language are transmitted to later generations of speakers of the prevailing language,
this is described as a substratum of that language. Typically, the substratum affects
the phonology of the adopted language, but generally little in the way of lexical
influence takes place. (2) When the newcomers are linguistically absorbed into the
indigenous population, the influence of their respective languages is typically most
obvious in the lexicon, although it can affect other parts of the grammar as well.
This is described as superstratum influence. Both types of contact, superstratum
and substratum, can result in the transfer of grammatical features.
Convergence occurs in cases of widespread and stable bilingualism and can
involve mutual interference, where features get shared by the converging languages.
The result is typological homogeneity (i.e. the languages become very similar in
structure). In some cases structural changes can be so extensive that languages can
no longer be regarded as genetically related to the rest of their former language
family; these languages include contact varieties like pidgins, creoles and mixed
languages.
FURTHER READING
A number of books focus on the linguistic outcomes of contact, including Myers-

Scotton (2002), Clyne (2003) and Winford (2003). For general information on con-
tact languages and contact induced changes, see Thomason and Kaufman (1991),
Thomason (1997) and Siegel (2008); contributions to the 2003 Annual Review of
Applied Linguistics (Vol. 23) focus on various themes to do with language contact
and change.
You’ll find a vast literature on specifically pidgins and creoles; recent works
include Baptista (2005), Singler and Kouwenberg (2008) and Velupillai (2015). For
specifically mixed languages, see Matras (2000) and Matras and Bakker (2003).
For more on cross-creole structures, we recommend the Atlas of Pidgin and Creole
Language Structures Online (Michaelis et al. 2013; available online at http://apics-
online.info); on the Language Bioprogram Hypothesis, see Bickerton (1981, with
a new 2016 edition available for download from Language Science Press @ http://
langsci-press.org/catalog/book/91, 1986).
The discussion on Gurindji Kriol comes from Meakins (2013a, 2013b), and that
on Light Warlpiri from O’Shannessy (2005, 2015). Domingue (1977) put the origi-
nal case for Middle English as a creole. On aspects of code-switching see Myers-
Scotton and Jake (2009), and on the Pennsylvania German-English contact see
Burridge (2007). Much has been written on language endangerment and the fragile
state of minority languages; see, for example, Nettle and Romaine (2000), Zucker-
mann and Walsh (2011) and Evans (2010).
EXERCISES
1 Word order change

Revisit what was written about English and German constituent order in Chapter
6 (in particular what we described as the “verbal brace”). Describe the order of
the major constituents in the following examples from Pennsylvania German, and
compare it to that of English. How would you characterize the influence of English
here?
a Mir hen alsemols heemgemacht eiscream g’hat do

we have before homemade icecream had here
‘We used to have homemade icecream here.’
b Er hot besser g’fielt wann er sich gut hewe hot kenne aeryets
he has better felt if he self good hold has can somewhere
‘He used to feel better, if he could get a good grip somewhere.’
c No verblatz ich weil ich bin ganz arrig vol

then burst I because I am totally very full
‘Then I’ll burst, because I’m completely stuffed.’
2 What borrowings can tell us

Borrowings may give us information about the technological or scientific status of
the speakers of languages in contact with others. In this chapter we described the
borrowings into English from American Indian languages and Australian Aborigi-
nal languages. Contrast these borrowings with those from Arabic: alchemy, alcohol,
algebra, alkali, cipher, nadir, zenith, zero.
What do you notice about the connotations of these different sets of borrowings,
and what do they tell us about the relationships between the languages and the cul-
tures of their speakers? (If you can, add to these lists of borrowings into English.)
3 Australian Kriol
The following two questions relate to Kriol, a variety of creole spoken by more than
20,000 speakers in the northern parts of Australia.
Part 1: The features of Kriol

State how the vocabulary and grammatical features of each sentence below differ
from those of Standard English. Identify which of these features might be found in
more colloquial varieties of English. (Note that close translations have been sup-
plied, with grammatical morphemes in capital letters.)
a Wi bin mekim big ampi

we PAST make+TRANS(ITIVE) big humpy
‘We made a big humpy [= hut, shelter]’
b I bin faindim im
I PAST find+TRANS him/her/it
‘I found him/her/it’
c Im kukimbat dat kanggaru
he/she cook+TRANS+CONTINUOUS that kangaroo
‘S/he is cooking that kangaroo’
d Mi na laikim go la kul
I NEGATIVE like+TRANS go LOCATION school
‘I don’t like going to school’
e Dat san blo mai sista im lib la Sydney
that son POSSESSION my sister he live LOCATION Sydney
‘My sister’s son lives in Sydney’
f Im blakpala
he black+ADJECTIVE MARKER
‘He’s black’
g Dey bin go indid
they PAST go TAG QUESTION
‘They went, didn’t they?’

h An’ mela new teacher gotta come
and our new teacher FUTURE come
‘And our new teacher is going to come’ [mela = several of us excluding you]
i Dad did get two turtle
Dad DO get two turtles
‘Dad got two turtles’
j E bin wipeim gotta tissue
he PAST wipe+TRANS with tissue
‘He wiped it with a tissue’
Part 2: Language endangerment

In Australia many people, both Aboriginal and non-Aboriginal, consider Kriol
varieties to be a threat both to Australian indigenous languages and to Standard
English. Why do you think they believe this – and are they correct in their view?
4 Pidginization/creolization and processes of language change

Describe the kinds of changes that take place in the creation of pidgins and creoles.
How do these compare with (a) normal internal historical processes and (b) exter-
nal change (i.e. the “forced” changes associated with language contact)?
5 Research essay
Take a language that has a considerable number of loanwords from other languages
(e.g. Japanese from Chinese and English; Indonesian loans from Arabic, Sanskrit
and Dutch; loans from French, Dutch, German, Latin and Greek into English etc.).
Begin with a list of 50–100 loanwords. Study the way in which foreign words are
adapted to the native linguistic system (phonologically, grammatically and seman-
tically). Also indicate what information the loans give us about the nature of con-
tact between the two languages.
NOTES
1 We should emphasize that there are other hypotheses regarding the emergence of pid-
gins and creoles (for example, some maintain that creoles emerge as second language
varieties of the superstrate, or lexifier, languages, and not out of pidgins); see the discus-
sion in Winford (2003), Siegel (2008) and Velupillai (2015).
2 Many thanks to Felicity Meakins for allowing us to use this map; the cartographer is
Brenda Thornley.
3 Naming these languages is controversial; we have adopted Corbett and Browne (2009)’s
use of “Serbo-Croat” as a cover term for Bosnian, Croatian, Montenegrin and Serbian.
4 Naturalness overlaps with “markedness”. Features are said to be natural if they are
unmarked across languages (e.g. have a wider distribution and are acquired earlier than
other features); we expect change to shift languages from a less natural to a more natural
state.
5 The creator of the term “diglossia”, Charles Ferguson, originally applied it to the pat-
terned use of formal and colloquial forms of the same language (for example by speak-
ers of Swiss-German or Arabic), but in a later work he extended the term to describe a
similar distribution between two different languages (see Ferguson 1959).
6 The overall patterns of influence here are similar to those recorded by Rayfield (1970) for
English-Yiddish contact.
9
Relatedness between languages
INTRODUCTION
The previous chapters have focused on linguistic change. We now turn our atten-
tion to the question of genetic affinity and how we might uncover the linguistic
relationships that arise as a consequence of change. To set the scene, go back to
a suggestion we made in the first chapter. Remember our example from the Ten
Commandments (Exodus 20:7) in French, Italian, Spanish, and Portuguese as well
as Latin (the “mother” of these four modern “daughter” languages). English again
only serves as a gloss. Here’s a snippet:
Latin: Non adsumes nomen Domini Dei tui in vanum

French: Tu ne prendras point le nom de l’Eternel, ton Dieu, en vain
Italian: Non usare il nome dell’Eterno, ch’è l’Iddio tuo, in vano
Spanish: No tomarás el Nombre del SEÑOR tu Dios en vano
Portuguese: Não tomarás o nome do Senhor teu Deus em vão
English: Thou shalt not take the name of the Lord thy God in vain
Clearly there are a number of lexical, even grammatical, resemblances here, and
Table 9.1 gives a sample.
Table 9.1
Lexical correspondences between different Romance languages
Latin French Italian Spanish Portuguese English (gloss)

non ne non no não not
nomen nom nome nombre nome name
Dei Dieu Iddio Dios Deus God
in vanum en vain in vano en vano em vão in vain
The similarities here are remarkable. How can we account for them? First, these
languages may have influenced each other. So one language may have borrowed
material from the other (as we discussed in Chapter 8). This is certainly true for
French and English. After the Norman Conquest in 1066, English borrowed many
Romance words such as vain. But a word like name is a problem. Yes, there is the
Latin word nomen, but there is also Old Frisian nama, Old High German namo, Old
Saxon namo and Old Irish ainm. All these appear to be related somehow. And yet it
seems implausible that they are all borrowed from other languages. The alternative
explanation is that these languages are somehow related, that they go back to the
same root, and that this is how their commonalities arise.
9.1 LANGUAGES, DIALECTS AND STANDARDS
Before we explore the idea of relationships between languages we need to discuss

what languages actually are. Most people seem to take the notion of “language(s)”
for granted. Yes, there are French, Japanese, Inuktitut, Walpiri and Navaho. But
what if we zoom in now? What about English English and American English? Are
these two languages, or two dialects of only one language? Bavarian and Dutch are
equally incomprehensible to a speaker of High German – are both independent
languages? Or dialects? Why is Dutch considered a language of its own and Bavar-
ian a dialect of German? The decision is, eventually, arbitrary. Back in the 1940s, the
wonderful Yiddish scholar Max Weinreich claimed, “A language is dialect with an
army and a navy” (at the Yiddish Scientific Institute Conference, 5 January 1945).
What he was trying to say is that there are no linguistic grounds for calling one
variety a language, and another one a dialect. Linguistically speaking, Dutch could
be a dialect of German, and Bavarian an independent language.
Matters get even worse when we consider what are called dialect continua, i.e.
fuzzy boundaries between dialects. Here speakers in communities located close-by
can understand one another (perhaps they live in neighbouring villages), but speak-
ers located at the geographical extremes of these continua cannot. So near the Ger-
man and Dutch border, German speakers using their own “home-grown” varieties
can communicate better with their Dutch neighbours than with German speakers
in the South of Germany, who speak southern varieties of German. So where does
German end and Dutch begin? There’s so much variance and indeterminacy here
that any attempt at dividing up such a continuum on the basis of purely linguistic
criteria (like mutual intelligibility) is problematic. Most of the time, languages shift
very slowly and gradually. The boundaries and language borders that we put up are
mostly political, cultural and arbitrary.
Our modern notion of “language” is closely tied to linguistic standards. We
believe that there is some Standard English, or German, or Japanese, which in itself
defines the language as such. From a linguistic point of view, this is not exactly
true. All standard languages are abstractions across idiolects, sociolects and dia-
lects, and most of the boundaries between languages are politically and arbitrarily,
rather than linguistically, defined. If we take the Dutch-German border and its
dialects in detail, there is no natural cut-off point where we can say that German
stops and Dutch begins. The political border between the two countries is the lin-
guistic border as well. We find the same gradual dialect diffusion between Hessen
(the German state to the north of Bavaria) and Bavaria itself. For political reasons
alone we recognize Bavarian as a dialect of German, and not as an independent
language. So, the definition of a language is closely linked to political and cultural
questions. But it is also a question of linguistic history, as we will see in the next
sections.
Relatedness between languages 217
9.2 ESTABLISHING GENETIC RELATIONSHIPS
Let’s assume that languages (and dialects) are somehow related, like family – we
would expect parents, brothers and sisters, cousins and other distant relatives. How
do we know who is related and who is not? And how do we know whether we are
dealing with a close brother or some long lost cousin? One of the easiest ways is to
simply compare the vocabulary of a set of languages. Take a look at Table 9.2.
Table 9.2
Comparison of four words in seven languages
English German Dutch Danish Swedish Italian Arabic

‘house’ [haʊs] [haʊs] [hœʏs] [hu:ʔs] [hʉ:s] [kasa] [bajt]
‘mouse’ [maʊs] [maʊs] [mœʏs] [mu:ʔs] [mʉ:s] [tɔpo] [fa:r]
‘louse’ [laʊs] [laʊs] [lœʏs] [lu:ʔs] [lʉ:s] [pidɔkkjo] [qamla]
‘out’ [aʊt] [aʊs] [œʏt] [u:ʔð] [ʉ:t] [fwori] [xa:ridʒan]
Five out of these seven languages show remarkable similarities, while two stick out.
The words for ‘house’, ‘mouse’, ‘louse’ and ‘out’ are almost identical in English, Ger-
man, Dutch, Danish and Swedish, but they appear to be very different in Italian and
Arabic. This leads us to suspect that these five languages may somehow be related to
each other, but not to the other two languages. On the basis of the vowel sounds, we
might even go further and hypothesize that English and German are more closely
related to each other, as are Danish and Swedish; Dutch could be somewhere in the
middle. So one might say that English and German, and Swedish and Danish, appear
to be sister languages, and that the two pairs are something like cousins.
Unfortunately, it is not that easy. It is not enough to simply compare words in
languages. If we just did that we would arrive at some very strange conclusions. For
example, the word for ‘eye’ in Modern Greek is μάτι [ˈmati], and in Malay it is mata
[mata]. Does this mean that Greek and Malay are related? The problem here is that
Greeks have never been in contact with Malay speakers (who live more than 5,000
miles, or 8,000 kilometres, away). So we can exclude borrowings here, but we can
also exclude historical relatedness for the same reason. There is simply no evidence
in history that Greeks and Malays go back to the same roots. But we do have this
kind of evidence for German, Dutch, English, Danish and Swedish.
When we compare languages in this way, an important concept is the symbolic
nature of words. Excluding sound symbolic expressions (like clang, jangle and
woosh), there is no natural or necessary connection between the shape of a word
and its meaning, and this means we can rule out the possibility that patterned regu-
larities are fortuitous. What are the chances that speakers of different languages
will independently arrive at the same or similar sound patterns to represent the
same object or concept? Certainly, coincidental similarities do arise (as between
Malay and Greek), but literally hundreds of sets like those in Table 9.2 permeate the
vocabularies of German, Dutch, English, Danish and Swedish, all showing the same
systematic similarities. This cannot be accidental.
So the tricky aspect of comparing languages in this way is identifying the actual
“cognates” (related words with a common origin); in other words, weeding out the
loans and the chance similarities. Generally, it is better to look for resemblances in
basic, common vocabulary, such as that for body parts, family relations or nature.
These tend not to be borrowed (though there are occasional exceptions, as we saw
in Chapter 8). The premise here is that basic (culturally neutral) words are more
resistant to borrowing, and it is these that are trusted when it comes to establishing
genetic relationships. When loans occur, they do need to be identified. Onomatopo-
etic words aren’t helpful either, even if they are rarely borrowed; because they imi-
tate natural sounds in some way, they may be similar (like English cockadoodledoo,
German kikeriki and Japanese kokekokko), and not because of a common origin.
Finally, it is always good to start with words that are obviously similar, but it is
also important to go beyond that stage and consider pairs of words that show little
or no similarity. Perhaps you can find a sound change rule that links two words
in a systematic fashion? Consider Latin quinque, Greek pénte, French cinq, Italian
cinque, German fünf, English five and Icelandic fimm. Greek and Icelandic hardly
look similar: pénte versus fimm! And yet they are cognates and go back to the same
root: Proto-Indo-European (PIE) *penkwe. How is that possible? Recall Grimm’s
Law outlined in Chapter 7. Table 9.3 summarizes the changes.
Table 9.3
Grimm’s Law
Proto-Indo-European Germanic
(voiceless stops) p, t, k became (voiceless fricatives) f, θ,
x (h)
(voiced stops) b, d, g became (voiceless stops) p, t, k
(breathy voiced stops) bh, dh, gh became (voiced stops) b, d, g
So the PIE */p/ regularly turned into /f/, hence fünf, five and fimm (note that
we are focusing here on the initial consonant only). When we looked at sounds in
Chapter 4, we mentioned that a cornerstone of historical linguistics is that changes
are not erratic but regular. This means that words like PIE *penkwe represent rela-
tionships between form and meaning that are carried into the daughter languages
(if items are retained), even where drastic sound changes have occurred.
However, we also mentioned at the time that there are non-phonetic forces work-
ing against the operation of regular sound change. Latin quinque is the fall-out of
one such force – quinque should be pinque, but via a kind of long-distance assimila-
tion, earlier pinque changed to quinque because it was affected by the numeral quat-
tuor ‘four’. Numerals are particularly susceptible to this kind of “contamination”
(English four should begin with [wh], but it changed in anticipation of the following
number five – another nice example of analogy at work!).
When we look at the other early Germanic languages (e.g. Gothic fimf, Old High
German fīmf, Old Saxon fīf ), we can suspect that, in Proto-Germanic, the mother
of the Germanic languages, the word for ‘five’ must have been something like *fimf.
Recall our discussion of compensatory lengthening in Chapter 4. When the [m]

was lost in Old Saxon and Old English (but not in German), the preceding vowel
was lengthened, hence fi:f. Long [i:] underwent the Great Vowel Shift in the later
Middle Ages and was turned into a diphthong [ai]. The result is Modern English
five [faiv].
So, to cut a long story short: words can be cognates even if they don’t look very
similar! We need to consider regular sound changes that can relate them to each
other. And we also need to consider the historical background for the evidence that
the supposedly related languages actually did have a chance of coming from one
and the same root. Again, it’s like looking at families. Because somebody has red
hair, like all the members of the Smith family, this does not automatically mean that
this person is related to the Smiths. We need to find historical (genetic) evidence
that this person can in fact be related to the Smiths in some way.
Problems for the regular operation of sound change

Throughout history people have attributed supernatural powers to names,
and naming taboos can occasionally cause problems for historical linguists
because personal names might be common words or derive from common
words. Simons (1982), for example, describes how on Santa Cruz (part of
the Solomon Islands), there is a taboo against using the name of certain rela-
tives (by marriage). Names consist of a common word (normally with a gen-
der marking prefix). So if a man’s mother-in-law is called ikio (i- ‘prefix to
female’s name’, kio ‘bird’), he cannot use the common word kio to refer to
birds. The effect of this is that something like 46% of the everyday vocabulary
is potentially taboo for some people on Santa Cruz; on the island of Malaita,
this figure is as high as 59%.
Euphemisms are thus created via the methods we described in Chap-
ter 3: semantic shifts of existing words, circumlocution and also phonologi-
cal modification. Sounds turn up in odd places; they mutate unexpectedly.
Words are often funny looking. (Here you might compare the situation where
the urge to swear in polite company drives an English speaker to sponta-
neously change fuck to fudge.) There is also a high rate of borrowing, even
among core vocabulary items that are not generally borrowed. Under normal
circumstances, why would speakers need to take kinship terms or pronouns
from another language? Yet, in this context even basic vocabulary of this
kind cannot be relied on to remain stable. Extensive borrowing and taboo
induced remodelling make it difficult to determine the chronology of lin-
guistic changes that have occurred. Irregular sound shifts have the effect of
accelerating vocabulary differentiation between genetically related languages
and can create a false impression of long divergences, in some cases even hid-
ing genetic connections.
9.3 THE FAMILY TREE MODEL
One of the ways in which we can represent relationships between languages (or
family members) is by developing a family tree. Family trees can show us parents,
children, grandchildren, cousins, uncles, nephews and all other possible relations.
But they also seem to assume that, like any other tree, there is a common root to
all of the family members. One such family tree is given in Figure 9.1. This is the
family tree of the McFly family from the Hollywood blockbuster Back to the Future.
In Figure 9.1 we can see that Marty McFly (the hero of the movie) is related to
George, his father, and that he also has kids, Marlene and Marty Jr. His grandfather
is Arthur McFly, and he has a large number of uncles and aunts (Milton Baines,
Sally Baines, Toby Baines, Joey “Jailbird” Baines . . .). With such a family tree we can
easily see who is more closely related to whom. So, for example, there is the McFly
lineage and the Baines lineage. David and Linda McFly are obviously more closely
related (as brother and sister) than David McFly and Sally Baines (as nephew and
aunt).
Figure 9.1
Family tree of the McFly family (Back to the Future)
Harold Jennivere
McFly McFly
Seamus Maggie Martin

McFly McFly McFly
William
Shawn ?
McFly
Sam Stella
Sylvia Arthur Baines Baines
McFly McFly
George
Lorraine Milton Sally Toby
Douglas
Baines Baines Baines Baines
McFly
David Linda Martin Seamus Jennifer

McFly McFly ‘Marty’ McFly Parker
Marlene Marty
McFly McFly Jr.
Figure 9.2
Family tree for the Germanic languages (simplified)
*Germanic
*North Germanic *East Germanic *West Germanic
East Norse West Norse Gothic High German Low German
Swedish, Norwegian, Modern German Yiddish

Danish Icelandic,
Faroese
Modern Low German Modern English Flemish, Dutch Frisian
Let us apply this idea to languages. In Figure 9.2 you can see a more or less stan-
dard family tree for the Germanic languages.
On the right hand side of Figure 9.2, we can see the West Germanic branch of the
family that sprouted the modern languages High German, Yiddish, Dutch, Flemish,
Frisian, Low German and English. This seems to reflect our initial thoughts on the
relatedness of German, Dutch and English on the one hand, and that of Danish and
Swedish on the other. This West Germanic group is separated from the North Ger-
manic languages and dialects, such as Icelandic, Norwegian, Faroese, Swedish and
Danish. The West and North Germanic languages can be distinguished from the East
Germanic languages, such as Gothic. Note that this group has died out, so that we can
only talk about historical documents here and reconstructed forms (as we mentioned
in Chapter 6). We should emphasize that we have considerably simplified the tree
here. Like actual families, the reality is more complex, and you will find there are many
different and considerably more complicated versions of the Germanic family around.
We can now reiterate this process and look at other large trees, such as that for the
Romance languages (Italian, Spanish, Portuguese, French, Latin . . .). Their (once
again simplified) tree is given in Figure 9.3.
Just like with the Germanic languages, we can see that the Romance languages
can be divided into eastern and western Romance languages. The eastern part com-
prises Romanian and Dalmatian, while the western part comprises Italian, on the
one hand, and French, Catalan, Spanish and Portuguese, on the other. These all
seem to go back to Continental Romance, which can be distinguished from the
dialects spoken in Sardinia, though both go back to Latin.
If we push further back now, we see that both Germanic and Romance languages
actually are related somehow. This is what we discussed in Chapter 7 when we
talked about Grimm’s Law and the First Germanic Consonant Shift that separated
Figure 9.3
Family tree for the Romance languages (simplified)
Latin
Classical Latin Vulgar Latin
Continental Romance Sardinian dialects
Eastern Romance Italo-Western Romance
Romanian Dalmatian Proto-Italian Western Romance
Italian dialects Gallo Romance Ibero Romance
French Occitano Romance Spanish Portuguese
Occitan Catalan
the Germanic languages from the Romance languages. Simplifying a lot, we might
say that the Romance languages and the Germanic languages form one group,
together with many other families, such as Hellenic (Greek), Anatolian, Balto-
Slavic and Indo-Iranian, to name only a few. In other words, in such a model we
can show that Modern German is historically related to Sanskrit, Hindi, Polish,
Russian, Portuguese and Icelandic!
The comparative method (buttressed by grammatical and other evidence) has
been used to classify the world’s languages into groups and subgroups. We now have
numerous established language families in addition to Indo-European. These include
(among many others) Austronesian (languages throughout the South Pacific and
Southeast Asia and into Madagascar); Sino-Tibetan (a large family of languages spo-
ken in East and Southeast Asia, and including the Chinese, Burmese and Tibetan
languages); the Finno-Ugric language family, assumed to include Hungarian as well
as Finnish and Estonian); Uto-Aztecan (languages of the western United States and
Mexico); and Pama-Nyungan (the major genetic grouping of Australian languages).
The number of language families is a matter of hot debate, in particular how many of
these families can be grouped into even larger ones, as we go on to discuss.
9.4 THE WAVE MODEL
As we mentioned briefly in Chapter 7, a number of scholars haven’t always been

convinced by the family tree model as a way of describing the development of lan-
guages – not only does it fail to accommodate the fuzziness of language and dialect
boundaries, but it also cannot capture the fact that languages (related or otherwise)
Figure 9.4
The Indo-European family tree
ndo-
European
Balto-
Anatolian ndo- ranian Tocharian Armenian Albanian Hellenic talic Germanic
Slavonic
Hite ndic ranian Balc Slavonic East North West
Lycian Sanskrit Ossec Lithuanian West South East Gothic celandic High Low
Old Church
Lydian Hindi Pashtu Latvian Polish Belarusian Danish Yiddish Frisian
Slavonic
Bengali Farsi Old Prussian Czech Serbian Russian Swedish German English
Marathi Kurdish Slovak Croaan Ukrainian Norwegian Dutch
Old Persian Polabian Bulgarian Flemish
Avestan Sorbian Macedonian Afrikaans
Slovenian
may through contact continue to influence each other over time. Trees like those
we gave earlier give the impression (a) that languages split off from a common
mother language by regular (sound) changes, (b) that the split between (daughter)
languages happens instantaneously and decisively, and (c) that there is little or no
contact between languages. So, they are rather abstract models of linguistic change.
But is this really how languages evolved?
A much more realistic scenario would be that dialects gradually move away from
each other. Every dialect develops its individual innovations, and they all begin to
group into new (standard) languages. The same applies to the Germanic languages.
There was no big bang that divided the Germanic languages into Dutch, German and
Danish. Rather, we would expect that dialects evolve slowly until we eventually see
our modern standard languages. As we mentioned before, there is no natural cut-off
point between the dialects of Dutch and German at the border. This national (and
linguistic) border is purely political, cultural and arbitrary. For this reason, Johannes
Schmidt in 1872 suggested a very different way of looking at language variation. His
idea was that languages develop in waves rather than trees. Imagine a pond, and
a quiet, peaceful surface of water. When somebody throws in a stone (a linguistic
innovation), waves ripple concentrically through the pond. When a second stone is
thrown in (another innovation), a new set of waves develops, which interacts with
the previous waves. The same happens with a third stone. And so on. Every stone
is linguistic innovation that can create at least one new language, but that can also
affect the languages around it. Figure 9.5 shows such an abstract diagram.
Figure 9.5
Wave model of diffusion (based on Schmidt 1856)
Innovation A
Innovation B
Innovation C
Innovation D
Figure 9.6
A wave diagram of the Germanic language family (McColl Millar 2015: 173), reproduced with permis-
sion of Taylor & Francis
McColl Millar (2015: 174) uses this model to describe the development of the
Germanic languages (Figure 9.6). This diagram may look very confusing at first,
but it is actually quite simple. He identifies 26 innovations that happened in one or
more of the Germanic languages. High German, for instance, is characterized by
the High German Consonant Shift (we discussed this in Chapter 7). This innova-
tion only applied to High German; it did not spread to other dialects or languages.
Similarly, Gothic alone is characterized by a change from /fl-/ to /θl-/. Now con-
sider innovation 12: the loss of dental fricatives /θ, ð/. This happened to a number of
languages: High German, Low German, Frisian, Norwegian, Swedish, Danish and
Dutch, but not English, Icelandic and Gothic. And sometimes innovations come in
bundles: innovations 2, 19, 21 and 22 characterize English and Frisian, but not the
other Germanic languages.
Wave diagrams like the one in Figure 9.6 are obviously very different from the tree
diagrams discussed before. Rather than assuming that a given language split from
a mother node at one point in time and developed independently from then on, it
takes into account that languages and dialects are in flux and in contact with each
other. Innovations don’t necessarily split off one language from the next. They may
happen across languages and dialects, and they may lead to a dynamic re-grouping
over time. Nevertheless, Figure 9.6 still gives an accurate picture of what tree dia-
grams represent: the Scandinavian languages somehow group together, as do the
West Germanic languages. Gothic sticks out as the most unusual Germanic lan-
guage, and English and Frisian also maintain their special status (some have argued
for an Anglo-Frisian sub-branch on the Germanic tree on the basis of shared fea-
tures between Old English and Old Frisian). So, in this respect, Figure 9.6 is really
not unlike the earlier Figure 9.2, but it is less a priori and top-down, and rather works
with changes across time, including those resulting from inter-language contact.
So why don’t we drop tree diagrams altogether and focus on wave diagrams
instead? Wave diagrams might be more realistic about actual developments, but
they are a lot worse when it comes to the representation of historical developments,
and reading them is a lot more complex. So, for example, it is very easy to tell which
language came first in Figure 9.2. You simply move “up” the tree to find the mother
node. This can’t be done in a wave diagram. Tree diagrams can be seen as the
abstract summary of what wave diagrams show. It is thus not surprising that most
historical studies today use trees to present genetic relationships. Nevertheless, we
should not forget that actual language histories are a lot more complicated than this.
9.5 QUANTITATIVE APPROACHES TO LANGUAGE DIVERGENCE
Much of what we have described in the previous sections stems from research that
is more than 100 years old. More recently, with the advent of computing power
and research into DNA genetics, researchers have become increasingly interested
in quantitative aspects of language. But, in fact, one of the earliest attempts at quan-
tifying linguistic typology and diversity was glottochronology.
The essential idea is that we can somehow pinpoint the date when a language split
off from its ancestor languages. For that, we first need a basic word list, i.e. words
that each and every language should have (such as I, bird, blood and so on), but that
are more or less culture free. There is a so-called Swadesh list (named after Morris
Swadesh) that lists the top 100 or 200 words of basic vocabulary. Note that slightly
different versions exist, for numerous reasons (it is, after all, quite hard to decide
which words are “culture free” and which are not!). Table 9.4 is therefore only one
example of such a Swadesh word list.
The procedure then is quite simple. You pick two languages and check whether
the words on the list are phonetically similar or not. So, Latin manus ‘hand’ is more
similar to Spanish and Italian mano than to English hand, which in turn is more
closely related to German and Dutch Hand, or Swedish hand. The words that are
similar enough are deemed cognates (on some caveats, see above), and the more
cognates you find, the shorter the distance in time since they separated. If you find
few similar words (or cognates) on the list for two languages that are probably
related, it is more likely that they split long ago. If they share lots of similarities,
they also share a lot of history.
Table 9.4
List of 100 basic words
1 I (1 Sg) 35 tail 69 stand (verb)

2 You (2 Sg) 36 feather 70 give
3 we 37 hair 71 say
4 this 38 head 72 sun
5 that 39 ear 73 moon
6 who? 40 eye 74 star
7 what? 41 nose 75 water (noun)
8 not 42 mouth 76 rain (noun)
9 all (of a number) 43 tooth 77 stone
10 many 44 tongue 78 sand
11 one 45 claw 79 earth (soil)
12 two 46 foot 80 cloud
13 big 47 knee 81 smoke (noun)
14 long 48 hand 82 fire
15 small 49 belly 83 ash(es)
16 woman 50 neck 84 burn (verb)
17 man 51 breasts (female) 85 path (road)
18 person 52 heart 86 mountain
19 fish (noun) 53 liver 87 red
20 bird 54 drink (verb) 88 green
21 dog 55 eat 89 yellow
22 louse 56 bite (verb) 90 white
23 tree 57 see 91 black
24 seed 58 hear 92 night
25 leaf 59 know (facts) 93 hot
26 root 60 sleep (verb) 94 cold
27 bark (of tree) 61 die 95 full
28 skin 62 kill 96 new
29 flesh 63 swim 97 good
30 blood 64 fly 98 round
31 bone 65 walk (verb) 99 dry
32 grease (fat) 66 come 100 name
33 egg 67 lie (recline)
34 horn 68 sit
In order to determine the time when languages split, we also need to assume
that there is a constant rate of retention through time. The assumption has been
that languages will retain about 86% of the words on the 100 word Swadesh list in
1,000 years; conversely, about 14% of the 100 word list will be lost every 1,000 years.
From this follows a fairly simple equation that helps us determine a split date, or
the “time depth” of a given language. In this formula, T is the time depth, or time
since the split in millennia; C is the percentage of cognates; and r is the expected
retention rate (86%).
T = log C/2 log r
So if a language still has 60 out of 100 possible cognates, this means that T =
1.77/3.87 = 0.46. This means a time depth of 460 years, or, in other words, the
language split off from its parent 460 years ago. With only 20 cognates left,
T = 1.3/3.87 = 0.33, or 330 years.
Needless to say, there are a number of problems with glottochronology, beginning
with the assumption that we can determine a culture-free basic vocabulary that all
languages should have. Most importantly, we find that many languages actually bor-
row words from other languages even for basic concepts from the top 100 list: grease,
mountain and person are not native English (or Germanic) words but are borrowed
from French, and skin might come from Old Norse. If we look at the 200 word
Swadesh list, the results are even worse, and we find lots of words borrowed from
other languages. The problem is that borrowed words may or may not look similar
to the words in the other language, but they do not belong to the particular language
we are gauging, and therefore they skew our results. There is, of course, something
like a basic word list, and concepts which are more universal than others. But, nev-
ertheless, even this list is not without problems. Languages may have more than one
word for every concept on the list – which one do we choose? Is it small or little? Is
it come or arrive? And what if two concepts on this list are represented by only one
term? Some languages do not distinguish between bark (no. 27) and skin (no. 28).
So, in a nutshell, there are a number of problems with any list of basic words.
Other difficulties arise with the rate of retention and loss. Icelandic is known to
be very conservative and has a much higher retention rate than English, for exam-
ple. So how can we be sure about the 86% retention rate? Moreover, few languages
have a documented history going back more than a couple of hundred years; so
how can we be so sure that they tend to replace 14% of their vocabulary? In sum,
we can say that glottochronology was a good first step in the right direction, but
that it is highly controversial in many respects. It is thus no surprise that scholars
soon began looking for alternatives. These new approaches sometimes work with
Swadesh-like word lists, but apply much more sophisticated computational meth-
ods and algorithms in order to establish relationships between languages. Some-
times, scholars try to go beyond the word list approach and include other features
that can be shared by languages and dialects. This is particularly interesting and
helpful when languages and dialects influence each other after they have split from
the mother languages. Campbell (2013: 447–492) gives an excellent overview on
current methods and ideas.
9.6 LONG-DISTANCE COMPARISON AND PROTO-WORLD
If we go back to the family trees developed in section 9.3 above we see that an
important part of this model is the idea that languages and language groups can
be traced back to their roots, the parent languages from which they split off at one
point in time. So all Romance languages (Italian, Romanian, French, Spanish, Por-
tuguese . . .) go back to their mother language, Latin. Similarly, all Germanic lan-
guages (German, Dutch, Danish, Swedish, Icelandic, Gothic . . .) go back to their
mother language, Proto-Germanic. Both Proto-Germanic and Latin (or rather
Proto-Italic) again go back to their mother language, PIE. Here they are joined by
their sister languages, Indo-Iranian, Anatolian, Armenian, Albanian, Tocharian,
Balto-Slavonic and Hellenic. Proto-Germanic probably developed in the fifth cen-
tury BCE, and with PIE we are already looking at a time depth of about 7,000 years.
In other words, linguists are attempting to reconstruct languages and language
developments that happened more than 7,000 years ago and for which we have no
material evidence (in a moment we will look at some methods of reconstruction).
The same can be said, in one form or another, about the other language families:
Uralic, Altaic, Afroasiatic, Dravidian, Sino-Tibetan, Austronesian and so on. It is
perhaps even possible to draw a map of the world’s languages at 10,000 BCE. This
alone is a remarkable achievement. But it also leads us to another interesting ques-
tion: where do these language families come from? There seem to be two possible
answers. Either they also go back to a common ancestor, the mother of all lan-
guages, or they developed independently.
Let us begin with the latter idea, the independent development of language fami-
lies in several different places. This would explain the sometimes fundamental dif-
ferences we see around the world. Superficially at least, it seems that an isolating
language like Turung, spoken in North-East India, has little to no commonalities
with a polysynthetic language like Tiwi, spoken in Australia. Consider the following
two sentences (thanks to Stephen Morey and Alice Gaby for these examples) (don’t
worry about the grammatical terminology in the Tiwi example – you just need to
take in that all this is captured in one word):
(9.1)
dai 3 naa 3 rii 2 git 1
that possessive thread tie
‘then (they) would tie thread’
(9.2)
ampiniwatuwujingimajirranirningiyangurlimayami
a-mpi-ni-watu-wujingu-ma-jirrakirningi-angurlimayi-ami
3MINS.F-NPST-LOC-MORN-DUR-COM-LIGHT-WALK-IPFV
‘She is walking over there in the morning with a light’
Languages can code information in very different ways, as we saw in Chapter 5.

In a ground-breaking and hotly debated article, Evans and Levinson (2009) show
that linguistic diversity is far greater than we used to think and that there is hardly
any feature that is common to all languages. This would fit in nicely with indepen-
dent developments all over the world (a multiple origin scenario).
However, there are problems with this idea. First, no matter how superficially dif-
ferent the languages of the world are, their fundamental building principles still seem
to be surprisingly similar. As trivial as it may sound, it is important to acknowledge
the fact that most, if not all, known languages have consonants and vowels, words and
word classes, perhaps even syntactic phrases and certain syntactic rules. Note that
this does not automatically mean there is something like an innate universal gram-
mar in Chomsky’s sense. Evans and Levinson claim that languages have developed
these properties independently, out of general biological and cognitive principles. For
example, as we saw in Chapter 6, many languages have a consistent head ordering.
This means that if a language has verbs followed by objects, it is also likely to have
prepositions before their nouns (and not after). Alternatively, when a language has
objects occurring before their verbs, it also tends to have postpositions, i.e. following
their nouns. Does this constitute a language universal that languages cannot have
developed independently? No, not necessarily, since a consistent ordering of head
modifiers in a language reduces the processing effort for speakers and hearers and
should thus be favoured on the basis of general cognitive properties. The correlation
might also be the fall-out of grammaticalization (pre- and postpositions often develop
from verbs, and thus they keep their order with respect to the noun phrase; another
source is head nouns in genitive constructions). So, in sum, both the enormous diver-
sity in languages and their similarities could be explained with multiple origins, com-
bined with evolutionary pressures from biology and cognition.
But, alternatively, we might also think about one common ancestor. This would
be in line with what we know about our biological evolution, where the most widely
accepted proposal says that humans, Homo sapiens, developed somewhere in East
Africa around 195,000 years ago and began to spread all across the world about
110,000 years ago (the so-called Out-of-Africa Hypothesis). If we want to link this
with language evolution, we would have to assume that some form of protolanguage
must have developed before Homo sapiens left Africa. In that case, the migrants took
this common protolanguage with them, and we would expect numerous innovations
in all parts of the world, which ultimately led to the diversity we see today. But we
would also expect some retention, leftovers that are common to all languages. But
is there any linguistic evidence for this? Can we go back more than 10,000 years and
look at the languages, and language families, that led to Indo-European, for example?
In 1903 the Danish linguist Holger Pedersen suggested that such a macro-
language family might exist, and he coined the term “Nostratic”. To this day,
there is no agreement among scholars as to which languages and language fami-
lies actually belong to Nostratic, but most approaches would now include Indo-
European, Uralic, Altaic and perhaps also Kartvelian. Many also add Dravidian
and Afroasiatic. Nostratic would have been spoken around 15,000 BCE.
But how exactly do the proponents of such an old macro-family arrive at their
conclusion? Their method is essentially the same as outlined above: they compare
lists of basic vocabulary from all these (reconstructed) language families. Table 9.5
gives some examples (note that V stands for an unspecified vowel).
Figure 9.7
The Nostratic family tree (after Bomhard 2008)
Nostratic
Afroasiatic Kartvelian Dravidian Eurasiatic
Indo-European Uralic Altaic
Table 9.5
Nostratic word list (adapted from Ruhlen 1994: 103)
Language Who? What? Two Water One/finger Arm Bend/knee Hair Smell/nose
Afroasiatic k(w) ma bwVr ak a
w
tak ganA bunqe put suna
Kartvelian min ma yor rtsˑqˑa ert tˑotˑ muql putˑ sun
Dravidian yāv yā irantu nı-ru birelu kaŋ menda počču čuntu
Eurasiatic kwi mi pālā akwa tik konV bu-k(ā ) pˑutˑV snā
When one looks at Table 9.5, it becomes clear that, despite some differences, there
are surprising similarities between these languages. The word for ‘what?’ is similar
in all four languages: ma, ma, yā, mi, and so are the words for ‘smell/nose’: suna, sun,
čuntu, snā. Some words are very similar in only two or three languages. Consider
‘who’, which is k(w) in Afroasiatic and kwi in Eurasiatic, or ‘finger’, which is tak in
Afroasiatic and tik in Eurasiatic. Akwa is the word for water in both languages. Is
that coincidence? Advocates of the Nostratic hypothesis would say no, and they
would claim that the similarities are strong enough (based on a much larger word
list) to warrant the idea that Afroasiatic, Kartvelian, Dravidian and Eurasiatic actu-
ally go back to one common ancestor, Nostratic.
In 1994 the American linguist Merritt Ruhlen pushed this idea one step further.
His initial question was a very simple one: if we assume Nostratic can be established
as one macro language family spoken about 15,000 years ago in some parts of the
world (mainly Europe, Asia and Africa), what about the other languages? Wouldn’t
it be plausible to assume that Nostratic itself comes from some ancestor language,
together with the other large families we haven’t looked at yet (the languages of
America and Australia, for example)? Ruhlen tried to reconstruct as many of the
other languages and families as possible and compared them to the material already
available. The results are summarized in Table 9.6.
Table 9.6 again reveals surprising resemblances. The word for ‘water’, for
instance, turns out to be the same for Eurasiatic and for Amerind (akwa), and the
Table 9.6
Proto-World word list (adapted from Ruhlen 1994: 103)
Language Who? What? Two Water One/finger Arm Bend/knee Hair Smell/nose
Afroasiatic k(w) ma bwVr ak a
w
tak ganA bunqe somm suna
Kartvelian min ma yor rtsˑqˑa ert tˑotˑ muql toma sun
Dravidian yā v yā irantu nı-ru birelu kaŋ menda pu-ta čuntu
Eurasiatic kwi mi pālā akwa tik konV bu-k(ā) punče sna-Khi
Khoisan !ku- ma /kam k···a- //kɔnu //ku- //gom /ˑu- ču-
Nilo-Saharan na de ball nki tok kani kutu sum čona
Niger-Congo nani ni bala engi dike kono boŋgo
Dené- k wi ma gnyis ʔoχwa tok kan pjut tsha-m suŋ
Caucasian
Austric o-ko-e m-anu ʔ(m) namaw ntoʔ xeen buku śya-m -ıjun
bar
Indo-Pacific mina boula okho dik akan buku utu sinna
Australian ŋaani minha bula gugu kuman mala buŋku mura
Amerind kune mana pˑal akwa- dikˑi kano buka summe čuna
word for ‘finger’ is remarkably similar in 7 out of 12 language families. The same
goes for ‘knee’ and ‘smell’. On the basis of such data, Ruhlen claims to have identi-
fied roots for at least some of these words (using methods of reconstruction we
examine below):
ku = ‘who’
ma = ‘what’
pal = ‘two’
akwa = ‘water’
tik = ‘finger’
kanV = ‘arm’
buŋku = ‘knee’
sum = ‘hair’
čuna = ‘nose, smell’
These words (or roots), Ruhlen argues, come from Proto-World, the language that
was spoken before Nostratic and that gave rise to all our modern languages. Proto-
World is the mother of all languages. It is hard to say when such a language might have
been spoken. The earliest fossils for Homo sapiens date back to about 195,000 years
ago, and our genetically common ancestor (“mitochondrial Eve”) dates back about
130,000 years. Some groups of Homo sapiens left Africa about 110,000 years ago. The
oldest signs of symbolic communication in Homo sapiens date back to the explo-
sion of culture in the Upper Palaeolithic, or the Late Stone Age, about 40,000 years
ago. Here we begin to find art and evidence for rituals. This is important because
language is also a symbolic system, so that we can assume that by 40,000 years ago
humans were probably able to use some kind of speech. Unfortunately, this is as far
as we can get. If we assume Nostratic was spoken about 15,000 years ago, all we can
say is that Proto-World must have been spoken about 150,000–120,000 years ago,
before Homo sapiens left Africa and spread across the world. It seems highly unlikely
that a later and independent development of Proto-World in different parts of the
world would have led to the kind of similarities we see in the data in Table 9.6.
The idea of Nostratic (and even more so Proto-World) has been fiercely debated.
If we look back at what we said above about the dangers and problems of finding
cognates, it is clear that identifying cognates in reconstructed languages dating back
more than 10,000 years is a highly speculative enterprise. Some of the apparent
cognates might be independent developments on the basis of onomatopoeia; so
the words for ‘smell’ might simply resemble sniffing sounds in the individual lan-
guages. Also, given the extremely fast pace of radical semantic changes over time,
it is anything but clear whether the reconstructed words or their cognates actually
mean what they are supposed to mean. Campbell and Poser (2008: 370–372), for
example, point out that there are at least three words in Spanish that can be traced
back to the root *kuna: cónyuge ‘wife, spouse’, china ‘girl’ and cana ‘old woman
(adjective)’. All these look as if they could be related to the root *kuna. However,
when we look at the detailed word histories, we see that this is far from the truth:
cónyuge comes from Latin and is related to the words ‘with’ and ‘yoke’. So this has
nothing to do with *kuna ‘woman’. China is a loanword from Quechua, a South
American language, where it means ‘young woman’, but loanwords don’t count, as
we mentioned above. Cana comes either from a Latin word for ‘white’, candidus, or
from Latin canna ‘reed’ and is again unrelated to the word ‘woman’.
In a nutshell, when we take together the dangers of cognate hunting and the very
concrete objections by Campbell and Poser, it becomes clear that the reconstruc-
tion of Proto-World, or even Nostratic, is a fascinating, but highly problematic and
speculative exercise. One need not go as far as Campbell and Poser and say that “the
search for global etymologies is at best a hopeless waste of time, at worst an embar-
rassment to linguistics as a discipline, unfortunately confusing and misleading to
those who might look to linguistics for understanding in this area” (p. 393), but it is
certainly advisable to take these ideas with more than just a pinch of salt.
9.7 RECONSTRUCTING THE PAST
In 1877 Thomas Alva Edison invented the phonograph, a machine that could
record the human voice and other sounds. This was the first time in human his-
tory that spoken language could be recorded. Whatever we know about languages
before 1877 must be based on written sources. Inevitably, these records are never
complete – depending on the time and language, we lack data from certain social
groups, styles and genres. Also, written language can never accurately represent
spoken language. So we need to reconstruct many aspects of language before 1877.
But what about languages or times for which we have no record at all? As just dis-
cussed, human language is at least 50,000 years old, but the first written records are
only about 5,000 years old. What about those years in between? And what about the
language stages for which we have only few or no records at all? If we want to say
anything about these, we also need to reconstruct what they could have been like.
9.7.1 Written sources as evidence of change

What can we say about a language if all we have are written sources? Can we say
anything about its pronunciation? And what can we say about informal styles and
registers? And how do we deal with those social groups that could not read or write?
Let us look at some of these questions in greater detail. What can we say about
Shakespeare’s pronunciation, for example? First, it should be clear that earlier pro-
nunciations sometimes have to be reconstructed; i.e. we do not have direct and
concrete evidence for them, but we need to develop arguments that support our
claims. In the case of Shakespeare, such arguments could come from a number of
sources – wordplay and puns or rhyming patterns for instance. Romeo and Juliet is
full of witty wordplay. Here is just one example:
MERCUTIO
Nay, gentle Romeo, we must have you dance.
ROMEO
Not I, believe me. You have dancing shoes
With nimble soles: I have a soul of lead
So stakes me to the ground I cannot move.
(Romeo and Juliet, I.4)
There seems to be some pun here on the words soles and soul. Mercutio, who
has “dancing shoes with nimble soles”, encourages Romeo to dance. But Romeo is
feeling sad and depressed; he has “a soul of lead” and thus does not want to dance.
This may lead us to suspect that in Shakespeare’s time already soul and sole were
pronounced in similar ways (but we still don’t know how).
Rhyming patterns can also be helpful. Some examples come from the famous
“wedding sonnet” no. 166:
Let me not to the marriage of true minds

Admit impediments. Love is not love
Which alters when it alteration finds,
Or bends with the remover to remove:
O no; it is an ever-fixed mark,
That looks on tempests, and is never shaken;
It is the star to every wandering bark,
Whose worth’s unknown, although his height be taken.
Love’s not Time’s fool, though rosy lips and cheeks
Within his bending sickle’s compass come;
Love alters not with his brief hours and weeks,
But bears it out even to the edge of doom.
If this be error and upon me proved,
I never writ, nor no man ever loved.
There are at least three word pairs that don’t seem to rhyme in this sonnet. Love is
made to rhyme with remove, come with doom and proved with loved. Such rhymes
no longer work, and they may point to some differences in the earlier pronuncia-
tion. But there are three problems. First, these could be so-called eye rhymes or
visual rhymes, i.e. rhymes not based on sound but on letters (e.g. modern bough and
through rhyme visually, but not in their pronunciation). With remove and love this
might actually be the case – but we will never know for sure. The second problem is
that there is no guarantee that Shakespeare always rhymed perfectly. The third and
most interesting problem is the fact that a rhyme only can tell us that two words
A and B rhymed, but it doesn’t tell us whether A sounded like B, or B sounded like
A. For that, we need more evidence, for example from other rhymes or from puns.
Sometimes we are lucky and find “indirect evidence”, such as contemporary
descriptions of how some words or sounds were pronounced. Ben Jonson, a con-
temporary of Shakespeare, wrote in his Grammatica Anglicana ‘English Grammar’
about the letter <o>: “In the long time it naturally soundeth sharp, and high; as in
chósen, hósen, hóly, fólly [. . .] In the short time more flat, and a kin to u; as còsen,
dòsen, mòther, bròther, lòve, pròve” (1640: 39). So love and prove probably sounded
like modern northern British English: [lʊv] and [pɹʊv]; by extension, we suspect
it was [rəmʊv]. So occasionally we can say something about the pronunciation of
earlier languages on the basis of what contemporaries said about the topic.
Nursery rhymes
Old nursery rhymes also give us clues about earlier pronunciations. Take
“Jack and Jill went up the hill to fetch a pail of water; Jack fell down and
broke his crown and Jill came tumbling after.” The words water and after are
awkward here, and it’s water that is the culprit. Like many similar words, the
vowel sound was coloured by the rounded lips of [w], and [watər] shifted
to [wɔtər]. So water originally rhymed with after. It wasn’t a perfect fit, of
course, but in non-standard pronunciations [f] was often left out (Dickens
occasionally spelled after as <arter>). So it was probably more a case that
“Jack and Jill went up the hill to fetch a pail of [wahter]; Jack fell down and
broke his crown and Jill came tumbling [ahter].”
What about Old Mother Hubbard who “went to the cupboard to get her
poor doggie a bone; but when she got there the cupboard was bare; so the
poor doggie had none.” The words bone and none look as if they ought to
rhyme, and they once did. Originally, both were pronounced with [a], though
at the time this rhyme was being sung, [a] would have shifted to something
closer to [o]. So the word none is the guilty party here. It later did its own
thing by shifting its vowel to rhyme with nun. None is historically ‘not + one’,
so it behaves in the same curious way as the number one.
There are all sorts of ways we can reconstruct the pronunciations of earlier
times. Our knowledge of sound change, dialect variation, early commentar-
ies and poetry all provide different clues. Rhymes are not always reliable, of
course, but they can provide vital pieces of the puzzle.
As we described in Chapter 6, written sources are particularly problematic when

it comes to the study of syntax. These days we can ask native speakers if a given
sentence is grammatical or not. If native speakers agree that it isn’t grammatical, we
have what is called negative evidence – confirmation that something isn’t possible
in a given language. When they say it is possible, we have positive evidence for its
existence. Alternatively, we could even check one of the many corpora with millions
of words to see whether that sentence (or a similar structure) occurs in real speech.
This, of course, only gives us positive evidence. Whatever is in the corpus must be
possible in that language (though there’s always the chance of accidental slips of the
tongue and brain!). However, if some structure isn’t in the corpus, this doesn’t mean
it is ungrammatical.
The problem is compounded in historical linguistics. All the data we have is nec-
essarily corpus data. We have no access to a time machine to ask native speakers
if some structure is acceptable or not. Even worse, as emphasized in Chapter 6, we
need so much more data to study past grammar, and yet most historical corpora are
small and limited in scope. If we never find a sentence in Shakespeare’s English that
has articles following their nouns (dog the), we might suppose this was not possible
in Shakespeare’s English. But who knows – we also might be wrong, and somebody
might find just such a phrase in a long lost manuscript.
Needless to say, matters get worse the further we go back in time, simply
because the amount and variety of sources usually becomes even more limited.
Early English Books Online (http://eebo.chadwyck.com/home), containing manu-
scripts from Shakespeare’s time (about 1500 to 1600), comprises an astonishing
350,000,000 words. This is almost as much as we have for some present-day Eng-
lish corpora such as the Corpus of Contemporary American English. The Innsbruck
Computer Archive of Machine-Readable English Texts (http://www.uibk.ac.at/ang
listik/projects/icamet/), one of the most comprehensive corpora of Middle Eng-
lish (ca. 1100–1500), only has about 7.8 million words. And if we go back another
500 years, we find that practically all Old English texts (ca. 800–1100) still available
to us today, and available through the Complete Corpus of Old English/Dictionary
of Old English (http://doe.utoronto.ca/pages/index.html), only amount to about
3 million words. This is 1% of the Early English Books Online, and about ten times
the complete works of Shakespeare. For languages other than English, the situa-
tion can be even worse. Few manuscripts have survived for Gothic, and for other
East Germanic languages, like Burgundian or Vandalic, we only have the name (as
mentioned in other documents), but not a shred of direct evidence is left. The same
applies to the many historical languages all around the world that did not have
written forms.
But let’s come back to those for which we have some evidence, such as Old and
Middle English. Even though there are a few million words of text still available, the
extant written sources are obviously not fully representative of the language. For
one, we can hardly say anything about informal language, as most documents are in
a rather formal and literary style. In Chapter 6 we mentioned the Old English leech-
doms (the Anglo-Saxon first aid kit), which are very different from the highly styl-
ized remains of Anglo-Saxon literature (such as the poems Beowulf and Caedmon).
They are the closest we come to the spoken idiom for that period, but they are still
rather formulaic. But if we are really lucky, every now and then we get a quick peek
at truly informal language. For Latin and Greek, we have some informal graffiti on
walls, vases and other ceramics. And for some languages (including Latin, Greek
and late Middle English) we have various personal letters, which are closer to infor-
mal, oral language structures than, say, formal letters, sermons or histories. So even
though these are written, they bring us nearer to what people may have said.
Nevertheless, historical corpora are necessarily deficient when it comes to the
linguistic representation of different social groups. Obviously, there are no writ-
ten records by the people who couldn’t read or write (though some may have dic-
tated, and current researchers try to uncover what they can from this language).
It depends a lot on the time and culture who these people were. In many cases,
female speakers didn’t leave written records. For example, there is not one single
manuscript from a female Anglo-Saxon; we have to wait for the Middle Ages.1 On
the other hand, we find many manuscripts by female authors from Latin, Greek and
even Ancient Egyptian (so a lot earlier than Old English!). In most societies and
cultures, the lower social groups (with low incomes, little learning and no social
power) were illiterate. This means we usually don’t have many written records of the
language of farmers, blacksmiths or fishmongers, or wet nurses, for example. For
English at least, this only gradually begins to change in the Early Modern period,
when we see the slow spread of literacy.
In sum, we can say that historical linguistics, once it goes back beyond 1877 in
time, is faced with challenges, mostly to do with the fact that there are only written
sources available. These bring with them their particular problems, but occasionally
also opportunities. For example, with written sources it is now possible to trace the
linguistic behaviour of single individual speakers across their lifetime, sometimes
over more than 50 or 60 years – provided of course that they were literate and actu-
ally wrote material across that time span. Surprisingly, this is much more common
in the Early Modern English period than we’d expect (https://www.uantwerpen.
be/en/projects/mind-bending-grammars/project/). So we really get a chance to
observe long-term changes in the individual and the linguistic community in real
time. Today such a study would be extremely costly and time consuming, as we
would have to follow and record speakers over several decades. Here, historical lin-
guistics with its written sources has a clear advantage over present-day data, and we
are not dealing with such “bad data” at all, as Labov (1994: 11) once complained.2
9.7.2 The comparative method

How can we reconstruct what happened in the past even if there is not enough
evidence? One of the most widely used approaches is the so-called comparative
method. You have already learned a lot about this earlier in this chapter. The idea
behind it is actually simple. Let us assume we have four languages, A, B, C and D,
which might go back to one common ancestor, Language E (a protolanguage for
which we have no direct evidence). Diagrammatically, this may look like a simple
language family tree, as in Figure 9.8.
Figure 9.8
A simple language family tree
Language E (protolanguage)
Language A Language B Language C Language D
Now let’s assume that Language A has a word for ‘hand’ which sounds like [mano]
<mano>, and Language B has a word for ‘hand’ which sounds like [mão] <mão>,
Language C has [mɛ]̃ <main>, and Language D has [mano] <mano> (represented
in Figure 9.9).
Let’s assume these are “cognates”, or related words, and that all four are “reflexes”
of a word in Language E; i.e. they can all be derived somehow from a common
ancestral form. Two things are interesting and important now. First, two of the
words are very similar to each other – in fact, they are the same: [mano] <mano>.
Second, we know from independent sources that it is more likely that consonants
get deleted and adjacent vowels get nasalized rather than vice versa (see also
Chapter 4). When we take these two pieces of evidence together, it makes sense
to assume that Language E, the parent language, probably had a word that began
with <ma> (all reflexes have this in writing, three in pronunciation) and had an
<n> following <ma>, so <man>. The rest of the word is more complicated. In this
particular case, we know Languages A–E. [mano] mano is Italian, [mão] mão is
Portuguese, [mɛ͂] main is French and [mano] mano is Spanish, and they all go back
to Latin (Language E). In Latin, the actual word for ‘hand’ is [manus] <manus> –
so we were really close with our reconstruction! This, in a nutshell, is the idea of
linguistic reconstruction by using the comparative method.
Needless to say, matters are a bit more complicated than this. First of all, in order
to reconstruct a language we need a lot more evidence. So we are not comparing
four or five words (or cognates) but hundreds of them. Second, the goal is not only
to reconstruct the vocabulary of that language, as we did with Latin, but to recon-
struct the whole linguistic system: phonology, morphology and syntax.
The first and maybe most difficult step along the way is the exclusion of non-
cognates, words which do not share a common ancestor. As we’ve just seen, this is
particularly tricky because sometimes words look stunningly similar even though
they are not related at all (remember Modern Greek μάτι [ˈmati] and Malay mata
[mata], both meaning ‘eye’), and sometimes words are hardly similar, and yet they
Figure 9.9
A simple language family tree for the word ‘hand’
Language E (protolanguage)
[mano] [mão] [mɛ̃] [mano]

are related (Greek πέντε [pente], German fünf [fynf]). We also need to exclude
apparent cognates which were borrowed at a later stage, simply because they don’t
point to a common ancestor, but only to some later language contact. Both Japanese
and Portuguese share the word pan ‘bread’, but they are not related at all. Japa-
nese adopted the word pan from Portuguese pão in Late Middle Japanese, between
about 1500 and 1600. At the same time, they also seem to have incorporated botan
‘button’ (Portuguese botão – originally a Germanic word butt), arokuuro ‘alcohol’
(Portuguese álcool – originally an Arabic word al-kuħuul) and tabako ‘tobacco’
(Portuguese tabako – originally Spanish and Caribbean Arawakan tabaco). This
doesn’t make Japanese in any way related to Portuguese or any of these languages,
and we have to be careful to exclude such adopted words (and hence non-cognates)
from our considerations.
What we need to do now, with all these caveats in mind, is to find a large list of
cognates and establish so-called sound correspondences. For the sake of simplicity,
let’s use some Romance again. Consider the lexical items in Table 9.7.
Table 9.7
Cognates in Romance languages
Lexeme Italian Spanish Portuguese French

‘goat’ kapra kabra kabra ʃɛvʁə
‘dear’ karo karo karu ʃɛʁ
‘head, top’ kapo kabo kabu ʃɛf
‘meat, flesh’ karne karne karne ʃɛʁ
In establishing sound correspondences, we look at the individual sounds of each

cognate in matching environments (this is important). The initial sound appears to
be /k/ in Italian, Spanish and Portuguese, but /ʃ/ in French. So we can simply write
down the sound correspondence:
k- = k- = k- = ʃ-
Similarly, /a/ seems the same in all cognates in Italian, Spanish and Portuguese
except the last one. This also seems different in French, where /a/ turns into /ɛ/:
-a- = -a- = -a- = -ɛ-/-ə (read: -ɛ- in the middle of a word, -ə at the end)
The sound /r/ is equally similar:
-r- = -r- = -r- = -ʁ-/-ʁ
/p/ or /b/ shows some variation:
-p- = -b- = -b- = -f/-v-
How about /o/?

-o = -o = -u = ø (ø means ‘nothing’)
And /e/ and /n/?
-e = -e = -e = ø
-n- = -n- = -n- = ø
Table 9.8 summarizes what we have found so far.
Table 9.8
Sound correspondences for goat, dear, head, meat, dog
k- = k- = k- = ʃ-
-a- = -a- = -a- = -ɛ-/-ə
-r- = -r- = -r- = -ʁ-/-ʁ
-p- = -b- = -b- = -f/-v-
-o = -o = -u = ø
-e = -e = -e = ø
-n- = -n- = -n- = ø
Can we draw any conclusions from this? In fact, we can, but we first need to con-
sider a few rules for comparative reconstruction (assuming, as we described earlier,
that sound changes are not capricious, but regular):
a When reflexes are the same, assume the ancestor language had that same
sound.
b Consider “natural” sound changes (see Chapter 4). It is more common for
/k/ to turn into /ʃ/, or for final vowels to get lost, than vice versa.
c Assume the winner takes all. In many cases, the most common form is the
source, and the rarer forms the innovations.
d Assume as few changes as possible.
e Never forget to consider where in the word particular sounds occur.
f It is sometimes easier to do consonants first.
So what does this mean for our data set? For k- = k- = k- = ʃ- and -n- = -n- = -n- = ø,
we can invoke rules (b), (c) and (d): /k/ to /ʃ/ is a very common, natural change
(lenition), and /k/ is the most common form among the four. And why would three
languages change to /k/ and only one keep /ʃ/? So let’s assume that /k/ is the original
sound. Similarly, three languages have nasals, so let’s assume that French is again the
aberrant one (besides, sounds are deleted much more often than they are added). It
is also very natural for stops like /p/ to become voiced between voiced segments (e.g.
in the middle of words). So it makes sense to assume /p/ as the original form, which
was then either voiced word medially in Spanish and Portuguese, or changed to a
fricative in French (also voiced in the middle of words). Rules (c) and (d) also help
with -r- = -r- = -r- = -ʁ-/-ʁ. We could argue that French underwent an innovation
that turned all /r/ into /-ʁ/. The majority of languages have /r/, and why would they
all change in the same direction, away from /ʁ/? This seems unlikely, so let’s assume
/r/. Rules (b) and (c) can help us with both /e/ and /a/. The reduction of vowels to
schwa /ə/ or zero at the end of a word is a common process, and the majority of
reflexes are /a/ and /e/; so let’s assume /a/ and /e/ for the protolanguage – we might
assume /ɛ/ occurs word medially in French instead of /a/ as a regular sound change.
The -o = -o = -u = -ø set is a little problematic, but it makes sense to invoke rules
(b) to (e) in this case. We are dealing with vowels at the end of the word, so erosion
or even loss is not unusual. This would explain French -ø. Out of the remaining
three, /o/ is the most common, and it would be hard to explain why two languages
changed in this particular direction, when one kept the original sound and one
eventually lost it. So, /o/ seems the most likely candidate for this set.
In sum, we arrive at the following sounds: /k/, /r/, /p/, /n/, /a/, /e/ and /o/, from
which we can attempt to reconstruct the lexemes on our list in (undocumented)
Proto-Romance.
Lexeme Italian Spanish Portuguese French Proto-Romance

‘goat’ kapra kabra kabra ʃɛvʁə *kapra
‘dear’ karo karo karu ʃɛʁ *karo
‘head, top’ kapo kabo kabu ʃɛf *kapo
‘meat, flesh’ karne karne karne ʃɛʁ *karne
Incidentally, the reconstructed forms are very close to Latin: capra ‘goat’, carus
‘dear’, caput ‘head, top’ and caro, carnis ‘meat, flesh’ (note that Latin spelling repre-
sents /k/ with <c>).
Digging for clues – internal reconstruction

As we’ve seen on several occasions in this book, language change usually
leaves behind relics. All sorts of things can get stuck in the linguistic sediment
over time. They might be words (earlier wer ‘man’ petrified in the compound
werewolf ), pronunciations (the old voicing rule preserved in modern alterna-
tions like wife-wives), meanings (the old meaning of wife ‘female’ buried in
compounds like midwife) and even grammatical features (original Object-
Verb word order embedded in expressions like woodcut and bloodshed).
Linguistic debris of this kind can help us discover historical language
structures. But the comparative method is only partly helpful here; we have
to use another method of reconstruction. This method is known as internal
reconstruction, and the basis of it is simple. Irregularities in the modern lan-
guage (like foot-feet) are the fall-out of changes, and if we undo these changes,
we can get back to a more regular set-up (recall our study of i-umlaut: irreg-
ular forms like foot-feet are relics of an earlier pronunciation rule whereby
the main vowel of the word assimilated to the high vowel of the lost plu-
ral ending). So with internal reconstruction, we infer earlier forms of a lan-
guage via material available from a synchronic description only. Whereas the
comparative method can help us draw linguistic family trees and lead us to
a common historical “mother language”, internal reconstruction only inves-
tigates earlier stages of a single language (e.g. Pre-Old English *føti > Mod-
ern English feet). Clearly the technique isn’t as powerful as the comparative
method, but it is a valuable tool in cases where languages have no known
siblings (language “isolates” like Basque). It is also usefully applied in prepa-
ration for the comparative method to bring forms back to a shape so that
comparisons are easier.
Modern languages are full of rubble, and hidden in the remains are partial
linguistic histories of these languages. So being a historical linguist is some-
times like being an archaeologist – we both dig stuff up in order to provide a
fuller picture of the past.
The techniques of phonological reconstruction give great insight into the nature
of phonological change, but they don’t transfer readily to syntax. Because the lexi-
cal shape representing a given meaning in language is completely arbitrary (the
symbolic nature of words), it follows that when lexical cognates are discovered
across languages they must be the result of shared history – a time of common
development when the cognates in question represented one lexical item in the
shared parent language. And we’ve seen how the relatedness between cognate items
is preserved, sometimes despite extreme sound change, because of the regularity of
sound change.
But what happens when we apply this methodology to syntax? Syntactic patterns
relate form to meaning, but they aren’t symbols in the same way. If, for instance, Lan-
guage A has the construct “Players hit the ball” (i.e. SVO) and Language B has the
construct equivalent to “Players the ball hit” (i.e. SOV), all we can compare are the
parts of the constructs, not the whole; i.e. the lexical but not the syntactic information.
To take this example further, if we want to reconstruct PIE word order, what we find
is that all three major word order types are represented in the daughter languages:
Dravidian languages (SOV) Celtic languages (VSO) Romance languages (SVO)
The proto word order may have been any three, all three or even none, i.e. free
word order. Indeed, all of these patterns have been suggested by linguists work-
ing on the ancestral word order. But even if we had three different corresponding
phones (a parallel case in phonology) in three related languages, we could still,
through our knowledge of phonological systems, and our understanding of pho-
nological change, attempt to reconstruct the shape of the proto-phoneme (the
sound correspondence p-f-h could plausibly evolve from *p). When syntactic cor-
respondences show completely diverging patterns, we can’t even be sure they are
cognate. The syntactic categories verb and object are extremely common across lan-
guages, and the comparison of OV and VO is meaningless. Another example is the
Indo-European passive. The syntactic category exists in all daughter languages, but
the syntactic constructions and incorporated lexical material are irreconcilable. We
assume PIE had some sort of passive, but what was its syntactic shape?
No one-to-one correspondences can be set up for syntactic constructs, because
they don’t evolve in the systematic way sounds do. The types of changes affecting
syntactic systems are the same ones that cause the comparative method to fail – as
we saw in Chapter 6, these are pattern replacement, analogy and reanalysis. But it’s
not all bad news.
One solution is to isolate language specific developments and reconstruct lan-
guages at the stage before the individual changes took place (so internal reconstruc-
tion), until we get patterns that are comparable. This method has had some success
in the reconstruction of past syntactic systems. Typology and language universals
can also assist. As we described in Chapter 6, our knowledge of syntactic types
indicates preferred directions of change, so we can use this as we do our knowledge
of phonological systems. Moreover, the growth of strong general theories of syntax
has given us insights into historical processes, which can be applied to the recon-
struction of sentence structures. Recall from Chapter 6 Talmy Givón’s now famous
catchphrase “Today’s morphology is yesterday’s syntax.” We know that independent
syntactic items become fossilized as morphology; hence, the synchronic morphol-
ogy of languages provides valuable clues about their early syntax. We also know that
grammar is created out of lexical items – we know the usual suspects, and we know
their paths of change. In short, we know a lot more than we did about how to go
about reconstructing the syntax of protolanguages.
SUMMARY
In this chapter we had a detailed look at the relationships between languages. After
a brief discussion of what languages actually are, we followed up on the venerable
and powerful idea that languages can be grouped in a family tree, so that we have
grandparent languages (like Indo-European), parent languages (like Italic) and
daughter languages (like Spanish, Italian, French etc.). As an alternative, we dis-
cussed the more dynamic wave model that captured the fact that new features of
a language can spread from a central point in continuously weakening concentric
circles (like waves created when something is thrown into water).
We examined “the comparative method”, the technique for studying the evolu-
tion of languages via the meticulous feature-by-feature comparison of two or more
languages, leading to the reconstruction of unattested ancestral forms. There were
two important concepts here: (1) the symbolic nature of words and (2) the prin-
ciple of regular sound change. Both tenets are what give linguists proper licence to
compare and reconstruct ancestral forms in this way. Here we looked far back in
time and discussed the pros and cons of a kind of Proto-World, the mother of all
languages. Finally, we also had a look at the (sometimes controversial) quantitative
methods of measuring language relationships, including approaches that combine
findings from evolutionary genetics and language typology.
FURTHER READING
There is a flourishing literature on the comparative method and related issues going
back a long way: we recommend Bloomfield (1933), Fox (1995) and Anttila (1989).
We briefly mentioned internal reconstruction as a second historical method; for
details see Hock (1991: Chapter 17); Fox (1995: Chapters 7–8) and McColl Millar
(2015: Chapters 8 and 9). Long-distance genetic relationships (taking in Nostratic
and Proto-World) have captured people’s imaginations, and the topic makes regu-
lar media appearances; you can find a discussion of this controversial aspect of the
comparative method in Campbell (2003), and of the genetic grouping of the world’s
languages in Comrie (2001). Glottolog (http://glottolog.org/) is also a good start on
the topic, as is the World Atlas of Language Structures (wals.info). For glottochro-
nology see Campbell (2013). Many researchers working on languages in Australasia
and the Pacific have written about naming taboos and noted the difficulty of finding
cognates: see Keesing and Fifiʔi (1969) and Holzknecht (1988).
EXERCISES
1 Investigating genetic relationships

Choose 20 random items from the Swadesh list in Table 9.4. Identify the words in
at least 15 different Germanic languages (choose across the different sub-branches)
for these 20 items. Is there any pattern? On the basis of what you find, identify the
language families and possible relationships for the 15 languages you investigated.
2 The comparative method

Examine the sets of Germanic words in Table 9.2 and answer the following questions:
a What is the name given to such sets of words?

b Provide the sound correspondences for the vowel sounds.
c Give the Dutch, German and Swedish words that would correspond to Eng-
lish brown.
d English has the word maroon corresponding to the Italian word marrone
‘brown’ – how do you explain the similarity to the Italian word?
e How do we account for the striking similarities between the vocabulary
items here? Why do you think we can rule out chance or borrowing?
f Draw a mini family tree to represent the relationship between these languages.
3 Reconstructing Proto-Algonkian (American Indian)

Look at the cognate sets for these languages of the Algonkian family. Suppose that,
on the basis of the two correspondence sets shown in lines (a) and (c), a protoseg-
ment *h were reconstructed. How could this reconstruction be questionable? Dis-
cuss. (Hint: think regularity of sound change.)
Gloss Fox Cree Menomini

a ‘he sits’ nahapiwa nahapiw nahapɛw
b ‘belch’ ʃekiwa sekiw sekeew
c ‘crow’ aho aha aʔa
d ‘he embarks’ posiwa posiw posew
e ‘knock down’ kaweni kawin kaeɛn
4 Reconstructing Proto-Polynesian
Examine the data below and answer the questions.
Gloss Hawaiian Samoan Māori Tongan

a ‘sky’ lani laŋi raŋi laŋi
b ‘road’ ala ala ara hala
c ‘house’ hale fale ɸare fale
d ‘rock’ haku fatu ɸatu fatu
e ‘sea’ kai tai tai tahi
f ‘canoe’ waʔa vaʔa waka vaka
g ‘new’ hou foou hou foʔou
h ‘turn over’ huli fuli huri fuli
i ‘fire’ ahi afi ahi afi
j ‘hand, arm’ lima lima riŋa nima
k ‘five’ lima lima rima nima
l ‘one’ kahi tasi tahi tahi
m ‘two’ lua lua rua lua ua
n ‘three’ kolu tolu toru tolu
o ‘four’ haa faa ɸaa faa
p ‘six’ ono ono ono ono
q ‘seven’ hiku fitu ɸitu fitu
r ‘eight’ walu valu waru valu
s ‘nine’ iwa iva iwa hiva
t ‘ten’ anahulu (day) aŋafulu ŋahuru hoŋofulu
u ‘mistake’ hala sala hara hala
v ‘cut’ maahele sele e here hele tele
w ‘cry’ kani taŋi taŋi taŋi
x ‘house fly’ nalo laŋo ŋaro laŋo/raŋo
y ‘fish’ iʔa iʔa ika ika
z ‘sennit’ ʔaha ʔafa kaha kafa
aa ‘sleep’ moe moe moe mohe
bb ‘mosquito’ makika name name name
cc ‘louse’ ʔuku ʔutu kutu kutu
dd ‘yam’ uhi ufi uɸi ʔufi
ee ‘rain’ ua ua ua ʔuha
ff ‘afraid’ makaʔu mataʔu mataku manavahee
gg ‘bird’ manu manu manu manu
hh ‘ear’ pepeiao taliŋa tariŋa taliŋa
ii ‘year’ kau tau tau taʔu

jj ‘deaf’ kuli tuli turi tuli
kk ‘pandanus’ hala fala ɸara fala
ll ‘light, moon’ — maalama marama malama
mm ‘moon’ mahina maasina — maahina
i What consonant correspondence sets are found?

ii What is the probable phonemic inventory of Proto-Polynesian?
iii What examples of metathesis are found?
iv What items do not appear to be cognate?
v Give the likely Proto-Polynesian reconstruction for each item.
5 Research essay
Write an essay (approx. 1,000 words) that explores the hypothesis that sound
changes operate “without exception”. Consider types of apparent exceptions, and
discuss how they might be dealt with. Your bibliography should have at least five
references.
NOTE
1 The project on women’s literary culture in England and Europe has an ongoing database
of texts and manuscripts, available at http://www.surrey.ac.uk/medievalwomen/index.
htm.
2 In this respect, the “Up” series offers wonderful opportunities for linguists. In many dif-
ferent places around the world people have been making long-term documentary films
that follow people’s lives from childhood into adulthood. There is now available a corpus
of audio files and transcripts from the original British “Up” series based on the inter-
views of individuals at seven year intervals over a period of 42 years (see http://www.
linguistics.berkeley.edu/~gahl/upInPress.pdf).
10
An end on’t
INTRODUCTION
In order to understand where languages have come from and where they might
be heading, historical linguistics has to draw on many aspects of the study of
living languages. The discipline embraces a range of different subfields within
linguistics, including those core areas that handle the structural features of lan-
guage (e.g. phonetics, phonology, morphology, syntax), those that deal with dif-
ferent aspects of language and communication (e.g. pragmatics and discourse),
language and society (sociolinguistics) and the mental make-up of human beings
(cognitive linguistics and psycholinguistics). More recently, it has accommodated
new topics to do with language endangerment, language change and new media,
corpora and computational applications. It’s not surprising, then, that research in
historical linguistics now encompasses a wide range of very different theoretical
frameworks.
In this final chapter we introduce some additional perspectives that we hope will
take you to what it is that lies at the very heart of the historical linguistics enterprise.
Our intention is to add a stronger theoretical dimension to some of the issues raised
in previous chapters, and to frame our discussion of some recent initiatives in the
discipline. We begin with the three obvious components to any investigation of
language change:
The “what” of change – possible and impossible changes

The “how” of change – processes and mechanisms of change
The “why” of change – explanations for change
At first, linguists concentrated their endeavours on identifying and classify-

ing the facts of language evolution. They concerned themselves with features
that had changed in the past (in particular, sounds) and the processes by which
these had changed; in other words, the “what” and the “how” of change. More
recently, the focus of research has shifted more to explanation, and the internal
and external forces that drive change. The main goals have become the col-
lection of historical information, the discovery of common patterns of change
and the testing of hypotheses about the causes of change. Ideally, these goals
lay the foundation of a general theory of language development with a set of
universal laws that regulate languages and compel them to proceed in particular
directions.
10.1 FIVE QUESTIONS FOR ANY THEORY OF CHANGE
In their seminal work “Empirical Foundations for a Theory of Language Change”,

Weinreich et al. (1968) pose five interrelated problems that any explicit theory of
language change must address:
1 The constraints problem (what determines the possible and impossible

changes and directions of change?)
2 The transition problem (how does change progress through the linguistic
system?)
3 The embedding problem (what is the linguistic and social context of lan-
guage change?)
4 The evaluation problem (what are the attitudes of a speech community
towards a given change, and the effect of these attitudes on change?)
5 The actuation problem (why do linguistic changes occur when they do and
where they do?)
To illustrate these five key questions, it is probably helpful to provide an example,

and one that is already familiar to you. Recall the developments in English negation
that we outlined in Chapter 6. Of course these represent but one type of change,
and other types relate differently to these problems; however, the various issues sur-
rounding diachronic negation are enough to give some idea of how these five issues
have been driving research in historical linguistics.
To summarize this development, take the following four sentences from differ-
ent stages in the history of English (note that these sentences all come from books
on health and good behaviour). What we see here are clearly identifiable phases in
the history of negation (the reality of course wasn’t quite this neat, and there was
considerable overlap between the stages):
(Old English) preverbal ne

1 Ne forlæt þu þæs blodes to fela ‘don’t let too much of the blood’ (Old English)
(Middle English) embracing ne-not
2 Ne blow not on thy drynke ‘don’t blow on your drink’ (Middle English)
(Early Modern English) postverbal not
3 Drinke not above four times ‘don’t drink more than four times’ (Early Mod-
ern English)
(Modern English) do-support
4 Don’t drink and drive (Modern English)
An end on’t 249
1 The constraints problem

First, in terms of the “constraints” problem, we are potentially looking at various
types of checks and controls here, such as those monitoring the direction of change,
the source of new negators, as well as their placement. These constraints involve
both diachronic aspects (i.e. negation across time), as well as synchronic aspects
(i.e. “universals” or (im)possible patterns of negation).
Most linguists now refer to this kind of negator renewal as “Jespersen’s Cycle”.
As we described in Chapter 6, cross-linguistic studies show that the change is
extensively attested across a wide range of languages. As a paradigmatic case
of grammaticalization, Jespersen’s Cycle (including the rise of do-support in
the last stage) qualifies as a “natural” or “normal” change. Moreover, the devel-
opment of not is in line with the first (and most common) of the three main
options that have been identified for new negative markers: namely nominal
minimizers such as (not) a bit (see Willis et al. 2013). In terms of positioning, we
know negative markers tend to precede the elements they negate, with preverbal
as the norm for standard clausal negation. What has come to be known as the
“Neg-first principle” (Jespersen 1917: 5; Hirt 1937: 73) receives strong support
from empirical studies, such as Dahl’s 1979 survey of negation (around 81% of
his 240 languages had negative particles before the verb, and the figure increases
if cases of prefixal negation are included; see also Dahl 2010). Both Jespersen
and Dahl suggest there is something natural about the preverbal placement of
negators, with evidence from second language learning and child language to
support this.
So the changes we see here in English negation illustrate well-trodden paths of
change, and yet they are not universal – nor would we expect them to be, for the
simple reason that “constraints” have to be violable for language change to ever take
place. A more useful formulation of this problem is therefore, as Labov (1982: 59)
has suggested, to locate “general principles that influence the course of linguistic
change without determining it absolutely”; in other words, to frame the problem
in terms of natural tendencies rather than rigid laws. As we have emphasized in
several places throughout this book, the aetiology (i.e. the study of the causes or
origins) of innovation and change is usually complex, with all sorts of interfering
psychological, physiological, linguistic, social and cultural pressures working to
coerce a language in a particular direction – linguistic structures don’t exist outside
these forces, and they militate against the operation of regular and utterly predict-
able changes.
Interestingly, in Germanic, Jespersen’s Cycle has resulted in a typologically less
common arrangement that places the negator after the verb in sentence negation
(Drink not). Hence, we would predict subsequent changes to restore a more “natu-
ral” situation. This is one way of viewing the Modern English system of negative
auxiliaries (don’t, can’t, won’t etc.); this restores the balance by once more ensuring
that negative elements appear before the lexical verb.
2 The transition problem

In terms of the “transition” problem (how changes disperse through the linguistic
system), we can view the remodelling of English negation as an orderly progression
through recognizable (but overlapping) stages (note that V = verb here):
(preverbal) ne-V > (embracing) ne-V-not > (postverbal) V-not > (preverbal) nega-
tive auxiliary-(main) V
Grammatical changes are gradual and (like phonological shifts) subject to the
same measured transmission. This progression did not affect all linguistic contexts
at once, but (akin to “lexical diffusion”, discussed in Chapter 7) spread gradually
through the system. As is typically the case in syntactic change, subordinate clauses
were conservative, preserving negative patterns long after they had been abandoned
in main clauses.
A crucial factor affecting transition is frequency. English (like its siblings) had
in common a group of frequent verbs that held onto older patterns of negation
for some time; for example, phrases such as I know not and I think not existed well
after dummy do had become a requirement. Throughout this book we’ve seen how
many different aspects of our linguistic behaviour are shaped by repetition – para-
doxically, the most frequent words and phrases are at the forefront of sound change
but prove to be the least adventurous when it comes to grammatical change. Over
time often-repeated structures become entrenched and therefore able to resist the
generalizing forces outside, hence the conservative nature of these high-usage verbs
(see Bybee 2003).
As shown vividly in Chapter 7, more recent formulations of the “transition”
problem have built in the routes of linguistic variants through speech communi-
ties (also incorporating the puzzle of how all this happens without interfering with
communication). To understand how these negative patterns were promoted and
dispersed, we need to study the social lives of English speakers, in particular the
relationships between individuals and their social networks. Relevant here is a cor-
nerstone of linguistic methodology known as the “uniformitarian principle”: devel-
opments within languages are subject to the same factors and controls at all times,
and hence we assume that the social structures that assist the transition of features
through speech communities today were also around in the past. Though the socio-
cultural settings for English speakers 1,000 years ago and English speakers in the
21st century are clearly very different (considers the rise of universal literacy, stan-
dardization, mass media and e-communication for starters!), uniformitarianism
assumes that changes today and those in the past are driven by the same sorts of
mechanisms.1 Indeed, as we’ll see in a moment, the discipline of socio-historical
linguistics rests on the viable application of sociolinguistic methods to historical
situations (though of course it uses written data to do this).
As the network studies of Lesley and James Milroy have shown, tight-knit commu-
nities with strong social networks and values are norm enforcers – speakers with few
external contacts will identify with and orientate towards those they interact with
An end on’t 251
most extensively. These networks correlate with conservative speech patterns and
lack of innovation (quite simply, members won’t want to speak differently from their
mates). On the other hand, where there is little social cohesion between members of
the speech community, such loose networks are more typical of larger, often urban
communities with many external contacts, and they are more open to innovation.
People have ties to other groups, and changes are spread via the weak ties between
them – these speakers are the conduits for change. Writing specifically on negation,
Nevalainen (1998) noted that “a Milroyan type of weak-ties network structure could
well have been the means of spreading the loss of multiple negation” (p. 281).
3 The embedding problem

A closely related problem involves how the change is embedded in the surrounding
linguistic and social setting. The changes in negation occur in the context of long-
term grammatical changes that have been taking place in the language. Crucial here
is the notion of “drift”, the cumulative unconscious selection of variants that propels
a language in a particular direction, or, as Sapir originally put it, “[l]anguage moves
down time in a current of its own making. It has drift” (1921: 160). Something we
haven’t mentioned here is that there is plenty of evidence suggesting that preverbal
ne formed a unit with the finite verb; hence, the transition from original ne to ne-
not to not is also a transition from morphological negation to an adverbial system of
negation, and therefore consistent with the drift towards increased analyticity, one
of the drifts Sapir identified for English (see also Chapter 5).
These changes have also taken place within another of Sapir’s drifts: the fixing
of constituent order and the shift from OV to VO structures. One way to see the
movement of the negative particle from the position before the finite verb to the
position after the finite verb in Germanic languages was a consequence of the gram-
maticalization of word order patterns, particularly verb-second order for declara-
tive main clauses. This is an illustration of where typological coherence has applied
additional pressure for change to have a causal role in the shift to postverbal nega-
tion (see Burridge 1993).
Any account of the changes to English negation must take in how variants were
embedded in the larger context of the speech community; it must also try to resur-
rect the social distribution of the various patterns and the motivations for spread-
ing certain favoured patterns. Assuming the uniformitarian principle, we take for
granted that the same dimensions of social structure relevant to linguistic change
in the modern language (age, sex, social class, ethnicity, race) were important when
these changes in negation were taking place – and that, like today, the innovative
forms would have acquired some prestige before being transmitted to the com-
munity as a whole. Also relevant here is an observation made by a number of early
linguists (such as Jespersen 1922: 259–61), namely that periods of dramatic linguis-
tic change appear to go hand-in-hand with social upheaval (for example epidem-
ics and civil wars). As we saw in Chapter 4, the Black Death in the 14th and 15th
centuries coincided with major changes in English, and this includes the transi-
tion from embracing to post-verbal negation. You can see now how social network
theory can make sense of this correlation. In some communities, the plague (made
worse by subsequent epidemics) killed up to three-quarters of the population. It
triggered psychic epidemics (see Gordon 1959: 545–79) and would have torn even
the densest of social networks apart. It is precisely at such tumultuous times that
variants are able to take off, spread and eventually embed themselves as long-term
changes in the language system.
4 The evaluation problem

It is usually the case that as soon as speakers become conscious of a linguistic
change, the response is negative (we addressed this in Chapter 1). For most speak-
ers, change is only acceptable if it takes place in the distant past – though fascinated
by the origins of words and the stories that lie behind the structures in their lan-
guage, they typically condemn changes that happen within their lifetime. This can
have an effect on the propagation of a variant.
Nevalainen (1998) looks at the stigma that attached to the use of multiple nega-
tion in the Early Modern period. These negative evaluations continue and may block
another Jespersenian scenario from unfolding based on the pattern You don’t know
nothing; more likely to take off are the less salient variants that can slip more easily
under the prescriptive radar such as never, at all, a bit (as in I don’t like him a bit).
However, prestige reversals do occur. Indeed, as we discuss later in this chapter, cur-
rent pressures on the standard language are endowing informal non-standard lan-
guage with a cachet and respectability (see Crystal 2006 on the “new sociolinguistic
climate” that is seeing the re-evaluation and elevation of non-standard features).
Non-standard variants, once condemned, now have a good chance of slipping out
of the local context into a bigger arena; multiple negation/negative concord is now
well on its way to being an established “vernacular universal” (see Kortmann and
Lunkenheimer 2013, who list it as having 80% attestation across 76 varieties of Eng-
lish worldwide). Moreover, with the increasing influence of newer varieties of Eng-
lish comes the diminishing clout of those native varieties from the so-called “inner
circle”. Linguistic innovations (including some very un-English-looking structures)
are now being started and propagated by second language speakers, foreign lan-
guage speakers and creole speakers (e.g. the current role of rapping).
5 The actuation problem

The preceding four problems all contribute to the big problem of “actuation”. As
Weinreich, Labov and Herzog originally expressed it (1968: 102): “Why do changes
in a structural feature take place in a particular language at a given time, but not
in other languages with the same feature, or in the same language at other times?”
In clarifying the actuation of the negation changes just outlined, linguists must
account for why the Germanic languages, and also French, display such different
chronologies. In English the exclusive use of not was the norm by the mid 1400s
(see also Jespersen 1917: 9), but for the Scandinavian languages, “Jespersen’s Cycle”
was completed at a much earlier date. What’s more, Scandinavian probably under-
went two negator renewals in a Jespersen-like progression: (1) reinforcement of ne
An end on’t 253
with -a/-at (ne disappeared by the 9th century); and (2) replacement of -a/-at by
eigi (completed by the 14th century) (cf. Willis et al. 2013). In German, nicht had
eliminated the old preverbal negative particle ni by 1300; in fact, even in the 13th
century, distribution of ni was limited to common usage verbs and special construc-
tions (see Paul 1959: 236–8). In Dutch the cycle was completed more slowly, and
there were dialect differences; by the mid 1600s, embracing negation had finally
disappeared from Hollandish but was still intact in Brabantish (only relics remain
today; see Burridge 1993). Only now is Modern French showing signs of losing
preverbal ne.
One fact we hope you will carry away from this book is that language change is not
straightforward. For a start, we have to consider what else is going on in the linguis-
tic system. For example, in English we see the development of do-support, and a set
of negative auxiliaries appearing before the main verb. So why have Modern Dutch
and German retained their postverbal negators? If you remember, in Chapter 7 we
described some of the older verb-final aspects of modern Dutch and German; this
so-called verbal brace means that German nicht and Dutch niet will frequently occur
before the main verb, and in subordinate clauses before all verb parts. We are looking
here at a complex of different intersecting pressures that have been influencing these
languages at different times: systemic (the linguistic system with interacting com-
ponents), cognitive/psychological (the mental make-up of speakers), physiological
(the production of language), social and political (the speech community and the
individual, the socio-political environment) and external (contact and borrowing).
And don’t underestimate the cultural preoccupations of speakers either. These
can be powerful triggers for change. If we compare human lives in Anglo-Saxon
times with those in modern times, we see a colossal transformation in human ways
of thinking – the earlier communal thought processes, once so gripped by natural
and supernatural outlooks, are a world apart from our modern secular sense of
identity, which presupposes understanding and control (think of personal man-
tras such as “doing your own thing”). What repercussions does this budding self-
awareness of the modern western individual have for linguistic structures? There is
no doubt that cultural forces shape language, even those features of language that
are more than skin deep (i.e. grammatical). With breakthroughs in science (say,
in neural networking) and advances in experimental tools will surely come better
understanding of this society, language and mind liaison (see Deutscher 2010; De
Busser and LaPolla 2015).
In short, what we’re saying here is that, like any change, these developments in
negation emerge from a complex interaction of factors. Analogous to the evolution-
ary processes of the “invisible hand” (see Keller 1994), language evolves through
processes of competing mechanisms and motivations.
10.2 ADDITIONAL THEORETICAL PERSPECTIVES
As hinted at the start of this chapter, historical linguistics draws on inspirations from
every one of the major branches of synchronic linguistics – a proper understanding
of sound change is impossible without a knowledge of how sounds are articulated

(phonetics), and so on. There have been significant inputs from fields outside lin-
guistics too, such as prehistory, philology, anthropology, psychology and sociology.
With such spread as this, all our small book has been able to do is to highlight
some of the fundamental issues, methods and results that characterize this vast and
very interdisciplinary field. In this section we outline three relatively new fields of
linguistic research that are now contributing to a fuller picture of language change.
10.2.1 Corpus linguistics

In various places in this book we’ve made reference to the use of corpus data in
the study of language change. (Note that corpus has a rather fancy plural corpora,
though you will also find people using the more mundane corpuses – another exam-
ple of analogy, as discussed in Chapter 5.) Basically, corpora are collections of texts,
either written or spoken (so recorded and transcribed). We can use them to find out
all sorts of things about how a language is being used or has been used. People have
always looked to texts as sources of language for analysis, but the advances in corpus
linguistics in recent years come with exciting possibilities, especially for examining
language in the past. We’ve already seen examples in this book of how historical
corpora can help to answer questions to do with the “what”, “how” and “why” of
change; for example, what new structures enter the language and how they spread
through the system (e.g. abruptly or slowly) and through the speech community,
how they relate to matters of style, as well as societal and cultural shifts, and so on.
Any collection of texts could be called a corpus, but within linguistics a corpus is
usually understood to be a structured, digitized collection of texts that is searchable
and that has been annotated to make it more useful. Corpora do vary in their degree
of annotation, but usually they include such things as morphological segmentation
and structural mark-ups such as clause and parts of speech tagging. Here histori-
cal corpora encounter complications, especially to do with word level annotation.
One reason is the considerable spelling variation in early texts. To give you some
idea of the extent of this variation, consider that in one short 14th century English
cookery manuscript (The Forme of Cury, or ‘Method of Cooking’) we find the word
ginger spelled nine different ways: gyngeuer, gyngyuyr, genger, gynger, gyngere, ginge,
gyngeuyr, gyngyuer, ginger.
There is a long tradition of studying the history of English, and as you’ve seen
from some of the case studies in this book, it relies heavily on texts that have sur-
vived from the earlier stages of the language. These days there are some wonderful
online resources available, notably the Helsinki Corpus of English Texts: Diachronic
and Dialectal (http://www.helsinki.fi/varieng/CoRD/corpora/HelsinkiCorpus/),
with texts covering the period from ca. 750 to ca. 1700 and a dialect part based
on transcripts of interviews with speakers of British rural dialects from the 1970s.
Nevalainen and Raumolin-Brunberg (2003), drawing on the Corpus of Early Eng-
lish Correspondence (http://www.helsinki.fi/varieng/domains/CEEC.html) (6,000
letters from 1417 to 1681), were able to reconstruct the social setting of late Middle
English and Early Modern English to investigate a number of linguistic changes,
An end on’t 255
including those in negation mentioned earlier. Even within the quite recent (and
more modest) Australian National Corpus (https://www.ausnc.org.au), there is a
historical collection, the Corpus of Oz Early English (CoOEE), comprising over two
million words of Australian, New Zealand and Norfolk Island origin between 1788
and 1900. It contains a combination of speech based texts (speeches, plays, court
proceedings, testimonies) and written texts (government legal documents, public
documents such as newspapers, and private documents such as letters and diaries).
We’ll have more to say on the uses of such databases below.
Corpora vary a lot with respect to the number of texts they contain and the over-
all size of the collection. Of course the desirable size for a corpus will depend on
what aspects of language you’re investigating; we saw in Chapter 6 how historical
syntax relies on generous and robust corpus data. Corpora also vary in their rep-
resentativeness. Some cover a wide range of text types and genres, and are broadly
representative of the wider community in terms of age, gender, dialect and socio-
economic group membership, while others are quite restricted in size, focusing on
one particular group. The size and composition of a corpus typically reflect the
resources that were available for compiling the corpus and the purposes it was
developed to meet. In the case of historical corpora, it can be a matter of salvaging
what we can, but even with a small collection much can be done (especially if you’re
investigating high-frequency items like pronouns or auxiliaries). Corpus based
research has many applications, and as Davies (2011: 63) puts it, any disadvantages
are far outweighed by the advantages.
The Corpus Resource Database gives an idea of just how many corpora are already
out there (http://www.helsinki.fi/varieng/CoRD/corpora/index.html), but new cor-
pora are becoming available all the time. A recent arrival is the Old Bailey Corpus
(http://www1.uni-giessen.de/oldbaileycorpus/index.html).2 It has transformed the
records of the almost 200,000 criminal trials held at the Old Bailey (London’s central
criminal court) into a massive diachronic corpus of spoken English over a 200 year
period (1720–1913) with approximately 14 million words (all annotated for parts of
speech and direct speech). Each text also provides socio-biographical information
on speakers (gender, age, occupation) and pragmatic information (the role of the
speaker in the courtroom: defendant, judge, victim, witness and so on). These texts
give an idea of what went on in English criminal courts in early times and, most
importantly for us, give us valuable access to the language of non-elite people. Since
the proceedings were taken down in shorthand by scribes in the courtroom, they
provide a rare glimpse of spoken English from the 17th through to the early 20th
century. (The scribes would have used quill, pen or pencil and probably used the
pen-writing method of shorthand – until stenotype machines became available in
the 1870s.) Being more colloquial in nature, these texts are especially useful – even
taboo words (notoriously absent from other writing) make an appearance. Tran-
scripts reveal that as early as the 1700s bugger could be used to strengthen negative
constructions; this is more than a century earlier than previously thought (It cannot
be helped now, I should not care a b – dy b – gg – r if I was going to be hung up for it;
see Musgrave and Burridge 2014 for details).
In recent years, the development of computer mediated communication and

improved possibilities for creating and disseminating high-quality recordings of
language have made for vast improvements in corpus design. Many modern cor-
pora now aim to be multimodal – so not simply text based collections, but high-
quality recordings of transcribed speech with aligned audio files as well as visual
records of interaction (video). With new corpora and new tools appearing all the
time, corpus based research can only get better.
10.2.2 Historical pragmatics

In Chapter 1 of this book, we had a look at Shakespeare’s language, so it’s only fair
if this last chapter also includes an example from Shakespeare. Look at the follow-
ing two excerpts from Much Ado about Nothing; in particular, pay attention to the
phrases (I) pray thee.
(10.1)
BENEDICK
Do you question me, as an honest man should do, for
my simple true judgment; or would you have me speak
after my custom, as being a professed tyrant to their sex?
CLAUDIO
No; I pray thee speak in sober judgment.
BENEDICK
Why, i’ faith, methinks she’s too low for a high
praise, too brown for a fair praise and too little
for a great praise: only this commendation I can
afford her, that were she other than she is, she
were unhandsome; and being no other but as she is, I
do not like her.
CLAUDIO
Thou thinkest I am in sport: I pray thee tell me
truly how thou likest her.
BENEDICK
Would you buy her, that you inquire after her?
(Much Ado about Nothing, I.1)
(10.2)
MARGARET
Troth, I think your other rabato [a piece of 17th century clothing] were better.
HERO
No, pray thee, good Meg, I’ll wear this.
An end on’t 257
MARGARET
By my troth, ’s not so good; and I warrant your
cousin will say so.
HERO
My cousin’s a fool, and thou art another: I’ll wear
none but this.
(Much Ado about Nothing, III.4)
First, it seems fairly clear that pray does not mean the same as pray in present-
day English. Very generally speaking, for Shakespeare it seems to mean something
like I ask you, or maybe I beg you. The first two occurrences (10.1) could be trans-
lated into Modern English as I beg you. As such, this is a good example for a so-
called performative verb, that is, a verb that performs the action it stands for (e.g.
in I quit, simply by saying that (in particular contexts), you perform the act of
quitting). Similarly, if you say I beg you, you are performing the act of begging.
English has undergone some changes here. We no longer use pray as a performa-
tive verb. In the Corpus of Contemporary American English (COCA; http://corpus.
byu.edu/coca/), which has 450 million words, we only find about 50 occurrences
of I pray you, and they are all either literary, or Shakespeare quotes, or deliberately
antiquated. I beg you, in contrast has more than 250 hits in the same corpus. And if
we consider simple please as a second alternative (‘please speak in sober judgment’,
‘please tell me truly’), then of course we find thousands of occurrences in Modern
English – but not a single one in Much Ado about Nothing (except for please as a
simple verb, e.g. “Father, as it please you”, “He both pleases men and angers them”)!
So one could say that here we see I pray thee as the Shakespearean equivalent of
modern please.
But now have a look at Hero’s use of the phrase in the second excerpt (10.2). This
appears to be slightly different. First, it does not have the subject pronoun I, it is just
pray thee. Second, it is not followed by a verb-initial clause as in the former two exam-
ples: “I pray thee speak . . .”, “I pray thee tell . . .”. And while this example of simple pray
thee could somehow also be translated as ‘please’ or ‘I beg you’, it also seems to func-
tion much more like a come on, in some sense, rather than a performative verb. In this
function, it helps to convey the speaker’s feelings and organize the conversation, rather
than conveying any actual information about the world. The same difference can also
be seen in the modern use of like in English. Look at examples (10.3) and (10.4):
(10.3)
I like pizza. I like McDonald’s. That’s me.
(COCA, Harper’s Bazaar, 2007, 3550: 530)
(10.4)
And she goes, “Oh, they’re in Paris. And I’m like Paris, California?” No, no, Paris,
France. And I’m like, what?
(COCA, NBC Dateline, 2012, 12017)
Very roughly speaking, the word like in (10.3) means ‘love’ (“I love pizza, I love
McDonald’s”). But you can’t substitute like in example (10.4) with love: “And I’m
love Paris, California? No, no . . . And I’m love, what?” sounds very weird. So like
in (10.3) is more like a full verb with a particular meaning, whereas like in (10.4) is
more like some conversation organizing element: it signals that the speaker is quot-
ing someone. Similarly, Hero’s use of “pray thee” is less like a full expression of her
wishes, but rather something to organize the conversation and the social relation-
ship between Margaret and Hero.
This is one example of a study that can be found in the field of historical prag-
matics; in other words, the study of language use and meaning making practices
in past contexts. Needless to say, such an enterprise is not without its challenges.
Present-day pragmatics has actual speakers and authentic (and spoken) language to
work with. So it becomes possible to talk about things like “speaker intention” for
present-day English utterances. But what about languages in the past? How do we
know what speakers actually intended? How do we know how they felt? Whether
they considered an utterance polite or impolite, bossy or shy? We do not have spo-
ken language and actual speakers available for past language stages. But there are
two ways to avoid or at least reduce this problem.
First, historical pragmatics usually takes a fairly broad stance towards what counts
as pragmatics. In line with a more European perspective, it includes almost any
kind of language resource that is used in the negotiation of meaning. Thus, written
language in all its forms also becomes part of the picture and is subject to pragmatic
studies.3 In other words, politeness strategies are relevant and interesting not in
spoken language alone, but also in written documents, such as letters. Here we can
look at salutations and closing formulae, for example. These can reflect different
levels of politeness or social relations. Consider the following examples, taken from
15th century English personal letters. Note that these are all the first lines of the
individual letters. (Don’t be thrown by the imaginative spellings here – remember at
this time there was no concept of a “correct way to spell”. Often reading the language
aloud helps comprehension, though one symbol might throw you – <ʒ>, or “yogh,”
appears instead of <y>.)
(10.5)
Ryth reuerend and worchepfull modyr, I recomand me on-to you as humbylly
as I can think, desyrinyng most hertly to her of youyr welfare and hertys ese [. . .]
‘Right revered and worshipful mother, I recommend myself to you as humbly as
I can think, desiring most heartily to hear of your welfare and heart’s ease [. . .]’
(John Paston III to his mother, Margaret Paston, 1468)
(10.6)
Ryth wyrchypful hwsbond, I recomawnd me to ʒw, desyryng hertyly to heryn
of ʒwr wel-fare [. . .]
‘Right worshipful husband, I recommend myself to you, desiring heartily to
hear of your welfare [. . .]’
(Margaret Paston to her husband, John Paston I, 1448)
An end on’t 259
(10.7)
I grete you wele, letying you wete that your brothere and his felesshep stond in
grete joparté at Cayster [. . .]
‘I greet you well, letting you know that your brother and his fellowship stand in
great jeopardy at Caister [. . .]’
(Margaret Paston to her son, John Paston II, 1469)
(10.8)
On Tuesday in the morwyn whas John Botillere [. . .] and Davy Arnald your
cook, and William Malthows [. . .] taken at Heylesdon be the balyf of Ey [. . .]
‘On Tuesday in the morning John Botillere [. . .] and Davy Arnold your cook,
and William Malthows [. . .] were taken at Heylesdon by the bailiff of Ey [. . .]’
(Margaret Paston to her husband, John Paston I, 1465)
First, it seems that letter openings were very much standardized and fairly com-
plex in the late Middle Ages – at least in comparison to our modern “Hi!” Second,
the examples (10.5) to (10.8) create the impression that they are somehow different
with regard to politeness and social roles. (10.5) seems to be the most polite, most
humble form to begin a letter. Not only does the writer say that he recommends
himself “most humbly” to the addressee, but he also uses the intensifier “most” in
his wish to know about the addressee’s wellbeing. (10.6) is very similar, but slightly
less polite and humble (perhaps we could call this a more or less neutral form of
address). It follows the same structure as (10.5) – “right worshipful husband, I rec-
ommend myself to you, I wish to know about your wellbeing” – but it does so less
emphatically. It does not mention the word “humble”, and it does not use boosting
superlatives such as “most heartily”. Example (10.7), then, is very different from
both (10.5) and (10.6). There is no formal address such as “my right worshipful
son”, nor any recommendations, nor any questions about the addressee’s wellbeing.
Rather, there is a simple impersonal greeting formula “I greet you well”, followed
directly by the actual message. In comparison to the first two examples, this must
have been rather blunt. Even worse, in some sense, is (10.8), which starts with the
message right away and does not offer any greetings or recommendations at all. So
we do get the impression that these letter openings can be ranked in terms of polite-
ness, from the most polite (10.5) to the least polite (10.8).
But how do we know? There are no speakers of late Middle English we can ask,
and only rarely have these ever commented in writing on the politeness or impo-
liteness of certain forms. The answer is that we cannot know for sure, but that we
can try to find some evidence for our conjectures. One thing we notice when we
study late Middle English letters is that they show highly formalized structures, as
we have seen above. This is documented not only by the high number of letters that
somehow follow these structures, but also by some more theoretical letter writ-
ing manuals and models still available today. What this means is that the major-
ity of letters follow the expected patterns, and that openings such as (10.7) and
(10.8) stick out – both in terms of numbers, but also with regard to explicit norms.
Second, when we look in greater detail at the actual contents of these letters (and
pragmatics is the study of meaning making, after all!), we discover that our first
impressions are actually confirmed. While (10.5) and (10.6) are fairly neutral, if not
even nice kinds of letters, (10.7) is an accusing, reprimanding letter from mother to
son (blaming him for failing to defend Caister, the family home, thereby putting his
brother in danger). Similarly, (10.8) seems to have been written in great agitation,
as it tells of the unjustified arrest of several men that worked for the addressee. In
other words, both forms of address seem to be related to some sort of emotional
distress. In (10.7) the curt greeting might reflect Margaret’s being angry with her
misbehaving son, and in (10.8) the lack of a greeting could reflect her great worries
about the situation. Eventually, what we see here is a complex and plausible analysis
of pragmatic features and verbal behaviour in written documents that are more
than 500 years old.
The second option in historical pragmatics to counter the lack of authentic spoken
data is to investigate materials which come as close as possible to spoken language.
We know spoken language can never be perfectly represented in written language,
but there are some genres and text types that come a bit closer than others. For
example, historical linguists can use drama to investigate at least some aspects of
spoken interaction, such as speech acts, discourse markers or some politeness fea-
tures. We have tried that above in our analysis of pray thee in examples (10.1) and
(10.2). It can be argued that drama to some extent mirrors or resembles spoken
interaction, and that some of the linguistic features that we find in drama reflect
what speakers say off stage. And, more importantly, we do not have to speculate
about the reactions of addressees. When we looked at the example of letter opening
formulae, we had to speculate a little bit about how people may have evaluated the
different structures. But if somebody gets insulted or flattered in a play, for example,
we often also see their reactions and can thus arrive at conclusions about the lin-
guistic triggers for these reactions.
Historical pragmatics necessarily relies on written corpora but will be informed
by research on how people today make sense out of messages. As we’ve been
emphasizing, historical linguistics by its nature is interdisciplinary, and it is con-
stantly being informed by breakthroughs in synchronic linguistics. Technological
advances in corpus design mentioned earlier are opening up all sorts of new ave-
nues for research possibilities by providing access to interactional phenomena that
go beyond what is actually being said to what is presupposed, what is implied and
so on; for example, in multimodal spoken corpora paralinguistic and non-verbal
elements are an inherent part of the database (see Haugh 2008). It’s all good news
for historical pragmatics!
10.3 HISTORICAL SOCIOLINGUISTICS
A second field of study that has gained wide prominence over the last 30 years or
so is historical sociolinguistics. Again, the idea is very simple. We touched on it
earlier: speakers in the past are social agents, just like speakers today. They associ-
ate and dissociate with certain social groups, just as present-day speakers do, and
An end on’t 261
their speech may depend in complex ways on language external factors such as age,
gender, education, place of living and the like. The only problem – again – is the
availability of data. Until the 1980s historical linguists worked with more or less
a-social data, i.e. data which either was not socially distinguishable in any way or
simply was not differentiated.
But in 1982, in a truly ground-breaking book called Socio-historical Linguistics,
Suzanne Romaine showed that even in historical linguistics, social information
on speakers and texts is available and can be used in fruitful ways in our analyses.
Romaine investigated the use of various strategies for forming relative clauses
in Middle Scots, the language spoken in Scotland in the 16th century. (Relative
clauses are a type of sub-clause, used to modify nouns and pronouns.) She discov-
ered that there is some complex, but systematic variability in the use of the differ-
ent markers that introduce the relative clause (the relativizers): that, the so-called
wh-relativizers, and the omission of the relativizer, also known as “zero”:
The thing that he remembered

The thing which he remembered
The thing he remembered
This variability depended on both language internal factors, i.e. factors that
belong to the language system, and language external factors, i.e. factors indepen-
dent of the language itself. In terms of language internal factors, she found that,
similar to Modern English, that was preferred in restrictive relative clauses (clauses
that identify or restrict the reference of the noun; e.g. I saw the man that stole the
cookie). In non-restrictive relative clauses (clauses that add extra information about
a noun whose reference is already well established; e.g. I met John, who was also
at the party), there is a preference for wh-relativizers and zero. Interestingly, and
unlike Modern (standard) English, zero was also possible for subject relativizers,
as in I saw a man stole the cookie. Today, this can only be heard in some dialects.
In terms of language external factors, Romaine was primarily interested in regis-
ter and style differences in the texts she analysed. Her data suggested that that was a
lot more common in informal texts, and texts representing some sort of oral speech.
The wh-relativizers, on the other hand, were prominent in the more formal texts.
This led Romaine to the conclusion that the wh-relativizers must have been intro-
duced through literate, learned styles in the most formal writings (an idea which is
also supported by the fact that wh-relativizers first occur in the more complicated
syntactic structures). Interestingly, at least in Scottish English, the wh-relativizers
have never been such a great success. To this day, the most common everyday rela-
tivizer for all purposes and functions remains that. All this goes to show that in his-
torical sociolinguistics we try to unify ideas from historical linguistics and modern
sociolinguistics.
All language change is ultimately tied up with language variation, and variation
in turn has both language internal and language external aspects. For quite some
time, it was assumed that historical data, because it is so sketchy and unreliable,
does not allow for any systematic study of language external factors. This is partly
true of course. We do not have first hand data from female speakers in the Old
English period, for example, or from illiterate peasants in 10th century France. But
this does not mean that we can’t discover meaningful and interesting patterns in the
data that we do have. As just described, new methods in the “digital humanities”
allow us to collect enormous amounts of data and to process these with the help of
powerful computers. We can apply sophisticated statistical tools to these data sets.
To be sure: this does not mean that now we can say something about female speak-
ers in Old English, or illiterate peasants in 10th century France. We still need to be
very critical about the actual data that is available to us (see Bergs 2012 for some
details). But modern tools now allow us to identify patterns in linguistic variation
which so far had either been ignored or deemed chaotic and not very revealing.
More than 30 years after Romaine’s ground-breaking monograph, historical
sociolinguistics can be regarded as a well-established discipline, just like historical
pragmatics. It has its own journal, the Journal of Historical Sociolinguistics (edited
by Gijsbert Rutten, Anita Auer and others), and an online network (HISON), which
also organizes an annual summer school, and we see the publication of an ever-
growing number of studies on numerous topics and languages.
In Chapter 7 we described how William Labov, and later linguists like Peter Trud-
gill and the Milroys, showed for the first time that linguistic variation was system-
atic, motivated and affected by social factors. From the viewpoint of correlational
sociolinguistics (as this approach came to be known), historical sociolinguistics has
successfully identified broad patterns of linguistic diffusion across social groups
in Early Modern English, for example. Earlier studies have established that the
development of do in questions and negation depends on the syntactic context (we
discussed Ellegård’s study in Chapter 7). It was quicker in negative questions (e.g.
Don’t you like cheesecake?) and slower in negative declaratives (e.g. You don’t like
cheesecake!). Studies in historical sociolinguistics have complemented these find-
ings by showing that, for example, the loss of do in positive statements (e.g. I do like
cheesecake) is something that was most prominent in London and the court, and
copied by other regions – except, it seems, for East Anglia and female writers, who
were lagging behind. Conversely, female speakers appeared to lead in the introduc-
tion of do in negative statements. So women’s linguistic behaviour, like that of men,
was by no means uniform (and nor would we expect it to be; see Cameron 2003).
The same kind of study also showed that the replacement of -th (as in he singeth) by
-s (he sings), as discussed in Chapter 7, was tied not only to language internal factors
(such as the final sound of the stem) but also to social factors. Third person -s came
from the North of England, together with huge waves of migrants in the 16th and
17th centuries moving to the South and in particular to London. This change was
also driven by female speakers, and in particular female speakers from the upper
gentry in the 17th century (see Nevalainen & Raumolin-Brunberg 1996: 84).
In a similar vein, other studies in historical sociolinguistics have looked at issues
such as historical dialectology (i.e. the study of linguistic variation as it depends on
region, including the perception and evaluation of dialectological differences) and
what is commonly termed sociology of language. The latter is not so much con-
cerned with correlational aspects but studies the role of language in society as such.
An end on’t 263
What kind of varieties or languages were speakers exposed to? What did they think
of these different varieties? How do modern national standard languages develop,
and how are minority languages treated over time? All these are questions that are
productively investigated under this label. One example is the study of the devel-
opments in 19th century Spain (see, e.g., Del Vallé 2013). There were some seri-
ous attempts in Spain (and elsewhere in Europe) in the 19th century to establish a
national standard language (like Spanish or German) in order to strengthen national
identity. For Spain this meant that other languages spoken there, like Catalan or
Galician, had to be suppressed in favour of Castilian Spanish, which was supposed
to become the national standard language. Needless to say, this provoked resistance
and fierce discussions. And, to this day, many regions in Spain (like Catalonia and
Galicia) maintain a strong local identity and still use their original language in many
contexts, rather than the official national language, Castilian Spanish. These develop-
ments illustrate one aspect that historical sociolinguistics studies from the viewpoint
of sociology of language. And there are many more. Consider the relationship of the
Gaelic languages and English in Ireland and Scotland, or of English and indigenous
languages in Australia, Africa, Asia and North America – or the development of
Standard Russian or Mandarin Chinese. All these are immensely interesting from a
sociological, political and cultural point of view, and issues such as language choice,
language identity, nation building, cultural identity and the protection of minority
languages are hot and pressing topics for historical sociolinguistics.
Busting language myths

Yet another area in historical linguistics is the study of language myths, or
“alternative histories” as they are sometimes called (Watts 2011, 2012). The
key idea is that, for the most part, the study of language history has been the
study of “official” or uniform languages, and how these developed. So, when
we look at the development of English, for example, we usually take a very
specific kind of perspective. We assume, for example, that at one point in time
(usually very long ago) English as a language was “born”. It then went through
some times where we find several English dialects and varieties, but not a
standard language as such until, ultimately, English became the uniform, offi-
cial standard language that we assume to exist today. This development has
usually been seen from either a “tunnel” or a “funnel” point of view (Watts
2011: 585). From a tunnel perspective, we only focus on what is inside the
tunnel, and we are moving towards the end of the tunnel, the light, as the end
point of the development. All variation, for example the different dialects,
lies outside the tunnel and does not come into view. With the funnel view on
language history, many related varieties have to pass through a funnel, which
gets much narrower at the bottom, and thus reduces all variability until we
have a uniform standard (see Figure 10.1).
Figure 10.1
The funnel view of language history (adapted from Watts 2011)
Varieties of English/Anglo-Saxon
Standard English
Watts points out that tunnel and funnel visions are problematic. The tun-
nel view simply ignores all variability and thus might miss important aspects
of language development – such as the fact that “third person -s” came into
Standard English from northern dialects (see above). Moreover, it margin-
alizes culturally and socially important facts, as, for example, the role and
development of Catalan or Galician in Spain. The fact is that these were and
still are parts of life and culture in Spain, and simply ignoring them does not
do justice to scientific investigations.
The funnel view, on the other hand, at least acknowledges that certain vari-
eties may have played a role in the development of the standard language
that stands at the bottom of the funnel, but it disregards the histories of these
varieties, as well as their continued development. The traditional account of
the Great English Vowel Shift (see Chapter 4), for example, focuses on some
sort of “standard” accent, ignoring other varieties where the shift didn’t occur
or did occur to varying degrees. And, as Watts points out, also at work here is
an associated language myth – the myth of greatness for the standard variety
(which underlies the name of course). In Chapter 4 we mentioned another
“great” vowel shift currently underway in the southern hemisphere varieties
of English, but what we really should emphasize is that vowels are shifting all
the time – and no one shift is any greater than the other.
10.4 THE RATE OF CHANGE – DIVERSITY AND STABILITY
In this book you have seen examples of changes at every level – orthography, phonol-
ogy, lexicon, semantics, pragmatics, morphology and syntax. We should not give the
impression, however, that all these linguistic elements are equally susceptible to inno-
vation and replacement. Lexical components are the most volatile, with vocabulary
An end on’t 265
addition and loss being particularly rapid; however, there are differences even within
the lexicon – culturally significant words are more prone to revitalization than basic
vocabulary, which can endure over centuries (as you saw in Chapter 9). Words of
high frequency are more prone to sound changes (of a reductive nature), but are
the most resistant to grammatical changes (of a levelling and regularizing nature).
In fact, grammatical aspects of the language (especially syntactic) exhibit the most
stability of all. However, as Janda and Joseph (2003: 88) remind us, if we are to truly
understand change, it is just as important, and interesting, to consider those aspects
of language that do not change as those that do. We saw in Chapter 7 that some
variation leads to dramatic change; other variation remains stable for long periods of
time. Nichols (2003) addresses precisely these questions, examining in detail differ-
ent kinds of stability and variability/renewal of linguistic elements over time.
It is also the case that individual dialects and languages do not change at a constant
rate. As the overall history of English shows, at least the history of the standard language
(we don’t want to be accused of tunnel/funnel vision), there can be periods of speeding
up and periods of slowing down. From the time of Old English through to the 18th cen-
tury, changes were complex and rapid. As a consequence, people in the 1700s could not
read with ease the literature of, say, roughly three centuries earlier; the poetry of Geof-
frey Chaucer presented difficulties, as it does today. And looking at language from a still
earlier time prompted even Chaucer to make the observation (in Troilus and Criseyde
II, 22ff): “[y]e knowe ek that in forme of speche is chaunge” (‘you know also that in (the)
form of speech (there) is change’), noting the “wonder nyce and straunge” (‘wonderfully
curious and strange’) nature of early English words. In his preface to the Eneydos (1490),
the printer William Caxton also observed how different and difficult early English was,
describing it as “so rude and brood that I coude not wele vnderstande it”.
And yet Modern English readers have little trouble reading texts of the 1700s; the
language of Jonathan Swift or Jane Austen is stylistically different and has some unfa-
miliar looking vocabulary, but it is recognizably Modern English. As Svartvik and
Leech (2006: 191) observe, such written texts might suggest very little has happened
to the language over the last few hundred years. True, this is writing, which is slow to
reflect changes, but a number of factors have also been putting the brakes on changes –
notably, the introduction of the printing press, the knock-on effects of standardiza-
tion and its linguistic straitjacketing, and the establishment of reading and writing
as educational necessities (rather than optional extras). These have had the effect of
slowing down changes, in some cases even reversing ones already well entrenched.
Spelling pronunciations
As Dwight Bolinger describes in Chapter 2 of his book Language the Loaded
Weapon (1980), so immersed are we in writing these days that it has taken
a hold on our minds. All this has had a conservative effect on the normal
processes of change. Influential written texts such as the Bible act as a kind
of artificial life-support system for obsolete expressions – words, images and
turns of phrase that would otherwise have shuffled off the lexical coil remain
in our mental lexicons. But the effects are even more obvious in phonology.
When spelling is there to act as a reminder, pronunciation is less apt to
change. Take the child who insists that remember is actually bemember (as one
of us experienced – a nice example of distance assimilation). This child will
eventually encounter the written representation, and with the visual image of
remember in his head is less likely to pronounce it bemember. Phonological
changes are generally reductive, and as we saw in Chapter 4, these kinds of
fast-speech phenomena are harbingers of change (they also feed grammatical
developments of the sort we saw in Chapter 6). These changes occur much
more readily if we have no notion of the written word but will slow down
considerably with a knowledge of writing, especially spelling. A conscious-
ness of spelling can even lead to the reversal of a sound change.
You probably pronounce the first syllable of the word comrade as [kɒm],
but well into the 19th century it was pronounced the same as the <o> vowel
you find in sponge and frontier – so [kʌm]. It was the pressure of spelling that
changed the pronunciation to [kɒm] – “let’s pronounce it like the spelling”.
Hence the variation in the pronunciation of <o> in words like comfrey and
constable. There are many such rearward shifts happening right under our
noses. The palatal pronunciation [ɪʃu] for the word issue has been around
since at least the 15th century. We know this because of early spellings with
<sh>: <isshewe>, <ishew>, <ishu>, <ishwe> and so on. However, the older
pronunciation [ɪsju] is making a comeback, probably again because of the
pressure of spelling. The clout of the written word is such that many older
pronunciations are returning from the dead. These days there’s a kind of tug
of war between [ɪʃu] and [ɪsju] – and [ɪsju] is getting a good bit of the rope.
However, disrupting forces are at work, promising once more major episodes of
change. In the case of English, the language has been described as “one of the most
hybrid and rapidly changing languages in the world” (Graddol 2006: 116). The pro-
cesses set in place by the electronic revolution, colloquialization, liberalization and
globalization are releasing speech from the conservative forces of the literary stan-
dard and its prescriptive ethos. The written tail has stopped wagging the spoken dog
(to use an image that nicely describes the control that writing has had over speech),
and features that have been lurking in the wings as variation now have a greater
chance of taking hold and being embedded in the language system as actual change.
10.4.1 The effects of new and emerging media

These days we are witnessing a linguistic revolution. The internet and the global
trend towards electronic communication is making an impact akin to some of the
major shifts in the past (from pen and ink, to printing, to the telephone, radio and
television broadcasting). It has blurred even further the (already rather blurry)
An end on’t 267
traditional distinction between speech and writing. Emails, tweets, text messages
and other social media are now routine aspects of most people’s lives. They involve,
of course, written language, but clearly they also share many of the features not just
of spoken language, but of actual conversation. In real-time online communication,
people exchange messages in much the same way as they would chatting face-to-
face. The spontaneity and speed of this sort of communication mean that we simply
don’t go in for the same careful organization and planning as we would with normal
writing. We end up writing very much as we speak, and this means a looser con-
struction, repetition, false starts, digressions, comment clauses and asides.
Electronic communication also has the expressive opportunities in word order
that we typically associate with spoken language, and there is lexical and grammati-
cal informality as well (I get called a lot of oder names coz my name is kinda hard
to pronounce so yeah I dont mind wat ever) – all things normally frowned upon in
writing. There are comment clauses and asides (oh by the way just so you know),
interjections (so wats happening with u huh), filled pauses (Umm about me), and
even repetition, false starts and digressions. There is also reduction, and plenty of it.
This kind of writing is full of the omissions, the contractions and the non-standard
spellings that represent the short cuts and assimilations of ordinary rapid speech
(cuzn ‘cousin’, pic ‘picture’, dno ‘don’t know’). Vowels are always less informative
than consonants, so these are typically sacrificed: because becomes cuz, bcuz, bcz
or bcos and be right back becomes brb. Single letters and numerals often replace
syllables or even full words: are becomes r, and ate becomes 8. It becomes compli-
cated when conventions are combined: Andrew becomes &ru. This last convention
is actually quite old, and examples like r ‘are’ are known as “rebuses” (letters or pic-
tures standing for whole words). In fact, Bergs (2009) points out that the linguistic
features of text messaging generally were well and truly around in earlier and more
established forms of communication.4
Of course e-communication does differ from spoken language in one obvious
respect – it lacks the vast repertoire of expressive devices that are available to speak-
ers. But it is developing its own. To some extent, unusual punctuation (including
capitalization, underlining, italics and bolding) and even creative spellings (such as
<soooooooo?!>) go some way to capturing these features. The use of “scare quotes”,
for example, or capital letters can show that a word has a special (non-standard)
sense, or can express the intonation and emphasis of spoken language. Strings of
non-alphabetic symbols such as ?#*! Have I got news!!!!!! is an effective way of avoid-
ing full-blown orthographic obscenity, while still getting the message across. Spe-
cial graphic devices such as emoticons and emoji also add to this written medium a
semantic dimension that places it closer to speech. These smiling, frowning, wink-
ing, crying etc. faces try to communicate something of the same meaning conveyed
by the prosodic and paralinguistic features of speech and make up for the fact that
we tend not to give e-messages the careful wording we might, say, in a snail mail
letter. Typically, we bang down the message, almost with the same speed and spon-
taneity of speech, but of course without the full support of the expressive devices
that speech can utilize. In e-conversations, people don’t observe the same politeness
conventions that go on in usual conversation. They don’t undertake the same time
consuming routines that are so important when it comes to establishing social rap-
port (greetings and leave-takings and so on). Without this social lubricant to oil
the interaction, messages can come across more brusquely and more directly than
originally intended.
Punctuation makes a comeback

Interesting things are happening to punctuation marks, with once neutral
little symbols taking on a whole heap of new significances. No longer aligned
to grammar or rhythm (two functions that have always been intertwined for
English), punctuation has broken free of the literary standard – with new
life now breathed into commas, colons, exclamation marks, question marks,
apostrophes, quotation marks and even full stops.
Not that long ago people were actually predicting the demise of these sym-
bols. In the new world of texting, instant messaging and tweets, fewer punc-
tuation marks means fewer keystrokes – less time and less effort. You might
receive a text:
on way home will call later
Why bother to indicate the start or end of sentences here? Indeed, some
years back, one of us recalls attending a workshop for editors where the mes-
sage was just this – “dump the punc” (was how the presenter put it).
But punctuation marks have far from faded away. Like some Schwar-
zenegger comeback, they are now bolder and braver than ever, no longer the
meek little symbols they once were. When people choose to use a full stop
now, there’s more to it than just ending the sentence. These writers are being
emphatic in some way – usually indicating a displeasure or frustration about
a situation. So, were you to reply to the previous text message (on way home
will call later) with
Thanks.
there’s now a whole lot of meaning lurking behind that little full stop – your
thanks’ becomes ‘thanks for nothing’. It’s punctuation with attitude.
Mark Liberman, a professor of linguistics at the University of Pennsylvania,
commented on this exact topic on Language Log: “Not long ago, my 17-year-
old son noted that many of my texts to him seemed excessively assertive or
even harsh, because I routinely used a period at the end” (http://languagelog.
ldc.upenn.edu/nll/?p=8667)
E-communication formats have been busy resuscitating the careers of
many moribund characters (like @ and _), bestowing upon them new tasks
and new responsibilities. Even ellipsis points are being used creatively. Like
An end on’t 269
“the curious incident of the dog in the night-time”, a succession of full stops
<. . .> no longer simply indicates missing text. And then there’s the inter-
robang <‽>. Brought back from the dead, this now “cool little punctuation
mark” (Urban Dictionary) is much loved by online communication channels –
social media requires all the emotion in the world, but not at the cost of
brevity.
Changes like these aren’t surprising. In this all-digital-all-the-time age
of Facebook, SnapChat, Instagram, chat rooms, web forums, online games,
blogs, tweets and so on, we’re looking at virtual speech communities where
people converse on a regular basis but in writing. And if their members are to
express their identity, personal and shared, they have no choice but to break
out of the straightjacket of the written standard.
We are not sure what societies like the Apostrophe Protection Society
(http://www.apostrophe.org.uk) think of all this. But surely it’s a matter for
celebration‽
So what might the future hold for a language like English in this era of technol-
ogy? Looking back, we have to consider the effects of several generations of televi-
sion viewing; it started with radio, of course, but it was television that really brought
different parts of the English speaking world into the average home. Varieties that
would never have been heard, even in the first half of last century, were suddenly in
everyone’s living room. Speech was being set free, at that time by audio and video
devices (or speech was shearing off from the anchor of writing, to use Bolinger’s
metaphor; 1980: 51).
Now, with the advent of e-communication, we have reached a stage which in
many ways is reminiscent of the situation in medieval times; in other words, before
prescription and before standardization – the time when people wrote as they spoke.
Chatting online is like chatting on the telephone or even face-to-face. Writers don’t
typically go through the sort of drafting processes and layers of editorial intervention
that reinforce the written standard. Instead, we find the sort of far-reaching varia-
tion that existed before standardization and before there was any autonomous prose
style. Regional and social variation is rampant on the net; texts and tweets can reveal
characteristics of accent or dialect in the spelling of words, as in <tawk> for <talk>
(in fact this is a return to a 17th century spelling – the start of l-drop!). There is
even idiolectal variation, where people use punctuation, spellings and abbreviations
to reflect their personality. The “boundless chaos of a living speech” (as Dr Samuel
Johnson put it in the preface to his famous dictionary) has broken through the lines
and now appears in writing, just as it did in the manuscripts of Middle English and
earlier. Getting used to seeing forms like gonna in writing undermines the stan-
dard and must inevitably speed up the acceptance of new grammar, as it will the
acceptance of new phonological forms (<gubment> for government) and new lexical
items (nomophobia ‘fear of being separated from one’s mobile phone’).
Because grammar
As you know, because is a conjunction (created out of the reanalysis of an
original prepositional phrase by + cause). An example of this use might be I
love grammar because it always comes up with surprises. It’s also part of a com-
pound preposition, thus one that occurs in a sequence with another preposi-
tion, as in because of grammar.
But something interesting is happening to because. On the internet there is
now a flourishing of examples where it appears as a preposition without of, as
in I’ve been missing out on sleep because the “Breaking Bad” series. It illustrates
the kind of reduction we’ve come to expect of casual e-speak, but in fact it’s
more interesting than this. Because not only appears before noun phrases
(as a standard preposition would) but also precedes adjective phrases (I’ve
been missing out on sleep because addicted), verb phrases (I missed the ending
because fell asleep) and even interjections (I was horrified because yikes!).
Early in this book we likened the transmission of linguistic forms to the
spread of thought contagions (or memes) – fads that spread from person to
person within a culture. Expressions are particularly infectious and dissemi-
nate rapidly through speech communities, especially virtual ones. Admit-
tedly, constructions like because grammar are still rather faddish creations,
but look out for them – it’s these sorts of jokey constructions that often pro-
vide the basis for real change in the language.
10.4.2 Social change

We’ve seen on many occasions in this book how societal and cultural shifts impact
on language, and of course all the changes we’ve been describing in this section so
far are also set against a backdrop of wide-ranging social change. It started in the
1970s with social and political movements pushing for clear and simple language.
The Plain English movement was a piece of social engineering that attempted to
redress inequalities in our societies – the intended audience was the “average” per-
son struggling with the complexities of law, government, banks and insurance (and
there were parallel cries heard all over Europe; e.g. the push for comprehensible
Swedish in place of byråkratsvenska). So-called political correctness also aimed to
ensure a fair go for all by getting English speakers to focus on the claims of differ-
ent groups. It prescribed and proscribed public language for ethnicity, race, gender,
sexual preference, appearance, religion, (dis)ability and so on. It’s not relative suc-
cess of either of these verbal hygiene operations that is at issue here, but the fact that
they are evidence of a new linguistic thinking.
Growing egalitarianism is now seeing the solidarity function of language gaining
over the status function. Many people are now trying to speak more “down-to-
earth” and with a more obvious stamp of the local. The vernacular has acquired
An end on’t 271
a brand new prestige. Features once condemned as ignorance and corruption are
now viewed with pride. We see this in pronunciation also.
The shift towards everyday spoken language is also very evident in the informal-
ization of television and radio. Perhaps you’ve heard examples of early broadcast-
ing – even the style of sporting commentary comes across as exceedingly formal in
matters of accent, vocabulary and grammar. Different regional and social accents
(and well away from posh end of the accent spectrum) are now commonplace in
these media. Broadcasters in places like Australia and New Zealand, for example,
have moved well away from the BBC-accented English that once dominated the
airways “downunder”; it is telling that when Brian Johns took over the A(ustralian)
B(roadcasting) C(orporation) in 1995, he was heard saying, “We don’t want an out-
dated accent” (by which he would have meant the local “cultivated” accents clos-
est to Received Pronunciation; see Bradley and Bradley 2001: 275). Many other
factors have been contributing to this more personal broadcasting style, including
changing technology (e.g. the introduction of the portable transistor radio in the
late 1950s, and the introduction of smaller and better microphones). Bear in mind
that up until quite recently it was even usual for radio presenters to dress in formal
clothing – a bow tie and dinner suit!
The short of it is, public speaking and writing are becoming progressively more
casual and everyday. We see this even in changes to the terminology that we use: lec-
tures are now more likely to be called talks, and terms like oratory (the art of public
speaking), rhetoric (the art and study of persuasive writing and speaking), elocution
(the art of public speaking where qualities of voice production, gesture and delivery
are emphasized) and recitation (the act of reciting memorized materials in public)
are simply no longer part of most people’s active vocabulary. Even the language we
use in this textbook is itself a good example of this informalization of expression.
The writing is more laid back and very much more personal than anything you will
find in earlier textbooks (at least up until the 1960s, when the changes began). Take,
for instance, the rather chatty way we constantly refer to you, the readers. If refer-
ence was ever made to reader(s) in the past, it was typically done using the third
person. This now comes across as very stilted; for example, “the astute reader [that’s
you!] will doubtless have noticed the familiar style adopted in this present book”.
Here’s an actual example from J.M.D. Meiklejohn’s A Brief History of the English
Language and Literature (published in 1887, but used well into the 1960s):
It is earnestly hoped that the slight sketches of the History of our Language
and of its Literature may not only enable the young student to pass his
examinations with success, but may also throw him into the attitude of
mind of Oliver Twist, and induce him to “ask for more”.
(Preface, p. 7)
To sum up, the processes put in place with the invention of the printing press
eventually fashioned a standard for languages like English. Dictionaries and
grammar books emerged; reading and writing became educational necessities.
All this had the effect of shackling speech to writing and applying the brakes
to the “fleet of juggernaut trucks” (this is Robert Burchfield’s (1985: 173) image
of English, but it works for all living languages) – retarding, and in some cases
even reversing, the normal course of change. But these processes that started with
Gutenberg are now being overturned, and informal, non-standard, unedited Eng-
lish has gone public. And with English now in almost every nook and cranny of
the globe, everything points to more diversity, more variety and more change (see
Crystal 2006).
10.5 WHERE TO FROM HERE?
As we have emphasized throughout this book, change schemas (like the one for nega-
tion outlined earlier) do not follow prescribed courses determined by exceptionless
laws or principles, but it is possible to talk about preferred pathways of change –
those “gutters” that channel language change, to use that image famously invoked
by Kuryłowicz (1945). Referring specifically to analogical change, Kuryłowicz lik-
ened these developments to episodes of rain. While we may not be able to predict
when it will rain, or even if it will, once the rain has fallen, we know the direction
the water will flow because of gutters, drainpipes and spouting. An important goal
of historical linguistics is thus to gain a clearer picture of these “gutters of change”,
in other words, to refine our notions of natural and unnatural change (à la the “con-
straints” problem).
Hence, more work must be done logging and classifying the changes that have
occurred within individual languages, especially those with well-documented
histories. The good news is that rapid advances in technology are making the job
much easier. We may have no time machine to take us back through history, but vast
improvements in corpus design are making available massive digitized collections
of texts that are annotated and searchable, and historical evidence better suited to
the study of language change is now more reliable and more readily available.
As we hope we have shown in this book, linguists are also getting better at the his-
torical dimension of the relationship between language and society; in other words,
at identifying how linguistic features are distributed socially, and how social factors
operate in partnership with linguistic mechanisms of change (à la the “transition”
and “embedding” problems). Trudgill (2011) offers a sociolinguistic typology based
on a huge range of contexts and languages; he reveals the socio-cultural phenom-
ena (e.g. social stability, size, contact, prestige, complexity and relative isolation of
a speech community) that are critically linked to the relative stability/replacement
of linguistic elements, the accelerating/decelerating stimuli for change, and other
issues to do with transmission and diffusion. Whereas tight-knit communities have
long been linked with linguistic stability, Trudgill shows that small speech commu-
nities with tight social networks “are more able, because of their network structures,
to push through, enforce, and sustain linguistic changes which would have a much
smaller chance of success in larger, more fluid communities – namely, changes of a
relatively marked, complex type” (p. 103). The closely integrated Anabaptist speech
community of North America offers robust support of this in the form of rapid
An end on’t 273
grammatical changes in Pennsylvania German (as described in Chapter 8); it is

clear that social context here has an accelerating influence (see Burridge 2015).
As Janda and Joseph (2003) conclude in their wonderfully wide-ranging over-
view of historical linguistics, it surely makes sense for historical linguists to study
language evolution by focusing on changes currently in progress – “building up
an inventory of well-studied present times which, as they accumulate into a store
of well-studied pasts, will slowly but inevitably provide a more solid database for
formulating and testing increasingly sophisticated hypotheses regarding language
change” (p. 128). For many of the world’s languages, this task is an urgent one. The
majority remain undescribed, and many of them are endangered (Native Ameri-
can languages, Australian Aboriginal languages and many Austronesian languages,
to name just a few). When it comes to mining the linguistic diversity around the
world, the discipline is again being greatly assisted by advances in computing and
technology. The developments within computer mediated communication and
improved possibilities for creating and disseminating high-quality recordings of
language have made for vast improvements in corpus design. As described earlier,
many modern corpora now aim to be multimodal – so not simply text based collec-
tions, but high-quality recordings, both audio and video. Ongoing breakthroughs
in speech technology, machine translation and transcription, information extrac-
tion, text mining, voice recognition and voice translation software, recording and
archiving technologies, and software development for (field) linguists, to name just
a few, are making easier the business of collecting, describing and cataloguing the
behaviour of language elements.
At the end of Chapter 6, we invoked the image of Tolkien’s cauldron, a metaphor
he used when he talked about fairy stories. His “Cauldron of Story” has been sim-
mering on the fire for a long, long time, with spicy bits of life continually being
added to the stock. And so it is with the story of language change. The pot continues
to bubble away, always being enriched, refined and enhanced. And with ongoing
contributions from within linguistics and many other fields, we have every reason
to be optimistic about future research in historical linguistics.
FURTHER READING
With the burgeoning interest in historical linguistics has come a flourishing of com-
prehensive textbooks and handbooks with accounts of the individual achievements
of the relatively new discipline areas and their perspectives on change; see indi-
vidual chapters in Joseph and Janda (2003). Bauer (1994) also has a good discussion
of Weinreich et al.’s five key problems, as does J. Milroy (1992); see Bergs (2012) and
Bergs and Pentrel (2015) for discussions of the “uniformitarian hypothesis”. Much
has been written on different perspectives on corpus linguistics: Viana et al. (2011);
the International Computer Archive of Modern and Medieval English (http://clu.uni.
no/icame/) also has an online journal which has reports and articles relating to the
area. Historical sociolinguistics is a growing field, and there is an abundance of read-
ing out there: Nevalainen and Raumolin-Brunberg (2003) and Bergs (2005) give
excellent overviews. Language and the internet is another huge and expanding field:
Crystal (2004); Rowe and Wyss (2009); Androutsopoulos (2014); and Georgopou-
lou and Spilloti (2015). Language@Internet (http://www.languageatinternet.org) is
an open-access electronic journal that publishes research on language and language
use mediated by the internet, the World Wide Web and mobile technologies.
EXERCISES
Write up a research project (approx. 1,500 words) on one of the following topics
(1–5). Think of structuring your report along the lines of the following (though you
don’t have to address every bullet point; it will depend on which project you select):
• Introduction (briefly say what you’re going to say)

• Literature review (provide a brief account of the relevant research to set the
scene)
• Methodology (briefly say what you did and how you did it)
• Results (give an account of what you found)
• Discussion (review the relevance of your results)
• Conclusion (provide a summary of results/findings; sometimes this can include
recommendations – what might now be done on account of what you’ve
uncovered)
• References (you should have at least six references)
• Appendix (not necessary, but you might want to include additional material
such as word lists)
1 A corpus investigation
Go to the Corpus of Historical American English: http://corpus.byu.edu/corpora.asp
and familiarize yourself with the web based user interface. Now search for the word
cool across time and registers. What can you say about the distribution and possible
developments? Try the same for the construction what with, as in “But you know
what with nursing and traveling with the newborn we said you need to take care”.
Again, is there any pattern in the distribution across the different registers? Does
this evolve over time?
2 Profiling a feature across time

Focus on a grammatical feature (such as passive, subjunctive, negation or do-
support) and compare its use diachronically across two different states. You can
be flexible in the time lapse (partly it will depend on the language and on the fea-
ture you select, but it would have to be long enough to view a change take place).
In the case of English, you could select, say, Early Modern English (or any earlier
period of course) through to modern times.
As part of your account, you need at least two sample texts for the two peri-
ods. One idea is to choose a text that has been rewritten many times. You can
find examples online, or there are textbooks that also have such passages (e.g.
An end on’t 275
Burnley 2000; Freeborn 2006). There are also many wonderful digitized texts read-
ily available; e.g. Frederick Furnivall’s (1868) collection from late Middle to Early
Modern English, all dealing with meals and manners (http://quod.lib.umich.edu/
cgi/t/text/text-idx?c=cme;idno=AHA6127.0001.001). Apart from the fascinating
insights these texts give into early etiquette, they are a treasure trove of language
for historical linguists.
3 The actuation of linguistic change

As outlined in this chapter, Weinreich et al. (1968) divided the task of explain-
ing linguistic change into five areas, with the first four aspects (constraints, transi-
tion, embedding and evaluation) contributing to the fifth (the actuation problem),
which they describe as being “the very heart of the matter”. Take some problem of
language change and examine it with respect to the issue of actuation (including
aspects to do with the social setting as well as the linguistic system).
You could take data sets involving changes from one language (e.g. vowel or
consonant shifts, evolution of grammatical categories and forms, lexical/semantic
variation and innovation).
4 W(h)ither the standard

As described in this chapter, recent developments are doing much to challenge
the authority of standard languages. Everything points to greater variety, less stan-
dardization and more changes coming (especially from newer varieties of English).
Comment on these developments and how you see them affecting Standard English
and its relationship to non-standard varieties.
5 Contributions to a theory of language change

Weinreich et al. (1968), Labov (1994) and Milroy (1992) all discuss the fundamen-
tal questions, principles and mechanisms of linguistic change. As described in this
chapter, Weinreich et al. present five different problems for linguistic change: con-
straints, embedding, transition, evaluation and actuation. What do Milroy (1992)
and Labov (1994) say about these five problems?
NOTES
1 The Neogrammarians of the 19th century borrowed this powerful dictum from geol-
ogy and biology (see Janda and Joseph 2003: 27–31). Classicist Hermann Osthoff and
Indo-Europeanist Karl Brugmann expressed it this way in 1878: “the psychological and
physiological nature of man as speaker must have been essentially identical at all epochs”
(Collinge’s translation; 1995: 205); Bergs (2012) offers an extensive discussion of the
principle of uniformity and the risk of anachronism in historical linguistics.
2 The full proceedings of the Old Bailey (1674–1913) are also freely available online:
http://www.oldbaileyonline.org.
3 The Journal of Historical Pragmatics, established in 2000 by Andreas Jucker and Irma
Taavitsainen, explicitly states that its “editorial focus is on socio-historical and prag-
matic aspects of historical texts in their sociocultural context of communication (e.g.
conversational principles, politeness strategies, or speech acts)”; https://benjamins.

com/#catalog/journals/jhp/main.
4 We might also add that the “horse and buggy” communities in North America, featured
in Chapter 8, were using emoticons in their personal letters well before they appeared
in electronic form; recall that the diglossic situation means their experience of English
is within more formal contexts, and presumably these symbols help add the emotional
spice to their writing.
References
Adams, Michael. 2009. Slang: The People’s Poetry. Oxford: Oxford University Press.
Aitchison, Jean. 2003. Words in the Mind. Oxford: Blackwell Publishing.
Aitchison, Jean. 2013. Language Change: Progress or Decay? 4th ed. Cambridge: Cambridge
University Press.
Algeo, John (with the assistance of Adele Algeo) (eds.). 1991. Fifty Years among the New
Words: A Dictionary of Neologisms 1941–1991. New York: Cambridge University Press.
Allan, Keith. 1986. Linguistic Meaning. vols. 1 and 2. London: Routledge & Kegan Paul
[Reprint edn: Beijing: World Publishing Corporation, 1991].
Allan, Keith. 2001. Natural Language Semantics. Oxford and Malden, MA: Blackwell
Publishing.
Allan, Keith & Kate Burridge. 1991. Euphemism and Dysphemism: Language Used as Shield
and Weapon. New York: Oxford University Press.
Allan, Keith & Kate Burridge. 2006. Forbidden Words: Taboo and the Censoring of Language.
Cambridge: Cambridge University Press.
Androutsopoulos, J. (ed.). 2014. Mediatization and Sociolinguistic Change. Berlin: de Gruyter.
Anttila, Raimo. 1989. Comparative and Historical Linguistics. Amsterdam: John Benjamins.
Anttila, Raimo. 2003. Analogy: The warp and woof of cognition. In Joseph & Janda (eds.),
425–440.
Aristotle. 1984. The Complete Works of Aristotle. The revised Oxford translation. Ed. by Jona-
than Barnes. Bollingen Series 71. Princeton: Princeton University Press.
Austin, John L. 1961. A plea for excuses. In James Opie Urmson & Geoffrey J. Warnock (eds),
Philosophical Papers, 175–205. Oxford: Clarendon.
Banfi, Emanuele & Giorgio Francesco Arcodia. 2008. The 生 shēng/sheng complex words in
Chinese between morphology and semantics. In Angela Ralli, Geert Booij, Sergio Scal-
ise, & Athanasios Karasimos (eds.), On-line Proceedings of the Sixth Mediterranean Meet-
ing of Morphology, 194–204 (Available online at http://www.lilec-mmm.it/wp-content/
uploads/2012/10/MMM6_Ithaca_Proceedings.pdf, Accessed on 2016–3–21).
Baptista, Marlyse. 2005. New directions in pidgin and Creole studies. Annual Review of
Anthropology 34: 33–42.
Barðdal, Jóhanna, Elena Smirnova, Lotte Sommerer, & Spike Gildea (eds.). 2015. Diachronic
Construction Grammar. Amsterdam and Philadelphia: John Benjamins.
Bauer, Laurie. 1983. English Word-formation. Cambridge: Cambridge University Press.
Bauer, Laurie. 1992. The second great vowel shift revisited. English World-Wide 13 (2): 253–268.
Bauer, Laurie. 1994. Watching English Change: An Introduction to the Study of Linguistic
Change in the Twentieth Century. London: Longman.
Bergs, Alexander. 2005. Social Networks and Historical Sociolinguistics: Studies in Morpho-
syntactic Variation in the Paston Letters (1421–1503). Berlin: Walter de Gruyter.
Bergs, Alexander. 2009. Just the same old story: The linguistics of text messaging and its
cultural repercussions. In Charley Rowe & Eva L. Wyss (eds.), Language and New Media:
Linguistic, Cultural, and Technological Evolutions, 55–73. New York: Hampton Press.
278 References
Bergs, Alexander. 2012. The Uniformitarian principle and the risk of anachronisms in lan-
guage and social history. In Juan Manuel Hernández-Campoy & Juan Camilo Conde-
Silvestre (eds.), The Handbook of Historical Sociolinguistics, 80–98. Oxford: Blackwell.
Bergs, Alexander & Meike Pentrel. 2015. Ælc þara þe þas min word gehierþ and þa wyrcþ . . . :
Psycholinguistic perspectives on early Englishes. In Michael Adams, Laurel J. Brinton
and Robert D. Fulk (eds.), Studies in the History of the English Language VI. Evidence
and Method in Histories of English (Topics in English Linguistics 85), 249–276. Berlin/
Boston: De Gruyter.
Bickerton, Derek. 1981. Roots of Language. Ann Arbor: Karoma Publishers.
Bickerton, Derek. 1986. Beyond roots: The five-year test. Journal of Pidgin and Creole Lan-
guages 1 (2): 225–232.
Bloomfield, Leonard. 1933. Language. Chicago, IL: The University of Chicago Press.
Bolinger, Dwight. 1980. Language the Loaded Weapon. London: Longman.
Bomhard, Allan R. 2008. Reconstructing Proto-Nostratic: Comparative Phonology, Morphol-
ogy, and Vocabulary. 2 volumes. Leiden: Brill.
Börjars, Kersti & Kate Burridge. 2011. From preposition to purposive to infinitival marker:
The Pennsylvania German fer . . . zu construction. In Michael T. Putnam (ed.), Studies
on German-language Islands (Studies in Language Companion Series 123), 385–411.
Amsterdam: John Benjamins.
Bradley, David & Maya Bradley. 2001. Changing attitudes to Australian English. In David
Blair & Peter Collins (eds.), English in Australia, 271–285. Amsterdam and Philadelphia:
John Benjamins.
Bradley, David & Maya Bradley (eds.). 2002. Language Maintenance for Endangered Lan-
guages: An Active Approach. London: Curzon Press.
Breen, Richard & David B. Rottmann. 2014. Class Stratification. Comparative Perspectives.
London: Routledge.
Burchfield, Robert. 1985. The English Language. Oxford: Oxford University Press.
Burnley, David. 2000. The History of English: A Sourcebook. London: Routledge.
Burridge, Kate. 1993. Syntactic Change in Germanic. Amsterdam: John Benjamins.
Burridge, Kate. 2007. Language contact and convergence in Pennsylvania German. In Alex-
andra Y. Aikhenvald & Richard M. W. Dixon (eds.), Grammars in Contact: A Cross-
linguistic Typology. (Explorations in Linguistic Typology, Volume 4), 179–200. Oxford:
Oxford University Press.
Burridge, Kate. 2013. 19th century study of sound change from Rask to Saussure. In Keith
Allan & Keith Brown (eds.), Oxford Handbook of the History of Linguistics, 344–365.
Oxford: Oxford University Press.
Burridge, Kate. 2015. The body, the universe, society and language: Germanic in the
grip of the unknown. In Rik De Busser & Randy J. LaPolla (eds.), Language Struc-
ture and Environment: Social, Cultural, and Natural Factors, 45–76. Amsterdam: John
Benjamins.
Bybee, Joan. 2003. Mechanisms of change in grammaticization: The role of frequency. In
Brian D. Joseph & Richard D. Janda (eds.), The Handbook of Historical Linguistics, 602–
23. Oxford: Blackwell Publishing.
Bybee, Joan. 2010. Language, Usage and Cognition. Cambridge: Cambridge University Press.
Bynon, Theodora. 1977. Historical Linguistics. Cambridge: Cambridge University Press.
Cameron, Deborah. 2003. Gender issues in language change. Annual Review of Applied Lin-
guistics 23: 187–201.
Campbell, Lyle. 2003. How to show languages are related: Methods for distant genetic rela-
tionship. In Brian D. Joseph & Richard D. Janda (eds.), The Handbook of Historical Lin-
guistics, 19–42. Oxford: Blackwell Publishing.
References 279
Campbell, Lyle. 2004. Historical Linguistics: An Introduction. 2nd ed. Edinburgh: Edinburgh
University Press.
Campbell, Lyle. 2013. Historical Linguistics. Edinburgh: Edinburgh University Press.
Campbell, Lyle & Richard Janda. 2001. Introduction: Conceptions of grammaticalization
and their problems. Language Sciences 23 (2–3): 93–112.
Campbell, Lyle & William J. Poser. 2008. Language Classification: History and Method. Cam-
bridge: Cambridge University Press.
Carstairs-McCarthy, Andrew. 2002. An Introduction to English Morphology. Edinburgh:
Edinburgh University Press.
Clyne, Michael. 2003. Dynamics of Language Contact. Cambridge: Cambridge University Press.
Cockayne, Oswald (ed.). 1865. Leechdoms, Wortcunning, and Starcraft of Early England.
3 volumes. London: Longman, Roberts and Green.
Coleman, Julie. 2012. Life of Slang. Oxford: Oxford University Press.
Collinge, N. E. 1995. History of historical linguistics. In E.F.K. Koerner & R. E. Asher (eds.),
Concise History of the Language Sciences: From the Sumerians to the Cognitivists, 2–212.
Oxford: Pergamon.
Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Oxford: Blackwell
Publishing.
Comrie, Bernard. 2001. Languages of the world. In Mark Aronoff & Janie Rees-Miller (eds.),
The Handbook of Linguistics, 262–282. Oxford: Blackwell Publishing.
Corbett, Greville G. & Wayles Browne. 2009. Serbo-Croat: Bosnian, Croatian, Montenegrin,
Serbian (Chapter 18). In Bernard Comrie (ed.), The World’s Major Languages, 2nd edi-
tion, 330–346. London: Routledge.
Crisma, Paola & Giuseppe Longobardi (eds.). 2009. Historical Syntax and Linguistic Theory.
Croft, William. 1990. Typology and Universals. Cambridge: Cambridge University Press.
Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Per-
spective. Oxford: Oxford University Press.
Crowley, Terry & Claire Bowern. 2010. An Introduction to Historical Linguistics. Oxford:
Oxford University Press.
Cruse, Allan. 2004. Meaning in Language. Oxford: Oxford University Press.
Crystal, David. 2004. Language and the Internet. Cambridge: Cambridge University Press.
Crystal, David. 2006. Into the twenty-first century. In Lynda Mugglestone (ed.), The Oxford
History of English, 394–414. Oxford: Oxford University Press.
Dahl, Östen. 1979. Typology of sentence negation. Linguistics 17: 79–106.
Dahl, Östen. 2010. Typology of negation. In Laurence R. Horn (ed.), The Expression of Nega-
tion, 9–38. Berlin: Mouton.
Davies, Mark. 2011. Interview with Mark Davies. In Vander Viana, Sonia Zyngier, &
Geoff Barnbrook (eds.), Perspectives on Corpus Linguistics, 63–80. Amsterdam: John
Benjamins.
De Busser, Rik & Randy J. LaPolla (eds.). 2015. Language Structure and Environment: Social,
Cultural, and Natural Factors. Amsterdam: John Benjamins.
Del Vallé, José (ed.). 2013. A Political History of Spanish. The Making of a Language. Cam-
Denison, David. 2003. Log(ist)ic and simplistic S-Curves. In Raymond Hickey (ed.), Motives
for Language Change, 54–70. Cambridge: Cambridge University Press.
Deutscher, Guy. 2010. Through the Looking Glass: Why the World Looks Different in other
Languages. New York: Metropolitan Books.
Dixon, Robert M. W. 1997. The Rise and Fall of Languages. Cambridge: Cambridge Univer-
sity Press.
280 References
Domingue, Nicole Z. 1977. Middle English: Another creole?. Journal of Creole Studies 1: 89–108.
Dorian, Nancy. 1994. Purism vs. compromise in language revitalization and language revival.
Language in Society 23: 479–494.
Dryer, Matthew S. 1992. The greenbergian word order correlations. Language 68: 81–138.
Dryer, Matthew S. 1997. On the six-way word order typology. Studies in Language 21: 69–103.
Dryer, Matthew S. 2011. The evidence for word order correlations: A response to Dunn,
Greenhill, Levinson, & Gray’s paper in Nature. Linguistic Typology 15: 335–380.
Dryer, Matthew S. 2013. Prefixing vs. suffixing in inflectional morphology. In Matthew S.
Dryer & Martin Haspelmath (eds.), The World Atlas of Language Structures Online.
Leipzig: Max Planck Institute for Evolutionary Anthropology. (Available online at http://
wals.info/chapter/26, Accessed on 2016–01–29.)
Eckert, Penelope. 1999. Linguistic Variation as Social Practice. Oxford: Blackwell.
Ellegård, Alvar. 1953. The Auxiliary do. The Establsihment and Regulation of its Use in Eng-
lish. Stockholm: Almqvist and Wiksell.
Elspaß, Stephan. 2005. Sprachgeschichte von unten: Untersuchungen zum geschriebenen All-
tagsdeutsch im 19. Jahrhundert. Berlin: Walter de Gruyter.
Emerson, Oliver F. 1921. The History of the English Language. New York: Macmillan.
Enfield, Nicholas. 2015. The Utility of Meaning: What Words Mean and Why. Oxford: Oxford
University Press.
Evans, Nicholas. 2012. Even more diverse than we had thought: The multiplicity of Trans-Fly
languages. In Nicholas Evans & Marian Klamer (eds.), Melanesian Languages on the Edge
of Asia: Challenges for the 21st Century, Language Documentation & Conservation Special
Publication No. 5, 109–149. (Available online at https://scholarspace.manoa.hawaii.edu/
bitstream/10125/4562/1/evans.pdf, Accessed on 2016-08-28.)
Evans, Nicholas & Stephen C. Levinson. 2009. The myth of language universals: Language
diversity and its importance for cognitive science. Behavioral and Brain Sciences 32 (5):
429–448.
Evans, Nick. 2010. Dying Words: Endangered Languages and What They Have to Tell Us.
Chichester: Wiley-Blackwell.
Ferguson, Charles. 1959. Diglossia. Word 15: 325–340.
Fischer, Olga, Anette Rosenbach, & Dieter Stein. 2000. Pathways of Change. Grammaticaliza-
tion in English. Amsterdam and Philadelphia: John Benjamins.
Fox, Anthony. 1995. Linguistic Reconstruction: An Introduction to Theory and Method.
Freeborn, Dennis. 2006. From Old English to Standard English: A Course Book in Language
Variation across Time. 3rd ed. London: Palgrave Macmillan.
Furnivall, Frederick (ed.). 1868. The Babees Book, The Bokes of Nurture of Hugh Rhodes and
John Russel. London: published for the Early English Text Society, Trübner and Co.
Georgopoulou, A. & T. Spilloti (eds.). 2015. The Routledge Handbook of Language and Digital
Communication. London: Routledge.
Ghomeshi, Jila, Ray Jackendoff, Nicole Rosen, & Kevin Russell. 2004. Contrastive reduplica-
tion in English. Natural Language & Linguistic Theory 22: 307–357.
Givón, T. 1971. Historical syntax and synchronic morphology: An archaelogist’s field trip.
Chicago Linguistic Society 7: 394–415.
Gleason, Jean Berko. 1958. The child’s learning of English morphology. Word 14: 150–177.
Gordon, Benjamin L. 1959. Medieval and Renaissance Medicine. New York: Philosophical
Library.
Graddol, David. 2006. English Next: Why Global English May Mean the End of English as a
Foreign Language. London: British Council.
References 281
Greenberg, Joseph H. 1960. A quantitative approach to the morphological typology of lan-

guage. International Journal of American Linguistics 26 (3): 178–194.
Greenberg, Joseph H. 1963. Some universals of grammar with particular reference to the
order of meaningful elements. In Joseph H. Greenberg (ed.), Universals of Language,
73–113. Cambridge: The MIT Press.
Grusky, David. 2014. Social Stratification. 4th ed. Boulder: Westview Press.
Halliday, Michael & Ruqayia Hasan. 1976. Cohesion in English. London: Longman.
Harlow, Ray. 2007. Māori: A Linguistic Introduction. Cambridge: Cambridge University Press.
Harris, Alice C. & Lyle Campbell. 1995. Historical Syntax in Cross-linguistic Perspective.
Cambridge: Cambridge University Press.
Haspelmath, Martin & Andrea D. Sims. 2010. Understanding Morphology. 2nd ed. London:
Routledge.
Haugh, Michael. 2008. Intention in Pragmatics. Berlin and New York: Mouton de Gruyter.
Hawkins, John A. 2002. Symmetries and asymmetries: Their grammar, typology, and pars-
ing. Theoretical Linguistics 28: 95–149.
Hawkins, John A. 2004. Efficiency and Complexity in Grammars. Oxford: Oxford University
Press.
Heine, Bernd. 2003. Grammaticalization. In Brian D. Joseph and Richard D. Janda (eds),
Handbook of Historical Linguistics, 575–601. Oxford: Blackwell Publishing.
Heine, Bernd, Ulrike Claudi, & Friedericke Hünnemeyer. 1991. Grammaticalization. Chi-
cago: University of Chicago Press.
Heine, Bernd & Tania Kuteva. 2003. On contact-induced grammaticalization. Studies in Lan-
guage 27 (3): 529–572.
Hickey, Ray (ed.). 2003. Motives for Language Change. Cambridge: Cambridge University
Press.
Hirt, Hermann. 1937. Indogermanische Grammatik Syntax 11. Die Lehre vom einfachen un
zusammengesetzten Satz. Heidelberg: Carl Winter.
Hitt, Jack. 2012. Words on Trial: Can linguists solve crimes that stump police. New Yorker,
July 23, 2012.
Hock, Hans Henrich. 1991. Principles of Historical Linguistics. 2nd rev. Berlin: Mouton de
Gruyter.
Hock, Hans Henrich & Brian D. Joseph. 1996. Language History, Language Change, and Lan-
guage Relationship: An Introduction to Historical and Comparative Linguistics. Berlin and
New York: Mouton de Gruyter.
Hofstadter, Douglas R. 1979. Gödel, Escher, Bach. New York: Basic Books.
Holzknecht, Susanne. 1988. Word taboo and its implications for language change in
the Markham family of languages, PNG. Language and Linguistics in Melanesia 18:
43–69.
Hopper, Paul J. & Elizabeth C. Traugott. 2003. Grammaticalization. 2nd ed. Cambridge:
Cambridge University Press.
Hughes, Geoffrey. 1989. Words in Time: A Social History of English Words. Oxford: Blackwell
Publishing.
Hyman, Larry. 1975. On the change from SOV to SVO: Evidence from Niger-Congo. In
Charles N. Li (ed.), Word Order and Word Order Change, 113–147. Austin, TX: Univer-
sity of Texas Press.
Janda, Richard D. & Brian D. Joseph. 2003. On language, change and language change – or,
of history, linguistics and historical linguistics. In Brian D. Joseph & Richard D. Janda
(eds.), Handbook of Historical Linguistics, 3–180. Oxford: Blackwell Publishing.
Jespersen, Otto. 1917. Negation in English and other Languages. København: Munksgaard.
282 References
Jespersen, Otto. 1922. Language: Its Nature, Development and Origin. London: George
Allen & Unwin Ltd.
Jonson, Ben. 1640. The English Grammar. New York: Sturgis & Walton Company.
Joseph, Brian D. & Richard D. Janda (eds.). 2005. The Handbook of Historical Linguistics.
Oxford: Blackwell Publishing.
Karaman, Burcu I. 2008. On contronymy. International Journal of Lexicography 21 (2): 173–192.
Katamba, Francis. 1994. English Words. London: Routledge.
Keesing, Roger M. & Jonathon FifiɁi. 1969. Kwaio word tabooing in its cultural context.
Journal of Polynesian Society 78: 154–177.
Keller, Rudi. 1994. On Language Change: The Invisible Hand in Language. London: Routledge.
Keller, Rudi. 2003. Sprachwandel. 3. Auflage. Tübingen und Basel: Francke.
Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The Electronic World Atlas of Variet-
ies of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. (Available
online at http://ewave-atlas.org, Accessed on 2014–05–31.)
Kuryłowicz, Jerzy. 1945. La nature des procès dits ‘analologiques’. Acta Linguistica 5: 15–37.
Kwon, Nahyun & Erich R. Round. 2015. Phonaesthemes in morphological theory. Morphol-
ogy 25 (1): 1–27.
Laberge, Suzanne & Gillian Sankoff. 1980. Anything you can do. In Gillian Sankoff (ed.), The
Social Life of Language, 271–294. Philadelphia: University of Philadelphia Press.
Labov, William. 1972. Sociolinguistic Patterns. Pennsylvania: University of Pennsylvania Press.
Labov, William. 1982. Building on empirical foundations. In Winfred Lehmann & Yakov
Malkiel (eds.), Perspectives on Historical Linguistics, 17–82. Amsterdam: John Benjamins.
Labov, William. 1994. Principles of Linguistic Change: Internal Factors. Oxford: Blackwell.
Labov, William. 2001. Principles of Linguistic Change: Social Factors. Oxford: Blackwell.
Labov, William. 2006. The Social Stratification of English in New York City. Cambridge: Cam-
bridge University Press.
Labov, William. 2010. Principles of Linguistic Change: Cognitive and Cultural Factors. Oxford:
Blackwell.
Lakoff, George & Mark Johnson. 1981. Metaphors We Live By. Chicago: University of Chi-
cago Press.
Langacker, Ronald. 1977. Syntactic reanalysis. In Charles N. Li (ed.), Mechanisms of Syntactic
Change, 59–139. Austin: University of Texas Press.
LaPolla, Randy J. 1994. An experimental investigation into phonetic symbolism as it relates
to Mandarin Chinese. In Leanne Hinton, Johanna Nichols, & John J. Ohala (eds.), Sound
Symbolism, 130–147. Cambridge: Cambridge University Press.
Ledgeway, Adam. 2012. From Latin to Romance. Morphosyntactic Typology and Change.
Leech, Geoffrey. 1981. Semantics: The Study of Meaning. 2nd ed. Harmondsworth: Penguin
Books.
Lehmann, Winfred P. 1973. A structural principle of language and its implications. Language
49: 47–66.
Levine, Rhonda (ed.). 2006. Social Class and Stratification. Classic Statements and Theoretical
Debates. Lanham: Rowman & Littlefield.
Liberman, Anatoly. 2005. Word Origins . . . and How We Know Them. Oxford: Oxford Uni-
versity Press.
Lieber, Rochelle & Pavol Štekauer (eds.). 2011. The Oxford Handbook of Compounding.
Lieber, Rochelle & Pavol Štekauer (eds.). 2014. The Oxford Handbook of Derivational Mor-
phology. Oxford: Oxford University Press.
References 283
Lightfoot, David. 1998. The Development of Language: Acquisition, Change and Evolution.
Oxford: Blackwell.
Lindblom, Björn. 1983. Economy of speech gestures. In Peter F. MacNeilage (ed.), The Pro-
duction of Speech, 217–245. New York: Springer-Verlag.
Lindblom, Björn. 1990. Explaining phonetic variation: A sketch of the H&H theory. In
William J. Hardcastle & Alain Marchal (eds.), Anonymous Speech Production and Speech
Modelling, 403–439. Dordrecht: Kluwer Academic Publishers.
Lockwood, William B. 1968. Historical German Syntax. Oxford: Clarendon Press.
Loebner, Sebastian. 2013. Understanding Semantics. London: Routledge.
Lutzeier, Peter. 2007. Wörterbuch des Gegensinns im Deutschen. Berlin and New York: Walter
de Gruyter.
Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University
Press.
Mailhammer, Robert. 2014. Etymology. In Claire Bowern & Bethwyn Evans (eds.), The Rout-
ledge Handbook of Historical Linguistics, 423–442. London: Routledge.
Mallinson, Graham & Barry J. Blake. 1981. Language Typology: Cross-linguistic Studies in
Syntax. Amsterdam: North-Holland.
Marmion, Doug, Kazuko Obata, & Jakelin Troy. 2014. Community, identity, wellbeing: The
report of the second national indigenous languages survey. Canberra: Australia: Australian
Institute of Aboriginal and Torres Strait Islander Studies (AIATSIS). (Available online
at http://aiatsis.gov.au/publications/products/community-identity-wellbeing-report-
second-national-indigenous-languages-survey; Accessed 2016-08-28.)
Martinet, André. 1955. Economie des changements phonétiques: traite de phonologie diachronique.
Berne: Francke.
Matras, Yaron. 1998. Utterance modifiers and universals of grammatical borrowing. Linguis-
tics, 16 (2): 281–332.
Matras, Yaron. 2000. Mixed languages: A function-communicative approach. Bilingualism:
Language and Cognition 3 (2): 79–99.
Matras, Yaron & Peter Bakker. 2003. The Mixed Language Debate: Theoretical and Empirical
Advances (Trends in Linguistics Series Volume 145). Berlin: Walter de Gruyter.
McColl Millar, Robert. 2015. Trask’s Historical Linguistics. 3rd ed. Abingdon: Routledge.
McConvell, Patrick & Felicity Meakins. 2005. Gurindji Kriol: A mixed language emerges
from code-switching. Australian Journal of Linguistics 25 (1): 9–30.
McLaughlin, Barry. 1984. Second-language Acquisition in Childhood: Preschool Children.
Mahwah, NJ: Lawrence Erlbaum Associates.
McWhorter, John H. 2003. Pidgins and Creoles as models of language change: The state of
the art. Annual Review of Applied Linguistics 23: 202–212.
Meakins, Felicity. 2013a. Gurindji Kriol. In Susanne Michaelis, Philippe Maurer, Martin
Haspelmath, & Magnus Huber (eds.), Contact languages based on languages from
Africa, Asia, Australia, and the Americas, 131–139. Oxford: Oxford University Press.
Meakins, Felicity. 2013b. Mixed languages. In Peter Bakker & Yaron Matras (eds.), Contact
Languages: A Comprehensive Guide, 159–228. Berlin and Germany: De Gruyter Mouton.
Meakins, Felicity & Carmel O’Shannessy. 2012. Typological constraints on verb integration
in two Australian mixed languages. Journal of Language Contact 5: 216–246.
Michaelis, Susanne Maria, Philippe Maurer, Martin Haspelmath, & Magnus Huber (eds.).
2013. Atlas of Pidgin and Creole Language Structures Online. Leipzig: Max Planck
Institute for Evolutionary Anthropology (Available online at http://apics-online.info,
Accessed on 2016–02–08.)
Miller, D. Gary. 1975. Indo-European: VSO, SOV, SVO or all three?. Lingua 37: 31–52.
284 References
Miller, D. Gary. 2010. Language Change and Linguistic Theory. Vol. 2. Oxford: Oxford Uni-
versity Press.
Milroy, James. 1992. Linguistic Variation and Change: On the Historical Sociolinguistics of
English. Oxford: Blackwell Publishing.
Milroy, James. 1993. On the social origins of language change. In Charles Jones (ed.), Histori-
cal Linguistics: Problems and Perspectives, 215–236. London and New York: Longman.
Milroy, James & Lesley Milroy. 1999. Authority in Language. Investigating Standard English.
3rd ed. London and New York: Routledge.
Milroy, Lesley. 1987. Language and Social Networks. New York: Blackwell Publishing. 2nd ed.
Oxford: Basil Blackwell.
Mithun, Marianne. 2001. The Languages of Native North America. Cambridge: Cambridge
University Press.
Modaressi, Yahya. 1978. A Sociolinguistic Analysis of Modern Persian. Kansas: University of
Kansas Press.
Müller, Peter O., Ingeborg Ohnheiser, Susan Olsen, & Franz Rainer. 2015–16. Word-Formation:
An International Handbook of the Languages of Europe. 5 vols. Berlin: Mouton de Gruyter.
Musgrave, Simon & Kate Burridge. 2014. Bastards and Buggers – historical snapshots of
Australian English swearing patterns. In Kate Burridge & Réka Benczes (eds.), Wrestling
with Words and Meanings: Essays in Honour of Keith Allan, 3–32. Melbourne: Monash
University Publishing.
Myers-Scotton, Carol. 2002. Contact Linguistics: Bilingual Encounters and Grammatical Out-
comes. Oxford: Oxford University Press.
Myers-Scotton, Carol & John L. Jake. 2009. Which language? Participation potentials across
lexical categories in codeswitching. In Ludmila Isurin, Donald Winford, & Kees de Bot
(eds.), Interdisciplinary Approaches to Code-Switching, 206–40. Amsterdam: Benjamins.
Nadkarni, Mangesh V. 1975. Bilingualism and syntactic change in Konkani. Language 51:
172–186.
Naro, Anthony J. & Miriam Lemle. 1976. Syntactic diffusion. In Sanford B. Steever, Carol A.
Walker, & Salikoko S. Mufwene (eds.), Chicago Linguistics Society 12 (2), Parasession on
Diachronic Syntax, 221–240. Chicago: Chicago Linguistics Society.
Nettle, Daniel & Suzanne Romaine. 2000. Vanishing Voices: The Extinction of the World’s Lan-
guages. Oxford: Oxford University Press.
Nevalainen, Terttu. 1998. Social mobility and the decline of multiple negation in early mod-
ern English. In Jacek Fisiak & Marcin Krygier (eds.), Advance in English Historical Lin-
guistics (1996), 263–291. Berlin: Mouton de Gruyter.
Nevalainen, Terttu & Helena Raumolin-Brunberg (eds.). 1996. Sociolinguistics and Language
History: Studies Based on the Corpus of Early English Correspondence. No. 15. Amster-
dam: Rodopi.
Nevalainen, Terttu & Helena Raumolin-Brunberg. 2003. Socio-Historical Linguistics: Lan-
guage Change in Tudor and Stuart England. London: Longman.
Nichols, Johanna. 2003. Diversity and stability in language. In Brian D. Joseph & Richard D.
Janda (eds.), Handbook of Historical Linguistics, 283–310. Oxford: Blackwell Publishing.
O’Connor, Joseph Desmond. 1973. Phonetics. Harmondsworth: Penguin.
Ogura, Mieko & William S-Y. Wang. 1996. Snowball effect in lexical diffusion. The devel-
opment of – s in the third person singular present indicative in English. In Derek
Britton (ed.), English Historical Linguistics 1994. Papers from the 8th Interational Con-
gress on English Historical Linguistics, 119–141. Amsterdam and Philadelphia: John
Benjamins.
Ohala, John J. & M. Grazia Busà. 1995. Nasal loss before voiceless fricatives: A perceptually-
based sound change. Rivista di Linguistica 7: 125–144.
References 285
O’Shannessy, Carmel. 2005. Light Warlpiri: A new language. Australian Journal of Linguistics
25 (1): 31–57.
O’Shannessy, Carmel. 2015. Typological and social factors influencing a new mixed language,
Light Warlpiri. In Gerald Stell & Kofi Yakpo (eds.), Code-Switching Between Structural
and Sociolinguistic Perspectives, 289–304. Berlin: De Gruyter.
Partridge, Eric. 1961. Adventuring Among Words. London: Andre Deutsch.
Paul, Hermann. 1959. Mittelhochdeutsche Grammatik. Tübingen: Max Niemeyer.
Peters, Pam. 2000. Paradigm split. Language and Computers 33: 301–312.
Pinker, Steven. 2000. The Language Instinct: How the Mind Creates Language. New York:
HarperCollins.
Pope, Jennifer, Miriam Meyerhoff, & D. Robert Ladd. 2007. Forty years of language change
on Martha’s Vineyard. Language 83 (3): 615–627.
Rayfield, J. R. 1970. The Language of a Bilingual Community. The Hague: Mouton.
Ringe, Don & Joseph F. Eska. 2013. Historical Linguistics. Toward a Twenty-First Century
Reintegration. Cambridge: Cambridge University Press.
Romaine, Suzanne. 1982. Socio-historical Linguistics. Its Status and Methodology. Cambridge:
Cambridge University Press.
Rowe, Charley & Eva L. Wyss (eds.). 2009. Language and New Media: Linguistic, Cultural, and
Technological Evolutions. New York: Hampton Press.
Ruhlen, Merritt. 1994. The Origin of Language: Tracing the Evolution of the Mother Tongue.
New York: Wiley.
Sampson, Rodney. 1999. Nasal Vowel Evolution in Romance. Oxford: Oxford University Press.
Sankoff, Gillian. 2005. Age: Apparent time and real time. In Keith Brown (ed.), Elsevier Ency-
clopedia of Language and Linguistics, 110–116. Amsterdam: Elsevier.
Sapir, Edward. 1921. Language: An Introduction to the Study of Speech. New York: Harcourt,
Brace and Company.
Scheler, Manfred. 1977. Der Englische Wortschatz. Berlin: Erich Schmidt Verlag.
Schmidt, Johannes. 1872. Die verwantschaftsverhältnisse der indogermanischen Sprachen.
Weimar: Böhlau.
Schmidt, Julian. 1856. Geschichte der deutschen Literatur im neunzehnten Jahrhundert: Die
Gegenwart. Leipzig: F. L. Herbig.
Schuchardt, Hugo. 1868. Slipe, slape, snorio, basilorio (zu zeitschr. XIV, 397–399). Zeitschrift
für vergleichende Sprachforschung auf dem Gebiete des Deutschen, Griechischen und
Lateinischen 17 (5): 396–400.
Schuchardt, Hugo. 1870. Über einige Fälle bedingten Lautwandels im Churwälschen. Gotha
(Schuchardts Hibilitationsschrift).
Schuchardt, Hugo. 1885. Ueber die Lautgesetze: gegen die Junggrammatiker. Berlin: Oppenheim.
Selting, Margret. 1999. Kontinuität und Wandel der Verbstellung von ahd. wanta bis gwd.
weil. Zur historischen und vergleichenden Syntax der weil-Konstruktionen. Zeitschrift
für germanistische Linguistik 27: 167–204.
Serjeantson, Mary S. 1936. A History of Foreign Words in English. New York: E. P. Dutton &
Company.
Sibielski, Rosalind. 2010. “What Are Little (Empowered) Girls Made Of?: The Discourse of
Girl Power in Contemporary U.S. Popular Culture”, Ph.D. Dissertation. Bowling Green
State University.
Siegel, Jeff. 2008. The Emergence of Pidgin and Creole Languages. Oxford: Oxford University
Press.
Simons, Gary F. 1982. Word taboo and comparative Austronesian linguistics. In Amran
Halim, Lois Carrington, & Stephen A. Wurm (eds.), Papers from the Third International
Conference on Austronesian Linguistics, Vol. 3, 157–226. Canberra: Pacific Linguistics.
286 References
Simpson, Adrian P. 2009. Phonetic differences between male and female speech. Language
and Linguistics Compass 3(2): 621–640.
Singler, John V. & Silvia Kouwenberg (eds.). 2008. The Handbook of Pidgin and Creole Stud-
ies. London: Blackwell.
Štekauer, Pavol & Rochelle Lieber (eds.). 2006. The Handbook of Word-formation. Dordrecht:
Springer.
Stockwell, Robert & Donka Minkova. 2001. English Words: History and Structure. Cam-
Sturtevant, Edgar H. 1942. Linguistic Change: An Introduction to the Historical Study of Lan-
guage. New York: Stechert.
Svartvik, Jan & Geoffrey Leech. 2006. English – One Tongue, Many Voices. New York: Palgrave.
Tagliamonte, Sali A. 2012. Variationist Sociolinguistics: Change, Observation, Interpretation.
Malden and Oxford: Wiley-Blackwell, 128–156.
Thibault, Pierette. 1991. La langue en mouvement: Simplification, régularisation, restruc-
turation. LINX 25: 79–92.
Thomason, Sarah Grey. 1997. Contact Languages: A Wider Perspective. Vol. 17 of Creole Lan-
guage Library. Amsterdam: John Benjamins Publishing.
Thomason, Sarah Grey, & Terrence Kaufman. 1991. Language Contact, Creolization, and
Genetic Linguistics. California: University of California Press.
Tomić, Olga M. 2006. Balkan Sprachbund Morpho-Syntactic Features. Dordrecht: Springer.
Traugott, Elizabeth C. 1982. From propositional to textual and expressive meanings: Some
semanticpragmatic aspects of grammaticalization. In Winfred P. Lehmann & Yakov
Malkiel (eds.), Perspectives on Historical Linguistics, 245–271. (Current Issues in Lin-
guistic Theory 24.) Amsterdam and Philadelphia: Benjamins.
Traugott, Elizabeth C. 1985. On regularity in semantic change. Journal of Literary Semantics
14 (3): 155–173.
Traugott, Elizabeth C. 2003. From subjectification to intersubjectification. In Raymond Hickey
(ed.), Motives for Language Change, 124–139. Cambridge: Cambridge University Press.
Traugott, Elizabeth C. & Richard B. Dasher. 2002. Regularity in Semantic Change. Cam-
Traugott, Elizabeth C. & Graeme Trousdale. 2014. Constructionalization and Constructional
Changes. Oxford: Oxford University Press.
Trotter, David. 2012. Middle English creolization. In Alexander Bergs & Laurel J. Brinton
(eds.), Historical Linguistics of English: An International Handbook, 1781–1792. Berlin:
Walter de Gruyter.
Trudgill, Peter. 1974. The Social Differentiation of English in Norwich. Cambridge: Cambridge
University Press.
Trudgill, Peter. 2011. Sociolinguistic Typology: Social Determinants of Linguistic Complexity.
Twain, Mark. 1880. The Awful German Language (published as an appendix in the second
volume of Twain’s A Tramp Abroad). Toronto: Robertson.
Van der Auwera, Johan. 2010. On the diachrony of negation. In Laurence R. Horn (ed.), The
Expression of Negation, 73–110. Berlin: Mouton.
Velupillai, Viveka. 2015. Pidgins, Creoles and Mixed Languages: An Introduction. Amsterdam:
John Benjamins.
Vennemann, Theo. 1974. Topics, subjects and word order: From SXV to SVX via TVX. In
John Anderson & Charles Jones (eds.), Historical Linguistics 1, 339–376. Amsterdam:
North-Holland.
Viana, Vander, Sonia Zyngier, & Geoff Barnbrook. 2011. Perspectives on Corpus Linguistics.
Amsterdam: John Benjamins.
References 287
Von Polenz, Peter. 1999. Deutsche Sprachgeschichte vom Spätmittelalter bis zur Gegenwart.
Bd. III: 19. Und 20. Jahrhundert. Berlin and New York: de Gruyter.
Wagner, Suzanne Evans. 2012. Age grading in sociolinguistic theory. Language and Linguis-
tics Compass 6 (6): 371–382.
Watts, Richard J. 2011. Language Myths and the History of English. Oxford: Oxford University
Press.
Watts, Richard J. 2012. Interdisciplinarity and historiography: Myths of the English lan-
guage; or, alternative histories of “English”. In Alexander Bergs & Laurel Brinton (eds.),
Handbook of English Historical Linguistics. Vol. II, 1267–1273. Berlin and New York:
de Gruyter.
Weinreich, Uriel, William Labov, & Marvin I. Herzog. 1968. Empirical foundations for a
theory of language change. In Winfred Lehmann & Yakov Malkiel (eds.), Directions for
Historical Linguistics, 95–188. Austin: University of Texas Press.
Willis, David, Christopher Lucas, & Anne Breitbarth. 2013. Comparing diachronies of nega-
tion. In David Willis, Christopher Lucas, & Anne Breitbarth (eds.), The History of Nega-
tion in the Languages of Europe and the Mediterranean: Volume I Case Studies, 1–50.
Winford, Donald. 2003. An Introduction to Contact Linguistics. Oxford: Blackwell Publishing.
Wolf, Clara & Elena Jiménez. 1979. El ensordecimiento del yeísmo porteño. In Ana María
Barrenechea (ed.), Estudios lingüísticos y dialectológicos, 115–145. Buenos Aires: Hachette.
Zipf, George K. 1949. Human Behavior and the Principle of Least Effort. New York: Hafner
Publishing Company.
Zuckermann, G. & M. Walsh. 2011. Stop, revive, survive: Lessons from the Hebrew Revival
applicable to the reclamation, maintenance and empowerment of Aboriginal languages
and cultures. Australian Journal of Linguistics 31 (1): 111–127.
Index
abbreviation 9, 35 – 6, 46, 47, 122, 126, American English 11, 39, 63 – 4, 117, 216
167, 269 American Speech 31
ABC (Australian Broadcasting Amharic 144
Corporation) 66, 271 analogical levelling, 112, 113
abduction 15, 110 analogical thinking 110 – 11, 130
Aboriginal languages 197, 200, 207 – 8, analogy 14 – 15, 24, 42 – 3, 64, 95 – 7,
212 – 13, 273 109 – 20, 128 – 30, 146, 149, 150, 154,
accent 14 – 17, 75, 90 – 1, 97, 194, 197, 209, 156 – 7, 243, 254; changes 95, 113 – 15,
264, 269, 271 118, 120, 128 – 30, 272; four-term 110;
accommodation 14, 17 – 18, 147, 222, 247 rules of 114 – 20
acquisition of language 15, 24, 110, 210 Ancient Greek 7, 55
acronym 36 – 7, 46 – 7 Anglo Saxon Chronicle 1, 3, 5
actualization 106 – 9, 130 Anttila, Raimo 56 – 7, 90, 109, 244
actuation 187, 248, 252 – 3, 275 aphesis 79
affinity between languages 215 – 43; apocope 79
common ancestor 229 – 33, 237 – 9; Apostrophe Protection Society 269
comparative method for determining apparent time 18 – 19, 21, 24, 176 – 9; see
217 – 19, 222, 237 – 43; family tree model also real time
220 – 2, 223, 231, 238; methods for Arcodia, Giorgio Francesco 126
reconstruction of relationship 233 – 43; Aristotle 64, 70
quantitative approaches to divergence assimilation 40, 81 – 4, 94 – 5, 113, 218,
determination 226 – 8, 243; wave model 266 – 7; anticipatory 82 – 2; distance 83,
168, 169, 222 – 6, 243; written sources as 84, 95, 113, 218, 266
evidence of change 234 – 7 Atlas of Pidgin and Creole Structures 16,
affixation 8, 31, 33 – 5, 42, 47, 50, 79, 106, 25, 212
118, 120, 127, 134, 154, 191 attitudes towards language change 21 – 3
age grading 19, 21, 25, 175 – 9 Auer, Anita 262
agglutinating languages 124, 125 – 6, 127, Australia 16, 34, 52, 92, 102, 158, 189,
130, 132, 134, 142, 143 193, 196 – 7, 200 – 1, 207, 213, 229, 231,
Aitchison, Jean 70, 101, 164 263, 271; Aboriginal languages 197,
Albanian 202, 223, 229 200, 207 – 8; Australian English 54,
Algeo, John 31, 35, 37, 50 100; contact languages in 197, 200 – 1;
allophones 6, 90, 92, 93 language death in 16, 207 – 8; mixed
alternative histories 263 – 4 language 190, 195, 198 – 201, 211 – 12;
amalgamation 43 pidgins in, 197, 200 – 1
amelioration 55 – 6, 70, 72 Australian English 54, 100
American Dialect Society 31, 100 Australian National Corpus 255
Index 289
backformation 34 – 5, 48, 50, 116 Castilian Spanish 263

backronyms 37 Catalan 207, 221 – 2, 263 – 4
Balkan Sprachbund 202 Caxton, William 25 – 6, 265
Bandwurmwörter ‘tapeworm words’ 32 Celtic 98, 143, 193, 207, 242
Banfi, Emanuele 126 chain shifts 99 – 100
Bantu 201 Chaucer, Geoffrey 1 – 2, 5, 11, 58, 75,
Basque 118, 119, 242 78 – 9, 90, 99, 151, 198, 265; Canterbury
Bauer, Laurie 9, 31, 48, 100 – 1, 142, 273 Tales, The 1 – 2, 75 – 6, 79; Troilus and
Bavarian German 216 Criseyde II 5, 265
Bell, Stephanie 47 Cheke, Sir John 22
Beowulf 76, 236 Chinese 123, 125 – 6, 207, 214, 222, 263;
Bergs, Alexander 183, 184, 187, 262, see also Mandarin Chinese
267, 273 Chomskyan linguistics 134, 230
Bickerton, Derek 196, 212 class, social 171, 172 – 5, 180, 182 – 3,
bilingualism 98, 190, 193, 195, 198 – 9, 185 – 6, 251
200 – 1, 203, 205 – 6, 211 Classical Egyptian 126
bipartite marker 114 – 15 Classical Latin 207, 222
Bislama 158, 196 clipping 35, 63 – 4, 106
Black Death 100, 162 , 251 clitics 154
blending 31, 37 – 8, 42 – 3, 47, 50, 61, code-switching 198 – 201, 209, 212
134, 164 cognates 107, 218, 219, 227 – 8, 233,
Bloomfield, Leonard 169, 244 238 – 9, 242, 244, 246
Bolinger, Dwight 265, 269 cognitive linguistics 66, 247
borrowing 10, 12, 31, 39 – 41, 48, 50, 52, commonization 38
55, 61, 64, 68 – 9, 80, 83, 94, 96 – 8, 100, communities of practice 185 – 6, 187 – 8
107, 109, 141, 190, 191 – 3, 203, 205 – 7, compensatory lengthening 78 – 9, 219
209, 212, 215, 217 – 19, 228, 239, complaint tradition 5, 21, 22, 24, 25 – 6
244, 253 complementary distribution 5 – 6, 168
Brabantish 253 Complete Corpus of Old English/
Brief History of the English Language and Dictionary of Old English 236
Literature, A (Mieklejohn) 271 compounding 31 – 3, 38 – 40, 42 – 3, 47, 50,
British English 22, 90, 235 106, 126, 137, 140, 236, 241, 270
broadening, semantic 38, 54, 57, 65, consonants 6 – 7, 36, 73, 76 – 9, 80 – 90,
69 – 70, 72, 74 93 – 4, 96 – 9, 104 – 5, 107 – 9, 111,
Bubonic Plague see Black Death 116 – 17, 119, 149, 153, 163 – 4, 168 – 9,
Bulgarian 202, 223 194, 203, 205, 218, 221, 226, 230, 238,
Burgundian 65, 236 240, 246, 267, 275; addition of 80 – 1;
Burridge, Kate 70, 101, 157, 204, 212, aspirated 7, 88; assimilation of 81 – 2,
251, 253, 255, 273 83; devoicing of 86; dissimilation of
88; English, evolution of 98 – 9; First
Campbell, Lyle 24, 101, 151, 157, 161, Germanic Consonant Shift 163;
228, 233, 244 French 149; High German Consonant
Canadian English 176, 202 Shift 168, 226; loss of 77 – 8, 109, 238;
Canterbury Tales, The (Chaucer) 1 – 2, nasalized 6
75 – 6, 79 constraints problem 248, 249, 272, 275
cascade effect 170 – 1, 186 construction grammar 134
290 Index
contact between different languages 15, diglossia 203, 206

98, 189 – 211; convergence 201 – 2; discourse marker 12, 70, 206, 260
discourse particles 206; language discourse particles 206
creation 190, 195 – 201; language dissimilation 7, 81, 87 – 8, 94
death 207 – 9; language maintenance dissociation 14, 17 – 18, 261
190 – 2; language shift 190, 192 – 4; Dixon, Robert 126, 131
substratum influence 194, 203, 204 – 6; do, periphrastic 20, 167
superstratum influence 203 – 4; types drift 251
of 190 – 202; see also creole languages; Dryer, Matthew 127, 143, 146 – 7,
pidgins 148
contamination 61 – 3, 218 Duden 21
contrastive focus reduplication 39 Duden, Konrad 21
contronyms 59, 70 Dutch 15, 50, 141, 163, 214, 216 – 17, 221,
convergence 201 – 2, 207, 210 – 11 223 – 4, 226 – 7, 229, 244, 253
conversion 35, 38, 48, 55, 100, 180, 190, Dyirbal 147
192 – 4, 205, 228, 262, 269
Corpus of Contemporary American Early English Books Online 236
English (COCA) 17, 236, 257 Early Modern English 8, 11, 104, 160 – 1,
Corpus of Early English 164, 166 – 7, 197, 198, 236, 237, 248,
Correspondence 254 254, 262, 274 – 5; pronouns in 11;
corpus linguistics 254 – 6, 273 verbal paradigm for 8, 166
Corpus of Oz Early English (CoOEE) 255 Earp, T. W. 62
Corpus Resource Database 254 – 5 Eastman, George 29
Costa Rican Creole 196 Eckert, Penelope 186 – 7
creation of language 29, 190, 195 – 201 e-communication (electronic
creole languages 16, 25, 80, 195 – 7, communication) 100, 250, 266 – 9
198 – 9, 207, 212 economy of gesture 77, 81, 86, 102
Cushitic 144, 201 Ellegård, Alvar 20, 167, 262
ellipsis 63 – 4, 70, 147, 268
Dahl, Östen 156, 249 embedding problem 2248, 251 – 2, 266,
Darwin, Charles 13 272, 275
Davies, Mark 255 “Empirical Foundations for a Theory of
death, language 16, 193, 207 – 9 Language Change” (Weinreich, Labov,
deterioration 54, 55 – 6, 60, 69 – 70, 72 Herzog) 248
diachrony 20, 77, 102 – 3, 105, 148, 157, Eneydos (Caxton) 25, 265
188, 248 – 9, 254 – 5, 275 Enfield, Nick 53 – 4, 66, 69 – 70
dialects 14 – 15, 16, 31, 40, 45, 87, 90, English 1 – 15, 25 – 7, 30 – 50, 54 – 5, 61, 63,
93, 100, 140, 170, 175, 207, 216 – 17, 67 – 8, 76 – 95, 98 – 110, 113 – 19, 121 – 7,
221 – 2, 224, 226, 228, 235, 253 – 5, 261, 131 – 8, 140 – 50, 151 – 5, 167, 176, 197,
263 – 5, 269 198, 202, 203, 204 – 7, 236, 237, 259,
dictionaries 7, 9, 16, 21, 26, 28 – 32, 35 – 6, 261; American English 11, 39, 63 – 4,
38, 40, 43, 46, 48 – 50, 54, 59, 61, 72, 117, 216; Australian English 54, 100;
74, 85, 106, 112, 115, 187, 209, 236, British English 22, 90, 235; Canadian
269, 271 English 176; New Zealand English
diffusion of change see spread of change 100; Philippine English 40; Scots
Index 291
English 90, 261; see also Early Modern from 40; compounding in 32 – 3, 35;
English; Middle English; Old English final consonant excrescence in 81;
epenthesis 80 First Germanic Consonant Shift 163;
etymology 29, 42, 46 – 8, 50, 63, 70, 73 – 4, High German 168, 225 – 6; inflections
95, 233 in 124; Low German 168; Luther
euphemism 52, 60 – 1, 65, 70, 219 German 203; metathesis in 89; Middle
evaluation of change 156, 248, 252 High German 86; nasal phonemes in
Evans, Nicholas 15, 208, 229, 230 93; negation 253; Old High German
evidence of change 18 – 21, 234 – 7 92, 114 – 15; Pennsylvania German
exaggeration 17, 59 – 60, 66 150, 202 – 6, 208 – 9, 210, 273; plurals
excrescence 80 – 1 114 – 15; Pre-Old High German 168;
extension see analogy Standard German 9, 21; Standard
High German 203; subordinate clause
family tree model 220 – 2, 223, 231, 238 in 9 – 10; Swiss German 203; syntax in
Farsi 18, 223 9; umlaut in 83; word order 141, 143
Finnish 124, 222 Germanic languages 13, 32, 83, 136, 140,
First Germanic Consonant Shift 148, 163, 218, 221 – 2, 224 – 6, 229, 236,
163 – 4, 221 244, 251 – 2; family tree for 221, 224 – 6;
folk etymology 42, 63, 70, 95 umlaut in 83 – 4, 114; see also separate
French 3, 4, 6, 10, 16, 31, 34, 47, 54, 62, entries for individual language
71 – 3, 79 – 80, 83, 88, 94, 103, 107, 112, varieties
123, 144 – 6, 149 – 151, 154, 175, 191 – 2, Gilliver, Peter 38
194, 198, 207, 214 – 15, 239 – 41, 253; Gleason, Jean Berko 110 – 11
Montreal French 175; negation in 154; glottochronology 226 – 8, 244
Norman French 6, 10, 191; Old French Gothic 136, 218, 221, 223, 226, 229, 236
54, 83, 112; vowels in 6; word order Graddol, David 266
144, 145, 149 Grammatica Anglicana (Jonson) 235
Frisian 215, 221, 223, 226 grammatical marker 83, 114 – 16, 120 – 1,
fusional languages 122 – 3, 124, 125 – 6, 123 – 4, 126 – 7, 134, 140 – 1, 148 – 50,
130, 132, 142 155 – 6, 200, 213, 249, 261
grammaticalization 8, 10, 12 – 13, 15 – 16,
Gaby, Alice 229 21, 35, 39 – 40, 42 – 4, 48, 67, 70, 85, 87,
Galician 263 – 4 106 – 8, 110, 114, 118, 121 – 2, 124 – 8,
Gascon 44 – 5 131, 134 – 8, 140 – 2, 144, 148 – 56,
gender, change and 171 – 2, 179 – 82, 186, 158 – 9, 161, 189 – 93, 195 – 6, 198,
188, 209, 255, 270 201 – 2, 204 – 5, 209, 211, 213 – 14, 215,
German 7, 9 – 11, 13 – 14, 21 – 2, 25, 222, 229 – 30, 236, 241, 249 – 51, 253,
32 – 6, 40, 45, 47, 53 – 4, 60, 70 – 2, 76, 265 – 7, 273 – 5; extension in 149, 154,
78 – 83, 85 – 90, 92 – 3, 95, 98, 102 – 3, 161, 235; phonological reduction
107, 114 – 16, 125, 127 – 8, 131, 136, 44, 153, 209, 270; reanalysis in 42 – 3,
140 – 1, 143, 146, 148 – 52, 157, 163 – 4, 108 – 9, 130, 149 – 50, 154, 156, 161,
168 – 9, 172, 176, 179, 186, 189, 191, 270; semantic changes 52, 153, 233;
193 – 4, 202 – 6, 208 – 10, 212, 214 – 19, structural changes 146, 153 – 4, 211
221 – 9, 236, 239, 244, 249, 251 – 3, Grassmann, Hermann 7, 88
263, 273; Bavarian 216; borrowing Grassmann’s Law 7, 88
292 Index
gravity model of diffusion 168 – 71; see Humboldt, Wilhelm von 122, 125

also Great Vowel Shift; wave model hyperarticulation 77
Great English Vowel Shift 99, 113, hyperbole 17, 59; see also exaggeration
219, 264 hypoarticulation 77
Greek 7, 16, 22, 30, 36, 42, 47, 61, 64 – 5,
70, 88, 96, 136, 144, 154, 202, 214, Iberian Romance 118 – 19
217 – 18, 222, 237 – 9; Ancient Greek 7 Icelandic 82, 218, 221 – 3, 226, 228 – 9
Green, Jonathon 30 imitation 14, 41, 95, 218
Greenberg, Joseph 131 – 2, 144, 145, 146, imitative expressions 41
148, 157 i-mutation 83
Gresham’s Law 56 Indonesian 207, 214
Grimm, Jakob 163 – 4, 218, 221 infixes 33 – 4
Grimm’s Law 163, 164, 218, 221 inflection 8 – 9, 13, 35, 50, 85, 106, 110,
Gullah 196 120 – 7, 129 – 30, 134, 136, 140, 142,
Gurindji 199 – 201, 212 146, 154, 160 – 1, 191, 198
Gurindji Kriol 200, 212 inflectional morphology 106, 127,
129 – 30, 136
Hall, Rich 30 inkhorn terms 22
Hanunoo 89 – 90 Innsbruck Computer Archive of
haplology 79 Machine-Readable English Texts 236
Hawkins, John 148, 157 internet 23, 28 – 9, 36 – 7, 48, 85, 100, 175,
Helsinki Corpus of English Texts: 266, 270, 274
Diachronic and Dialectal 254 intolerable homonymy 44 – 6
Herzog, Werner 252 Irish 79, 83, 103, 210, 215
High German 168, 216, 221, 225 – 6 isofunctional morphological marker
High German Consonant Shift 114 – 15
168 – 9, 226 isolating languages 123 – 4, 125 – 6, 132,
Hindi 16, 207, 222 – 3 134, 142, 196, 229
historical linguistics 7, 23 – 4, 48, 76, 94, Italian 3, 4, 79, 82, 102, 145, 163, 207,
101, 106, 134 – 7, 163, 198, 208, 218, 215, 217 – 18, 221 – 2, 227, 229, 238 – 9,
236 – 7, 247 – 8, 250, 253, 260 – 1, 263, 241, 243 – 4
272 – 3, 275
historical pragmatics 23, 256 – 60, 262 Jamaican Creole 196
historical sociolinguistics 23, 260 – 4, 273 Janda, Richard 161, 265, 273
Historical Sociolinguistics Network janus-words 59
(HiSoN) 260 – 3, 273 Japanese 12 – 13, 55, 70 – 2, 80, 138, 157,
Hock, Hans Heinrich 45, 83, 101, 115, 194, 214, 216, 218, 239
131, 244 Jespersen, Otto 99, 152, 154, 249, 251 – 2
Hofstadter, Douglas R. 14, 109, 110 Jespersen’s Cycle 152 – 3, 154, 249, 252
Hollandish 253 Johns, Brian 271
Hopkins, Gerald Manley 22 Johnson, Samuel 28, 70, 269
Hopper, Paul 131, 134, 153, 155, 157 Jonson, Ben 90 – 1, 235
hothouse words 30 – 1, 45 Joseph, Brian 265, 273
Howard, David 47 Journal of Historical Sociolinguistics
Hughes, Geoffrey 44, 48, 192 262
Index 293
Keller, Rudi 10, 17, 25, 253 Light Warlpiri 200 – 1, 212

Knight’s Law 56 Linguistic Atlas of New England 178
Krio 196 Low German 168, 221, 226
Kriol 80, 102 – 3, 196, 200 – 1, 207, 212 – 13 Luther German 203
Kuryłowicz, Jerzy 114, 115, 116, 118, 119,
120, 130 – 1, 148, 272 Ma’a 201
McColl Millar, Robert 24, 86, 101, 112,
Laʒamon’s Brut 115 118, 119, 225, 244
Laberge, Suzanne 175 McConvell, Patrick 200
Labov, William 134, 170 – 1, 172, 173, 175, Macedonian 202, 223
177 – 8, 180, 182, 237, 249, 252, 262 Macquarie Dictionary 54
Langacker, Ron 78, 104 – 5 McWhorter, John 195
language acquisition see acquisition of malapropism 62 – 3
language Malay 84, 144, 217, 238
language creation 29, 190, 195 – 201 Mali 144
language death 16, 207 – 9 Mańczak, Witold 120, 131
Language the Loaded Weapon Mandarin Chinese 16, 123, 125, 207, 263
(Bolinger) 265 Maori 38 – 9, 94, 108 – 9, 197, 210
Language Log 268 Māori Pidgin English 197
language maintenance 22 – 3, 190 – 2, Matisoff, Jim 153
203, 211 Matras, Yaron 206
language myths 25, 210, 263 – 4 Meakins, Felicity 200
language planning and policy 209 – 10 meaning associations 53 – 4
Latin 3, 4, 11, 16, 22, 26, 30, 45, 54, 56, 64, Media Lengua 201
66, 68, 71 – 3, 80, 82, 88, 92, 95 – 6, 118, Meiklejohn, J.M.D. 271
121 – 3, 125, 127, 146, 162 – 3, 188, 192, metanalysis see reanalysis
207, 214 – 15, 218, 221 – 2, 227, 229, metaphor 27, 59, 64 – 5, 70, 110, 114, 126,
233, 237 – 8, 241; Classical Latin 207, 269, 273
222; inflections 122, 125; Law of Latin metathesis 81, 88 – 90, 94, 167, 246
Rhotacism 92, 96; word order 121, 122 metonymy 65
Law of Latin Rhotacism 92, 96 Middle English 2, 8, 11, 95, 98 – 9, 104 – 5,
Law of Semantic Change 56, 66 108, 115, 151, 160 – 2, 164, 187, 197,
Leech, Geoffrey 52, 70, 265 198, 212, 236 – 7, 248, 254, 259, 269;
Lehmann, Winfred P. 145, 157 boundary shift in 151 – 2; pronouns in
Lemle, Miriam 141 11; resources 236; verbal endings in 8
lenition 81, 85 – 7, 92, 96, 240 Middle High German 86
levels of change 5 – 13 Middle Scots 261
Levinson, Stephen 229, 230 Milroy, James 21, 22, 250, 262
lexemes 106, 239, 241 Milroy, Lesley 21, 22, 183, 250, 262
lexicon 4, 16, 23, 28 – 51, 52, 54, 59, 61, Modaressi, Yahya 18 – 19
163, 166 – 7, 190 – 2, 195, 198, 200 – 2, Montreal French 175
209, 211, 264 – 6; additions to 9, 29 – 43; Moran, Caitlin 85
changes to 28 – 48; deletions from 43 – 6 Morey, Stephen 229
leximania 30 morphemes 42, 89, 109, 111, 123 – 6,
Liberman, Mark 48, 63, 268 131 – 2, 153 – 4, 191, 213
294 Index
morphology: changes in 5, 8 – 9, 106 – 31, 242, 248, 262, 265; inflections in

134, 136 – 7, 149, 151, 154, 166, 247; 13; merger in 91; negation in 151;
inflectional 106, 127, 129 – 30, 136; resources 236; as rhotic 163; verbal
reasons for 127 – 30; see also analogy; paradigm of 8; Viking Norse influence
reanalysis; typology 192; voiced fricatives in 94; word order
morphosyntax 134, 159 – 61 in 137 – 40
mountweazels 30 – 1, 38 Old French 54, 83, 88, 112
myths, language 25, 210, 263 – 4 Old High German 81 – 2, 86, 88, 92,
114 – 16, 168, 215, 218
Nahuatl 124 onomatopoeia 41, 95, 218, 233
names 29, 36 – 8, 40, 49, 54, 58, 65, 78, 79, Oxford English Dictionary (OED) 7,
101, 102, 108 – 9, 123, 193, 219 – 20, 267 10 – 11, 16, 30, 38, 54, 59, 112, 115
Naro, A. J. 141
narrowing, semantic 55, 60, 69 – 70, 72, 74 palatalization 82 – 3, 89, 95, 103, 175, 266
National Indigenous Languages paradigms 8, 85 – 6, 111 – 14, 119 – 20, 150,
Survey 208 153, 249
“Neg-first principle” 249 Partridge, Eric 55
negation 151 – 7, 161, 176, 180, 196, 248, Paston, John 258 – 9
249 – 53, 255, 262, 272, 274 Paston, Margaret 258 – 9
Neogrammarian school of linguistics 95, Paul, Hermann 13 – 14
101, 163 – 4 Pedersen, Holger 230
neologisms 29, 31, 34, 50 Pennsylvania German 71 – 2, 150, 202 – 6,
Nevalainen, Terttu 251, 252, 254 208 – 9, 210, 212, 273; discourse
New Guinea 15, 196, 207 particles in 206; substratum influence
New Oxford American Dictionary 31 203, 204 – 6; superstratum influence
New Zealand 52, 92, 94, 100, 197, 255, 203 – 4
271; New Zealand English 100 periphrastic do 20, 167
Nichols, Johanna 265 Philippine English 40
NoCussing Club 186 Phillips, Betty 165
nonce words 34 phonemes 6, 82, 90 – 4, 98, 136, 153, 194;
Norman Conquest 6, 191 – 2, 215 merger of 91 – 2; split of 92 – 4
Norman French 6, 10, 191 phonesthemes 41, 48, 61
Norwegian 144, 221, 223, 226 phonetics, changes to 5 – 7, 247; erosion
Nostratic 230 – 1, 232 – 3, 244 127 – 9; phonemic change and 90 – 4
nursery rhymes 116, 235 phonology, changes to 5 – 7, 75 – 101, 127,
247; complementary distribution 5 – 6;
obsolescence 43 – 4, 48 irregular 94 – 7; loss of sounds 77 – 9;
O’Connor, Joseph 101 modification of sounds 81 – 90; new
Ogura, Mieko 166 sounds 80 – 1; phonetic change and
Old Bailey Corpus 255 90 – 4; processes of 76 – 90; reasons for
Old English 1 – 6, 8, 10, 13, 43 – 5, 52 – 3, 97 – 101; reduction 153; segmental 5;
63, 73, 77 – 80, 82 – 3, 91, 94, 98, 103 – 4, suprasegmental 7
107 – 8, 112 – 13, 116 – 17, 127, 132 – 3, pidgins 16, 40, 190, 195 – 6, 197 – 9, 201,
137 – 8, 140, 142, 151, 155, 159 – 61, 211 – 12, 214
163, 192, 197 – 8, 207, 219, 226, 236 – 7, playfulness 3, 17, 38, 71, 156
Index 295
polysemy 53 reanalysis 42 – 3, 106 – 9, 130, 149 – 51,

polysynthetic languages 124 – 5, 229 154, 156, 161, 243, 270; definition
Pope, Jennifer 178 – 9 of 108
Portuguese 3, 4, 163, 215, 221 – 2, 229, rebracketing see reanalysis
238 – 41 Received Pronunciation (RP) 163, 271
Poser, William 233 reconstruction, linguistic 21, 233 – 43;
pragmatic marker 12, 206 comparative method of 217 – 19,
pragmaticalization 12 – 13, 206 237 – 43; internal 241 – 2; written
pragmatics 5, 11 – 13, 23 – 5, 131, 206, 247, sources in 234 – 7
255, 256 – 60, 262, 264, 275; changes in reduction 43 – 4, 69, 77 – 8, 85, 100 – 1, 134,
11 – 12; historical 256 – 60 153, 193, 209, 241, 265 – 7
Pre-Old High German 168 reduplication 36, 38 – 9, 61, 84, 88, 196
prefixes 33 – 4, 42, 66, 88 – 90, 103, 127, regularity hypothesis 95, 120, 164, 247
219, 249 Rhenish Fan 168, 169
prejudice 53, 55, 58 – 9, 69 Romaine, Suzanne 136, 212, 261 – 2
pronunciation 4 – 5, 7, 13 – 14, 41, 43, Romance languages 3 – 4, 79, 145, 163,
62, 78, 80 – 3, 85, 87 – 91, 93 – 96, 207, 221 – 2, 229, 239, 242; see also
101 – 3, 105, 111, 114, 148 – 9, 163 – 5, French; Italian; Latin; Portuguese;
168, 172 – 3, 175 – 7, 180 – 1, 186 – 7, Spanish
193 – 4, 203, 234 – 5, 238, 241, 265 – 6, Romanian 83, 202, 221 – 2, 229
271; accent 15, 16, 75, 90 – 1, 97, 194; Ruhlen, Merritt 231 – 2
reconstruction of 234 – 5; spelling 95; Russian 16, 65, 82, 87, 123, 143 – 4, 194,
stress 4, 7; variation in 13 – 14 222 – 3, 263
prothesis 80 Rutten, Gijsbert 262
Proto-Germanic 79, 82 – 3, 89, 102 – 3,
140, 141, 163 – 4, 218, 229 Sankoff, Gillian 175
Proto-Indo-European (PIE) 7, 163 – 4, Sanskrit 7, 132, 163, 214, 222 – 3
218, 229, 242 – 3 Sapir, Edward 251
Proto-Romance 241 scare quotes 267
Proto-World 229 – 33, 243 – 4 Scheler, Manfred 16
psycholinguistics 106, 247 Schlegel, August Wilhelm 122, 125
pull chain 99, 100, 102 Schmidt, Johannes 168, 224
Pullum, Geoff 63 Schuchardt, Hugo 164, 168
punctuation 209, 267, 268 – 9 Scots English 90, 261
purism 209 – 11 semantic halo 55 – 6
push chain 100, 102; see also chain shifts; Semantics (Leech) 52 – 3
pull chain semantics, changes to 10 – 11, 52 – 70;
chain reaction 56 – 7; consequences
Quechua 201, 233 of 54 – 7; linguistic reasons for 61 – 5;
psychological reasons for 58 – 61;
rate of change in language 5, 264 – 72 socio-cultural reasons for 57 – 8;
Raumolin-Brunberg, Helena 166, theories of 66 – 9
254, 273 Serbo-Croat 202
real time 18, 20 – 1, 24, 176 – 9, 237, 267; Shakespeare, William 2 – 4, 8, 11, 17, 23,
see also apparent time 48, 99, 118, 142, 234 – 6, 256 – 7; Much
296 Index
Ado about Nothing 2 – 3, 8, 17 – 18, Sturtevant’s paradox 114

256 – 8; Romeo and Juliet 234 – 5 subjectification 68, 70
shifting see conversion subordinate clauses 9 – 10, 22, 135, 137,
Shorter Oxford English Dictionary 16 139 – 41, 143, 150, 250, 253
Shuy, Roger 154 subreption 58, 69
sibilants 111, 166 substratum influence 193 – 4, 203,
Simons, Gary 219 204 – 6
Simpson, John 30 suffixes 13, 33 – 6, 42 – 3, 78, 83, 103,
sneaky diffusion 141 – 2 109, 114 – 16, 119 – 20, 126 – 7, 148 – 9,
social class 171, 172 – 5, 180, 182 – 3, 158, 166, 191; Basque 118 – 19, 242;
185 – 6, 251 inflectional 13; Maori 38 – 9, 94, 108,
social networking platforms 28 – 9, 269 109, 197, 210; Spanish 119 – 20
social networks, change and 182 – 5, superstratum influence 190 – 1,
250 – 2 203 – 4, 211
Socio-historical Linguistics (Romaine) Supplement to the Oxford English
250, 261 Dictionary 9
sociolinguistics 14, 21, 23, 172, 179 – 80, Svartvik, Jan 265
182 – 3, 209, 247, 250, 252, 260 – 3, Swadesh, Morris 226 – 8, 244
272 – 3; historical 260 – 4 Swadesh list 226 – 7, 228, 244
“Some Universals of Grammar Swahili 79, 124, 125 – 6
with Particular Reference to the Swiss German 40, 203
Order of Meaningful Elements” synaesthesia 64 – 5
(Greenberg) 144 syncope 79
sound laws 95 – 6, 164 synecdoche 65
sound symbolism 41, 52, 70, 217 syntax, changes in 9 – 10, 107, 131,
South Seas Jargon 197 134 – 57, 163, 247; historical 134 – 7;
Spanish 3, 4, 16, 40, 79 – 80, 119 – 20, 157, reconstruction of 236 – 7, 242 – 3
163, 194, 201, 207, 215, 221 – 2, 227,
229, 233, 238 – 9, 240 – 1, 243, 263; taboo words 11, 45, 47 – 8, 60 – 1, 65, 70,
Castilian Spanish 263 96, 155, 219, 244, 255
spelling 6, 18, 40 – 1, 43, 46, 62, 78, 82, Tagalog 40
86 – 7, 90 – 1, 93 – 5, 100, 114, 117, 149, Telegraph 33
209, 241, 254, 258, 266, 267, 269 Thai 144
spheres of usage 115 – 16 Thibault, Pierette 175
spread of change 162 – 86; age in 175 – 9; Tiro, Marcus 139
communities of practice in 185 – 6; tokens 129 – 30
gender in 179 – 82; within linguistic Tok Pisin 158 – 9, 196, 207
system 163 – 7; patterns of 168 – 71; Tolkien, J.R.R. 62, 117, 157, 273
social class in 172 – 5; social factors in transition problem 248, 250 – 1, 272, 275
171 – 82; social networks for 182 – 5; Traugott, Elizabeth Closs 66 – 7, 68, 70,
within social structures 167 – 86 131, 134, 153, 155, 157
Standard German 9, 21 Trudgill, Peter 168 – 70, 172 – 3, 174, 187,
Standard High German 203 262, 272
standardization 136 – 7, 250, 265, 269, 275 types 20 – 1, 28, 31, 39, 43, 54, 69, 76 – 7,
Sturtevant, Edward 96 81, 91, 125 – 6, 129 – 30, 132, 136, 138,
Index 297
141, 143 – 4, 146, 150, 179, 190 – 202, 264, 267; addition of 80; Great English
205, 211; pure 145 Vowel Shift 99, 113, 219, 264; harmony
typology 106, 121 – 7, 130 – 1, 134, 83; lengthening of 39, 78 – 9; loss of
142, 143 – 8, 156 – 7, 226, 243, 272; 6, 77, 78, 79, 94; merger of 91 – 2;
morphosyntactic 106, 121 – 7; word nasalized 6, 238; reduction of 153, 241;
order change and 121, 143 – 8 umlaut 83 – 4
umlaut 83 – 4, 113 – 14, 241 Wang, William S-Y. 166

UNESCO 207 Warlpiri 199 – 201, 212
uniformitarian principle 250 – 1, 273 Watts, Richard 263 – 4
universal grammar 134, 230 wave model 168, 169 – 71, 222 – 6, 243
Urban Dictionary 26, 30, 32, 85, 269 Weinreich, Max 24, 216, 248, 252,
273, 275
Vandalic 236 Welsh 143
variation in language 13 – 18, 21 – 4, Wiktionary 34
172 – 4, 179 – 80, 187 – 8, 195 – 6, 224, Willis, David 155, 157, 249, 253
261 – 3, 265 – 6, 269 word order 9 – 10, 13, 121 – 4, 127, 136,
Vennemann, Theo 145, 157 137 – 42, 143 – 5, 147 – 9, 156 – 7, 196,
verb-final ordering 10, 138 – 41 198, 204, 212, 241 – 2, 251, 267: change
verbal brace 140, 141, 212, 253 in 137 – 42; typology and 121, 143 – 8
verbicide 44 words see lexicon
Verein deutsche Sprache e.V. 186 World Atlas of Linguistic Structures
vernacular universal 252 Online (WALS) 127, 143, 148, 244
Vietnamese 123
Viking Norse 192 Yulparija 147
vocabulary see lexicon Yupik 124 – 5
vowels 5 – 6, 76 – 7, 79 – 80, 82 – 3, 85 – 6,
91 – 2, 94, 96, 99 – 100, 113, 117, 153, Zipf ’s principle of least effort 77,
163, 194, 203, 205, 230, 238, 240 – 1, 153

(Understanding Language) Kate Burridge - Alexander Bergs - Understanding Language Change-Routledge (2017)

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

(Understanding Language) Kate Burridge - Alexander Bergs - Understanding Language Change-Routledge (2017)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Understanding Language) Kate Burridge - Alexander Bergs - Understanding Language Change-Routledge (2017)

Uploaded by

Copyright:

Available Formats

Understanding Language Change

Understanding Language Change offers a complete and accessible introduction to

• A step-by-step approach introducing language change at the sound, word and

Written by two experienced teachers, Understanding Language Change is essen-

Kate Burridge is Professor and Chair of Linguistics at Monash University, Australia.

Other titles in the series:

1 Setting the scene 1

2 Changes to the lexicon 28

3 Changes to the semantics 52

4 Changes in sound structure 75

4.4.5 Frequency factors 100

5 Changes in word structure 106

6 Changes in sentence structure 134

7 The spread of change 162

8 Languages in contact 189

8.1.3 The creation of new languages 195

9 Relatedness between languages 215

10 An end on’t 247

1.1 Change in apparent time in Ghazvin (an) 19

7.1 The Rhenish fan 169

1.1 Some formal correspondences between different Romance languages 4

9.4 List of 100 basic words 227

Map 7.1 by Hans Erren (2010) https://en.wikipedia.org/wiki/High_German_con-

Map 8.1 by Brenda Thornley, reproduced with kind permission of Felicity Meakins.

Setting the scene

Languages change – constantly, inevitably. Unfortunately, this is a little bit hard to

“Thou hast yhad five housbondes,” quod he [Jesus],

Men may divine and glosen up and down,

“Now you have had five husbands,” he [Jesus] said,

Latin French Italian Spanish Portuguese English (gloss)2

in vanum en vain in vano en vano em vão in vain

1.1 CHANGE OCCURS AT ALL LEVELS

1.1.1 Phonetics and phonology

complementary distribution we treat them as allophones of a single phoneme /f/.

ca. 13th century ca. 14th–16th century ca. 17th–18th century

Suprasegmental phonology (phonology beyond the segment, i.e. beyond the

1st person I sing

1st person we sing

1st person ic· singe

1st person we- singaþ

teach-es. And, again, variation is what underlies linguistic change. Ultimately, we

(1.1) a Coco Chanel erfand das kleine Schwarze.

Interestingly enough, recent informal spoken varieties of German show a vari-

Here we see syntactic change in progress. German subordinate clauses seem to

(1.4) I like Halloween.

a kono heya no denki hiruma demo itsumo tsuite-iru

The development of demo is thus from less personal to more personal, or

1.1.6 Change across different levels

1.2 VARIATION AND CHANGE – TWO SIDES OF THE SAME COIN

characterized by a personal idiolect: his or her individual use of language. Some

(1.8) Rosa Luxembourg was a terribly nice name for a revolutionary.

1.3 EVIDENCE OF CHANGE IN PROGRESS

Tehran 10–19 Tehran 20–29 Tehran over 50

Affirmative declaratives (statements): I eat cheesecake vs. I do eat cheesecake.

Negative Questions Affirmative Questions

1.4 ATTITUDES TO CHANGE

1.5 HOW THIS BOOK IS STRUCTURED

1 Setting the scene 1

2 Changes to the lexicon 28

3 Changes to the semantics 52

4 Changes in sound structure 75

4.4.5 Frequency factors 100

5 Changes in word structure 106

6 Changes in sentence structure 134

7 The spread of change 162

8 Languages in contact 189

8.1.3 The creation of new languages 195

9 Relatedness between languages 215

10 An end on’t 247

1.1 Change in apparent time in Ghazvin (an) 19

7.1 The Rhenish fan 169

1.1 Some formal correspondences between different Romance languages 4

9.4 List of 100 basic words 227

k coast ‘sea shore’ (< ‘side’)