The Linguistics

of the History
of English
Remco Knooihuizen
The Linguistics of the History of English
Remco Knooihuizen

The Linguistics
of the History of English
Remco Knooihuizen
University of Groningen
Groningen, The Netherlands

ISBN 978-3-031-41691-0 ISBN 978-3-031-41692-7 (eBook)

https://doi.org/10.1007/978-3-031-41692-7

© The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature
Switzerland AG 2023

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

Cover illustration: Contributor: tbkmedia.de/Alamy Stock Photo

This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland
AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

This is an introductory textbook on the history of the English language. At the

same time, it is also an introductory textbook on historical linguistics in English.
The book has as its starting point some of the major questions in the study
of language change—how do different types of language change originate and
spread?—and discusses these with a particular focus on examples from the history
of English. Compared to many other introductory textbooks, which take students
through the different stages of the English language chronologically, the focus here
is less on history and more on linguistic theory. Textual evidence is offered in sup-
port of the theoretical discussion. Text boxes on empirical studies go into more
depth on specific details, and other boxes give a more cross-linguistic perspective
by looking at comparable developments in other languages.
Although this is an introductory textbook, some prior knowledge is assumed.
Students should ideally have followed one or more introductory linguistics courses
and should be familiar with basic concepts and terminology; advanced knowledge
of linguistic theory is not required. It is also assumed that students are familiar
with the broad strokes of the history of the English language and the history of
Anglophone countries (especially England). The overview of these histories in
Chapter 1 is meant as a refresher that will bring readers up to speed, but it does
not pretend to be a comprehensive presentation.
At the end of the book, my goal is that students will be equipped with
enough theoretical and factual knowledge to conduct their own independent or
semi-independent research on topics from the history of English. They should
also be well-equipped to engage in public debate about language change and to
dispel many of the myths we see in public discourse, such as the idea that lan-
guage change entails the decay of a perfect Standard English, brought on by the
exceptionally lazy youth of today.
The textbook has been developed as teaching material over a decade of teaching
the course ‘English Language Change’ at the University of Groningen. It should
hopefully fit courses in other degree programmes equally well. My thanks go out
to many cohorts of students for their feedback on various iterations of the text,
and to many colleagues for their support. Comments from an anonymous reviewer
have been extremely helpful in clarifying the focus in some of the chapters, and in
dealing with smaller issues throughout. My own educational and research history
at the Universities of Groningen and Edinburgh is as a generalist with a focus on
v
vi Preface

Germanic languages more widely, and I have had to make myself familiar with the
norms and traditions of ‘Anglistics’ separately. I hope that this mixture of inside
and outside perspectives on the field will inspire new generations of students to
work on historical linguistics both in English and in other languages.

Groningen, The Netherlands Remco Knooihuizen

Contents

Part I Introduction
1 So What Had Happened Was . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Explaining Language Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Part II Sound change

3 Phonetic Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4 Phonological Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5 Track Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

Part III Contact-Induced Change

6 Language Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
7 Dialect Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

Part IV Structural Change

8 Analogy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
9 Grammaticalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
10 Syntactic Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
11 The Loss of Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
12 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

vii
List of Figures

Fig. 2.1 Schematic representation of the S-curve pattern

of the spread of a linguistic change through a speech
community . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Fig. 3.1 Sound changes in lenition (Adapted from Bauer 2008,
p. 686) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Fig. 4.1 Schematic representation of the Great Vowel Shift . . . . . . . . . . . . 51
Fig. 4.2 The Northern Cities Shift (Adapted from Labov 1994,
p. 191) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Fig. 4.3 The Canadian Vowel Shift (Adapted from Boberg 2019,
p. 93) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Fig. 4.4 The California Vowel Shift (Adapted from Podesva et al.
2015, p. 159) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Fig. 4.5 The Southern Hemisphere Shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Fig. 4.6 Schematic representation of the foot/strut split
and related changes, after Wells (1982, p. 199) and Turton
and Baranowski (2021, p. 166) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Fig. 4.7 Schematic representation of the trap/bath split and related
changes, after Wells (1982, pp. 232–234) . . . . . . . . . . . . . . . . . . . . . 64
Fig. 4.8 Schematic overview of, from left to right, primary split
(trap/bath), secondary split (foot/strut), and merger
(square/nurse) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Fig. 6.1 Distribution of Scandinavian place names in England
(Adapted from Fellows-Jensen 2011, p. 70) . . . . . . . . . . . . . . . . . . . 100
Fig. 6.2 Uptake of French loan words in English from the eleventh
to nineteenth centuries, based on data by Jespersen
and Baugh (Baugh 1935) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Fig. 7.1 Restructuring along a spectrum of contact varieties
(Adapted from Siegel 2001, p. 193) . . . . . . . . . . . . . . . . . . . . . . . . . . 135

ix
x List of Figures

Fig. 10.1 Tree diagram for The black cat found cheese
from Germany in the pantry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Fig. 10.2 Verb movement and the change from OV to VO order . . . . . . . . . 174
Fig. 10.3 A negated subordinate clause before the loss of V-to-I
movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Fig. 10.4 A negated subordinate clause after the loss of V-to-I
movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
List of Tables

Table 3.1 Regular correspondences in five Germanic languages . . . . . . . . 35

Table 4.1 Consonant inventories in Old English, Middle English,
and Modern English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Table 4.2 Changes to the pronunciation of long vowels in the Great
Vowel Shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Table 5.1 Sound changes in the history of the English word comb . . . . . . 72
Table 5.2 Changes to the pronunciation of long vowels in the Great
Vowel Shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Table 6.1 Thomason and Kaufman’s borrowing scale . . . . . . . . . . . . . . . . . . 92
Table 6.2 Old English paradigms for to be: wesan and bēon . . . . . . . . . . . 97
Table 6.3 Overview of contact situations in English . . . . . . . . . . . . . . . . . . . 110
Table 8.1 Paradigms for the verb ‘to choose’ . . . . . . . . . . . . . . . . . . . . . . . . . 145
Table 8.2 Strong verb classes in Old English . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Table 11.1 Cases in German . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Table 11.2 Some examples of case patterns in Old English . . . . . . . . . . . . . . 187
Table 11.3 Comparison of case patterns in Old English and Old
Norse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

xi
 Part I
Introduction
So What Had Happened Was
Fifteen Centuries of English-Language Change
1

Introduction

To make sense of the different forms that English has taken throughout its history,
it is customary to divide that history into different periods. The major divisions that
are conventionally made are Old, Middle, and Early Modern English, before we
get to present-day English. There is variation in the dates given to these different
periods, and the years given in this chapter are meant as indicative only.
The periods in the history of English are ideally defined by means of linguis-
tic characteristics. This is necessarily imprecise, as language change is a gradual
process. A medieval peasant in Kent did not go to bed one day having spoken Old
English and wake up the next day speaking Middle English; these are our own
projections, and we use different factors in deciding when one period ends and
the next starts, such as historical events, the reigns of kings or dynasties or sim-
ply round numbers. There is a lot of variation within periods, and there are many
similarities between periods as well. Ultimately, the periodisation is as much a
socio-historical as it is a linguistic decision.
The bird’s-eye view of the history of English in this chapter focuses on both
these factors: it sketches the historical context for the development of English in
the past 1500 years and gives brief illustrations of what the language looked like
in the different periods. The illustrations are based on a short bible passage (Luke
9: 12–17) as the bible is a text that has been translated many times at different
periods in the history of English, and we can therefore easily compare the forms
of the language at different times. Neither the socio-historical nor the linguistic
overview aim to be exhaustive. They serve as background for the discussion later
in the book of the processes by which English has changed.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 3

R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_1
4 1 So What Had Happened Was

Old English (500–1150)

Socio-Historical Context

The Romans were gone, to begin with.

The Old English period starts shortly after the settlement of Angles, Saxons, and
Jutes in England from the mid-fifth century. These were Germanic peoples stem-
ming from the coastal areas of what is now the Netherlands, northern Germany,
and southern Denmark. Their settlement in England is part of a period of large-
scale migrations within Europe surrounding the fall of the Western Roman Empire.
There is debate among historians about the number of settlers, with estimates
ranging from 20,000 to as high as 200,000 (Ward-Perkins 2000, p. 524).
The Germanic settlers did not arrive in an unpopulated country, of course. The
Roman occupation of Britain may have ended in 410 AD—some of the earliest
Germanic settlers may have encountered some Roman stragglers—but there was a
native Celtic population, the Britons, still in place. The exact interactions between
the Germanic settlers and the Britons are unclear because of a lack of historical
sources, but what is clear is that from the earliest history of English, the language
developed in a situation of language contact (see section “Influence from Celtic
Languages” in Chapter 6).
The settlers spoke different Germanic dialects that over time coalesced into
a variety that was distinct and ‘English’ (see Chapter 7 for a discussion of this
process). We have very few sources for the earliest forms of Old English. Most
manuscript evidence dates from the ninth century onwards, by which time we
can already distinguish a number of regional dialects: Mercian, Northumbrian,
Anglian, and West Saxon. This last dialect was the most prominent and at times
served as a sort of standard written language—both because the kingdom of
Wessex was relatively powerful and because a lot of West Saxon material was
written and subsequently copied, also in other parts of England (Timofeeva 2022,
pp. 175–176). This does not imply the uniformity of language that the concept of
a ‘standard’ suggests today, however. Although some features show almost com-
plete uniformity (Faulkner 2020), the variation found in written texts is enough that
Timofeeva (2022) could successfully use sociolinguistic methodology to uncover
the influence of social networks and communities of practice on the language in
this period.
Despite the political fragmentation into smaller kingdoms, the term the people
of early medieval England used for themselves and their language was Englisċ
or Ænglisċ: simply ‘English’. The appearance of the term ‘English’ suggests that
regardless of political allegiance, the people came to see themselves as speakers
of the same language no later than the early eighth century (Smyth 1998).1

1 The terminology surrounding language and culture in early medieval England has been subject
to heavily debated change over the years. The term Anglo-Saxon has long been used to refer to the
language, culture, and people at the time. However, in public discourse in some Anglophone groups
the term has developed connotations with a racialised imagined indigenous heritage in Britain,
Old English (500–1150) 5

There are two further historical developments that led to language contact in
the Old English period. The first of these is the Christianisation of England in the
seventh century. Initiated by Pope Gregory I in 597, this process ended with most
royalty and nobility being converted before 700. The church used a good deal of
Latin terminology, which was borrowed into Old English (see section “Influence
from Latin” in Chapter 6).
The other language contact situation involved Vikings, who invaded England
regularly from the eighth century (e.g., the raid on the monastery of Lindisfarne in
793) and who settled in the north of England from the ninth century. By 878, much
of Northern England was under Viking control. Although armed conflict contin-
ued until the eleventh century, Viking settlers and the already existing English
population co-existed and mixed during these centuries (Hadley 2000, pp. 1–17).
The Vikings spoke Old Norse, a language closely related to Old English, which
brought with it a particular type of language contact (section “Influence from Old
Norse” in Chapter 6 and Chapter 7). The Old Norse influence on Old English is
therefore extensive but difficult to see.
The conquest of England by the Normans, starting in 1066, is the beginning of
the end for the Old English period. A changing sociolinguistic status and increased
contact with French changed the language almost beyond recognition. The result
of that process is generally seen as Middle English.

Linguistic Sketch

Old English was characterised by an almost exclusively Germanic lexicon, which

makes it very difficult for speakers of present-day English to read. It also had a
very extensive inflectional system, with many different case endings for nouns and
adjectives and different verb endings to indicate person, number, tense, etc. This
is exemplified by the translation of the bible fragment from the Wessex Gospels,
written around 990:

12 Þa geƿ at se dæg forð. 7 hig tƿ elfe him genealæhton 7 sædon him; Læt þas menego þ̄ hig
farun on þas castelu 7 on þas tunas þe her abutan synt: 7 him mete findon. forþam þe ƿ e synt
her on ƿ estere stōƿ e; 13 Ða cƿ æð he to him. sylle ge him etan; Ða cƿ æðon hı̄g ƿ e nabbað
buton fı̄f hlafas 7 tƿ egen fixas. buton ƿ e gan 7 ūs mete bicgon 7 eallum þissum ƿ erede; 14
Þar ƿ æron neah fif þusenda ƿ era; Ða cƿ æþ he to his leorningcnihtun: Doþ þ̄ hig sitton. þurh
gebeorscypas fiftegum. 15 7 hig sƿ a dydon 7 hi ealle sæton; 16 Ða nam he þa fı̄f hlafas 7 þa
tƿ egen fixas. 7 on þone heofon beseah 7 bletsude hig 7 breac. 7 dælde his leorningcnihtum.

which has led to the term being seen as exclusionary. The International Society of Anglo-Saxonists
changed their name to International Society for the Study of Early Medieval England in 2019 in
response to these connotations. The term Old English, which is the term used in this book, had ear-
lier already been used for the language only: Anglo-Saxon people spoke Old English. This term
is sometimes criticised because it is felt to over-emphasise the continuity between Old and Mid-
dle English, for many of the same reasons of imagined heritage (see Watts, 2011, pp. 30–32, for a
discussion).
6 1 So What Had Happened Was

17 Þa æton hig ealle 7 ƿ urdon gefyllede. 7 man nam þa gebrotu þe þar belifon tƿ elf cypan

fulle;

The most obvious thing about this passage is the use of letters we no longer use
in English today: <æ> (which represented the vowel /æ/), <þ> and <ð> (which both
interchangeably represented the dental fricatives /θ/ and /ð/), and <ƿ > (which stands
for the consonant /w/). There are also two frequent abbreviations: <þ̄> stands for
þæt ‘that’, and <7> stands for and ‘and’.
There are many words in the fragment that are no longer used in English today:
forþġewı̄tan ‘to proceed’, ġenēalǣċan ‘to approach’, and ġebēorsċip ‘company’
(literally ‘beer-ship’) are some of the most striking ones.2 Others have only a
residual existence or a different meaning in present-day English, such as stōw
‘place’ (in stowaway), wer ‘man’ (in werewolf ), cȳpa ‘basket’ (now coop ‘hen-
house’), and mete ‘food’ (not just ‘flesh’). And we see compound words typical
of Old English, such as leorning-cniht ‘disciple’ (literally ‘learning-knight’). It is
no coincidence that the vast majority of present-day English translations of these
words are Romance loanwords, but in the Old English period, these were very few
and far between, with only castel ‘castle, town’ in this fragment.
The fragment also shows evidence of the case system of Old English: compare
eallum (a dative form for the indirect object) in verse 13 with ealla (a nominative
form for the subject) in verses 15 and 17. Verb inflection was much more extensive
than today, as can be seen from he cwæþ ‘he said’ (v. 13) and hı̄e cwæþon ‘they
said’ (v. 13). All these inflectional endings showed a variety of unstressed vowels:
<a>, <e>, <o>, and <u> in this text. The pronoun they is hı̄e (spelled hi, hig or hı̄g in
the fragment; dative him) with an initial h-.
The syntactic structure of Old English is more difficult to glean from the
untranslated passage, but the differences can be seen in the glossed version of
v. 12 in (1):

(1) Þa geƿ at se dæg forð. and hig tƿ elfe him genealæhton and sædon him; Læt þas menego
þæt hig farun on þas castelu and on þas tunas þe her abutan synt: and him mete findon.
forþam þe ƿ e synt her on ƿ estere stōƿ e;

Then went the day forth and they twelve him approached and said him Let the crowds
that they go to the castle and to the towns that here about are and them food find because
that we are here on desert place
‘Then the day proceeded, and the twelve of them approached him and said to him, “Let
the crowds (go), so that they go to the castle and the towns that are around here, and find
them food, because were are in a deserted place.’

2 Examples in the text are given in a standardised dictionary reference form of Old English, and
therefore differ slightly from the forms that are found in the fragment.
Middle English (1150–1500) 7

The two most striking syntactic features in this sentence are the position of the
verb last in the relative clause þe hēr abūtan sind ‘that are about here’, and the
place of the indirect object him ‘him’ before rather than after the verb ġenēalǣhton
‘approached’.
All in all, Old English was a very different type of language from present-day
English. Speakers of other Germanic languages often find it slightly easier (but
still difficult!) to read a text in Old English than speakers of English who do not
speak another Germanic language. The changes that would turn Old English into
Middle English meant that a lot of Germanic vocabulary was lost, as well as some
grammatical features that Old English shared with other Germanic languages.

Middle English (1150–1500)

Socio-Historical Context

The Norman Conquest of 1066 brought about considerable change. The ensuing
migration of Norman French speakers was not at an incredibly large scale, but they
did take up key leadership positions, essentially replacing the existing nobility.
The courts became Francophone, with English being relegated to the lower social
classes. But just as there was no population replacement after 1066, there was no
language shift either: everyone continued to speak their own language, Norman
French for the elites or English for the common man, with bilingualism restricted
to those people whose professions meant they had to deal with both population
groups (see section “Influence from French” in Chapter 6).
English played second fiddle to French for two to three centuries. In the thir-
teenth, and especially in the fourteenth century, however, we see that the balance
shifts again in favour of English. There are a number of reasons for this. One is that
the political relations between England and France became increasingly difficult.
The continental possessions of the English king were lost under John Lackland in
the early thirteenth century, although contacts of course remained. The Hundred
Years War (1337–1453) between England and France, however, did much to whip
up anti-French sentiment and nudge the anglicisation of the elites. Another rea-
son is the Black Death of 1349. Mass death and population decline caused labour
shortages, which meant that the Anglophone working classes held a stronger social
position. The same can be said for the emergent merchant middle class.
The demise of French and the emancipation of English can be seen in the rein-
statement of English as the language of the law (in the Statute of Pleading, 1362)
and of official government (under Henry V, r. 1413–1422). There was also a revival
of English-language literature in this period. Most of the English writing since the
Norman Conquest had been religious in nature—while French was the language of
government, religion had continued as an English-language domain—but the four-
teenth century saw literary highlights as Geoffrey Chaucer (The Canterbury Tales)
and William Langland (Piers Plowman), as well as an English bible translation by
John Wycliffe.
8 1 So What Had Happened Was

Some changes in the language are visible in the language already shortly after
the Norman Conquest in 1066. We can partly explain these by scribes no longer
being educated in the West Saxon pseudo-standard, so that they would once again
write the language as they would speak it. Some of this, therefore, reflects changes
that had already happened, and that only appeared in manuscripts with some delay,
and not anything to do with Norman French influence. By the end of this period,
we again see an increasing standardisation of the language, based on the dialect of
the most populous region, the East Midlands, and especially the capital, London.
The introduction of the printing press in the 1470s drastically accelerated this
development.

Linguistic Sketch

Over time, the Old English inflections were simplified and mostly eroded away.
The language that we call Middle English, then, has far fewer inflections, and as a
consequence of that, a more fixed word order. There are also significant changes in
its lexicon. In the Early Middle English period, the lexicon was still predominantly
Germanic, although there had been many Old Norse loanwords due to the Viking
settlements in the North of England in the ninth and tenth centuries. But by the
Late Middle English period, the lexicon had changed profoundly, and many Old
English words had been replaced by loanwords from French (or more properly,
Anglo-Norman). This is illustrated by the bible translation by John Wycliffe (c.
1380):

12 And the dai bigan to bowe doun, and the twelue camen, and seiden to hym, Leeue the

puple, that thei go, and turne in to castels and townes, that ben aboute, that thei fynde mete,
for we ben here in a desert place. 13 And he seide to hem, Yue ye to hem to ete. And thei
seiden, Ther ben not to vs mo than fyue looues and twei fischis, but perauenture that we
go, and bie meetis to al this puple. 14 And the men weren almost fyue thousynde. And he
seide to hise disciplis, Make ye hem sitte to mete bi cumpanyes, a fifti to gidir. 15 And thei
diden so, and thei maden alle men sitte to mete. 16 And whanne he hadde take the fyue
looues and twei fischis, he biheeld in to heuene, and blesside hem, and brak, and delide to
hise disciplis, that thei schulden sette forth bifor the cumpanyes. 17 And alle men eeten, and
weren fulfillid; and that that lefte to hem of brokun metis was takun vp, twelue cofyns.

The lexical developments are probably the most striking in this passage:
puple ‘people’, desert, place, peraventure ‘maybe’, disciples, cumpanyes, cofyns
‘coffins’, … These are all French loanwords that in some cases was simply added
to the lexicon, but also very often entirely replaced the Germanic words from Old
English.
The majority of morphological endings had disappeared, and what remained
was spelled with an <e>, which suggests a pronunciation with /∂/. The endings
may even no longer have been pronounced at all. There is no longer any evidence
of case, apart from in the pronominal system—where in this fragment, the subject
form they had been introduced, but the object form was still hem. The only remnant
Early Modern English (1500–1700) 9

of verbal morphology is the plural ending {-en} in the past tense (camen, seiden,
weren, etc.) but not in the present (turne, finde; but ben ‘are’).
The syntax in this passage does not strike a modern reader as very odd. Subor-
dinate clauses are no longer verb-final, and constituent order is more or less fixed
as it is in present-day English. This means that the greatest amount of syntactic
change happened in the transition from Old to Middle English.

Early Modern English (1500–1700)

Socio-Historical Context

The Early Modern period, the period of the Renaissance, saw another emancipa-
tion of English. This time it was not an emancipation from French, but English
joined in a European-wide process in which vernacular languages took over func-
tions from Latin, in particular in religion (driven by the Reformation) and science
(Burke 2004). In order to be able to use English in these new domains, the lan-
guage needed to expand its vocabulary significantly. This meant that there were
many Latin loanwords, so many in fact that there was an ultimately unsuccessful
puristic pushback. These Latin loanwords often ended up in higher, more for-
mal registers. This leaves us with a number of English-French-Latin triplets of
increasing formality, e.g., kingly, royal, regal.
In this period, the written language is becoming much more standardised in both
spelling and grammar, partly as a result of the introduction of the printing press,
a growing market for books, and the authoritative bible translation commissioned
by King James VI and I (1611). This means that it becomes more difficult to
spot language change as it happens in the many documents we have from this
period. Many private documents survive from this period as well, however, and
they do allow for a more detailed tracking of changes in progress (Nevalainen and
Raumolin-Brunberg 2003).
An important development is that this period sees the beginning of nation build-
ing, although perhaps with a less strong ideological link between language and
nation than we come to see from the nineteenth century onwards (Anderson 2006;
Burke 2004, pp. 160–172). We also see the first contours of a growing British
Empire, with settlement colonies in North America from the early seventeenth
century. Although these colonies at first retained close connections to Britain,
they soon grew, became increasingly self-sufficient, and developed social group
identities separate from those in the British Isles.3 This meant that we can see
the beginnings of growing numbers of varieties of English, which develop along
separate paths.

3 The colonisation of Ireland from the twelfth century, and the incorporation of Scotland and Eng-

land under one ruler in 1603, were earlier beginnings of Empire. The smaller distance and closer
connections, however, meant that the dynamics of development of these new varieties of English
were different from those in the Early Modern period and later.
10 1 So What Had Happened Was

Linguistic Sketch

The form of English in the Early Modern period can be exemplified by this frag-
ment from the Geneva Bible (1560). This is relatively early in the period; the
translation pre-dates the King James Version by about fifty years, and is less than
200 years younger than the Late Middle English translation by Wycliffe.

12 And when the day began to weare away, the twelue came, and sayd vnto him, Sende the
people away, that they may goe into the townes and villages round about, and lodge, and
get meate: for we are here in a desart place. 13 But he sayd vnto them, Giue ye them to
eate. And they sayd, We haue no more but fiue loaues and two fishes, except we should go
and buy meate for all this people. 14 For they were about fiue thousand men. Then he sayde
to his disciples, Cause them to sit downe by fifties in a company. 15 And they did so, and
caused all to sit downe. 16 Then he tooke the fiue loaues, and the two fishes, and looked vp
to heauen, and blessed them, and brake, and gaue to the disciples, to set before the people.
17 So they did all eate, and were satisfied: and there was taken vp of that remained to them,

twelue baskets full of broken meate.

We can see that the last remaining verb endings have disappeared: came and sayd
for earlier camen and seiden. The spelling in this fragment is considerably closer
to what we are used to today: the only major difference is the distribution of <v>
word-initially and <u> elsewhere, where present-day English has reassigned these
symbols to represent a consonant and a vowel, respectively. The fragment also
sometimes uses <y> where we would write <i>. Finally, there are still some cases
of <e> at the end of words, but these were almost certainly not pronounced.
What we cannot learn from the spelling itself, but what is clear from meta-
linguistic comments as well as rhyming patterns in the literature, is that English
underwent a significant change in the pronunciation of long vowels, known as
the Great Vowel Shift. This sound change has been given almost mythical status,
perhaps undeservedly so (see section “The Great Vowel Shift” in Chapter 4). It is,
however, one of the earliest examples in English of a standardised spelling system
obscuring a major sound change.
In addition to this change in the phonology, there are also a few important
grammatical changes in this period. One is the introduction of the auxiliary verb
do in questions, negations, and emphatic statements. In the early stages of this
change, it was also used in simple affirmative statements, as in they did all eat
(v. 17), but it disappeared from those contexts again before the usage had gained
currency.
Another change, which unfortunately is not exemplified in the fragment, is
the disappearance of the informal second-person singular pronoun thou. This was
replaced by the second-person plural pronoun ye (see v. 13), which was also used
to address individuals in more formal contexts. The form ye itself was eventually
replaced by you, originally the object form of the pronoun. The singular pronoun
thou was still in regular use when the King James Version of the bible was trans-
lated. According to the conventions of the time, informal thou was the appropriate
form to address God by, as it indexed trust, intimacy, and affection. Now that we
Late Modern English (1700–Now) 11

no longer use thou, the archaic and formal nature of the KJV has given the pronoun
an air of formality that is completely anachronistic.

Research Highlight
CONTACT AND CHAOS CAUSE CHANGE
Language change is inevitable, but different languages change at different
speeds, and the rate of change can also change at different time periods in
the history of a single language. Linguists have been trying to find out what
causes these differences.
One explanation is that languages with less drastic change tend to be
geographically isolated. The Old Norse spoken around the year 1000 was
roughly the same across all of Scandinavia, but if we look at present-day
Scandinavian languages, geographically remote Icelandic is much more sim-
ilar to Old Norse than Danish, spoken close to the European heartland, is.
Within Norwegian and Swedish, it is the dialects spoken in more isolated
valleys that retain archaic features like the dative case.
The driver of change in this scenario is language contact, in particular
contact scenarios that involve adult learners of the language. Trudgill (2012)
has argued that complex features disappear in adult second-language acqui-
sition, and he posits that this sociolinguistic situation may also lead to faster
language change (Trudgill 2020).
Another explanation is that language change accelerates in situations of
social upheaval. This view sees language as a ‘punctuated equilibrium’: it is
roughly stable until some event disturbs the balance and a lot of language
change happens suddenly. This would explain why there was rapid change in
English in the fourteenth and fifteenth centuries—caused by the Black Death
and the Hundred Years’ War—but relative stability in the eighteenth century
(Trudgill 2020, p. 8). An analysis of language change across many features in
Middle and Early Modern English suggests that there are indeed periods of
rapid change that can be temporally linked to major social events. In addition
to the Black Death and the Hundred Years’ War, the Norman Conquest also
triggered a period of rapid language change. With social upheaval as well as
language contact, how could it not have?

Late Modern English (1700–Now)

Socio-Historical Context

The standardisation processes that were initiated in the Early Modern period con-
tinued into the most recent period, which despite its fairly long time span we will
call Late Modern English, or sometimes ‘present-day’ English. In particular the
eighteenth century was an important time for standardisation, with the publication
12 1 So What Had Happened Was

of Samuel Johnson’s Dictionary of the English Language (1755) and, at least as

importantly, a whole host of grammars and elocution manuals which promoted
the prestigious standard or ‘polite’ language to the growing, upwardly mobile
and increasingly literate and educated middle classes (Watts 2011, pp. 204–206).
For the first time, we see an organised effort to decrease linguistic diversity and
change. This prescriptivist movement has greatly influenced people’s attitudes to
language, and although linguists typically favour a descriptive approach, the strong
standard language ideology that prescriptivism has engendered may even have to
power to itself change people’s language habits (Curzan 2014).
The United Kingdom was also the earliest location of the Industrial Revolu-
tion, which led to the urbanisation of settlement patterns; this meant more people
speaking fewer distinct varieties of English. Easier communication between urban
centres, first by rail, later by road, air, or by ever increasing telecommunication
technologies, facilitates contact, which is a force for linguistic homogeneisation as
well. On the other hand, Anglophone mediatised culture and education has also
created more and more localised second-language varieties of English, each with
their own characteristics (see section “English as a Lingua Franca” in Chapter 7).
The eighteenth and nineteenth centuries also saw a marked increase in colonisa-
tion efforts. Settlement colonies were established in South Africa, Australia, New
Zealand, and a number of smaller locations; the predominantly British migrants
to these colonies in time developed their own varieties of English (see section
“New-Dialect Formation in Colonial Englishes” in Chapter 7). The same is true
for the extraction colonies set up in this period in Africa, South and South East
Asia, and the Caribbean, where the number of British settlers was smaller, but
where they did take control of government, the economy, and education. The slave
trade, predominantly but not exclusively across the Atlantic Ocean, led to many
enslaved people from Africa being displaced to the Americas, where they devel-
oped English-lexifier creoles (see section “Koinéisation in Creole Languages” in
Chapter 7). Empire, therefore, was a force for linguistic diversification.
The keyword in this most recent period, then, is contact; but whether the contact
results in greater or lesser linguistic diversity depends on the individual circum-
stances. The differences between varieties of English give us many clues about
how English must have changed in the past few centuries, and also tell us a lot
about how language changes in general.

Linguistic Sketch

The standardisation of spelling and grammar also makes it difficult to trace lan-
guage change in the period since 1700, although more and more work is being
done on informal private letters and (in the past twenty years) social media posts,
which can show a much more variable and fragmented picture of language in this
period. While none of the language from c. 1700 and later will look exceptionally
strange to us, there are constant small incremental changes so that a more recent
bible translation still looks different from the Early Modern version.
Sources for English Historical Linguistics 13

12 The day was drawing to a close, when the twelve came up to him, and said, “Send the
crowd away, so that they may make their way to the villages and farms around about, and
find themselves lodgings and provisions, for we are in a lonely spot here.” 13 But Jesus said,
“It is for you to give them something to eat.” “We have not more than five loaves and two
fish,” they answered. “Unless indeed we are to go and buy food for all these people.” 14 (For
the men among them were about five thousand.) “Get them seated in companies,” was his
reply, “about fifty in each.” 15 This they did, and got all the people seated. 16 Taking the five
loaves and the two fish, Jesus looked up to heaven and said the blessing over them. Then he
broke them in pieces, and gave them to his disciples to set before the people. 17 Everyone
had sufficient to eat, and what was left of the broken pieces was picked up – twelve baskets.

This translation, from the Open English Bible (2010), is in a relatively formal
register which alludes to the language from Early Modern translations. In addition
to the standardised spelling, it is therefore difficult to spot language change in this
fragment. The only clear change is that the plural fishes has been replaced by fish;
but this is a detail in the grand scheme of things. But it would be a mistake to
say that there has been little to no language change in English since the 1700s!
There have been major changes in phonology (see Chapter 5), as well as more
subtle changes in morphology (e.g., the development of new second-person plural
pronouns as y’all and you guys to disambiguate the multiple meanings of you)
and in syntax (e.g., the development of the going to future marker, or the be like
quotative marker). We will revisit some of these in later chapters.

Sources for English Historical Linguistics

If we want to study language change in the history of English, we have to start

with the linguistic facts. The Bible passages in this chapter do a decent enough
job illustrating the major differences between the forms of English in different
periods, but we will typically want to look beyond them for a more in-depth study
into specific changes. If nothing else, a wider scope allows us to investigate more
linguistic contexts, as well as social and genre effects. This section will give a
brief overview of sources that have been used in the systematic study of English-
language change. There are of course a great many sources, and this overview
simply cannot be anywhere near comprehensive; it does, however, aim to cover
some of the more widely-used sources and source types, many of which are freely
available online.
A good starting point are dictionaries, chief among them the Oxford English
Dictionary (https://oed.com/). The OED offers an overview of the development
of forms and meanings for each lemma, with many examples from a variety of
sources. To this we can add dictionaries of specific varieties of English such as
A Dictionary of the Older Scottish Tongue (DOST) and The Scottish National Dic-
tionary (snd, both at Dictionaries of the Scots Language, https://dsl.ac.uk/), the
Dictionary of American Regional Englishes (DARE, https://www.daredictionary.
com/, see Adams 2013), the Dictionary of Canadianisms on Historical Principles
(DCHP-2, https://www.dchp.ca/dchp2/, see Dollinger 2022). Such sources allow
14 1 So What Had Happened Was

us to chart, for example, the uptake of words from Aboriginal languages in the
Australian National Dictionary (Dixon 2008), or the lexical influence of German
on American English in regions with a high proportion of German immigrants (von
Schneidemesser 2002, on the basis of DARE). For Old English, the online version
of Bosworth Toller’s Anglo-Saxon Dictionary (Bosworth et al. 2014, https://bos
worthtoller.com/) offers many examples and links to primary sources.
Linguistic atlases are another good set of sources. Three major atlas projects
from the University of Edinburgh’s Institute for Historical Dialectology (now the
Angus McIntosh Centre for Historical Linguistics, http://www.amc.lel.ed.ac.uk/)
deserve particular attention: the electronic version of the Linguistic Atlas of Early
Middle English (eLAEME), the Linguistic Atlas of Late Middle English (LALME),
and the Linguistic Atlas of Older Scots (LAOS). These are based on large collec-
tions of diplomatically transcribed manuscripts and have greatly helped the study
of earlier stages of English since the first inception of laeme in the 1950s. Among
the linguistic atlases for more recent stages of the language, we can mention those
based on the 1950s Survey of English Dialects (SED): The Linguistic Atlas of
England (Orton et al. 1978) and An Atlas of English Dialects (Upton and Widdow-
son 2006). The Atlas of North American English (Labov et al. 2006) is based on
telephone surveys from the 1990s.
For primary research, the most useful tool are corpora: searchable collections
of texts or, in more recent times, speech recordings. Many of the dictionaries and
atlases mentioned above also give access to the underlying corpora, in addition to
providing more processed and ready-made interpretations. Two corpora from the
University of Helsinki have been extremely informative for our understanding of
language change in English. The Helsinki Corpus of English Texts (led by Matti
Rissanen) contains texts from c. 730 to 1710, and the Corpus of Early English
Correspondence (led by Terttu Nevalainen) spans the period from 1400 to 1800.
Information on both corpora can be found at https://varieng.helsinki.fi/CoRD/. Cor-
pora of legal proceedings also provide a good record of variation in written orality,
e.g., the Old Bailey Corpus (1720–1913; Huber et al. 2016), An Electronic Text Edi-
tion of Depositions 1560–1760 (Kytö et al. 2011), and the Records of the Salem
Witch-Hunt (Rosenthal 2009; Grund 2021).
Widely-used corpora of contemporary English are available at https://www.
english-corpora.org/: the British National Corpus (BNC, 100 million words from
the 1980s and 1990s), the Corpus of Contemporary American English (COCA,
one billion words from 1990 to 2019), the Corpus of Global Web-based English
(GloWbE, 1.9 billion words from the 2010s), and more. The Vienna-Oxford Inter-
national Corpus of English (VOICE 2021) focuses on English as a Lingua Franca
(see section “English as a Lingua Franca” in Chapter 7), with linguistic produc-
tion of non-native speakers of English. Although not publicly available, William
Labov’s Philadelphia Neighborhood Corpus (see Labov et al. 2013, pp. 33–34
for details) also merits a mention here: it was primarily recorded over a span of
more than 40 years by Labov’s students, each adding to the corpus as part of their
coursework.
References 15

All in all, English is an extremely high-resource language, with many corpora

and an excellent research infrastructure in place even for the oldest stages of the
language.

From Sources to Explanations

This first chapter has given a general overview of the development of the English
language in the c. 1500 years from the settlement of Angles, Saxons, and Jutes in
England to its current status as the default international language. It was of course
impossible to chart every single change in detail; it is clear that there has been a
major restructuring of the language’s lexicon, phonology, morphology, and syntax.
The socio-historical context for these changes has often been one of social flux and
of contact—between English and other languages, or between different varieties
of English.
But the purpose of this book is not to describe individual changes in English,
but rather to focus more on the general linguistic processes by which these changes
can be explained. Part II of the book deals explanations for sound change, both
on the phonetic level of pronunciations and on the phonological level of sound
systems. Part III discusses theory of contact-induced change, with an account of
various language and dialect contact situations in the history of English. Part IV,
finally, focuses on structural change in the language, with potential explanations
for changes in morphology and syntax.
This introductory Part I continues in the next chapter with an exploration of
what it means to find an explanation for language change, and where such answers
may be found.

References

Adams, Michael. 2013. Introduction: DARE from now on. American Speech 88 (3): 257–260.
https://doi.org/10.1215/00031283-2413957.
Anderson, Benedict. 2006. Imagined communities: Reflections on the origin and spread of
nationalism, 2nd ed. London: Verso.
Bosworth, Joseph, Thomas Northcote Toller, Christ Sean, and Ondřej Tichy, eds. 2014. An Anglo-
Saxon dictionary online. Prague: Faculty of Arts, Charles University. https://bosworthtoller.
com/.
Burke, Peter. 2004. Languages and communities in Early Modern Europe. Cambridge: Cambridge
University Press.
Curzan, Anne. 2014. Fixing English: Prescriptivism and language history. Cambridge: Cambridge
University Press. https://doi.org/10.1017/CBO9781139107327.
Dixon, R.M.W. 2008. Australian aboriginal words in dictionaries: A history. International Journal
of Lexicography 21 (2): 129–152. https://doi.org/10.1093/ijl/ecn008.
Dollinger, Stefan. 2022. Canadian English lexis and semantics: A historical-comparative resource
in contrastive, real-time perspective, 1683–2016. In Earlier North American Englishes, ed.
Merja Kytö and Lucia Siebers, 205–230. Amsterdam: Benjamins. https://doi.org/10.1075/veaw.
g66.09dol.
16 1 So What Had Happened Was

Faulkner, Mark. 2020. Quantifying the consistency of “standard” Old English spelling. Transac-
tions of the Philological Society 118 (1): 192–205. https://doi.org/10.1111/1467-968X.12182.
Grund, Peter J. 2021. The sociopragmatics of stance: Community, language, and the witness
depositions from the Salem witch trials. Amsterdam: Benjamins. https://doi.org/10.1075/pbn
s.329.
Hadley, Dawn M. 2000. The Northern Danelaw: Its social structure, c. 800–1100. London:
Leicester University Press.
Huber, Magnus, Magnus Nissel, and Karin Puga. 2016. Old Bailey corpus 2.0. http://hdl.handle.
net/11858/00-246C-0000-0023-8CFB-2.
Kytö, Merja, Peter J. Grund, and Terry Walker. 2011. Testifying to language and life in Early
Modern England. Amsterdam: Benjamins.
Labov, William, Sharon Ash, and Charles Boberg. 2006. The Atlas of North American English:
Phonetics, phonology and sound change. Berlin: Mouton de Gruyter. https://doi.org/10.1515/
9783110167467.
Labov, William, Ingrid Rosenfelder, and Josef Fruehwald. 2013. One hundred years of sound
change in Philadelphia: Linear incrementation, reversal, and reanalysis. Language 89 (1): 30–
65.
Nevalainen, Terttu, and Helena Raumolin-Brunberg. 2003. Historical sociolinguistics: Language
change in Tudor and Stuart England. London: Longman.
Orton, Harold, Stewart Sanderson, and J.D.A. Widdowson, eds. 1978. The Linguistic Atlas of
England. London: Routledge.
Rosenthal, B., ed. 2009. Records of the Salem Witch-Hunt. Cambridge: Cambridge University
Press. https://doi.org/10.1017/9781107589766.
von Schneidemesser, Luanne. 2002. Settlement history in the United States as reflected in DARE:
The example of German. American Speech 77 (4): 398–418. https://doi.org/10.1215/00031283-
77-4-398.
Smyth, Alfred P. 1998. The emergence of English identity, 700–1000. In Medieval Europeans, ed.
Alfred P. Smyth, 24–52. London: Macmillan. https://doi.org/10.1007/978-1-349-26610-4_2.
Timofeeva, Olga. 2022. Sociolinguistic variation in Old English: Records of communities and
people. Amsterdam: Benjamins. https://doi.org/10.1075/ahs.13.
Trudgill, Peter. 2012. On the sociolinguistic typology of linguistic complexity loss. In Potentials of
language documentation: Methods, analyses, and utilization, ed. Frank Seifart, Geoffrey Haig,
Nikolaus P. Himmelmann, Dagmar Jung, Anna Margetts, and Paul Trilsbeek, 90–95. Honolulu:
University of Hawai‘i at Mānoa.
Trudgill, Peter. 2020. Sociolinguistic typology and the speed of linguistic change. Journal of
Historical Sociolinguistics 6 (2): 20190015. https://doi.org/10.1515/jhsl-2019-0015.
Upton, Clive, and J.D.A. Widdowson. 2006. An atlas of English dialects. London: Routledge.
VOICE. 2021. The Vienna-Oxford International Corpus of English (version VOICE 3.0 Online).
https://voice3.acdh.oeaw.ac.at.
Ward-Perkins, Bryan. 2000. Why did the Anglo-Saxons not become more British? The English
Historical Review 115 (462): 513–533. https://doi.org/10.1093/ehr/115.462.513.
Watts, Richard J. 2011. Language myths and the history of English. Oxford: Oxford University
Press.
Explaining Language Change
How Change Starts and Spreads
2

Introduction

The approach in this book is to take the facts of language change in the history of
English as a starting point for an exploration of why the language changed. How
did we get from one stage of the language to another? What kinds of language
changes do we encounter, and how do they come about or take root? The socio-
historical context of language change is definitely very important here, but we also
need to look at the machinery of language change and find out how we ended up
where we are now.
One thing that is important to realise is that we could just as well have ended
up with a version of the English language that looked completely different. Espe-
cially the older and more established textbooks on the history of English present
that history as a process that inevitably results in Received Pronunciation (RP) or
Standard Southern British English (SSBE). This is called a teleological approach
(from Greek télos ‘purpose’): the idea that natural occurrences can be explained
through purposeful design. But in fact, even if there are strong tendencies in how
language change works, in the end it operates fairly randomly. We can even see
this within English, as there are many varieties of English that are not SSBE. It
would be quite a coincidence if language change had missed its pre-defined tar-
get for almost all speakers of the language except a politically powerful privately
educated elite in the south of England…
Broadly speaking, the study of language change can be captured in two over-
arching questions. The first is how language change starts; this is known as the
actuation question. The second is how language change spreads; this is variably
known as the transition question (Weinreich et al. 1968, p. 101), the implementa-
tion question (Trask 1997, p. 281), or the propagation question (Croft 2000, p. 5).
These questions have kept linguists busy since the establishment of linguistics as
a modern science a century and a half ago. We are going to tackle them here,
starting with the easier one before we move to the more difficult one.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 17

R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_2
18 2 Explaining Language Change

Throughout the book we will be looking at individual changes from the history
of English to exemplify the linguistic processes under discussion, with the tacit
understanding that these examples may be extrapolated to other, similar cases.
This may seem risky: can a discussion of a change in twentieth-century Australian
English really inform our view of a different change that took place in twelfth-
century England? But according to the uniformitarian principle (Labov 1994,
p. 22), it is possible to use our knowledge of language variation and change in the
present to shed light on similar processes in the past. After all, there is no reason
to believe that the physiology or cognition of past speakers were meaningfully
different from those of present-day speakers, and for all the social change over the
centuries there has been ‘English’, human social relations are also underlyingly
very similar to what they once were. This means that whatever present-speakers
do, past speakers may also have done, and an extrapolation of current studies to
past situations has more advantages than it has disadvantages.

The Spread of Language Change

The question of how language change spreads is comparatively easy, because we

are dealing with an observable phenomenon. Whether we are looking at a new
word, a new syntactic construction, or a change in pronunciation, it is theoretically
possible to chart each use of the innovation: who uses it, where are they, and when
are they? This way we can see a language change spread through a society, through
space, and through time. Some studies using (geo-tagged and timestamped) data
from social media can do this in a very detailed way, charting change through
geographical areas (e.g., Grieve et al. 2018, 2019).
In the time before big data and computational linguistics, however, there were
already theories about how language change spread. These were modelled in the
form of trees or waves.

Trees

One of the better-known visualisations in historical linguistics is that of language

family trees. The analogy with family trees is intentional and pervasive: what these
trees purport to show are the ‘genetic’ relationships between languages. We talk
about ‘mother’, ‘daughter’, and ‘sister’ languages. For example, Old Norse is the
‘mother language’ of the modern Scandinavian languages, and English and Frisian
are ‘sisters’. Persian and Armenian are some of the ‘cousins’ of English, whereas
Turkish and Arabic belong to different families altogether.
The defining points in the tree model are splits, break-away moments when
a group of languages undergo a change that other, related languages do not
undergo. The defining change for Germanic languages (the language sub-family
that English belongs to, together with German, Dutch, Frisian, and the Scandi-
navian languages) was the (First) Germanic Consonant Shift, or Grimm’s Law; a
The Spread of Language Change 19

change that took place only in this set of languages some two millennia ago (see
section “Neogrammarian Change” in Chapter 3).
The splits in the tree model are not only linguistic, but also to some extent
demographic. The reasons why some languages did and other languages did not
undergo a change—or rather, why some groups of speakers did and other groups
of speakers did not—is that they must have split off from an earlier group, lost
contact, and continued developing the language in separate ways. There is some
evidence for this in the geographical patterning of language (sub-) families, as
more closely related languages are often spoken in areas close to each other. In
the context of the nineteenth century, when this theory was developed, the ideal
of coherent nations that shared a common language also supported the model of
trees and splits. However, the tree model is an abstraction and does not take into
account the smaller nuances of variation in humans and language. Essentialising
the family relationships between languages and, by extension, population groups
can have far-reaching consequences, as the misuse of this view of language and
history in the Third Reich shows (Hutton 1999).
Language family trees function fairly well as a broad-strokes model for the
ancient history of languages, to the extent that we can know this. But there are
also aspects of language change that cannot be captured in an approach that relies
entirely on diverging populations. Take Frisian, for example, a Germanic language
spoken in the northern Netherlands. If we look at the earliest history of the lan-
guage, Frisian shares many developments with English; this is why it is often
regarded as the closest relative of English. But if we look at what the language
looks like now, Frisian looks a lot more like Dutch than like English, as a result
of centuries of more or less intensive contact with Dutch after the initial split
from English. The tree analogy cannot account for this. Similarly, Norwegian
is often given an intermediate position between (older) West Scandinavian and
(newer) East Scandinavian, because of centuries of intensive contact with Danish.
Even without that contact, the transitions between Norwegian (West Scandinavian)
and Swedish (East Scandinavian) dialects are much more gradual than the all-or-
nothing splits-based tree model would dictate. The tree model also fails to account
for the large proportion of Romance-origin words in the English lexicon. English
is (correctly) seen as a Germanic language, but a pure splitting account cannot
explain why so many English words look more like French and Latin than like
German, Dutch, or Danish.

Waves

A contemporary rival for the family tree model is the wave model. This model
visualises the geographical spread of language change as waves in a pond, rippling
out from the centre if you throw a stone in the water. (Seismic waves from an
earthquake epicentre would be an equivalent but less rustic metaphor.) Of course
these do not need to be pure concentric circles as there might be obstacles in the
20 2 Explaining Language Change

way, such as bodies of water, mountain ridges, marshland, or social barriers to

communication.
The maximum geographical extent of a linguistic feature is bounded by a lin-
guistic boundary or isogloss. Sometimes we have bundles of isoglosses running
closely together in the same area, which would give us good reason to conceptu-
alise a ‘language’ boundary in that area, but just as often isoglosses go across
language boundaries. There are several features that Norwegian dialects share
with Swedish dialects, but not with other Norwegian dialects, for example. As
an extreme example, there are several features (e.g., the use of a definite arti-
cle suffix) that are shared by Romanian, Bulgarian, and Albanian, languages that
belong to different sub-families but which are spoken in a contiguous area in the
Balkans.
The wave model, therefore, emphasises contact over genetic similarity. Because
of its reliance on geography, it is less able to deal with migration and other popu-
lation upheaval: it cannot account for the differences between colonial varieties of
English, and that it works so well for Germanic and Romance languages is mainly
because these populations have been relatively sedentary since the migration period
of the fourth to sixth centuries.
Dialectological research from the second half of the twentieth century showed
that the waves-in-a-pond metaphor for the geographical spread of language change
was inaccurate. Changes would spread from one large population centre to another
large population centre, skipping the more sparsely populated area in between.
The smaller towns and countryside would adopt a change only after it had already
been firmly established in the nearby larger towns. Dialectologists Peter Trudgill
and Jack Chambers even attempted to catch this city-hopping pattern, or ‘grav-
ity model’, in a mathematical formula (Chambers and Trudgill 1998, p. 179).
According to the model, the linguistic influence of one place on another depends
on the population of both places and the distance between those places, and on
the pre-existing linguistic similarities between the places. We should probably not
place too much mathematical trust in this formula, and Chambers and Trudgill
themselves discuss some of its shortcomings. The formula does not deal well
with competing influences from different population centres and cannot account
for small villages being influenced by local towns more than metropolises such
as London that swamp them in population size. Moreover, their operationalisa-
tion of linguistic similarity was rather crude, and although there have been major
advances in quantifying linguistic distance (‘dialectometry’, see Nerbonne 2009,
for an accessible introduction) it is unclear whether the spread of language change
can really be quantified like this. The social factors that Cahmbers and Trudgill
included in their equation, however, do turn out to be relevant also in subsequent
research.
The Spread of Language Change 21

Research Highlight
NOT ALL DISTANCES ARE EQUAL
According to a pure wave model, language change would spread with an
equal speed in all directions. It follows that the linguistic distance between
two varieties should be more or less correlated with the geographical dis-
tance between these varieties. With a gravity model, the relative size of the
locations where the varieties are spoken should also play a role.
Nerbonne and Heeringa (2007) tested these hypotheses on the basis of a
dataset from over 50 locations in the north-east of the Netherlands. They
showed that there was a strong correlation between linguistic and geo-
graphical distance. Adding population size to their statistical model did not
meaningfully improve results.
But these results may be specific to the Netherlands, where the flat coun-
tryside makes travel unproblematic in all directions. In Norway, on the other
hand, there are barriers to travel in the form of mountains and fjords. There,
geographical distance correlated less well with linguistic distance, and a bet-
ter predictor was travel time—in particular, historical travel time from around
the year 1900 (Gooskens 2005). After all, more than the distance between
places it is the effort it takes to get somewhere that determines contact.
One of the latest additions to the distance measurements is cognitive dis-
tance: how far people think a place is. It turns out that, again for dialects in
the northern Netherlands where geographical distance is a very strong pre-
dictor, adding the idea that people have about distance results in a stronger
correlation with perceived linguistic distance (Sekeres 2022).

Social Networks

Research in sociolinguistics since the late 1970s has identified the role that social
networks play in the propagation of linguistic change; see, e.g., Milroy and Milroy
(1985) for an empirical account by two of the key researchers in this area. Social
networks are characterised by the interpersonal relationships we have with each
other. Network links can be weak or strong, depending on the number of different
types of social relationship we have with a person (as a family member, a neigh-
bour, a sports team mate, etc.) and the importance we place on the relationship.
Networks themselves can be dense or sparse, depending on how many people a
person has social relationships with, also have social relationships with each other.
It was found that stronger links and denser networks function as an enforcement
mechanism for (linguistic) conformity, while weaker links and sparser networks
are more open to outside (linguistic) influence.
We can link this to the wave and gravity models and to the elements in Trud-
gill’s mathematical formula. Smaller villages tend to have denser social networks
than larger cities and thus are somewhat shielded from linguistic change. Frequent
22 2 Explaining Language Change

contact between weak links in different cities, on the other hand, makes it easy
for an innovation to spread from one city to the other. Once a change has been
introduced on the periphery of a network, it can make its way to the core; if core
members of a network pick up on a change, it can spread rapidly like wildfire.
Smaller towns and villages that are slower to pick up on language change simply
do not have the necessary number of contacts with the places of innovation; they
do not have enough weak links to introduce changes into the core of the network,
which itself is potentially more resistant to change in the first place.

Spread Through the Speech Community

Like the geographical spread of a change is gradual, so too is the spread of a

change within a speech community. Whenever a change is introduced in a com-
munity—either because it originated there or because it arrived there through a
weak social network link—there first follows a period of variation during which
the new (or more advanced) form co-exists with the old (or less advanced) form.
The field of variationist sociolinguistics, which started as a search for the ‘empir-
ical foundations’ of language change (Weinreich et al. 1968), has now collected
evidence on the connection between language variation and language change for
over half a century, and has uncovered a number of patterns that recur so often
that they may be seen as foundational to how a change spreads through a speech
community.
Although a change spreads through a community gradually, the pattern is not
linear. Instead, it follows what is known as an S-curve. When a change is first
introduced, it spreads slowly, and the new variant is only used by an avant-garde of
speakers and/or in very specific contexts. At some point, the change gains traction,
and once it reaches critical mass we see a rapid spread of the new variant (of course
coupled with an equally rapid demise of the old variant). Towards the end of the
process, the change slows down again, and the old variant sees some residual use
before it eventually disappears. A schematic representation of such an S-curve is
given in Fig. 2.1. During the period of variation, the new variant may be used in
certain contexts or by certain speakers more than in other contexts or by other
speakers. The variation is ‘constrained’ by various factors in what Weinreich et al.
(1968) called ‘orderly heterogeneity’.
One such factor is, almost naturally, speaker age. As most language change
happens when language is generationally transmitted (Labov 2007), we can expect
younger speakers to use relatively more of the new variant, and older speakers to
use relatively more of the old variant. Historical linguistics has found the S-curve
pattern in changes in Early Modern English (Nevalainen and Raumolin-Brunberg
2003), but such ‘real-time’ approaches to the variationist study of language change
actually came after more practical ‘apparent-time’ approaches in which speaker
age was used as a proxy for time. Taking the speech of older people as a reflection
of the speech of earlier years in fact works remarkably well, and charting language
The Beginnings of Language Change 23

Fig. 2.1 Schematic representation of the S-curve pattern of the spread of a linguistic change
through a speech community

change against speaker age instead of time gives the same S-curve pattern as his-
torical real-time studies. There are of course speakers who change their speech
significantly during their lifespan and who may therefore disrupt such an S-curve,
but they are exceptional and the influence of lifespan change can be accounted for
in an apparent-time study (Sankoff and Blondeau 2007; Sankoff 2019).
Studies have also attempted to uncover who the leaders of language change
are, and what traits they share. At the level of individual speakers, this is as yet
an unanswered question (Tamminga 2021), but when it comes to groups that lead
language change, there is a clear pattern: in the western societies in which such
research is usually done, it is typically women, rather than men, who are the inno-
vators in language change (Labov 2001, pp. 274–275). Why this should be the
case, and in particular why women should be the leaders of language change both
towards prestigious standard variants and towards low-prestige non-standard vari-
ants, is a question that has never been satisfactorily answered. Perhaps the global
overarching category of ‘women’ hides a diversity in localised feminine identities
that may draw speakers towards one or the other variant; see Meyerhoff (2019,
pp. 232–249) for a discussion.

The Beginnings of Language Change

Whereas the spread of change is an observable phenomenon, the actuation of

change is not (Yu 2023). With the possible exception of some phrases that develop
into internet memes, it is nearly always impossible to tell when, where, and by
24 2 Explaining Language Change

whom an innovation was first used, let alone what caused the innovation. We have
already seen that much uncertainty remains about the transition question, so this
would suggest that we should be completely in the dark about the actuation ques-
tion. However, we have come a long way, and possible answers to the actuation
question will pop up frequently in the next chapters. The following serves as a
preview of what is to come.

Early Approaches

Like with the propagation question, scholars have been engaging with the actuation
question for a very long time. Inevitably, this means that there have now been quite
a few theories that have been proved to be wrong, or that are at least extremely
unlikely. Unfortunately, some of these persist in the wider population. We will
briefly deal with them here so that you are better equipped to do your civic duty
and can correct people when they talk nonsense about linguistics.
One set of pseudo-explanations sees climate and geography as a driving force in
language change. The cold and humid North Sea climate was blamed for a change
from /A / to /O / in Old English: it requires less opening of the mouth, so you could
▼
▲
▼
▲

avoid breathing in the chilly air (Sweet 1900, p. 32). A similar explanation is used
for Grimm’s Law, in the earliest stages of Germanic languages, when voiceless
stops changed to voiceless fricatives, thought to have occurred around the time of
a climate change towards cooler weather.1 An alternative explanation may be (or
not!) that the early Germanic peoples lived in hilly or mountainous areas and were
constantly running out of breath from walking uphill, which automatically turned
their stops into fricatives—while neighbouring Slavic peoples on the plains stuck
to stops (Meyer 1901, pp. 118–119).
Another set of explanations sees speakers’ personalities as the cause of sound
change. Sweet (1900, p. 20), who came up with the North Sea climate expla-
nation as well, notes that even children are generally able to acquire even ‘the
difficult Semitic throat-sounds’; the mispronunciations leading to language change
must be ‘the result either of carelessness or sloth’. Sometimes, these qualities are
attributed to entire groups of people. A whole range of sound changes and loss
of morphological complexity in English-based creoles and African American Ver-
nacular English has been explained with reference to personal characteristics of
the speakers, in a not even subtly racist standard language ideology that persists
to the present day (Rickford and King 2016). At the same time, the ‘courageous’
stops-to-fricatives change in Proto-Germanic was linked by the German linguist

1 There appears to have been a period of cooling around 500 BC, so that part of the theory holds
up. But fricatives actually require a more open mouth than stops, so the theory does not make a
lot of sense. Also, you would be equally likely to close your mouth for sand storms in hot desert
climates, which makes the climate approach rather vacuous (Schrodt 1974, p. 201).
The Beginnings of Language Change 25

and fairy-tale collector Jacob Grimm to the ‘unstoppability of [the] advance into
all parts of Europe’ of ‘the invincible German race’ (Lightfoot 2013, p. e25).2
A final theory but at the same time one of the earliest ideas about language
variation and language change is the story of the Tower of Babel (Genesis 11: 5–9).
The people in the city of Babel were building a tower that reached into the heavens.
In order to restrain what people would be able to do, God made it so that they all
spoke different languages, could no longer understand each other, and moved to
different places. Many monotheistic religions, even some that did not originate in
the Middle East, have a story along similar lines. The story has mockingly been
re-named ‘wrathful dispersion theory’, in reaction to the attempted sciencification
of creationism as ‘intelligent design’. The Tower of Babel does not purport to be
an explanation for language change; it claims to explain language diversity, which
it correctly claims is related to a lack of regular communication between people.
It is simply not very precise about the direction of the link between the two.

Language Change, Language Use, and Language Users

Modern approaches to the actuation question (and in fact the transition question)
see language change as the direct result of language use, specifically language use
by human beings. This has physiological, cognitive, and social aspects. These will
be dealt with in much more detail in coming chapters, and the brief overview that
follows should serve as a preview of those discussions.
For simple physiological articulatory reasons, there is a lot of variation in
speech production. From an articulatory point of view, speech is nothing more
than a string of sounds in succession, so a string of articulatory gestures:3 raise
the tip of your tongue, round your lips, start vibrating your vocal folds, lower the
tip of your tongue again, etc. Because we are not machines, we will never be able
to produce exactly the same token of a sound twice. Try saying the sound /æ/
twenty times: while we simply hear twenty /æ/s, when you analyse them closely
you will see that they differ in precise height, frontness, pitch, and duration. As
we will see, even this mundane variation may be a trigger for sound change. Artic-
ulation becomes even less precise when you string individual sounds together and
have to deal with the transitions between sounds. Mistiming these transitions is
also a frequent cause of sound change.

2 Grimm almost seems to imply that the change was deliberate, although examples of (success-
ful) deliberate language change are exceptionally rare (cf. Thomason 2007). And where Grimm
described this change as ‘strengthening’, we shall see in later weeks that current linguistic thought
describes the same change as weakening, whatever that says about pre-historic Germanic tribes.
3 Outside articulatory phonetics, it may seem unusual to think of tongue movements as ‘gestures’,

which are more associated with signed languages. But the same argument applies to signed lan-
guages as to spoken languages: in that case, there are small variations in hand shape, position,
movement, etc., that may lead to change.
26 2 Explaining Language Change

The other physiological aspect that influences language change is speech per-
ception. Our hearing is not always optimal, and there may be disturbances such as
background noise that could cause misunderstanding.4 If these misunderstandings
stick, they are another source of change. Speech perception also brings us to more
cognitive explanations of change: while at one level we can often perceive small
differences in pronunciation, at another level our perception is categorical. The
twenty [æ]s are twenty different instantiations of the same category /æ/. How we
perceive boundaries between categories, and the different factors that play a role
in this, can also bring about sound change.
The main cognitive factor that we deal with in language is that humans tend
to categorise events and try to spot patterns. These become shortcuts in our brains
that make it easier and quicker to deal with the world. But sometimes we spot
patterns or make categories that do not exist for other language users, or even
force some structure onto language that was not previously there. We may do this
on an individual basis, but when we then produce new utterances based on that
conceptualisation of how language works, we add to the evidence that our patterns
and categories are the correct ones, and more people may adopt them.
Language learning, then, is also a cognitive activity. It is predominantly chil-
dren who learn language who are engaged in the conceptualisation of patterns and
categories, which is why we often think of language change as generational. But
we can also add or change patterns and categories in adulthood. This happens
sometimes in our first language (‘lifespan change’) but perhaps more frequently
also in our second language. When patterns and categories from different language
are active in our brains, this may lead to contact-induced language change.
A final cognitive factor in language change is that humans are inherently cre-
ative. Not only do we use the building blocks at our disposal to create new
utterances on a daily basis that have never been used in the history of humankind,
we also sometimes creatively change the building blocks themselves, or the rules
for putting these blocks together.

Language Change as Evolution

Since the turn of the twenty-first century, historical linguists have increasingly also
made use of theories from biological and cultural evolution. Although Charles Dar-
win already noted that his theory of evolution could apply to languages as well, the
historical linguistic theories in this framework are more properly seen in the light
of the work in cultural evolution, most notably Richard Dawkins’s The selfish gene
(Dawkins 1976). Evolutionary approaches to language change have been proposed
by Croft (2000, 2008), Ritt (2004), and Mufwene (2008); Kretzschmar’s (2015)
work on language as a complex system takes a related approach. While they differ
in details, the underlying principles are the same. The presentation here is based

4 For signed languages, the same applies to sight.

The Beginnings of Language Change 27

on Ritt (2004), who uses examples from the history of English to illustrate the
theory.
Darwinian evolution is based on variation in genetic structures, originating, for
example, from copying errors in replication. Certain structures from a varied pool
replicate better than others because they are ultimately more suited to their envi-
ronment. Linguistic evolution should be seen in the same way: certain variants
replicate better than others, which is how variation in language leads to change.
Ritt (2004, p. 89) stresses that he does not propose this as a metaphor for lan-
guage change (as, e.g., in Blevins 2004) but that he really sees evolution as the
mechanism through which language changes: languages exist in humans as ‘hosts’
and develop on their own. The question for linguistics, then, is what the unit of
replication is, and what pressures there are on the selection of one variant over
another.
In Dawkins’s view of evolution, the unit of replication is a meme. These are
networks of associations (with a neural basis in the brain) that we may think
of linguistically as existing of different components. What Ritt calls a ‘phone-
meme’ is linked in our minds to articulatory gestures, an auditory impression, and
the various lexical or morphological items in which it occurs. A ‘morph-meme’
contains phonological, syntactic, semantic, and suprasegmental information (Ritt
2004, pp. 171, 173). So a linguistic meme, in this sense, is not a picture of a cat
on the internet, but can be, for example, the phoneme /i /, the concept of a long
▼
▲

vowel, consonant-vowel syllable structure, or a strong-weak rhythmic structure.

Such linguistic memes replicate and evolve, not in language acquisition as some
earlier accounts of language change would have, but in language use (Croft 2008,
p. 222). Not all memes have the same changes of replication, and this ‘survival of
the fittest’ depends on a number of selective pressures. We have seen a number of
these pressures already: they can be, for example, the human articulatory system
and human auditory perception, or the frequency and social meaning of variants.
Memes are also influenced by surrounding memes (Ritt 2004, pp. 221–227). They
can even evolve together with other memes in order to increase their chances of
replication, in a process of meme co-adaptation (Ritt 2004, p. 237). This, just like
errors in replication, can be an evolutionary answer to the actuation question.
As an illustration of his evolutionary theory of language change, Ritt (2004,
pp. 240–288) discusses a set of changes to vowel length in Late Old and Early
Middle English. These are discussed in this book in Chapter 5 following a tra-
ditional rule-based account as four separate changes—homorganic lengthening,
pre-cluster shortening, open syllable lengthening, and trisyllabic shortening—each
with a fair number of exceptions. In Ritt’s account, these four changes are ulti-
mately one and the same process: ‘two memes [the long and short versions of the
same vowel] competing for membership in higher-level mimetic configurations’
(p. 241). The rhythmic structure of English in this period was strongly trochaic,
a strong syllable followed by a weak one. In order to best fit this structure, and
to increase their chances of replication in the syntactic and rhythmic contexts in
which they occurred most frequently, some words (morph-memes) teamed up with
a phone-meme of a different length. This also explains the many exceptions to the
28 2 Explaining Language Change

rule-based account: different combinations of pressures led to different statistical

likelihoods of outcomes (Ritt 2004, p. 274).
As this evolutionary account of language change shows, there are many possible
reasons why language may change, and at any given time a variety of them might
be at play simultaneously. We will return to these reasons both explicitly and
implicitly throughout this book as we look at concrete changes in the history of
English. At times the actuation and transition questions may take centre stage,
but we may also sometimes discuss the smaller details of a change and then it is
up to the reader to think about where the changes fit into the larger questions of
historical linguistics.

References

Blevins, Juliette. 2004. Evolutionary phonology: The emergence of sound patterns. Cambridge:
Cambridge University Press.
Chambers, J.K., and Peter Trudgill. 1998. Dialectology, 2nd ed. Cambridge: Cambridge University.
Croft, William. 2000. Explaining language change: An evolutionary approach. Harlow: Longman.
Croft, William. 2008. Evolutionary linguistics. Annual Review of Anthropology 37: 219–234.
https://doi.org/10.1146/annurev.anthro.37.081407.085156.
Dawkins, Richard. 1976. The selfish gene. Oxford: Oxford University Press.
Gooskens, Charlotte. 2005. Travel time as a predictor of linguistic distance. Dialectologia et
Geolinguistica 13: 38–62. https://doi.org/10.1515/dig.2005.2005.13.38.
Grieve, Jack, Andrea Nini, and Diansheng Guo. 2018. Mapping lexical innovation on American
social media. Journal of English Linguistics 46 (4): 293–319. https://doi.org/10.1177/007542
4218793191.
Grieve, Jack, Chris Montgomery, Andrea Nini, Akira Murakami, and Diansheng Guo. 2019. Map-
ping lexical dialect variation in British English using Twitter. Frontiers in Artificial Intelligence
2: 11. https://doi.org/10.3389/frai.2019.00011.
Hutton, Christopher M. 1999. Linguistics and the Third Reich: Mother-tongue fascism, race and
the science of language. London: Routledge.
Kretzschmar, William A., Jr. 2015. Language and complex systems. Cambridge: Cambridge Uni-
versity Press.
Labov, William. 1994. Principles of linguistic change. Volume 1: Internal factors. Oxford:
Blackwell.
Labov, William. 2001. Principles of linguistic change. Volume 2: Social factors. Oxford: Blackwell.
Labov, William. 2007. Transmission and diffusion. Language 83 (2): 344–387. https://doi.org/10.
1353/lan.2007.0082.
Lightfoot, David W. 2013. Types of explanation in history. Language 89 (4): e18–e38.
Meyer, Heinrich. 1901. Über den Ursprung der germanischen Lautverschiebung. Zeitschrift für
deutsches Altertum und deutsches Litteratur 45: 101–128.
Meyerhoff, Miriam. 2019. Introducing sociolinguistics, 3rd ed. London: Routledge.
Milroy, James, and Lesley Milroy. 1985. Linguistic change, social network and speaker innovation.
Journal of Linguistics 21 (2): 339–384. https://doi.org/10.1017/S0022226700010306.
Mufwene, Salikoko S. 2008. Language evolution: Contact, competition and change. London:
Continuum.
Nerbonne, John. 2009. Data-driven dialectology. Language and Linguistics Compass 3 (1): 175–
198. https://doi.org/10.1111/j.1749-818x.2008.00114.x.
Nerbonne, John, and Wilbert Heeringa. 2007. Geographic distributions of linguistic variation
reflect dynamics of differentiation. In Roots: Linguistics in search of its evidential base, ed.
References 29

Sam Featherston and Wolfgang Sternefeld, 267–298. Berlin: Mouton de Gruyter. https://doi.
org/10.1515/9783110198621.267.
Nevalainen, Terttu, and Helena Raumolin-Brunberg. 2003. Historical sociolinguistics: Language
change in Tudor and Stuart England. London: Longman.
Rickford, John R., and Sharese King. 2016. Language and linguistics on trial: Hearing Rachel Jean-
tel (and other vernacular speakers) in the courtroom and beyond. Language 92 (4): 948–988.
https://doi.org/10.1353/lan.2016.0078.
Ritt, Nikolaus. 2004. Selfish sounds and linguistic evolution: A Darwinian approach to language
change. Cambridge: Cambridge University Press.
Sankoff, Gillian. 2019. Language change across the lifespan: Three trajectory types. Language 95
(2): 197–229. https://doi.org/10.1353/lan.2019.0029.
Sankoff, Gillian, and Hélène. Blondeau. 2007. Language change across the lifespan: /r/ in Montreal
French. Language 83 (3): 560–588. https://doi.org/10.1353/lan.2007.0106.
Sekeres, Hedwig. 2022. Out of sight, out of mind? A cognitive geographic approach to dialectology.
Rijksuniversiteit Groningen MA thesis, Groningen.
Schrodt, Richard. 1974. Die germanische Lautverschiebung und ihre Stellung im Kreise der
indogermanischen Sprachen. Wien: Halosar.
Sweet, Henry. 1900. The history of language. London: Dent.
Tamminga, Meredith. 2021. Leaders of language change: Macro and micro perspectives. In Lan-
guage variation—European perspectives: Selected papers from the Tenth International Confer-
ence on Language Variation in Europe (ICLaVE 10), Leeuwarden, June 2019, ed. Hans Van de
Velde, Nanna Haug Hilton and Remco Knooihuizen, 269–289. Amsterdam: Benjamins. https://
doi.org/10.1075/silv.25.12tam.
Thomason, Sarah G. 2007. Language contact and deliberate change. Journal of Language Con-
tact—Thema 1: 41–62.
Trask, R.L. 1997. Historical linguistics. London: Arnold.
Weinreich, Uriel, William Labov, and Marvin I. Herzog. 1968. Empirical foundations for a the-
ory of language change. In Directions for historical linguistics: A symposium, ed. Winfred P.
Lehmann and Yakov Malkiel, 95–188. Austin: University of Texas Press.
Yu, Alan C. L. 2023. The actuation problem. Annual Review of Linguistics 9: 215–231. https://doi.
org/10.1146/annurev-linguistics-031120-101336.
 Part II
Sound change
Phonetic Change
Why Sound Change Is Both Regular and Random
3

Introduction

∫traight to the Point

Consider the words state and straight. At first glance, the only difference between
the two is that straight has an /r/ sound while state does not. But if you listen
closely, you will find that for many speakers of English there is an additional dif-
ference: whereas state is just [steIt], straight sounds more like [∫tôeIt]. In other
words, it sounds like the /s/ in straight is pronounced as [∫]. Perception exper-
iments and acoustic measurements (Rutter 2011; Stevens and Harrington 2016)
show that the [∫] in straight is very similar to or even indistinguishable from the
[∫] in shake /∫eIk/, where you would expect it in the first place.
If we want to understand why the /s/ in straight is often pronounced as [∫],
we need to take a much closer look at what happens in our mouth when we
articulate this word. Let us first look at the articulation of the sequence /tr/, as
in tray. To articulate [t], you raise the tip of your tongue towards your alveolar
ridge. To articulate the next sound [ô], which is the most common realisation of
/r/ in English, the tip of your tongue is retracted and perhaps curled back but still
roughly at the same height as your alveolar ridge. As your tongue transitions from
the [t] to the [ô], it briefly passes a position just behind your alveolar ridge: the
exact place of articulation of the post-alveolar fricative [∫]. This means that for
many speakers of English, the word tray begins with the sequence [t∫ô]. Now add
an /s/ to the beginning of that sequence, as you would do in straight. It is very
likely that your brain is already preparing your tongue for that [t∫ô] that follows,
and you do not quite hit the exact position just below your alveolar ridge that is
necessary to articulate an [s]. Instead, you end up slightly short of that position,
again in the place of articulation of [∫]. Rather than [stô], the word straight actually
begins with [∫t∫ô] for many speakers.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 33

R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_3
34 3 Phonetic Change

This so-called s-retraction is a feature found in many speakers of English around

the world. It has been reported for the United States (Lawrence 2000; Baker et al.
2011; Rutter 2011), the United Kingdom (Lindsey 2019), Australia (Stevens and
Harrington 2016), and Trinidad and Tobago (Ahlers and Meer 2019) and likely
occurs in other varieties as well. Because we can pinpoint the exact articulatory
processes that give rise to s-retraction, it serves as a prototypical example of sound
change that we will return to at several places in this chapter.

Chapter Overview

There are several reasons why we begin our search for an answer to the actuation
question by looking at sound change. The first is that sounds are the most elemen-
tary building blocks of language, and sound change has historically been seen as
the purest form of language change. By starting here, we simply follow tradition.
A more important reason is that, as we have seen for s-retraction, sound change
can often be quite transparently linked to physiological factors, giving us clear and
tangible explanations that may not be available for lexical and syntactic change.
As we will see in this chapter and the next, however, physiological factors play a
role in phonetic change, i.e., the type of sound change where it is mostly irrele-
vant that sounds together form a linguistic system of phonemic contrasts. When
we look at phonological change, the other type of sound change, physiological
factors will take a back seat to cognitive factors.
The main thought that will be presented in this chapter is that sound change
is regular. We will continuously be chipping away at this notion, but with a few
important caveats, regularity is still a cornerstone of the study of sound change.
Once it happens, that is, because the one thing that is unpredictable about sound
change is when it occurs.

Sound Change Is Regular

Historical Linguistics as a Science

The long nineteenth century was a period of great progress in our understanding of
language change. A lot of this was due to our improving understanding of science
in general and the application of the scientific method to linguistic research. In
their search for the oldest, most original form of their language—a quest jointly
inspired by the nineteenth-century ideologies of national-romanticism and colo-
nialism—linguists used the comparative method. By systematically comparing
languages, they aimed to, firstly, establish whether similarities between languages
pointed to a ‘genetic’ relationship between them, and secondly, try to reconstruct
what a proto-version of the language(s) would have been like.
Sound Change Is Regular 35

Table 3.1 Regular correspondences in five Germanic languages

English German Dutch Norwegian Icelandic
fish Fisch vis fisk fiskur
find finden vinden finne finna
foot Fuß voet fot fótur
freeze frieren vriezen fryse frjósa
thin dünn dun tynn þunnur
thorn Dorn doorn torn þorn
three drei drie tre þrír
thief Dieb dief tyv þjúfur
hound Hund hond hund hundur
house Haus huis hus hús
high hoch hoog høy hár
hand Hand hand hånd hönd

The evidence we need for this comes from regular correspondences: a sound in
language must generally and systematically correspond to a sound in another lan-
guage, whatever word you are looking at. Take, for example, the words in Table 3.1
from five languages in North-West Europe. If you look at the initial consonants in
these words, the similarities are striking and the patterns are almost immediately
clear. English f - corresponds to f - in German, Norwegian, and Icelandic, and to
v- in Dutch; English th- [θ] corresponds to d- in German and Dutch, to t- in Nor-
wegian, and to þ- [θ] in Icelandic; and English h- corresponds to h- in all four
other languages. There are many more words that can be added to this table, and
many more sets of consonants for which such a table can be set up.
The systematicity in these correspondences suggests that these five languages
share a common ancestor. Wherever there was a sound change, such as the change
[θ] > [d] in Dutch, it applied regularly. In the relatively infrequent cases where the
correspondence does not work, this is very often because we are talking about a
loan word. For example, the German and Dutch equivalents for throne and theatre
do not begin with /d/ but with /t/ because these are both loans from late medieval
French where the word was pronounced with initial /t/. (The /θ/ in English appears
to be a learned over-etymologising spelling pronunciation based on the Greek
origin of these words; when English borrowed this word from French, it was orig-
inally also with /t/.) An alternative reason for a lack of correspondence, especially
if the exceptions themselves also seem to be regular and systematic, can be that
subsequent sound changes have obscured the regularity (more on this below). It
is also important to note that in order to find such regular correspondences, you
cannot always use direct translations of the words. The best English translation of
German Hund would be ‘dog’, but it is clear that ‘hound’ is the word that is the
‘cognate’, the word that is genetically related. The semantic shifts that we find in
36 3 Phonetic Change

cognates can sometimes be greater than in ‘dog’ ~ ‘hound’, but they are always in
some way transparently related.
The similarities between closely related languages (such as the Germanic lan-
guages in Table 3.1, but also Romance and Slavic languages, for example) are
obvious and were of course well known already before the nineteenth century.
The broader, more distant relationships were not, however. The study of Sanskrit,
in comparison with Latin and Ancient Greek, was one of the factors that helped
bring to light these distant relationships. Bengal-based British judge Sir William
Jones suggested in a 1786 lecture that the similarities were so striking that this
could not be a coincidence. Just like Romance languages have a common ancestor
in (Vulgar) Latin, Sanskrit, Latin, and Ancient Greek must have had a common
ancestor too. This ancestor language is now known as Proto-Indo-European (PIE).
One of the goals of historical linguistics, or ‘philology’ as it was often called at
the time, became to reconstruct the lexicon, phonology and grammar of PIE, and
chart the language changes that led to the split into the Indo-European language
sub-families. The overviews of regular correspondences in related languages form
a synchronic record of the diachronic changes that have taken place in these lan-
guages. With some logical thinking, we can then hypothesise in what order these
changes must have taken place and follow the path back to an earlier version of
the language.

Problems for the Regularity Hypothesis

Neogrammarian Change

A group of linguists, predominantly writing in German, went to work try-

ing to reconstruct these different earlier stages of the languages of Europe:
Proto-Indo-European, Proto-Germanic, Proto-Celtic, etc. They are known as the
Neogrammarians or Junggrammatiker, which roughly translates as ‘new-style lin-
guists’. They used the idea of completely regular sound changes as a guiding
principle and were able to capture the major divergences that occurred between
language sub-families in a series of ‘sound laws’; there is a wide variety of regu-
lar historical sound changes that are now known as Someone’s Law. The influence
of the neogrammarians on the study of sound change was so large that you often
see the term ‘neogrammarian change’ for any regular, exceptionless sound change.
The most important sound change that sets Germanic languages apart from
other Indo-European languages is the change known as Grimm’s Law, named after
Jacob Grimm, who is otherwise known as a fairy-tale collector. (It was actually
Friedrich von Schlegel and Rasmus Rask who described the sound change first,
but Grimm today gets the credit for it, in part because Rask wrote in Danish and
his publications were not as widely known. In Denmark, the change is typically
known as Rask-Grimm’s Law.) Grimm’s Law consists of three sets of changes:
Problems for the Regularity Hypothesis 37

1. The PIE voiceless stops /p t k/ become voiceless fricatives /f θ h/ in Proto-

Germanic, and this change does not happen in other Indo-European languages.
Compare Latin piscis ‘fish’, trēs ‘three’ and canis ‘hound’ with their English
translations as examples.
2. The PIE voiced stops /b d l/ become voiceless stops /p t k/ in Proto-Germanic,
and do not undergo this change in other Indo-European languages. Examples
here are Latin duo ‘two’ and genū ‘knee’. (There is no evidence of word-initial
/b/ in PIE, but that change has been attested in different positions in a small
number of words.)
3. The PIE voiced aspirated stops /bh dh lh / become voiced stops /b d l/ in Proto-
Germanic. Here we have to compare Sanskrit bhrātar ‘brother’ and dhā- ‘put’,
cognate with English ‘do’, as examples. Comparisons with Latin are less obvi-
ous, because these voiced aspirated stops also did not make it into Latin: the
equivalents are frāter and fēci ‘I made’, both with /f/.

Germanic languages all show Grimm’s Law, while non-Germanic Indo-European

languages do not. The regularity of the pattern is sometimes hidden by subsequent
sound changes. For example, the English cognate (and translation) of Latin crūdus
is ‘raw’, which does not start with /h/ as we would expect based on Grimm’s Law.
The Old English form, however, was hrēaw, with an /h/ that was lost from such
initial consonant clusters in the twelfth century.

In other languages
GRIMM’S LAW: THE SEQUEL AND THE REMAKE
The outcomes of Grimm’s Law are also obscured in German, remarkably by
a repeat of Grimm’s Law. This is known as the Second Germanic consonant
shift (Grimm’s Law being the first) or the High German consonant shift. The
changes that constituted it took place in Old High German between the 4th
and 10th centuries, in some cases not long after Grimm’s Law itself. The
voiceless stops that resulted from Grimm’s Law again change to voiceless
fricatives, so we have modern German hoffen ‘hope’, Wasser ‘water’, and
machen ‘make’. In some positions, they remain at an intermediate stage as
affricates: Apfel ‘apple’, Zaun [tsa*n] ‘fence’ (cognate with town). Voiced
stops change to voiceless stops, as in Tier ‘animal’ (cognate with deer).
These changes happened most consistently in southern German varieties,
such as Bavarian and Swiss German. If we go further north, more and more
parts of the Second Germanic consonant shift do not apply. The different
isoglosses for these changes run close together in the eastern part of the
German language area, but in the west, they fan out and spread a much
larger area along the river Rhine. This ‘Rhenish fan’ is probably the most
famous isogloss bundle outside the English-speaking world.
38 3 Phonetic Change

We also find Grimm’s Law-type changes in non-Germanic languages. The

variety of Italian spoken in Tuscany, for example, is famous for the ‘Tuscan
gorgia’, where stops (in standard Italian) are pronounced as fricatives: i piedi
[i FjEdi] ‘the feet’, la torta [la θorta] ‘the cake’, and la casa [la xaza] ‘the
house’. This change also applies to voiced stops, e.g., e beve [e βeve] ‘(s)he
drinks’ (examples from Montemagni et al. 2013). This suggests that a change
from a stop to a fricative is fairly common cross-linguistically.

Seemingly Irregular Change

Grimm’s Law is the most famous example of a neogrammarian sound change.

However, there also appeared to be many examples where the law did not apply.
In some of these examples, a voiceless stop changed to a voiced stop instead of
the voiceless fricative that we expected: Latin patēr corresponds to Old English
fæder, not *fæþer.1 In other examples, the voiceless stop does not change at all:
Latin spūo is English ‘spew’, not *sfew; and octō corresponds to ‘eight’ (Old
English ahta, not *ahþa). So does this mean that the ultimate neogrammarian
change is not in fact regular and exceptionless, and that it, therefore, maybe is not
a neogrammarian change? Does the neogrammarian hypothesis even hold at all?
It appears that Grimm’s Law does not apply if the stop is the second element in
a consonant cluster: stops do not become fricatives after /s/, and if there are two
stops in a row, only the first becomes a fricative. This looks like a problem for
the neogrammarian hypothesis, but this exception itself also constitutes a regular
pattern. Sound changes do not have to apply across the board, but they can be
conditioned: as long as the change applies in every instance where the conditions
are met, we still have regular and exceptionless change. The s-retraction from the
beginning of this chapter is an example of such a conditioned sound change: /s/
changes to /∫/ only when it is in a consonant cluster that also contains /r/. We often
write these conditioned sound changes in the form of a phonological rule:

/s/ → /∫/ / __Cr

Unfortunately for Grimm’s Law, not all exceptions were covered by the condi-
tion about consonant clusters: we still have words like Old English fæder where
Grimm’s Law should have applied, but did not. Once again this endangered the

1 Later change in English caused father to be pronounced with /ð/ in present-day English. Regular
sound change should have given /fæd∂/ for father. The /d/ changed to a fricative because that fit
the pattern found in other kinship terms such as brother. This is called analogy; see Chapter 8.
Brother had /θ/ in early Old English, in accordance with Grimm’s Law, which later changed to /ð/
in regular sound change.
Physiological Explanations for Sound Change 39

validity of the neogrammarian hypothesis. The solution came from Karl Verner,
who noted that the difference between the words for ‘brother’ and ‘father’ in PIE
(as well as Sanskrit and Latin) was not the consonant, but word stress: frāter
had stress on the first syllable, but patēr on the second. For voiceless stops
between two vowels, the original Grimm’s Law only applied when the vowel
before the consonant was stressed in PIE—as in ‘brother’. Other intervocalic
voiceless stops instead changed into voiced stops—as in ‘father’. This second
option is now known as Verner’s Law. After all this, if we also take word stress
into account, the changes of Grimm’s Law become completely regular again, and
the neogrammarian hypothesis is saved.
Verner could never have figured out what conditioned the difference between
Grimm’s Law and Verner’s Law if he had only looked at Germanic languages.
Another change that happened in Germanic languages was that the variable word
stress from PIE shifted uniformly to the first syllable. That meant that the condi-
tioning factor for Verner’s Law disappeared from the language and could only be
recovered by comparison with Sanskrit or Latin.

Physiological Explanations for Sound Change

The discussion of neogrammarian change has so far given us two insights about
sound change: firstly, that sound change is regular and without exceptions, and sec-
ondly, that sound change can be phonetically conditioned. Neither of these should
be a surprise if we focus on the physiological and articulatory processes under-
lying sound change. Our articulators will in principle behave similarly in similar
circumstances. If your default way of pronouncing a /t/ becomes a little bit fur-
ther forward (a dental pronunciation) or a little bit further back (alveolar), the
automatism of your muscle memory will ensure that this applies to all instances
of /t/, so it is logical that sound change should be regular. For the conditioning
of sound change, we can again look at s-retraction: the backing of /s/ in this
change is the direct result of the /s/ begin pronounced in combination with a
following /t/ and /r/. In any other context, your tongue movement would take a
different trajectory which would not result in s-retraction. The conditioning of
sound change, therefore, also has a clear physiological explanation.
A third piece of the sound change puzzle that may also be explained by artic-
ulatory processes is that some sound changes occur much more frequently than
other sound changes.

Lenition

Cross-linguistic evidence of sound change appears to indicate that some directions

of change are preferred over other directions. This is visualised in Fig. 3.1, where
the IPA symbols should be read as indicating primarily voicing and manner of
articulation, not so much place of articulation. In words, the diagram shows the
following preferred directions of change:
40 3 Phonetic Change

Fig. 3.1 Sound changes in lenition (Adapted from Bauer 2008, p. 686)

• long consonants change to short consonants.

• voiceless consonants change to voiced consonants.
• stops change to fricatives.
• stops and fricatives change to approximants.
• sounds involving tongue movement change to sounds not involving tongue
movement.
• sounds disappear (rather than appear out of nowhere).

We often use lenition, or ‘weakening’, as an umbrella term for these often-attested

changes, and we talk about sounds higher up in the diagram as being relatively
strong ( fortis) in articulation, and sounds lower down in the diagram as being rel-
atively weak (lenis). Lenition is much more common than the opposite, ‘fortition’
or ‘strengthening’, and is therefore also thought to be more natural. So what is it
about lenition that supposedly makes this direction of change more natural?
An earlier definition of lenition phrased the relative strength of sounds, and
therefore, the likely direction of sound change, as follows:

A segment X is said to be weaker than a segment Y is Y goes through an X stage on its way
to zero. (Bauer 2008, p. 607)

But this definition of lenition is difficult to work with: it hinges on a clearly

described end point, and therefore, we cannot be certain that a sound change counts
as lenition until the end point—zero, the disappearance of the sounds—is reached.
If we have a change, say, from /t/ to /θ/, we have no way of knowing that there
will be subsequent sound changes that will eventually make that sound disappear
completely.
The definition is also strongly unidirectional: it only leaves room for movement
down the diagram, with no changes up the diagram. It suggests that we have to
define fortition as a change ‘away from zero’, whatever that means. Ideally, an
Physiological Explanations for Sound Change 41

explanation for sound change should explain why lenition is more common than
fortition, but should also be able to explain why fortition can sometimes happen.
Because clearly it does happen: Proto-Germanic /θ/ changed to /d/ in Dutch, for
example. And a change from /j/ (very low in the diagram) to something like /dZ/
(an affricate, so logically relatively high in the diagram) is regularly attested, such
as in Latin maior > Italian maggiore ‘bigger’, or in the variation in Spanish calle
‘street’ [kaʝe] ~ [kaɟ͜ʝe]. So all in all, such a definition of lenition as a frequent
form of sound change is a valid observation and generalisation, but not a valid
explanation.
A more likely explanation for lenition lies in the articulatory processes that
we discussed earlier. This had already been acknowledged before: for example,
an earlier definition of lenition was as ‘articulatory effort reduction’ (Bauer 2008,
p. 606), suggesting that typically, sounds lower in the diagram require less articu-
lation when spoken in a string of other sounds. Although often used, this definition
has the problem that it suggests that speakers seek to exert less effort when speak-
ing, in other words, that they are lazy. Bauer’s (2008, p. 611) own approach defines
lenition as ‘the failure to reach a phonetically specified target: articulatory under-
shoot or underachievement’. This removes the idea that some articulations are
easier than others, but introduces undertones of failure, which perhaps also is not
desirable.
What is implied in the generalisations about lenition is that these are the
expected directions of sound change for intervocalic consonants. We, therefore,
need to look at the articulatory gestures that are required for, say, a stop in between
two vowels. In [ata], the transition from [a] to [t] involves partly closing your
mouth, raising the tip of your tongue to your alveolar ridge, and switching off
vocal fold vibration. The transition of [t] to [a] reverses all these gestures again.
It is difficult to time these gestures to exactly co-occur, and you might not actu-
ally have finished a gesture that is part of pronouncing the [t] before you move
on to preparing for the second [a]. If you don’t quite get to the closure between
your tongue and the alveolar ridge, you may end up actually producing a fricative
[θ]; if you have not switched vocal fold vibration off and back on again, and it
keeps running throughout, you produce a voiced stop [d]. This is what is meant by
‘articulatory undershoot’, and its results are exactly what we would expect from
the generalisations under lenition.
But the effects of articulation on sound change go beyond lenition. The exact
opposite, fortition, can also often be explained as a mis-timing of articulatory ges-
tures. If you re-start vocal fold vibration late, you may end up with [ath a], and
keeping your tongue at your alveolar ridge gives [at a]. Going the other way, we
▼
▲

can arrive at [ata] from [ada] by pausing vocal fold vibration, and from [aθa] by
accidentally touching the alveolar ridge with your tongue instead of merely bring-
ing it to a position nearby. Although some of these may be articulatory overshoot
rather than undershoot, the underlying principle remains the same: speaking is
hard. Because humans are not machines, we sometimes make mistakes or take
shortcuts, with potential sound change as a consequence.
42 3 Phonetic Change

Research Highlight
LENITION IN SPITE OF ARTICULATORY EFFORT
(Non-)rhoticity, the presence or absence of /r/ in syllable codas, is one of
the most striking differences between varieties of English. Scottish English
varieties are typically rhotic, but /r/ is in the process of being lost from
syllable codas, especially for working class speakers. With new techniques
such as ultrasound tongue imaging, researchers are trying to find out what
underlies the process of rhoticity loss, which also may shed light on the loss
of rhoticity from southern England English a few centuries earlier.
Experiments with adolescents from Glasgow show that although many
working class speakers often have no audible /r/ in coda position; the ultra-
sound videos made of their speech show that they do in fact make the
tongue movements that are consistent with a typical /r/ pronunciation. Rather
than undershooting an articulatory target, they are hitting the target with no
audible result. This is probably because the tongue movements are timed
relatively late in the syllable, and the speakers have already ended vocal fold
vibration so there is not much sound left for a typical approximant realisation
of /r/ (Lawson et al. 2018).
Middle class speakers do still have an audible /r/. Lawson and Stuart-
Smith (2021) hypothesise that this may be because the typical middle-class
/r/ is a ‘bunched’ approximant, which for articulatory reasons is closer to
the syllable nucleus, and therefore, more likely to be voiced than the typical
working class ‘retroflex’ or ‘tip-up’ /r/, where tongue movement can be more
delayed. We need to look at every small detail of articulation, co-articulation,
and timing to understand how sound change happens.

Other Types of Sound Change

There are other types of sound change that can be linked to articulation, but that
are difficult to classify in terms of the definition of lenition given earlier. In the
word hamster /hæmst∂/, for example, when you transition between the /m/ and the
/s/, you need to time the movement of your velum, tongue, and lips, as well as the
vibration of your vocal folds. If that is mistimed, and there is a brief period where
you’ve moved your velum up and stopped vibrating your vocal folds but you’ve not
yet moved your tongue and lips, you will end up with something like [hæmpst∂].2
The same goes for prince being pronounced as [prInts]. In these examples, we

2Let’s just spell this out for the sake of clarity. [m] is a voiced nasal bilabial stop. If we raise the
velum (change nasal to oral) and stop vibrating our vocal folds (change voiced to voiceless), but
not move our tongue and lips (no changes to place and manner of articulation), we end up with a
voiceless oral bilabial stop, or [p].
 43

have made a new sound appear where there was none before, by the same process
of articulatory undershoot.
Assimilation is another example of the same phenomenon: the pronunciation
of the nasal in in in phrases like in Poland [m], in France [ɱ], in Turkey [n],
in Germany [ñ], or in Canada [ŋ] is also an example of speakers ‘skipping’ the
canonical place of articulation of /n/ and going straight to the place of articulation
of the next consonant. This kind of allophony can often lay the foundations for
sound change as well. If we ‘prepare’ for the pronunciation of the /i / in key by
▼
▲

pronouncing the /k/ slightly further front in the mouth as [c] (so key [ch i ]) then
▼
▲

that opens the way for listeners of the language to start aiming for this fronted
pronunciation as a target. This is ultimately how we went from Old English ċiriċe
[kirike] > [t∫irit∫e] > church [t∫f t∫].
▼
▲

Against the ‘Effort Reduction’ Hypothesis

The idea that lenition—or any sound change—is the result of effort reduction can
be seen as problematic because it potentially paints not the articulatory process,
but the speakers themselves as lazy, opening the doors to reactionary negative
attitudes towards a sound change. But a much bigger problem is perhaps that it is
not clear where the effort reduction actually is. A reduced-effort articulation of a
plosive may result in a fricative, but if the fricative becomes the new target, that
actually is more difficult to articulate than the original plosive. After all, a plosive
‘simply’ requires closure of the vocal tract, any degree of closure with any degree
of force, while a fricative requires exactly close approximation with no over- or
undershoot which would result in a different type of sound. In other words, there is
a difference between process and outcome when we look at lenition as articulatory
effort reduction.
More evidence against the ‘effort reduction’ hypothesis comes from experimen-
tal research. Kaplan (2010, pp. 25–68) tried to use alcohol intoxication as a way to
artificially force speakers to exert less effort speaking. The results show that with
increasing consumption of vodka and orange juice, speakers do not necessarily
produce more fricatives for stops, or voiced sounds for voiceless sounds. In other
words, what drunk speakers do does not correspond to the standard lenition-type
changes (but is instead fairly chaotic and unpredictable). Smaller experiments on
different types of intoxicants show similar results. A possible objection to this
type of experiments, apart from the ethics of feeding participants intoxicants, is
that they may not actually artificially induce effort reduction at all. If you make it
more difficult for people to control their muscles, they actually have to try harder
to speak, rather than slacking off.
44 3 Phonetic Change

Overshoot and Undershoot in Listeners

The variation that is inherent in our speech production, with overshooting, under-
shooting, and assimilation as described above, gives a lot of opportunity for
language change. And in strict terms, there is probably a great deal of language
change being actuated every single day. Very little of that ever reaches the tran-
sition stage where it can spread to other speakers—or fail to spread, if the social
conditions for the change are unfavourable. Speakers of English today will have a
fair chance understanding the spoken English of three centuries ago, which, despite
the often conservative nature of speech communities of a standardised language,
is remarkable given the amount of potential change there has been. In order to
understand why there is so little language change, we need to shift our attention
from speakers to listeners. Like our speech production, also our speech perception
is variable and fallible. There are obviously mishearings that occur due to hearing
problems or background noise, but we will leave those for what they are and focus
on the interaction between variation in speech production and variation in speech
perception instead.
As listeners, we are generally very good at dealing with variable speech pro-
duction (Ohala 2003). We can use the overall speech rate of a sentence and the
duration of a sound, e.g. /p/, to decide whether a speaker said topic or top pick. We
use our knowledge of how a speaker has produced other vowels to decide whether
a vowel sound is an /E/ or and /I/—and we make different decisions for different
speakers with the exact same vowel stimulus. And we know that following back
vowels drags the articulation of sibilants back a bit, so the exact same sibilant is
perceived as /∫/ when it played with a following /a/, but as /s/ when it is played
with a following /u/.3 Variation in production is built into our speech perception.
But of course our perception is not infallible. Even in a simple task such as the
perception of the consonants /p/, /t/, and /k/ we appear to make many mistakes.
Often we can use semantic information to prevent misunderstanding, but the full
scale of variation in perception can be seen from an experiment with nonsense syl-
lables (Ohala 2003). The researchers played participants a syllable, e.g., /pi/, and
the participants had to decide whether they heard /pi/, /ti/, or /ki/. On average, they
got it wrong 17% of the time, but there were some syllables that were misheard
much more frequently than that. For example, /ki/ was heard as /ti/ by 47% of
participants, a higher proportion even than the 38% who heard it correctly as /ki/.
It is no wonder that this sound change /ki/ > /ti/, roughly what we saw in church
above, is so frequent.

3 Apart from our subconscious knowledge of how /∫/ and /s/ are affected by surrounding sounds, we
also use social information in deciding which sibilant we hear. Men typically pronounce sibilants
a bit further back than women do, which is probably to a large extent a physiological process; at
least in English, there is also a sexual orientation dimension with gay men pronouncing /s/ further
to the front than straight men do—a purely social process, as far as we can tell. These tendencies
all play a role simultaneously in speech perception.
Overshoot and Undershoot in Listeners 45

Our speech perception is moreover multi-modal, so we do not just use sound,

but also visual information when figuring out what others say. This is usually ben-
eficial, because it allows us to still understand people on the basis of lip-reading
when the sound signal is somehow disturbed. How important the visual informa-
tion is can be seen from the so-called McGurk effect: in experiments where the
same audio clip of someone saying /bababa/ is played over video of someone say-
ing a different sequence of sounds (such as /dadada/ or /lalala/), listeners will
overwhelmingly ‘hear’ the sounds that match what they see, rather than what is
actually played to them.
When we as listeners are confronted with variable production and variable per-
ception, there are essentially three things we can do. In most cases, we will correct
for that variation and decide that what we heard as one sound is actually another
sound. To do this we use not only our phonetic knowledge about co-articulation
and our sociolinguistic knowledge about speaker patterns, but also (and perhaps
most importantly) semantic information. Some possible solutions simply make a
lot more sense in context than others, and we also engage in predictive listening.
This is very apparent in misheard song lyrics, and also in the recasting of obscure
expressions with more common vocabulary. The 17% error rate in understanding
syllables in the experiment above was so high exactly because the syllables were
played in isolation, so no semantic information and no context was provided.
Correction is by far the most common outcome, which is why there is rel-
atively little language change despite the large variability. But there are other
options. Sometimes we may think that the misspoken or misheard pronunciation
was the intention, and we do not correct for it. Instead, we store it in our brains
as a possible valid pronunciation of a word or phrase and may use it ourselves at
some later occasion. Not correcting our perception when we should have is called
hypocorrection and is a possible cause of language change.
The opposite is of course also possible: hypercorrection is when we think we
misheard and correct for this, when in fact there was no pronunciation or mishear-
ing. Hypercorrection is more commonly known than hypocorrection, especially in
the context of prescriptivist grammar rules and spelling, but the same principle is
at play in speech perception and can lead to language change if the hypercorrected
pronunciation is stored in the brain.
The processes of correction, hypocorrection, and hypercorrection are especially
important in the context of language acquisition, which is where intergenerational
language change happens. Adult speakers generally have good control of their
articulatory systems (for what that is worth) and they are experienced users of
the language with a lot of subconscious knowledge about all the factors influenc-
ing production and perception, so they are often able to correct mispronunciations
and misperceptions appropriately. Children, on the other hand, are not. They are
still learning the language, with each individual trying to figure out for themselves
how the language works on the basis of data available to them—and this data,
everything that is being said around them, including all the mispronunciations,
half-finished sentences, etc., is not of very great quality. The result is that children
are not as successful at correction, and that they may in fact store many incorrect
46 3 Phonetic Change

(or, to use a more neutral term, ‘changed’) versions in their brain. Some of these
stored versions may be corrected at a later point when more, new, and better evi-
dence becomes available, but some may not. These remaining changed versions
further drive language change, when they are spread to other language users.

References

Ahlers, Wiebke, and Philipp Meer. 2019. Sibilant variation in new Englishes: A comparative
sociophonetic study of Trinidadian and American English /s(tr)/-retraction. In Proceedings of
Interspeech 2019, 291–295. https://doi.org/10.21437/Interspeech.2019-1821.
Baker, Adam, Diana Archangeli, and Jeff Mielke. 2011. Variability in American English s-
retraction suggests a solution to the actuation problem. Language Variation and Change 23
(3): 347–374. https://doi.org/10.1017/S0954394511000135.
Bauer, Laurie. 2008. Lenition revisited. Journal of Linguistics 44 (3): 605–624. https://doi.org/10.
1017/S0022226708005331.
Kaplan, Abby. 2010. Phonology shaped by phonetics: The case of intervocalic lenition. University
of California (UCSC) Dissertation, Santa Cruz.
Lawrence, Wayne P. 2000. /str/ → /∫tr/: Assimilation at a distance? American Speech 75 (1): 82–87.
https://doi.org/10.1215/00031283-75-1-82.
Lawson, Eleanor, and Jane Stuart-Smith. 2021. Lenition and fortition of /r/ in utterance-final posi-
tion: An ultrasound tongue imaging study of lingual gesture timing in spontaneous speech.
Journal of Phonetics 86: 101053. https://doi.org/10.1016/j.wocn.2021.101053.
Lawson, Eleanor, Jane Stuart-Smith, and James M. Scobbie. 2018. The role of gesture delay in
coda /r/ weakening: An articulatory, auditory and acoustic study. Journal of the Acoustical
Society of America 143 (3): 1646–1657. https://doi.org/10.1121/1.5027833.
Lindsey, Geoff. 2019. English after RP: Standard British pronunciation today. London: Palgrave
Macmillan.
Montemagni, Simonetta, Martijn Wieling, Bob de Jonge, and John Nerbonne. 2013. Synchronic
patterns of Tuscan phonetic variation and diachronic change: Evidence from a dialectometric
study. Literary and Linguistic Computing 28 (1): 157–172. https://doi.org/10.1093/llc/fqs057.
Ohala, John J. 2003. Phonetics and historical phonology. In The handbook of historical linguistics,
ed. Brian D. Joseph and Richard D. Janda, 669–686. Oxford: Blackwell.
Rutter, Ben. 2011. Acoustic analysis of a sound change in progress: The consonant cluster /stô/ in
English. Journal of the International Phonetic Association 41 (1): 27–40. https://doi.org/10.
1017/S0025100310000307.
Stevens, Mary, and Jonathan Harrington. 2016. The phonetic origins of /s/-retraction: Acoustic and
perceptual evidence from Australian English. Journal of Phonetics 58: 118–134. https://doi.org/
10.1016/j.wocn.2016.08.003.
Phonological Change
Cognitive Pressures on the Sound System
4

Introduction

We will now move on from a focus on physiological explanations for sound change
and consider how the way that we process sounds influences change. This means
that we will look at more cognitive factors and how they interact with the phys-
iological explanations. Many of the changes discussed here are also described
as being driven by the linguistic system itself (Martinet 1952). As language is
not actually a tangible thing, just the outcome of human behaviour, the linguistic
system is necessarily an effect of human cognition, and system-driven change is
actually cognition-driven change.

Cognitive Factors in Sound Change

Maximisation of Phonological Space

As discussed earlier, our speech production is highly variable. It is also continuous:

we can, for example, theoretically make any vowel, i.e., any possible combination
of minute differences in tongue height and tongue frontness, as well as tiny dif-
ferences in vowel length. But we don’t see the same continuity in our speech
perception: our speech perception is categorical. We perceive a vowel as either
/æ/ or /E/, and vowels that are ambiguous between /æ/ and /E/ get forcibly assigned
to one of the two categories.
Our sound system is designed to contrast sound categories, so that we can use
those contrastive sound categories (phonemes) to distinguish words with different
meanings. An ideally optimised phoneme system makes the best possible use of
the available ‘space’ to distinguish sounds. Abkhaz is a language with only two
vowels: /a/ and /ɨ /, where one is maximally low and the other maximally high. A
language with three vowels, like Arabic, is likely to have /a/, /i/, and /u/, which are

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 47

R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_4
48 4 Phonological Change

situated as far away from each other in the vowel space as possible. A language
with five vowels, like Spanish, adds /e/ and /o/ to these three, which still gives
us an evenly spaced system of contrastive sounds. This even spacing ensures that
the risk of confusion is minimised. As a guiding principle, we call this the max-
imisation of phonological space (Moulton 1962). This is the (cognitive) principle
that will be the focus of the vast majority of the changes discussed in the first
part of this chapter. Of course, when a language has a very high number of vow-
els, it is difficult to maintain an evenly spread vowel system: Swedish (18 vowel
phonemes), for example, has relatively many front high vowels, while English
(15–20 vowel phonemes, depending on accent and definition) has comparatively
many back vowels.

Symmetry and Regularity

Another systemic factor that is often discussed is that of symmetry or regular-

ity: it is cognitively advantageous for language systems to be as regular and as
symmetrical as possible (Martinet 1952). An often-cited example of a symmetry-
driven change is the consonant phoneme system of English at different diachronic
stages (see Table 4.1). The Old English consonant system was regular: it had a
set of stops that occurred in voiced-voiceless pairs, and a set of fricatives that
occurred in voiceless form only.1 In the Middle English period, due to English
taking up loanwords from French, /v/ and /Z/ became distinguishing phonemes.
But this resulted in an asymmetrical system, where some voiceless fricatives had
a voiced counterpart, and others did not. The system intervened by allowing other
voiced fricatives to become phonemes as well, restoring the symmetry (The only
exception to this pattern is /h/, but there are many h-dropping accents in English,
which, therefore, are symmetrical).
Again, language is not a material thing, let alone an independent actor that can
intervene in language change of its own accord, so the drive for symmetry must
be found in cognitive factors. One way of thinking about this is to look at which
features are activated in our mental representation of these categories. For Old
English, only stops were marked for the feature [±voice]; we do not say that the
OE fricatives were marked as [–voice], but rather that they were underspecified for
[±voice]. The feature just did not matter at all. With the introduction of /v/ and /Z/
as distinct phonemes, the feature [±voice] also becomes activated for fricatives,
so that all fricatives must now be specified as either [+voice] or [–voice]. Such
an approach can be quite helpful to understand why language change occurs, and
even why some sounds in a particular language are more or less likely to be subject

1Note here that we are talking about the phoneme system of Old English. The voiceless fricative
phonemes were pronounced as voiced fricatives in certain contexts, but as these contexts were pre-
dictable and there were no contrasts between voiced and voiceless fricatives as there were between
voiced and voiceless stops, we say that Old English did not have voiced fricatives as a separate
category.
Chain Shifts 49

Table 4.1 Consonant inventories in Old English, Middle English, and Modern English

OLD ENGLISH

Voiceless stops p t t∫ k
Voiced stops b d dZ g
Voiceless fricatives f θ s ∫ x
Voiced fricatives
MIDDLE ENGLISH

Voiceless stops p t t∫ k
Voiced stops b d dZ g
Voiceless fricatives f θ s ∫ h
Voiced fricatives v Z
MODERN ENGLISH

Voiceless stops p t t∫ k
Voiced stops b d dZ g
Voiceless fricatives f θ s ∫ h
Voiced fricatives v ð z Z

to change (Natvig and Salmons 2021), but it requires a more advanced knowledge
of phonetics and phonology, so we will not be going down this track any further.

Chain Shifts

The distribution of phonemes as a result of the maximisation of phonological space

is in fact a delicate balance. If for whatever reason that balance is disturbed, we
will try to reorganise the system in such a way that once again we have optimised
the contrasts between sounds.
Let us take the changes from Grimm’s Law as an example: we have voiceless
stops turning into voiceless fricatives (/t/→/θ/), voiced stops turning into voiceless
stops (/d/→/t/), and voiced aspirated stops turning into regular voiced stops (/dh /
→/d/). Assume for the sake of argument that there were physiological reasons why
the aspiration on /dh / was lost. The result of that change was /d/, but there was
already a phoneme /d/. In other words, the pronunciation of /dh / got too close to
/d/ in phonological space, so in order to preserve the existing distinctions, /d/ had
to change as well. The result of that change was /t/, which was also a phoneme that
already existed. Once again, in order to preserve phonemic distinctions, a sound is
forced to change, and /t/ became /θ/.
Such a series of connected changes driven by the usefulness of being able to
make phonemic distinctions as efficiently as possible is known as a chain shift.
These can theoretically take two forms. In a push chain, one sound changes to
become more similar to another sound, which then is ‘pushed’ out of the way
to maintain the distinction. The alternative is also possible, a pull chain or drag
50 4 Phonological Change

Table 4.2 Changes to the

before after
pronunciation of long vowels
in the Great Vowel Shift rise [ri z∂]
▼
▲ [raIz]
cheese [t∫e z∂]
▼
▲ [t∫i z]
▼
▲

team [tE m]
▼
▲ [te m] (→ [ti m])
▼
▲
▼
▲

cake [ka k∂]▼

▲ [ke k]
▼
▲

goat [gO t] ▼
▲ [go t] ▼
▲

blood [blo d] ▼
▲ [blu d] ▼
▲

mouth [mu θ] ▼
▲ [ma*θ]

chain, where a sound changes to become less similar to another sound. The second
sound can then move in the direction of where the first sound used to be without
compromising phonemic distinctions and making better use of the available space.
In such a scenario, the second sound is being ‘pulled’ or ‘dragged’ along with the
first. A drag chain scenario for Grimm’s Law would start with /t/→/θ/, followed
by /d/→/t/ and finally /dh /→/d/.2

The Great Vowel Shift

One of the most famous sound changes from the history of English is such a chain
shift: the Great Vowel Shift. This sound change, which took place roughly in the
Early Modern period, affected the pronunciation of long vowels, as can be seen
in Table 4.2 and Fig. 4.1. As a rough generalisation, you could say that the long
vowels were all raised by one position in the vowel diagram. As /i / and /u / had ▼
▲
▼
▲

no higher position to raise to, they diphthongised to /aI/ and /a*/, respectively.
The sound changes that make up the Great Vowel Shift were commented upon
by orthoepists at the time—people who gave guidance on the (in)correct pronun-
ciation of English according to an emerging spoken standard—but the connection
between the changes was a later discovery. The term ‘The Great Vowel Shift’ was
first coined by Otto Jespersen, or maybe it was the Austrian linguist Karl Luick
who talked about die große Lautverschiebung in his historical phonology of English
around the same time. Ever since, the Great Vowel Shift has been seen as an almost
mythical reputation, and perhaps not deservedly so (Watts 2011, pp. 139–156; see
section “Do Vowel Shifts Really Exist?”).
The Great Vowel Shift is generally timed to very broad periods, such as the
Early Modern period, or ‘between Chaucer and Shakespeare’, but this is a gross
overgeneralisation. It is possible to see when the GVS took place by looking at
spelling evidence. Professional printers tended to regularise spelling more and

2Because Grimm’s Law happened so long ago, and the end result of a push chain or drag chain is
ultimately the same, we simply do not know for sure what type of chain shift Grimm’s Law really
was.
Chain Shifts 51

Fig. 4.1 Schematic representation of the Great Vowel Shift

more, even though there was no fixed standard spelling and we find variation
in printed work as well, but private letters give much better evidence for pronun-
ciation. Even educated writers sometimes used ‘phonetic’ spellings which are an
indirect source of knowledge about changes in pronunciation. For example, writ-
ing ceme for came suggests a GVS-type change in pronunciation: the writer has
a new pronunciation [ke m], but uses the old phoneme-to-grapheme correspon-
▼
▲

dences where /e /=<e>. Conversely, someone who writes gine for join has probably
▼
▲

acquired new phoneme-to-grapheme correspondences where <i> stands for a diph-

thong. These kinds of misspellings are found well into the seventeenth century, so
the Great Vowel Shift was not finished by the time of Shakespeare.
The GVS is generally seen as a chain shift, and much of the academic debate
about the GVS revolves around the question of whether it was a push chain or a
drag chain. In other words, which sound change set the whole GVS in motion?
There are two options, with 15th-century spelling evidence for both (Lass 1992,
p. 150). In the first scenario, high vowel diphthongisation, the first change is the
diphthongisation of /i / and /u /, which then drags the other vowels up. There are
▼
▲
▼
▲

some <ey> spellings for ME ı̄ (such as feynd for find). There are few spellings
clearly suggesting the diphthongisation of ME ū, but this sound was already spelled
<ou> which works remarkably well for a diphthong, especially one that is a bit
less wide than present-day /a*/. The alternative scenario is mid vowel raising,
where /e / and /o / move up, push the higher vowels out of the way and force
▼
▲
▼
▲

them to diphthongise. Spellings of <i> or <y> for ME ē, and of <ou> (the French-
inspired spelling for /u /!) for ME ō occur sporadically in the fifteenth and sixteenth
▼
▲

centuries.
Lass (1992) clearly comes down on the side of mid vowel raising being the trig-
ger for the chain shift. This is because in Northern English and Scottish dialects,
the diphthongisation of ME ū never took place—traditional Scots dialects have
moose, hoose, oot, etc., with /u /. On the other hand, there are no dialects where
▼
▲

ME ı̄ does not diphthongise. The front/back asymmetry for the GVS in Northern
English and Scots can be explained by an earlier fronting of /o / to /ø / in these
▼
▲
▼
▲
52 4 Phonological Change

dialects; it is exactly in the dialects that did not have a back /o / vowel that /u /
▼
▲
▼
▲

failed to diphthongise. This suggests that the GVS started with the raising of the
mid vowels: with no /o / to push /u / out of the way, /u / had no reason to change
▼
▲
▼
▲
▼
▲

to /a*/ (Smith 2007, for this reason discusses a Southern and a Northern version
of the GVS separately).
For the GVS as a whole, this would mean that it is simultaneously a push chain
(leading to the diphthongisation of /i / and /u /) and a drag chain (leading to the
▼
▲
▼
▲

raising of the low and low-mid vowels).

In other languages
HIGH VOWEL DIPHTHONGISATION IN GERMANIC LANGUAGES
The diphthongisation of /i / and /u / in English is not uncommon in West
▼
▲
▼
▲

Germanic languages. In German, these vowels diphthongised to /aI/ and /a*/,

just like in English, and in Dutch they end up as /EI/ and /oey/, slightly
narrower and (for /u /) fronted diphthongs. The German diphthongisation
▼
▲

took place in the 12th to 16th centuries, but the Dutch diphthongisation was
roughly contemporaneous with the Great Vowel Shift and was accepted as
a standard Dutch feature from the 17th century (van Loey 1970, pp. 83–93;
Howell 2006). Unlike in English, though, high vowel diphthongisation in
German and Dutch was not part of a larger chain shift.
High vowel diphthongisation did not happen in Frisian and Low Saxon,
two other West Germanic languages, nor do we find it in Scandinavian lan-
guages. Only in Faroese do they change to /*I / and /ʉ u /, respectively, as
▼
▲
▼
▲

part of a set of atypical diphthongisations of long vowels with unclear his-

torical timing (Lindqvist 2003). Swedish has taken a different approach to
the tendency of long vowels rising, but high vowels having nowhere to rise
to: they try to raise them anyway. In an ongoing change, /i / is pronounced
▼
▲

with a fronted tongue body which gives the acoustic effect of a fricative. This
so-called Viby i(named after a town, not because this /i / vibes in any partic-
▼
▲

ular way) is used most by young urban women and has social connotations
of poshness (Westerberg 2019).

Principles of Vowel Shift

Vowel chain shifts occur all the time, and in many languages. Based on an analysis
of chain shifts in the history of several languages, as well as some chain shifts in
progress, Labov (1994, p. 116) has formulated three principles of vowel shifting.
There are four of them:

• principle 1: In chain shifts, long vowels rise.

• principle 2: In chain shifts, short vowels fall.
• principle 2a: In chain shifts, the nuclei of upgliding diphthongs fall.
• principle 3: In chain shifts, back vowels move to the front.
Chain Shifts 53

There is some evidence for Principle 1 in the Great Vowel Shift: the mid vowels
/e / and /o / were raised and dragged the vowels below them along. The GVS can-
▼
▲
▼
▲

not give evidence for Principle 2, as the GVS applies to long vowels only. Some
other vowel shifts in English, which we will discuss below, do apply to short vow-
els, but these shifts do not unambiguously support this principle. Principle 2a is
derived from Principle 2: if you see a diphthong as a short vowel plus a glide, then
it follows that the nuclei of the diphthongs should fall. The gradual diphthongi-
sation of high vowels in the GVS could be seen to support this principle. If we
interpret /i / as a diphthong /ii/, then a gradual fall of the nucleus would give some-
▼
▲

thing like /ii/>/Ii/>/∂I/>/2I/>/aI/.3 There is some evidence that the diphthongisation

of /i / followed roughly this trajectory. The fronting of back vowels (Principle 3)
▼
▲

obviously did not happen in the GVS, but there are current changes in English
where back vowels move to the front: goose-fronting and goat-fronting. These
have some characteristics of a chain shift in that goose-fronting always comes
first, and there is no goat-fronting without goose-fronting. A production-related
reason for Principle 3 may be that there is less space in the back of the mouth
than in the front (Torgersen and Kerswill 2004, p. 31) and the back of your tongue
is less flexible than the front; it is therefore easier to make phonemic distinctions
with front vowels than with back vowels.
Labov has later rephrased his principles slightly, so that instead of long vowels,
he talks about ‘tense nuclei’ rising ‘along a peripheral track’, and instead of short
vowels, we have ‘lax nuclei’ falling ‘along a non-peripheral track’ (Labov 1994,
p. 176). He links these principles to speech production:

[V]owels pronounced with greater length and energy [i.e., long and tense vowels] will show
a tendency to move toward more exaggerated, or more extreme, positions farther from the
position of rest. (Labov 1994, p. 261)

Essentially, this boils down to an exaggeration of ‘tenseness’ for long vowels and
an exaggeration of ‘laxness’ for short vowels.

Vowel Shifts in Progress

The Great Vowel Shift is of course not the only vowel shift that has taken place in
the history of English. There are several vowel shifts currently ongoing in different
varieties of English. Whereas the relatively poor historical record obscures the
exact mechanism of how the GVS progressed, we do have very good records of
both the production and perception of vowels in these varieties where a vowel

3 In present-day broad Australian English, /i / is phonetically produced as [Ii], providing some evi-
▼
▲

dence for this first step. There are potentially articulatory reasons for this: it is difficult to sustain
precisely the same tongue position over a ‘longer’ period of time (some 100–200 milliseconds), and
the slightly more lax start of the vowel represents the stage before the tongue reaches the (tense)
target position.
54 4 Phonological Change

Fig. 4.2 The Northern Cities Shift (Adapted from Labov 1994, p. 191)

shift is underway. This means that we can get a much better understanding of how
vowel shifts are actuated and progress. What is notable is that unlike the GVS,
which affected long vowels, the current vowel shifts in English affect the short
vowel system, which until the twentieth century appears to have been relatively
stable.

Northern Cities Shift

The current vowel shift that has received the largest amount of scholarly attention
is the Northern Cities Shift, a series of changes to vowels in varieties spoken in the
‘Inland North’, the area around the Great Lakes in the north-eastern United States.
The changes were first recorded in the early twentieth century in urban dialects in
this area (Chicago, Milwaukee, Cleveland, Buffalo, etc.), which gives the shift its
name.
The NCS consists of a series of changes to five short monophthongs: first, the
raising and fronting of trap; second, the fronting of lot; third, the lowering of
thought; fourth, the backing of strut; and fifth, the lowering and backing of
dress (and, for some speakers, kit). A speaker with a NCS vowel system may
pronounce the words block and bus very similarly to how a non-NCS speaker
would pronounce black and boss.4 The visualisation in Fig. 4.2 shows how these
changes are connected in the vowel space and strongly suggests that the NCS is a
pull chain. Tense trap raises while lax vowels lower, which is in accordance with
Labov’s generalisations.
The discussion of the NCS in the Atlas of North American English (Labov et al.
2006, pp. 187–205) lends more credence to the idea of a pull chain as the different

4These examples come from a short (2:27) YouTube video from an early-2000s documentary
where William Labov explains some of the experiments he has done around the NCS: https://www.
youtube.com/watch?v=9UoJ1-ZGb1w.
Chain Shifts 55

steps in the shift were discovered sequentially: the first evidence of trap raising
and lot fronting date from the late 1960s, while a later stage as strut back-
ing was first described in the mid-1980s. The bigger picture shows these changes
co-occurring, which is necessary evidence for a chain shift. At the level of indi-
vidual cities, groups within cities, or individuals within groups, the order is less
strict. While trap raising is extremely widespread, some speakers may for exam-
ple have strut backing without thought lowering. An earlier analysis by Labov
(1994, p. 195), therefore, suggested that the NCS was made up of three separate,
independent chain shifts.
The timing, geographical spread, and speaker profile of the NCS have meant
that the change has been linked to a kind of ideological ‘white flight’—white
speakers in cities that saw their racial demographics change rapidly during the
Great Migration of African Americans from the South to the North in the first
half of the twentieth century may have used the NCS to linguistically distance
themselves from African American speakers. This may also be why the NCS is not
found in neighbouring varieties of Canadian English (Boberg 2008, p. 144). There
is, however, no work linking racial attitudes to NCS vowel patterns to confirm this
hypothesis. Moreover, since the 2010s, there are indications that the NCS is in the
process of being reversed through dialect contact, as speakers increasingly orient
themselves towards a non-local identity (D’Onofrio and Benheim 2020).

Low-Back Merger Shift

Another vowel shift, again involving short vowels, is ongoing in a number of vari-
eties of English in North America. It was first described as a feature of Canadian
English. The ‘Canadian Vowel Shift’ was one of the features that distinguished
Canadian English from varieties of English from the United States (Clarke et al.
1995; Boberg 2008; see Fig. 4.3). However, Clarke et al. (1995, p. 213) also noted
that a similar vowel shift was ongoing in Southern California. This ‘California
Vowel Shift’ (Podesva 2011; Podesva et al. 2015) is visualised in Fig. 4.4. In
more recent years, vowel shifts like this have been described for a range of other
varieties of English in North America as well, and if we want to see these as
instantiations of the same phenomenon, a descriptor linking the change to a spe-
cific region does not do the change justice. Because a key factor in the shift seems
to be the merger of the lot and thought lexical sets, the term ‘Low-Back Merger
Shift’ (LBMS) seems to be the most appropriate (Boberg 2019, p. 94).
Although we call the LBMS a ‘vowel shift’ and usually describe it as a chain
shift, it is not always entirely clear how the component changes are related to each
other. Parts of the LBMS do look like a traditional chain shift: the lowering of
dress, the lowering and backing of trap, and the backing and raising of lot
(merging with thought/cloth which itself may lower a bit) seem to be related
changes. However, the fact that lot is allowed to merge with thought/cloth
already goes against the concept of a chain shift, which is supposed to prevent
mergers.
56 4 Phonological Change

Fig. 4.3 The Canadian Vowel Shift (Adapted from Boberg 2019, p. 93)

Fig. 4.4 The California Vowel Shift (Adapted from Podesva et al. 2015, p. 159)

A problem for a full chain-shift interpretation of the Californian version of

the LBMS are the kit and trap vowels. kit raises before /ŋ/, but lowers else-
where; trap raises before all nasals, but lowers elsewhere. (With dress lowering
to around /æ/ and pre-nasal trap raising to around /E/, this means that in some
cases the two vowels appear to have flipped positions.) It is difficult to fit this to
the idea of maximisation of phonological space.
Similarly, the fronting of the long vowels goose and goat is not logically
related to changes in the short vowels, simply because they differ in length. Even
if any of the short vowels would come close to goose or goat, there is no danger
of a merger because they are distinguished by vowel length and slight diphthongi-
sation. The fronting of foot, which is a short vowel, and the centralisation of
strut, which is not visualised in the diagrams but which has been described
for both Canada and California, are not transparently linked to the core chain
Chain Shifts 57

shift either. The apparent incoherence of these changes does call into question the
concept of a chain shift.

Research Highlight
MAKING SENSE OF THE LOW-BACK MERGER SHIFT
The changes described in the Low-Back Merger Shift are found across North
America: in Canada from coast to coast, and in the United States in many
states except in the East and South. What these varieties all have in common
is that the LOT and THOUGHT/CLOTH lexical sets are merged. Instead of
the many individual local vowel shifts that have previously been described,
we now prefer to see these as a single shift with minor local variations: the
Low-Back Merger Shift (Becker 2019).
A structural account in which the LOT–THOUGHT/CLOTH merger can
trigger a reorganisation of the vowel system (in this case, perceived as a
vowel shift) was given by Roeder and Gardner (2013). In this account, all
vowels in the phonemic system are categorised with the features [±TENSE],
[±HIGH] and/or [±LOW], and [±FRONT]. In dialects with the low-back
vowel merger, the only low vowels in the system are TRAP and the merged
vowel. These are already distinguished from each other by the feature
[±TENSE], which means that the feature [±FRONT] becomes irrelevant
for low vowels.
The implication of this is that TRAP no longer needs to be a front vowel,
and it is free to move to the back. This allows for a more symmetrical system,
among others (Roeder and Gardner, 2022). Ongoing research will need to
unify this account with the movement of STRUT and FOOT, as well as
explain the local variations in the progression of the Low-Back Merger Shift.

Southern Hemisphere Shift

The third ongoing vowel shift to discuss is the Southern Hemisphere Shift
(Fig. 4.5), a chain shift involving short vowels in the Southern Hemisphere vari-
eties of English (Bauer 1992). In this vowel shift, the trap vowel is raised, which
causes dress to raise as well, which in turn causes kit to move. In Australian
English, kit raises to around [i]; in New Zealand, it backs and centralises to [∂];
and in South African English, it can do either of those things depending on pho-
netic context (the so-called kit split) (Fig. 4.5). This shift has all the characteristics
of a push chain. It is possible to see the chain begin with the fronting of the strut
vowel (strut is generally transcribed as /2/ but is phonetically more like [5] in
these varieties, so fronting would bring it close to [æ]) but Bauer (1992) argues
against this as there is no cross-dialectal evidence that a fronted strut correlates
with a raised trap. The other three vowels do move together, as an analysis of
58 4 Phonological Change

Fig. 4.5 The Southern Hemisphere Shift

New Zealand English speakers over a more than 100-year time span shows (Brand
et al. 2021).
The Southern Hemisphere Shift raises a problem for Labov’s generalisations
about chain shifts: according to the generalisations, short vowels in a chain shift
should fall (Principle 2), but in this shift they raise. One solution to this (Bauer
1992) is to claim that although these vowels are short in Southern Hemisphere
Englishes; they are situated on the periphery of the vowel space and may have
laxer counterparts in, e.g., near and square. We may, therefore, see the shift
as one of tense nuclei rising along a peripheral track, which matches Labov’s
generalisations.
A remaining question is why three different speech communities (Australia,
New Zealand, and South Africa) would undergo roughly the same change at
roughly the same time, even though there was no significant contact between these
three communities. Trudgill et al. (2000) claim that this may be a case of ‘drift’:
an incipient sound change would have moved with British settlers from the UK
to the colonies in the Southern Hemisphere and come to full bloom in the three
communities separately. This, however, is more of a sociolinguistic question.

Do Vowel Shifts Really Exist?

After the presentation of these four chain shifts—one historical and three from
the present—we can ask whether vowel shifts really exist. The concept of a chain
shift neatly fits our theories about sound change being regular and about phonemes
being optimally distant from each other in phonological space, but in reality, some
squinting may sometimes be necessary to see the coherence that the term ‘chain
shift’ implies. Labov’s analysis of the Northern Cities Shift sees it as three separate
Chain Shifts 59

chains glued together, and also in the Californian Vowel Shift we find that not all
individual changes seem logically connected to each other.
Two things are necessary to identify a chain shift. The first of these is systematic
co-variation: the individual changes that make up the shift need to be connected
to each other in time, so that vowels move up or down together. A possible lag
between the different changes would then give indications for the shift being a
push or a pull chain. The second criterion is that there is a plausible structural
reason for the co-variation: they need to be logically connected to each other in the
vowel space. Clusters of variables may pattern together and share social meaning
(a linguistic ‘style’; Eckert 2001, p. 123) but we must be able to link the co-
variation to the maximisation of phonological space in order to support a chain
shift analysis.
The Northern Cities Shift looks like it meets these criteria: the temporal order of
vowel changes matches their relative position in the vowel space. But note that this
was an idealisation, and that there were so many differences in the order in which
vowels changed that Labov (1994) suggested cutting the vowel shift in three. The
ongoing reversal of the NCS is not a unitary phenomenon either, with only trap
and lot, but not dress, reversing together (D’Onofrio and Benheim 2020). The
California Vowel Shift is framed as a chain shift, but involves mergers and place
swaps, and also affects the long vowels goose and goat which need not logically
be implicated in the shift. The Southern Vowel Shift does show simultaneous and
connected change in the three sets trap, dress, and kit, but this change patterns
together with lot backing, for which there is no clear phonological explanation
(Brand et al. 2021). It seems therefore that maximisation of phonological space
is at least in part a driver of chain shifts, in particular in their actuation, but that
other, more social factors are also involved in the propagation of these changes.
We can apply the same criteria to the Great Vowel Shift. It has been argued that
the GVS cannot in fact be seen as a single change (Stockwell and Minkova 1988).
Apart from the raising of Middle English mid-high vowels (fleece and goose)
and the diphthongisation of high vowels (price and mouth), which do appear to
be connected in a chain shift-like fashion, the rest of the GVS could be a collection
of independent sound changes that look like, but are not really, a coherent whole
(see also Lass 1992). More precisely, the argument goes, the other vowels merged
with already-existing diphthongs like /EI/ (as in day, weight) and /o*/ (as in know,
grow) (Stockwell and Minkova 1988, pp. 368–369). In hindsight, the GVS looks
as if all long vowels of English were raised in unison in a neatly choreographed
chain shift; even though it is most often not a problem to pretend that they were,
it is useful to be aware of the fact that they very possibly were not.
If the GVS may not in fact have been a chain shift, or even a single phe-
nomenon, it is curious that it has become known as the most famous sound change
from the history of English. The reason for this is probably part wishful thinking,
part neogrammarian doctrine (but not necessarily wilful deception; Stockwell and
Minkova 1988, p. 376). Watts (2011, pp. 139–156) discusses the historiography of
the GVS in some detail and argues that it needs to be seen in the context of 19th-
century historical linguistics. The term ‘great’ refers to ‘the grandiose proportions,
60 4 Phonological Change

the uniqueness […] and the historical significance of those shifts’ (Watts 2011,
p. 155), and the GVS was dated with reference to two famous authors (Chaucer
and Shakespeare). We now know that there have been many such fundamental
restructurings of vowel systems both in English and in other languages, and that
the real time span of the GVS lies between the 13th and early eighteenth centuries.
The concept of the Great Vowel Shift is a product of its time, when historical lin-
guistics was making major advances, but was still stuck in the idea that they needed
to write the history of the purposeful emergence of Standard English. English had
reached its destination of greatness through a mythical sound change, which was
complete just in time for a major author (Watts 2011, p. 144).

Change in Phonemic Categories

Throughout our discussion of phonological change, we have so far taken the main-
tenance of contrastive phonemic categories as a guiding principle. Vowel shifts
(and also consonant shifts) occur because we need to be able to tell apart one set
of sounds from another, so a change in one sound can set off a domino effect
of subsequent changes, simply to preserve existing contrasts. But there are also
changes that affect the categories themselves. Sometimes a distinction is given up,
and two categories merge; or alternatively, a new distinction can be introduced
through a phonemic split.

Mergers

An ongoing sound change in many varieties of English in the United Kingdom

is th-fronting, the pronunciation of the dental fricatives /θ/ and /ð/ as /f/ and
/v/, respectively.5 This change started in the Cockney accent of East London
in the nineteenth century, but has since spread to most areas of the UK, largely
irrespective of educational background or socio-economic class, and while stigma-
tised, it is frequently heard in formal speech as well (Lindsey 2019, pp. 75–76).
Th-fronting is variable for some speakers, as is common for a sound change
in progress, but for those speakers who th-front invariably, there is no longer a
distinction between /f/ and /θ/, or between /v/ and /ð/.
From the point of view of category preservation, the loss of a distinction is
problematic. We could easily expect th-fronting to set in motion a chain shift where
the labiodental fricatives change into some other sounds. But this does not happen,
so the reduced ability to make a distinction is apparently (subconsciously) seen to
be less important than the potential gains in ease of articulation. The distinction
between dental and labiodental fricatives seems to only have a fairly marginal

5 Word-initial /ð/ may also surface as /d/, so that /dIs/ and /vIs/ are both alternative pronunciations
for this.
Change in Phonemic Categories 61

position in English anyway: there are very few minimal pairs for /f/ vs. /θ/, and
for /v/ vs. /ð/, and although some of these involve very frequent words (deaf and
death, free and three), other relevant words are less frequent. In most cases, context
will give enough semantic clues to which of the words is used. These distinctions,
therefore, do not do a lot of important work in the language, or in other words,
they have a relatively low functional load.6
A distinction with a high functional load, which is used to differentiate many
high-frequency words, is much more likely to be maintained than a distinction
that is used to differentiate only few lower-frequency words. It would appear that
when we are presented with a language change that threatens existing categories;
we assess whether a merger or the maintenance of the distinction offers the most
cognitive benefits. Despite the loss of categories, this means that mergers do occur
relatively frequently in language change.7 Especially in the context of dialect con-
tact, moreover, a number of studies suggest that it is easier to ignore a distinction
that exists in your first dialect when speaking a second dialect than it is to acquire
a distinction in your second dialect that does not exist in your first dialect (Cham-
bers 1992; Nycz 2013; see also Siegel 2010, p. 128). It is therefore that Herzog’s
Principle states, ‘Mergers expand at the expense of distinctions’.
In the case of th-fronting, the phonetic result of the merger is identical to the
pronunciation of one of the categories involved—all the /θ/ words simply join
the pre-existing /f/ category. There are also examples where the end result of
the merger is different from the original categories that are involved in it. The
nurse lexical set, for example, is the result of a three-way merger in Early Mod-
ern (Southern) English between what was originally /Ir/, /*r/, and /Er/. All three
came to be pronounced as /∂r/, different from any of the pre-merger pronuncia-
tions; that vowel has since lengthened under the influence of the /r/, which was
subsequently lost, giving us present-day SSBE nurse /z / (This merger did not
▼
▲

happen in conservative Scottish and Irish accents, where you can find two or even
three different vowels in the words bird, word, and heard).

Splits

Just as we can decrease the number of phonemic categories in a language through

a merger, we can introduce new categories through a process called phonemic split.
There are two different kinds of phonemic split, which can be illustrated by the
foot/strut split and the trap/bath split. The absence of these two splits from

6 See Martinet (1952, pp. 3–12) for an early discussion in English of the concept, albeit as ‘func-
tional yield’. The concept had earlier been described in French and German by, among others, Jules
Gilliéron, Roman Jakobson, and Nikolaj Trubetzkoy (Wedel et al. 2013).
7 The most spectacular example of a merger is the Modern Greek phoneme /i/, which is the end

result of historical mergers involving what were originally nine different vowels in Ancient Greek
(Johnson 2010, p. 1).
62 4 Phonological Change

Northern English accents constitutes the most salient North/South differences in

English in England.

The FOOT/STRUT Split

The foot/strut split, which affected short /u/, started as an allophonic alterna-
tion. The short /u/ vowel was pronounced as a more lax [*] and was changing
to a slightly more open realisation [7]. This vowel, usually named ‘ramshorn’, is
an unrounded version of [o]; it later changed to [2] which is the vowel usually
associated with the strut lexical set. After bilabial and labiodental sounds, how-
ever, the higher and rounded version [*] remained. This is probably an effect of
articulation, as bilabial and labiodental sounds are pronounced using the lips, so
that the lip rounding gesture is more easily preserved.
The distribution of rounded [*] and unrounded [7] was more complicated than
that, however, as other surrounding sounds also played a role. An analysis of the
(present-day) pronunciation of foot and strut words by speakers from Manch-
ester—who have a single category for these lexical set and whose accent therefore
parallels the situation before the foot/strut split, invoking the uniformitarian
principle—shows a tendency for a higher, backer, and rounded vowel with a fol-
lowing plosive or /l/ or a preceding bilabial sound, and for a lower, fronter, and
unrounded vowel with a following nasal (Turton and Baranowski 2021). These
characteristics are found overwhelmingly in foot and strut words, respectively.
This means that a small phonetic tendency could over time have developed into an
allophonic alternation.
If this was just an allophonic alternation, we would not have a split, because this
would still be a single phoneme, with predictable alternative realisation [*] or [7]
depending on neighbouring sounds. But the factors that cause a speaker towards
[*] or [7] do not always occur in isolation: what to do with pun with a preceding
bilabial but a following nasal? It was therefore not always easy to predict which
vowel would occur in which word, and the split was likely never a clean division
into two neatly defined categories.
What made matters more difficult is that another change interfered with the
foot/strut split. The vowel in some words with long /u / shortened to /*/. Again
▼
▲

there are tendencies for the phonetic environments that favoured shortening—
monosyllabic words in which /u / was followed by /d/, /v/, /t/, /θ/, or /k/ (Minkova
▼
▲

2014, p. 272)—but this is not a failsafe rule. Think, for example, of food /fu d/ ▼
▲

(without shortening) and good /g*d/ (with shortening). Moreover, the vowel in
some words appears to have shortened in time to participate in the foot/strut
split, while others shortened after foot and strut had already established them-
selves as separate categories. Here you can contrast blood /bl2d/ (early shortening
and following foot/strut split) with good /g*d/ (late shortening).8 Because of
these two sound changes interacting, the alternation between [7] and [*] is no

8Long /u / itself was the result of the Great Vowel Shift applied to Middle English /o /. This is why
▼
▲
▼
▲

many words with [2] or [*] are spelled with <oo>.

Change in Phonemic Categories 63

Fig. 4.6 Schematic representation of the foot/strut split and related changes, after Wells (1982,
p. 199) and Turton and Baranowski (2021, p. 166)

longer anywhere near allophonic. There are even true minimal pairs such as tuck
[t2k] ~ took [t*k]. We can no longer predict which of these two pronunciations we
get on the basis of phonetic context, which means that we now have two different
phonemes. A summary of the processes involved in the foot/strut split is given
in Fig. 4.6.
The foot/strut split can be dated to the seventeenth and eighteenth centuries.
Crucially, this is early enough for the founding wave of English settlers in North
America to take with them a variety with separate foot and strut categories.
Also Scottish English has the foot/strut split—or at least, it has a strut vowel
that is different from the foot = goose vowel. It is a bit counterintuitive to have a
change that originated in Southern England spread to Scotland, but somehow skip
Northern England along the way. At least part of the explanation lies in the fact
that Southern England-style English was taught and learned as a second language
(or dialect) in Scotland after the Unions of the Crowns (1603) and Parliaments
(1707). In Irish English, however, this change appears not to have caught on and
we see a lot of variation in the foot and strut lexical sets.

The TRAP/BATH Split

The other phonemic split, which in England is not only regionally distributed
but also has class connotations, is the trap/bath split. This split affected short
/a/. In the eighteenth century, words which had a short /a/ in Middle English were
pronounced with a vowel somewhere in the range of [æ~a]. This is equivalent to
how our current trap lexical set is pronounced. There were also words which had
a long /a / in Middle English, and these were pronounced [æ ~a ] in the eighteenth
▼
▲
▼
▲
▼
▲

century (the palm lexical set).

The words with a short vowel then underwent what looks like a conditioned
sound change, in which the vowel was lengthened if it was followed by a voiceless
fricative. This change affected words such as pass, glass, class, and brass. Length-
ening also took place before a nasal plus stop or fricative (dance, answer, branch,
example) although that change appears to have happened slightly later than after
just a voiceless fricative (Minkova 2014, pp. 240–241). After these changes, trap
words with a lengthened vowel (these constitute what we now call the bath lexi-
cal set) were indistinguishable from palm words, which already had long [æ ~a ]. ▼
▲
▼
▲
64 4 Phonological Change

Fig. 4.7 Schematic representation of the trap/bath split and related changes, after Wells (1982,
pp. 232–234)

So from a diachronic perspective, bath words are ex-trap words that have joined
the palm set.
Throughout the nineteenth century we have some additional changes affecting
this group of words. First, the loss of rhoticity and compensatory lengthening of
words with original /ar/ meant that the start lexical set was added to the words
pronounced with long [a ]. After this we have another sound change called start
▼
▲

backing, which turns start [a ] into [A ]. This change generally also applies to palm
▼
▲
▼
▲

and bath words. The vowel in trap words is still variable in the [æ~a] range; we
have seen higher realisations in the discussion of vowel shifts above, while the
realisation in SSBE has been lowering in the second half of the twentieth century
(e.g., Harrington et al. 2000).
A schematic overview of the development of a-type vowels is given in Fig. 4.7.
Again, a cross-dialectal comparison allows us to date this change to the mid-
to-late eighteenth century. The trap/bath split did not happen in North American
varieties, although there is a lengthening effect in bath words in varieties from
New England. This suggests that the trap/bath split happened after American
English was established as a separate variety from British English. But it did hap-
pen in Southern Hemisphere varieties, so it must have happened before large-scale
migration to Australia, New Zealand, and South Africa. Interestingly, however,
colloquial Australian English often has a trap-like short /æ/ in words with a nasal
plus stop or fricative, such as dance (noticed by Mitchell [1958], and investigated
empirically by Horvath and Horvath [2001]). This may suggest that the vowel was
lengthened in these words later than in simple pre-fricative contexts.

Lexical Effects in the Spread of Sound Change

In both the foot/strut and the trap/bath splits, a crucial factor in the loss
of predictability of a pronunciation was that not all words appear to have been
affected by a sound change at the same time. For the lengthening of trap words,
we have just seen that the two conditions for lengthening (a following voiceless
fricative vs. a following nasal plus obstruent cluster) applied at such different times
that one set was exported to Australia while the other was not. But even within the
conditions, there are differences. The lengthening of trap words never reached
words like gas, crass, ass, cancel, fancy, mansion, hand, etc. Some words are
Change in Phonemic Categories 65

variable even in today’s ‘standard’ SSBE: Wells (1982, p. 135) mentions graph,
Basque, drastic, blasphemy, transfer, substantial, and others.
These differences in timing are seemingly at odds with the neogrammarian
hypothesis, which sees sound change as completely regular and without excep-
tions. From a neogrammarian point of view, a sound change can be phonetically
gradual—the trap vowel becomes a little bit longer and the strut vowel a lit-
tle bit lower with each generation of speakers—but crucially it must be lexically
abrupt: all vowels that meet the criteria of the conditioned sound change are
affected at the same time. What we see here is something different: some words
are affected by a change before other words. This lexically gradual view of how
sound change progresses through a language is known as lexical diffusion (Wang
1969; Cheng and Wang 1977).
That lexical diffusion exists should not come as a surprise considering what we
know about the many factors constraining language variation (which is often an
effect of language change in progress). Sociolinguistic studies often find frequency
effects, and these may lay at the basis of lexical diffusion. The exact influence of
frequency on the diffusion of sound change is complicated and still not entirely
understood. One theory is that higher-frequency words undergo a sound change
faster because the ‘new’ pronunciation simply occurs more often than that of
lower-frequency words, so both listeners and speakers have more opportunities to
notice the change and update their mental representations of these words. Studies
of lexical effects in sound change often find no effect at all, however, and occa-
sionally lower-frequency words even lead a sound change (Hay et al. 2015). One
factor that may be involved here is whether a sound change increases or decreases
the risk of a changing phoneme being confused with another phoneme, but this
hypothesis still needs to be empirically tested (Todd et al. 2019).
What we see in the foot/strut split and especially in the trap/bath split, is
a lexically diffused sound change that stopped before it was entirely completed.9
It is possible that both changes are still (slowly) ongoing, although the strong
ideology of Standard English is more likely to preserve this messy intermediate
situation than to promote the completion of these changes.

Two Kinds of Phonemic Split

The trap/bath and foot/strut splits represent two different kinds of phonemic
split (Campbell 2004, p. 25). The first of these is called primary split, exemplified
by the trap/bath split. This kind of split does not produce a new phoneme, so
the number of phonemes in the language remains unchanged. All that happens is
that an allophone of a phoneme abandons its original phoneme and joins another
phoneme instead. This is illustrated by the left diagram in Fig. 4.8. The other type
of split is called secondary split, where the split does result in a new phoneme,

9 Around the same time as the trap/bath split, words from the lot lexical underwent lengthen-
ing in similar conditions, resulting in the cloth lexical set. This also appears to have progressed
with lexical diffusion and was later partially reversed, so that the outcome is even more haphazard
(Wells 1982, p. 234).
66 4 Phonological Change

Fig. 4.8 Schematic overview of, from left to right, primary split (trap/bath), secondary split
(foot/strut), and merger (square/nurse)

and which creates a new phonological contrast that did not exist in the language
before. This is what happened with the foot/strut split and is illustrated by the
centre diagram in Fig. 4.8. For the sake of completeness, the right diagram in the
figure shows a merger, where the number of phonemic contrasts is reduced, as
exemplified by the North West English square/nurse merger.

Are Mergers Reversible?

When two phonemes have merged—and have been acquired by someone as a

single category—it should theoretically be impossible to reverse that merger. Of
course, the new single phoneme can be split again in some other conditioned
sound change, but it should not be possible to split it back into the original two
groups. After all, the original pre-merger categories were separate phonemes, and
there are likely to have been minimal pairs (now homophones) that phonological
conditioning of a sound change cannot tease apart anymore.
Still, such a merger reversal appears to have happened in the Great Vowel Shift.
Middle English had three separate phonemes mate /a /, meat /E / and meet /e /.
▼
▲
▼
▲
▼
▲

These were the input for the Great Vowel Shift. From rhyming patterns in Shake-
speare and his contemporaries, we can conclude that for them, mate and meat
were the same, probably /e /, but meet /i / was different. This suggests that in this
▼
▲
▼
▲

period, we have a merger between the mate and meat lexical sets. In present-day
English, however, mate /eI/ is different from meat /i /, and meat is now the same
▼
▲

as meet. So how did the original meat set get un-merged from mate, so that it
could merge with meet instead?
To understand this, we can look at the properties of some other mergers. The
first example is the merger between the foot and goose lexical sets in Mid-Ulster
English (Maguire et al. 2013, p. 231). In this variety, both sets are pronounced as
[ʉ ]. But Maguire, who is a native speaker of this variety, states that this picture
is a simplification. While goose is always [ʉ ], foot varies between [ʉ ] and [ ].
This means that at one, phonetic, level of analysis, the foot and goose sets are
clearly merged, because the [ʉ ]s in both sets are exactly the same. It is entirely
legitimate to rhyme good (a foot word) and food (a goose word). But at another,
phonological, level of analysis, we do still have a contrast. We do still have two
separate groups: one that can only be [ʉ ], and one that can also be something else.
Because this kind of variable merger scenario maintains the distinct categories, it
is not a true merger, and the categories can still develop in separate ways. Whatever
linguistic and social factors determine the choice for [ʉ ] and [ ] in the Mid-Ulster
Change in Phonemic Categories 67

English speakers’ foot vowel may change to make [ ] the overwhelmingly more
frequent variant, at which point a learner’s analysis of foot and goose as two
clearly distinct categories becomes much more likely.
Another piece of evidence comes from the square/nurse merger in Lancashire
(Barras 2006). An acoustic analysis of square and nurse vowels from speakers
from different places in Lancashire and Greater Manchester showed that some
speakers clearly have two different vowels, [E ] and [z ], some speakers clearly
▼
▲
▼
▲

have the same vowel in both sets, usually [z ], and some speakers have two vowels
▼
▲

that are very close together in the vowel space. Some speakers from this last group,
even though they have ever so slightly different vowels, were unable to tell their
own vowels apart. That is, we can acoustically see a difference between their
pronunciation of fair and fur, but when they were asked to identify a sound file
of themselves saying a word as either fair or fur, they performed at chance levels.
This is an example of a near-merger: speakers produce a distinction they cannot
perceive (Maguire et al. 2013, p. 232). We can see the same phenomenon in the
foot=strut unsplit speakers from Manchester, who have different vowels visible
in a close linguistic analysis, but who did not perceive the difference themselves
(Turton and Baranowski 2021).
The mate/meat merger in Early Modern English may have been such a vari-
able merger or near-merger. Perhaps the two sets only shared one allophone, like
foot and goose in Mid-Ulster English. This would allow Shakespeare and his
contemporaries to produce perfect rhymes, but there would still have been two dif-
ferent mental categories which would allow for meat to diverge from mate and
later merge with meet. Alternatively, Shakespeare’s mate and meat sets could
have been extremely similar but not identical, like the square and nurse sets for
some speakers from Lancashire. Shakespeare himself might not even have been
able to perceive the difference, but others might have, which again allows for
different categories and a separate development of the meat set.
Another way to ‘unmerge’ merged sounds is through dialect contact and
second-dialect acquisition. This is the mechanism for the unmerger of the cot-
caught (=lot/thought) merger by Canadian migrants to New York City (Nycz
2013). Canadian English has a single category /A/ for both these lexical sets, but
if Canadian speakers hear some words from that category pronounced with differ-
ent vowel by New York speakers in their new environment, who have lot /A/ but
thought /O/, they may eventually learn to both perceive and produce two different
categories themselves as well. The mechanisms that play a role here are also rele-
vant to the dialect contact situations that we will talk about in Chapter 7. Dialect
contact is also how speakers from the North of England can acquire a strut /2/
despite having a single category themselves, albeit with variable success (Turton
and Baranowski 2021).
What we have seen about apparent cases of the reversal of merger ultimately
still supports Garde’s Principle, ‘Mergers are irreversible by linguistic means’. In
most cases, we were not dealing with a complete merger, so that the old cat-
egories were still somewhat intact. The Canadians who learned to unmerge the
lot/thought merger were only partially successful at this, but more importantly
68 4 Phonological Change

did not do this by entirely linguistic means: there are also social factors at play
here, which lie outside the purely cognitive linguistic system.

Physiological and Cognitive Factors in Sound Change

In Chapter 3, we investigated the physiological factors underlying sound change:

the effects of mistimings or shortcuts in our articulatory systems or simply the side-
effects of co-articulation that may set a sound change in motion. The takeaway
message was that, provided the conditions in which a sound change applies are
described accurately, sound change is regular and without exceptions.
In this chapter, we have had to adapt our understanding of sound change after
also investigating some cognitive factors in sound change. These involved, in the
first instance, a realisation that sounds are part of a phonemic system. Our cog-
nition has a preference for systems that are symmetrical, but more importantly, it
wants to maintain (abstract) systems by preserving existing categories. This may
often keep change from happening at all, or it may set in motion a chain of related
changes.
But this is still an idealised picture. Sometimes the drive to preserve cate-
gories loses out to an articulatorily driven change, and two categories merge.
Similarly, our categorical perception can be overridden and a co-articulatory effect
can develop into a phonemic split.
What we have also seen in this chapter is an alternative to regular neogram-
marian sound change: lexical diffusion. Whereas articulatory effects and category
preservation are in theory universal factors, they are influenced by non-universal,
language-specific factors that cause sounds in some words to change faster than
in other words. The most important language-specific factors are lexical frequency
and functional load. To describe these non-universal factors as language-specific
is again a simplification. The frequency of lexical items and the functional load of
distinctions may also differ within a language: between varieties at different places
and in different time periods. Such differences may in fact be why a system that
seems to have been stable for centuries suddenly undergoes change. The linguistic
system is a delicate balance between conflicting pressures, and subtle changes in
non-universal factors may tip the balance and set a change in motion (Sóskuthy
2015).

References

Barras, Will. 2006. The square--nurse merger in Greater Manchester: The impact of social and
spatial identity on phonological variation. MA thesis, University of Edinburgh.
Bauer, Laurie. 1992. The second Great Vowel Shift revisited. English World-Wide 13 (2): 253–268.
https://doi.org/10.1075/eww.13.2.04bau.
Becker, Kara, ed. 2019. The Low-Back-Merger Shift: Uniting the Canadian Vowel Shift, the Califor-
nia Vowel Shift, and short front vowel shifts across North America (Publications of the American
Dialect Society 104). Durham, NC: American Dialect Society.
References 69

Boberg, Charles. 2008. Regional phonetic differentiation in Standard Canadian English. Journal
of English Linguistics 36 (2): 129–154. https://doi.org/10.1177/0075424208316648.
Boberg, Charles. 2019. A closer look at the short front vowel shift in Canada. Journal of English
Linguistics 47 (2): 91–119. https://doi.org/10.1177/0075424219831353.
Brand, James, Jen Hay, Lynn Clark, Kevin Watson, and Márton Sóskuthy. 2021. Systematic co-
variation of monophthongs across speakers of New Zealand English. Journal of Phonetics 88:
101096. https://doi.org/10.1016/j.wocn.2021.101096.
Campbell, Lyle. 2004. Historical linguistics: An introduction, 2nd ed. Edinburgh: Edinburgh
University Press.
Chambers, J.K. 1992. Dialect acquisition. Language 68 (4): 673–705. https://doi.org/10.2307/
416850.
Cheng, Chin-Chuan, and William S.-Y. Wang. 1977. Tone change in Chao-zhou Chinese: A study
in lexical diffusion. In The lexicon in phonological change, ed. William S.-Y. Wang, 86–100.
Den Haag: Mouton. https://doi.org/10.1515/9783110802399.86.
Clarke, Sandra, Ford Elms, and Amani Youssef. 1995. The third dialect of English: Some Canadian
evidence. Language Variation and Change 7 (2): 209–228. https://doi.org/10.1017/S09543945
00000995.
D’Onofrio, Annette, and Jaime Benheim. 2020. Contextualizing reversal: Local dynamics of the
Northern Cities Shift in a Chicago community. Journal of Sociolinguistics 24 (4): 469–491.
https://doi.org/10.1111/josl.12398.
Eckert, Penelope. 2001. Style and social meaning. In Style and sociolinguistic variation, ed.
Penelope Eckert and John R. Rickford, 119–126. Cambridge: Cambridge University Press.
Harrington, Jonathan, Sallyanne Palethorpe, and Catherine Watson. 2000. Monophthongal vowel
changes in Received Pronunciation: An acoustic analysis of the Queen’s Christmas broadcasts.
Journal of the International Phonetic Association 30 (1–2): 63–78. https://doi.org/10.1017/S00
25100300006666.
Hay, Jennifer B., Janet B. Pierrehumbert, Abby J. Walker, and Patrick LaShell. 2015. Tracking
word frequency effects through 130 years of sound change. Cognition 139: 83–91. https://doi.
org/10.1016/j.cognition.2015.02.012.
Horvath, Barbara M., and Ronald J. Horvath. 2001. A geolinguistics of short A in Australian
English. In English in Australia, ed. David Blair and Peter Collins, 341–355. Amsterdam:
Benjamins.
Howell, Robert B. 2006. Immigration and koineisation: The formation of Early Modern Dutch
urban vernaculars. Transactions of the Philological Society 104 (2): 207–227. https://doi.org/
10.1111/j.1467-968X.2006.00169.x.
Johnson, Daniel Ezra. 2010. Stability and change along a dialect boundary: The low vowels of
Southeastern New England (Publications of the American Dialect Society 95). Durham, NC:
Duke University Press.
Labov, William. 1994. Principles of linguistic change. Volume 1: Internal factors. Oxford:
Blackwell.
Labov, William, Sharon Ash, and Charles Boberg. 2006. The Atlas of North American English:
Phonetics, phonology and sound change. Berlin: Mouton de Gruyter. https://doi.org/10.1515/
9783110167467.
Lass, Roger. 1992. What, if anything, was the Great Vowel Shift? In History of Englishes: New
methods and interpretations in historical linguistics, ed. Matti Rissanen, Ossi Ihalainen, Terttu
Nevalainen, and Irma Taavitsainen, 145–155. Berlin: Mouton de Gruyter.
Lindqvist, Christer. 2003. Thesen zur Kausalität und Chronologie einiger färöischer Lautgesetze.
Arkiv för nordisk filologi 118: 89–178.
Lindsey, Geoff. 2019. English after RP: Standard British pronunciation today. London: Palgrave
Macmillan.
Maguire, Warren, Lynn Clark, and Kevin Watson. 2013. Introduction: What are mergers and can
they be reversed? English Language and Linguistics 17 (2): 229–239. https://doi.org/10.1017/
S1360674313000014.
70 4 Phonological Change

Martinet, André. 1952. Function, structure, and sound change. Word 8 (1): 1–32. https://doi.org/
10.1080/00437956.1952.11659416.
Minkova, Donka. 2014. A historical phonology of English. Edinburgh: Edinburgh University Press.
Mitchell, Alexander G. 1958. Spoken English. London: Macmillan.
Moulton, William G. 1962. Dialect geography and the concept of phonological space. Word 18
(1–3): 23–32. https://doi.org/10.1080/00437956.1962.11659763.
Natvig, David, and Joseph Salmons. 2021. Connecting structure and variation in sound change.
Cadernos de Linguística 2 (1): 1–20. https://doi.org/10.25189/2675-4916.2021.V2.N1.ID314.
Nycz, Jennifer. 2013. New contrast acquisition: Methodological issues and theoretical implica-
tions. English Language and Linguistics 17 (2): 325–357. https://doi.org/10.1017/S13606743
13000051.
Podesva, Robert J. 2011. The California vowel shift and gay identity. American Speech 86 (1):
32–51. https://doi.org/10.1215/00031283-1277501.
Podesva, Robert J., Annette D’Onofrio, Janneke Van Hofwegen, and Seung Kyung Kim. 2015.
Country ideology and the California Vowel Shift. Language Variation and Change 27 (2):
157–186. https://doi.org/10.1017/S095439451500006X.
Roeder, Rebecca V., and Matt Hunt Gardner. 2013. The phonology of the Canadian Shift revisited:
Thunder Bay Cape Breton. University of Pennsylvania Working Papers in Linguistics 19 (2):
161–170. https://repository.upenn.edu/pwpl/vol19/iss2/18.
Roeder, Rebecca V., and Matt Hunt Gardner. 2022. A unified account of the low back merger shift.
Paper presented at Methods in Dialectology XVII, Mainz, August 1.
Siegel, Jeff. 2010. Second dialect acquisition. Cambridge: Cambridge University Press.
Smith, Jeremy J. 2007. Sound change and the history of English. Oxford: Oxford University Press.
Sóskuthy, Márton. 2015. Understanding change through stability: A computational study of sound
change actuation. Lingua 163: 40–60. https://doi.org/10.1016/j.lingua.2015.05.010.
Stockwell, Robert P., and Donka Minkova. 1988. The English vowel shift: Problems of coherence
and explanation. In Luick revisited: Papers read at the Luick-Symposium at Schloß Liechtenstein,
15.–18.9.1985, ed. Dieter Kastovsky and Gero Bauer, 355–394. Tübingen: Narr.
Todd, Simon, Janet B. Pierrehumbert, and Jennifer Hay. 2019. Word frequency effects in sound
change as a consequence of perceptual asymmetries: An exemplar-based model. Cognition 185:
1–20. https://doi.org/10.1016/j.cognition.2019.01.004.
Torgersen, Eivind, and Paul Kerswill. 2004. Internal and external motivation in phonetic change:
Dialect levelling outcomes for an English vowel shift. Journal of Sociolinguistics 8 (1): 23–53.
https://doi.org/10.1111/j.1467-9841.2004.00250.x.
Trudgill, Peter, Elizabeth Gordon, Gillian Lewis, and Margaret MacLagan. 2000. The role of
drift in the formation of native-speaker Southern Hemisphere Englishes: Some New Zealand
evidence. Diachronica 17 (1): 111–138. https://doi.org/10.1075/dia.17.1.06tru.
Turton, Danielle, and Maciej Baranowski. 2021. Not quite the same: The social stratification and
phonetic conditioning of the foot–strut vowels in Manchester. Journal of Linguistics 57 (1):
163–201. https://doi.org/10.1017/S0222226720000122.
van Loey, A. 1970. Schönfelds historische grammatica van het Nederlands: Klankleer, vormleer,
woordvorming, 8th ed. Zutphen: Thieme.
Wang, William S.-Y. 1969. Competing changes as a cause of residue. Language 45 (1): 9–25.
https://doi.org/10.2307/411748.
Watts, Richard J. 2011. Language myths and the history of English. Oxford: Oxford University
Press.
Wedel, Andrew, Abby Kaplan, and Scott Jackson. 2013. High functional load inhibits phonologi-
cal contrast loss: A corpus study. Cognition 128: 179–186. https://doi.org/10.1016/j.cognition.
2013.03.002.
Wells, J.C. 1982. Accents of English. Cambridge: Cambridge University Press.
Westerberg, Fabienne. 2019. Swedish “Viby-i”: Acoustics, articulation, and variation. In Proceed-
ings of the 19th International Congress of Phonetic Sciences, Melbourne, Australia 2019,
ed. Sasha Calhoun, Paola Escudero, Marija Tabain, and Paul Warren, 3696–3700. Canberra:
Australasian Speech Science and Technology Association Inc.
Track Changes
Sound Change in the History of English
5

Word Histories

Just from looking at a text in Old English and comparing it to the form of the
language today, it is clear that there has been a lot of change in the history of
English. At the level of the word, a large part of these changes have been sound
changes. In the previous chapters, we have looked at a number of major sound
changes in the history of English, but the selection of sound changes in those
chapters was driven by theoretical points. The sound changes served as illustration
and there was no attempt to place them in chronological order.
This chapter does the opposite. It offers a chronological overview of selected
sound changes in English from the Old English to the present. The selection has
by necessity been restricted to ‘highlights’ or ‘greatest hits’, as it would be impos-
sible to give a full account. It also omits a number of small details with the
sound changes that are discussed and smooths over some complications: for all
the emphasis placed on the regularity of sound change, there appear to be particu-
larly many irregularities in the history of English.1 The selection is made so that it
becomes possible to take most words in Old English and, by following the sound
changes in this chapter in chronological order, arrive at the present-day form of
the word.
As an example of such a sound history of an individual word, let us look at
the Old English word camb and how it transformed into its present-day English
equivalent, comb, in Table 5.1. We can see that the Old English pronunciation,

1 It is not clear why this should be the case, but in most other Germanic languages there are far
fewer lexical exceptions to regular sound change. It is possible that this may have to do with a large
degree of dialect diversity and dialect and language contact at crucial periods in the history of the
language, around the time of the Norman Conquest—the transition from Old to Middle English—
and the reestablishment of English and its development as a standard language—the transition from
Middle to Early Modern English.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 71

R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_5
72 5 Track Changes

Table 5.1 Sound changes in the history of the English word comb
Old English camb kAmb
homorganic lengthening 10th century A→A ▼
▲ =⇒ kA mb
▼
▲

changes to low vowels 11th century A →O

▼
▲
▼
▲ =⇒ kO mb
▼
▲

Great Vowel Shift 15th–17th centuries O →o

▼
▲
▼
▲ =⇒ ko mb
▼
▲

consonant cluster reduction 17th century -mb → -m =⇒ ko m

▼
▲

diphthongisation of goat 19th–20th centuries o → ∂*

▼
▲ =⇒ k∂*m
Present-day English comb k∂*m

kAmb, first underwent a change in which the vowel became lengthened: kA mb. ▼
▲

This vowel was then raised to kO mb. Because the vowel was now long, it par-
▼
▲

ticipated in the Great Vowel Shift, and the word changed to ko mb. Then, the
▼
▲

consonant cluster at the end of the word was simplified (ko m), and finally, with
▼
▲

the diphthongisation of the goat vowel we arrive at our present-day pronunciation,

k∂*m.
The important takeaway here is that sound changes happened in a particular
order. The word comb could only participate in the Great Vowel Shift because an
earlier sound change had resulted in a long vowel; the original short vowel would
not have participated in the GVS otherwise. The output from one step is therefore
the input for the next step.
Two comments are necessary at this point. The first is that there are no brackets
around the IPA transcriptions in this example, or in the transcriptions elsewhere in
this chapter. Instead, IPA transcriptions are given in bold. This is because in some
cases, it is necessary to work at a more precise level than phonemic transcription,
so slanted brackets are inappropriate. On the other hand, these transcriptions do not
show a great deal of phonetic detail either, and sometimes phonetic differences are
overlooked, so square brackets are inappropriate as well. By not writing brackets,
we avoid making any claims about exact levels of precision.
The other comment is that this chapter takes Standard Southern British English
(SSBE) as its reference accent. It is necessary to focus on a single variety of
English in order to maintain the overview in the chapter, and SSBE allows us
to include a number of sound changes that other reference accents do not. For
example, the reference variety for American English did not undergo the loss of
rhoticity or the trap/bath split, and Scottish Standard English has forgone even
more changes. Where relevant, the chapter will note differences between a number
of standard reference accents. The focus on SSBE should not be read to mean that
it is taken to be the ‘best’, let alone the ‘only allowable’ accent of English.
Word Histories 73

Evidence for Change

Of course there are many uncertainties around the pronunciation of English in

earlier stages of the language. Recording equipment has only been available for
just over a century, which gives audio-based evidence a time depth to the middle of
the nineteenth century. Some of the earliest recordings were made by the German
linguist Wilhelm Doegen, who collected dialect samples from British prisoners of
war during World War I. The informants were born between 1870 and 1900 and
their speech reflects English (dialect) pronunciation of the time. Some recordings
that were made later focused on older informants and therefore go a little bit
further back. The NORMs (non-mobile older rural males) in the Survey of English
Dialects, recorded in the 1950s, were born around the same time as the World War
I prisoners of war.2 The Origins of New Zealand English project (see section
“New-Dialect Formation in Colonial Englishes” in Chapter 7) used recordings
from the 1940s, with the oldest informants being born in the 1840s. Individual
radio and television recordings survive of speakers also born in the middle of the
19th century, but for earlier English speech there is no direct evidence.
Another type of evidence comes from commentary on pronunciation. In par-
ticular from the Early Modern period, with the incipient standardisation of not
only written but also spoken English, there were many pronunciation guides that
described the ‘polite’ pronunciation of English (Watts, 2011, pp. 204–206), some-
times in contrast with ‘impolite’ pronunciations. These publications form valuable
evidence for variation and change in pronunciation. Some commentary was writ-
ten by non-native speakers and contrasts English pronunciation with that of other
languages, e.g., French.
The bulk of our evidence, however, is reconstruction. Here, we can base our
knowledge of pronunciation on spelling—although standardised spelling from the
Early Modern period and pseudo-standardised spelling in the Old English period
mean that we do not always get a complete and reliable picture from existing
documents. Where there is spelling variation, however, this may often give us an
indication of how words were pronounced. Rhymes are particularly useful for vow-
els, but see the discussion about mergers in section “Are Mergers Reversible?” in
Chapter 4; because poetry also makes use of alliteration, we can extend the same
principles (and problems) to consonants. The development of loanwords (English
loanwords in other languages or loanwords from other languages in English) allow
us to compare with other languages, and of course we can work with the cross-
linguistic evidence that lay at the basis of neogrammarian endeavours. All in all,
historical linguists have been able to reconstruct the pronunciation of earlier stages
of English in remarkable detail. Of course, the very fine nuances that can be dis-
covered with audio recordings are not available for these earlier stages, but they
are not necessary for our purpose anyway.
A more complete account of sound changes in English, including possible the-
oretical explanations for the changes, can be found in, e.g., Jones (1989), Smith

2 A selection of recordings from the Survey of English Dialects and the prisoner of war recordings
from the Berliner Lautarchiv is available from the British Library website, at https://sounds.bl.uk/
Accents-and-dialects/.
74 5 Track Changes

(2007), and Minkova (2014). Wells (1982) includes discussion of sound change
from the Middle English period onwards in his treatment of individual lexical
sets.

The Pronunciation of Old English

The starting point for our journey is Old English. The dialectal variation in Old
English, with four main dialects (West-Saxon, Kentish, Mercian, and Northum-
brian), is relatively well understood. The majority of documents is written in a
West-Saxon variety, but other varieties of Old English have had at least as much
if not more influence on the development of English phonology from the Middle
English period onwards (Minkova, 2014, p. 184). The overview of Old English
pronunciation here is necessarily an idealised simplification.

Vowels

Old English had seven monophthongal vowels: a A, æ æ, e e, i i, o o, u u, and y y.

This means that the vowels were pronounced more or less according to their IPA
values. Most of these vowels are relatively unproblematic for present-day English
speakers, with the exception of y, which is a high front rounded vowel (as in
French plus or German dünn).
The monophthongs occurred as both short and long vowels. In modern editions,
the long vowels are often indicated with a macron: ā A , ǣ æ , ē e , ı̄ i , ō o , ū u , and
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲

ȳ y . It is possible that (some of) the long and short vowels may also have differed
▼
▲

in quality, along the same lines of the long/short and tense/lax alternations we find
in present-day English: long e alternates with short E, i with I, o with O, u with *,
▼
▲
▼
▲
▼
▲
▼
▲

and y with ʏ. Some phonetic transcription of Old English that you may encounter
▼
▲

take such alternations into consideration, while others simply use the same symbol
with or without a length mark.
In addition to the monophthongs, Old English had three diphthongal vowels,
which again were pronounced roughly according to their IPA values: io io, eo
eo, and ea eA. These diphthongs also came in short and long versions, where the
length of the longer diphthongs was carried mostly by the first element: ı̄o i o, ēo ▼
▲

e o, and ēa e A.
▼
▲
▼
▲

Consonants

Of the Old English consonants, b, p, d, t, l, m, n, and w were pronounced as in

Modern English (and according to the IPA values of the letter symbols). The r was
an alveolar trill r. As in Modern English, n is pronounced as ŋ when followed by
a velar stop. Unlike in most varieties of Modern English, the velar stop is still
pronounced, e.g., hring hrIŋl ‘ring’.
Sound Changes in English 75

The velar plosives c k and g l had palatal allophones, t∫ and j respectively,

when they occurred adjacent to front vowels. In modern editions, the palatal ver-
sions are often indicated in spelling by placing a dot above the letter: <ċ> and
<ġ>.
The h is pronounced as h in most cases. It was also pronounced in consonant
clusters at the beginning of words (hl, hn, hr, and hw) as in hring hrIŋl ‘ring’
above. When it comes to actual pronunciation, hw is likely to have been [ʍ],
as it still is in words like what in some varieties of English today. The other
clusters were possibly voiceless sonorants [l̥ ], [n̥ ], and [r̥ ], which is how they are
pronounced in modern Icelandic. In coda position, h was pronounced as x, as in
feoh feox ‘money, cattle’.
The fricatives f f, þ θ, and s s were voiceless, but they had voiced allophones
v, ð, and z when they occurred between voiced sounds within words. This is never
indicated in the spelling: if these fricatives occurred intervocalically, they were
always voiced. Voiceless fricatives between voiced sounds are indicated with two
letters: <ss>, etc. The letters <v> and <z> were not used in Old English. We do find
<ð>, but it was used interchangeably with <þ>; both symbols, therefore, stand for
both voiceless and voiced fricatives according to the allophonic rules.
Finally, the letter combination <sc> (also <sċ> with dots to indicate the spe-
cial pronunciation) was pronounced ∫, as in disċ dI∫ ‘dish’. Similarly, the letter
combination <cg> or <ċġ> stood for dZ, as in hryċġ hrydZ ‘ridge’.

Example Transcription

To illustrate the pronunciation of Old English according to these simplified guide-

lines, let us look back at a few lines of Old English text from section “Linguistic
Sketch” in Chapter 1, here given in a standardised form:

Þā ġewāt sē dæġ forþ and hı̄e twelf him ġenēalǣhton and sæġdon him, lǣt þās meniġu þæt
hı̄e faren on þās castel and on þās tūnas þe hēr abūtan sind and him mete finden, forþam þe
wē sind hēr on wēstre stōwe.
θA jewA t se dæj forθ And hi e twelf hIm jene Alæ xton And sæjdon hIm, læ t θA s meniju
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲

θæt hi e faren on θA s kAstel And on θA s tu nAs θe he r Abu tAn sInd And hIm mete fInden,
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲

forðAm θe we sInd he r on we stre sto we.

▼
▲
▼
▲
▼
▲
▼
▲

Sound Changes in English

This section gives a chronological overview of major sound changes in the history
of English, with brief annotations. Examples are given with changes indicated by
an arrow →.
76 5 Track Changes

Old and Early Middle English

Homorganic Lengthening (10th C.)

Short vowels become long when they are followed by any of the consonant clusters
ld, nd, rd, or mb. These clusters were called ‘homorganic consonant clusters’
because they consist of a sonorant followed by a voiced stop at the same place of
articulation, so they were pronounced with the ‘same organs’. An example is ċild
t∫ild ‘child’ → t∫i ld. This change did not happen when more consonants followed
▼
▲

the cluster, so the (irregular) plural ċildru t∫ildru ‘children’ retained a short vowel.

Pre-cluster Shortening (10th C.)

Before other (non-homorganic) consonant clusters, long vowels become short, as
in fȳst fy st ‘fist’ → fyst. The consonant cluster st is somewhat unpredictable in
▼
▲

this change, as there are also many examples where a long vowel did not become
short, e.g., māst mA st ‘most’⟶ mAst (Minkova, 2014, p. 213).
▼
▲

We see the effect of this change regularly in present-day English because many
of these clusters were the result of morphological endings, such as the past tense
ending -te or the derivational suffix -th. For example, the verb cēpan ke pAn ‘keep’
▼
▲

had a regular past tense form cēpte ke pte ‘kept’ → kepte; and from the adjective
▼
▲

dēop de op ‘deep’ we can make a noun dēopþ de opθ ‘depth’ → deopθ. Because of
▼
▲
▼
▲

later sound changes, what was once regular morphology has turned into seemingly
irregular alternations.
The change also applied across morpheme boundaries in compound words,
such as in wı̄sdōm wi sdo m ‘wisdom’ → wisdo m—as opposed to wı̄s wi s ‘wise’,
▼
▲
▼
▲
▼
▲
▼
▲

which retained the long vowel.

Diphthong Smoothing (11th C.)

The Old English diphthong eA becomes a monophthong æ, as in ċealc t∫eAlk
‘chalk’ → t∫ælk. The long version of the diphthong becomes a long monophthong:
brēad bre Ad ‘bread’ → bræ d. A similar thing happened to the diphthong eo,
▼
▲
▼
▲

which became a monophthong ø. Examples here are dēop de op ‘deep’ → dø p for

▼
▲
▼
▲

the long version, and deopþ deopθ ‘depth’ → døpθ for the short version.
We can see the monophthong [æ] as a combination of the features from the
original diphthong: the frontness from [e] combined with the low vowel height
of [A]. Similarly, the frontness and mid vowel height of [e] combined with the
roundedness of [o] gives [ø], which is the smoothed outcome of the other diph-
thong. Note that æ already existed in the Old English vowel inventory, while ø
was a new addition. Neither vowel lasted very much longer, as they changed quite
quickly afterwards.

Changes to Low Vowels (11th C.)

The low vowels æ and A change into other vowels, depending on whether they are
long or short.
Long æ raises to E , as in brēad bræ d ‘bread’ → brE d. Long A also raises, to
▼
▲
▼
▲
▼
▲
▼
▲
▼
▲

O , so we get hām hA m ‘home’ → hO m (This last change did not happen in the
▼
▲
▼
▲
▼
▲
Sound Changes in English 77

north of England and in Scotland. Alternations such as home ~ hame ultimately go

back to whether or not this change happened in the eleventh century).
The short vowels æ and A both changed to a: ċealc t∫ælk ‘chalk’ → t∫alk, and
ramm rAm ‘ram’ → ram. This is a merger between these two vowels, where the
front–back opposition in the low vowel is given up. Throughout the rest of the
history of English, the exact quality of this short a is hypothesised to oscillate
between [a] and [æ] (we can even see such an oscillation in twentieth-century
SSBE) but this is not relevant at the level of detail we are working at here. For
words with original æ it is equally fair to say that their phonological represen-
tation remained roughly the same throughout the history of English: sæp sæp
‘sap’ → SSBE sæp.

Loss of Front Rounded Vowels (11th C.)

The front rounded vowels ø and y are unrounded, usually by merging with the
front unrounded vowel at the same height. For ø this is fairly straightforward, for
example dēop dø p ‘deep’ → de p. The y usually merges with i as expected, as
▼
▲
▼
▲

in pytt pyt ‘pit’ → pit, but there is a lot of variation in outcomes. i is definitely
most common, but in different regions we also find u or even e, and some of these
forms have eventually made their way into SSBE.
The status of these front rounded vowels in Old English is uncertain. The short-
lived existence of ø is hypothesised as an intermediate step between the original
diphthong eo and the eventual outcome e in parallel with the development of
the diphthong eA. There is a lot of documentary evidence for the history of y,
where different spellings with <y>, <i>, <u>, and even <e> suggest that the vowel
had been unstable for a long time (Minkova, 2014, pp. 193–198). This need not
be a surprise, given that it was the only front rounded vowel in Old English, and
systemic pressures would have contributed to its disappearance.
The long front vowels ø → e and y → i change in parallel with the short
▼
▲
▼
▲
▼
▲
▼
▲

vowels.

Reduction of Unstressed Vowels (11th C.)

Vowels in unstressed syllables are pronounced with less extreme realisations than
in stressed syllables. This results in them all being ‘reduced’ to schwa ∂. We
see this reflected in the spelling, as more and more vowels that used to be
spelled with different symbols becomes spelled with <e>. An example is sacu
saku ‘sake’ → sak∂. This reduction of unstressed vowels had enormous conse-
quences for the morphology and syntax of English as well; see Chapter 11 for a
discussion.

Loss of h from (Most) Initial Consonant Clusters (12th C.)

The h in the word-initial consonant clusters hl, hn, and hr disappears, leaving
just the sonorant. Examples are hnecca hnek ∂ ‘neck’ → nek ∂, and hring hriŋl
▼
▲
▼
▲

‘ring’ → riŋl.
There is documentary evidence that this change also affected the consonant
cluster hw, as words like what and white are regularly found with a simple <w>
78 5 Track Changes

spelling (as opposed to <hw>, <wh>, or even <quh>) (Minkova, 2014, pp. 108–
112). However, hw appears to have stood stronger than the other clusters. It only
disappeared from SSBE in the twentieth century (see below) and there are several
varieties of English that maintain a phonological distinction between /w/ and /ʍ/
today.

Open Syllable Lengthening (13th C.)

Short vowels become long in open syllables. This is another sound change where
the conditioning needs to be described accurately, and it takes into account the
syllabic structure of the word. In a word like sacu sak∂ ‘sake’ → sa k∂ the change
▼
▲

happens, because the input form can be divided into syllables as sa.k∂ and the a
is therefore at the end of an open syllable. In hnecca nek ∂, on the other hand, the
▼
▲

syllabification is nek.k∂. The e is the nucleus of a closed syllable and therefore

does not lengthen to eː.
Minkova (2014, pp. 221–224) notes that the change happened less consistently
with high vowels, except in Northern dialects. Words with a second syllable con-
sisting of simply a consonant and ∂ show open syllable lengthening with very
few exceptions; these words are pronounced with a single syllable in present-
day English. Words that have retained two syllables in present-day English
overwhelmingly do not show open syllable lengthening.

Trisyllabic Shortening (13th C.)

Another change involving vowel length as a consequence of syllable structure is
trisyllabic shortening, where long vowels in stress syllables become short when
they are followed by two or more syllables. Since stress in non-loanwords in
English is generally on the first syllable, this change affects long vowels in
words of at least three syllables. There are, however, very few non-loanwords of
this length. One of the few examples is sūþerne su ð∂rn∂ ‘southern’ → suð∂rn∂.
▼
▲

Compare this with the base word sūþ su θ ‘south’ which does not undergo tri-
▼
▲

syllabic shortening. With subsequent sound changes, we arrive at the present-day

alternation between south with a* and southern with 2.
With very few examples with non-loanwords, trisyllabic shortening would be a
relatively obscure sound change in the history of English. But Minkova (2014,
pp. 216–219) notes that it is especially in loanwords from Romance (French
and Latin) that we find evidence for the change: divine divi n∂, but divinity
▼
▲

divi niti → diviniti. Such alternations only happen in English, not in French or
▼
▲

Latin, but because of the lack of evidence from non-loanwords it is still poorly
understood.

Loss of Unstressed Vowels (14th C.)

The schwa ∂ that was the result of the reduction of unstressed vowels disappears, as
a result of the difference in stress between the first syllable (stressed) and the sylla-
ble containing a ∂ (unstressed). The word sacu sa k∂ ‘sake’ → sa k is an example.
▼
▲
▼
▲

We can also see how the different changes work together: with the reduction of
Sound Changes in English 79

u to ∂, the word initially retains two syllables, which allows it to undergo open
syllable lengthening before the ∂ eventually disappears.
If the ∂ was followed by a consonant and it was part of the stem of the word, it
usually remained, as in brōþor bro θ∂r ‘brother’ ⟶ broːθr . If the ∂ was part of an
▼
▲

ending, such as the plural ending -∂z or the past tense ending -∂d, it did disappear,
unless this would cause two identical consonants to end up next to each other. We
see the effects of this in present-day English endings -z in dogs but -∂z in houses;
-d in climbed but -∂d in defended.
The ∂ in the infinitive ending -∂n deserves a special mention here. Doc-
umentary evidence shows that the final n disappeared first, so that the
infinitive ending changed to -∂. We then see regular loss of the unstressed
vowel, leaving the infinitive ending at ‘zero’. As an example of a full tra-
jectory of the infinitive ending from Old English, we can take cēpan ke pAn ▼
▲

‘keep’ → ke pan → ke p∂n → ke p∂ → ke p.

▼
▲
▼
▲
▼
▲
▼
▲

Late Middle and Early Modern English

At this point, there have been so many sound changes that the Old English form of
the word is often unrecognisable and using Old English spelling would be confus-
ing. In the remainder of this section, examples will be given with their present-day
English spelling instead. As these are changes that occurred after the process of
developing a standard spelling for English started, the majority of these changes
are not reflected in spelling anymore.

Loss of /x/
The voiceless velar fricative x, which occurs only post-vocalically and often in
clusters, disappears. This usually also results in a simultaneous lengthening of the
immediately preceding vowel, so-called ‘compensatory lengthening’, as in night
nixt → ni t. In the majority of cases, x is lost early enough for the new long vowel
▼
▲

to participate in the Great Vowel Shift.

After back vowels, or in consonant clusters (often rx or xt), there was an alter-
native outcome of the loss of x, namely f. We see this in tough tux → tuf, dwarf
dwerx → dwerf, and draft draxt → draft. At the same time, we may also sim-
ply see the loss of x, as in through θrux → θru . Given the existence of spellings
▼
▲

with <f > in the 17th century, such as <dofter> for daughter, the loss of x in these
words may not always have led to a long vowel in time for the Great Vowel
Shift. The present-day outcomes of words in this category are not always regularly
predictable on the basis of their Old English form.
80 5 Track Changes

Research Highlight
CHANGE THE OTHER WAY: /ft/>/xt/ IN DUTCH
The change in English from /xt/ to /ft/, as in draught and (temporarily)
daughter, is difficult to explain on articulatory grounds. Perhaps the two
sequences sound similar enough so that they were confused in perception
instead. That becomes even more likely if you consider that the exact oppo-
site change, from /ft/ to /xt/ (spelled <cht>), happened in Middle Dutch (van
Loey 1970, pp. 99–100). This gives Dutch lucht ‘air’ (German Luft), kracht
‘force’ (German Kraft), achter ‘after’, and zacht ‘soft’. The word gracht
‘canal’ is related to the verb graven ‘to dig’; the earlier form graft shows
a more intuitive correspondence between /v/ and /f/. There are even some
Middle Dutch sources that write vijchtig ‘fifty’, but that turned out to be
too different from the base number vijf ‘five’, and the modern Dutch vijftig
resisted that change with the help of analogy.

Vowel Shortening in Monosyllabic Words (15th and 17th C.)

Another irregular and poorly understood sound change is the shortening of long
vowels E and u in monosyllabic words. These are separate changes that happened
▼
▲
▼
▲

at different times.
E was shortened to E in some words in the fifteenth century: head hE d → hEd,
▼
▲
▼
▲

bread brE d brEd, and dead dE d → dEd did undergo the change, but phonologi-
▼
▲
▼
▲

cally similar words such as lead (verb) lE d ⟶ lɛd and read rE d ⟶ rɛd did not
▼
▲
▼
▲

(Note that traditional Scottish varieties have head hi d, bread bri d, and dead di d,
▼
▲
▼
▲
▼
▲

which suggests that this change did not occur in those varieties). The shortening
of E happened predominantly before -d, in later stages also before -t, -θ, and -f
▼
▲

(Minkova, 2014, pp. 271–272).

The other irregular shortening occurred in the seventeenth century and affected
u ; we have already seen this change in the discussion of the foot/strut split in
▼
▲

section “Thefoot/strutSplit” in Chapter 4. These words typically had o in Mid- ▼

▲

dle English, with the u being the result of the Great Vowel Shift (see below). Also
▼
▲

for this change the following consonant played a role in the timing: it happened
first with a following -d, later also before -v, -t, -θ, and -k.

Great Vowel Shift (15th–17th C.)

In the Great Vowel Shift, long vowels are raised, with the high vowels i → aI and
▼
▲

u → a* becoming diphthongs. The Great Vowel Shift is discussed in some detail

▼
▲

in section “The Great Vowel Shift” in Chapter 4; for an overview of changes,

Table 4.2 is repeated here as Table 5.2.

Consonant Cluster Reductions (17th C.)

A particular set of consonant clusters is simplified by omitting the first element
from word-initial clusters kn, gn, and wr, and by omitting the last element from
word-final clusters mb, mn, and ŋl.
Sound Changes in English 81

Table 5.2 Changes to the

before after
pronunciation of long vowels
in the Great Vowel Shift rise [ri z∂]
▼
▲ [raIz]
cheese [t∫e z∂]
▼
▲ [t∫i z]
▼
▲

team [tE m]
▼
▲ [te m] (→ [ti m])
▼
▲
▼
▲

cake [ka k∂]

▼
▲ [ke k]
▼
▲

goat [gO t]▼

▲ [go t] ▼
▲

blood [blo d] ▼
▲ [blu d] ▼
▲

mouth [mu θ] ▼
▲ [ma*θ]

Modern English

With a few exceptions for non-standard varieties of English, the sound changes
up to this point took place in all (reference) varieties of the language. By the sev-
enteenth century, English-speaking settler colonies had been established in North
America, and colonial settlement would continue throughout the following cen-
turies. This means that from the seventeenth century onwards; sound changes
in (what would develop into) SSBE were not always transported to varieties of
English on the other side of the world, or even, surprisingly, to other varieties
within Britain.

FOOT/STRUT Split (17th-18th C.)

In the foot/strut split, the short vowel u lowers in some words to 7 and later 2
(the strut lexical set). Other words with u (the foot lexical set) retain the vowel,
which later laxes to *. There are tendencies for a phonological conditioning of the
foot/strut split, but they are not without exceptions.
The shortening of long u interferes with this change, as vowels that short-
▼
▲

ened earlier and that were in the relevant phonological environment participate in
the lowering to 7, whereas the vowel in other words shortened later and did not
participate in this lowering.
The foot/strut split did not apply in Northern English varieties. It is discussed
in more detail in section “Thefoot/strutSplit” in Chapter 4.

Vowel Changes Before /r/ (18th C.)

Although there is some uncertainty about the exact realisation of r in Early Mod-
ern English—it was probably still an alveolar trill [r] but it has also been suggested
that an approximant such as [ô] was a possible realisation—there were clear effects
on the preceding vowel that are likely the result of co-articulatory effects. The
common element in these changes was lengthening and centralisation.
Short ir, Er, and ur became centralised as ∂r (later lengthened to ∂ r = f r).
▼
▲
▼
▲

The words affected by this change now form the nurse lexical set, so this process
is sometimes called the nurse merger. Other short vowels generally lengthened,
as in cart kart → ka rt. Long vowels generally changed into centring diphthongs,
▼
▲

e.g., pure pju r → pj*∂r or near ni r → nI∂r.

▼
▲
▼
▲
82 5 Track Changes

These vowel changes also happened in American varieties of English. It is only

in some varieties in Scotland and Ireland that the original vowels are (to some
extent) maintained.

Research Highlight
AN/r/ BY ANY OTHER NAME…
The rhotic consonant /r/ is rather variable. For the majority of the history of
English, we think it was pronounced as an alveolar trill [r]. It is now most
commonly an approximant [ô], but other realisations also occur: a tap [R]
in Scotland, a uvular fricative [Ѕ] in traditional dialects in the northeast of
England (the ‘Northumbrian burr’), or a labiodental approximant [V] in the
southeast of England. This is still nothing compared to the twenty realisa-
tions of /r/ that have been identified in Dutch (Sebregts 2014, p. 56). So what
makes /r/ so variable?
David Natvig (2020) argues that this is because phonologically, /r/ does
not have to be anything specific. According to the theory of ‘modified
contrastive specification’ (see also Natvig and Salmons 2021), sounds are
defined inside a system by contrastive features. Based on these features, we
can make hierarchical successive divisions based on the presence or absence
of features. In such a system, /r/ is defined as not a vowel, not an obstruent,
not a nasal, not a lateral… it is defined by what it is not, and is not bound
to any specific characteristics of its own. That gives /r/ a greater freedom
to vary or change without having to change its place in this contrastively
specified sound system.

TRAP/BATH Split (18th C.)

In some words with short a, the vowel is lengthened and not much later backed
to A . This happens especially if a is followed by (a cluster containing) a voiceless
▼
▲

fricative, as in bath baθ → bA θ. Words affected by the change are now the bath
▼
▲

lexical set; remaining words with short a (now transcribed as æ) are the trap
lexical set. The trap/bath split did not occur in Scotland, Northern England, and
American varieties of English; it did make its way to Southern Hemisphere vari-
eties of English, although some bath words there retain æ. The split is described
in more detail in section “Thetrap/bathSplit” in Chapter 4.

Loss of Rhoticity (Early 19th C.)

Now that vowels before r are audibly different from vowels not before r, it is
possible to imply the following r instead of actually pronouncing it. This leads to
post-vocalic r actually being dropped from the pronunciation of the word: pure
pj*∂r → pj*∂, etc.
This is not so much the loss of r as a phoneme in English, or even from the
words that are affected by the change. Instead, it was a change in the phonotactics
of English; the rules of which sounds can go where in the syllable. The possibility
Sound Changes in English 83

of having r in the syllable coda disappeared, that is, you could no longer have
r in non-prevocalic position. The phoneme /r/ has been maintained in the mental
representation of these words, and it resurfaces when followed by a vowel, e.g.,
the car is [ð∂ kh A ô Iz]. This phenomenon is called ‘linking r’. It has even been
▼
▲

extended to cases where there was never an /r/, but where the vowel matches a
vowel that could appear before /r/. An example of this ‘intrusive r’ is law [lO ] ~ the
▼
▲

law is [ð∂ lO ô Iz].

▼
▲

There is little evidence of large-scale loss of rhoticity before 1700, and it

appears to have become part of the spoken standard in Southern England only after
1800, with some residual rhoticity found in SSBE as late as the 1870s (Minkova,
2014, p. 126). Rhoticity was lost in the reference varieties from England and the
Southern Hemisphere, and some varieties of American English. Reference vari-
eties from North America, Scotland, and Ireland have maintained rhoticity. In
traditional dialects of English, the loss of rhoticity is continuing to the present
day, with rhoticity still found in some areas in Western England such as Devon
and Lancashire.

Wine/Whine Merger (20th C.)

A change that has is roots in the Middle English period but was only accepted
as part of the SSBE standard in the twentieth century is the loss of the dis-
tinction between w and hw (phonetically [ʍ]). This meant that whine hwaIn
([ʍaIn]) → waIn was now homophonous with wine.
The wine/whine merger ends the exception that hw had as the only possible
cluster with initial h since the loss of the other h-initial clusters in the twelfth
century. It is very widespread in varieties across the world, with Scottish and Irish
English being the only reference varieties without the merger. The non-merger is
also found, however, in varieties and idiolects elsewhere, especially in careful (or
mock-careful) speech.

Diphthongisation of FACE and GOAT (19th–20th c.)

The final sound change on this chronological list is the diphthongisation of the
face and goat vowels, as in face fe s → feIs and goat lo t → l∂*t (lo*t in
▼
▲
▼
▲

American English). Again, this is a change that must have started earlier, as we
see these diphthongisations also in North American and Southern Hemisphere
Englishes. Varieties in Scotland, Ireland, and the north of England typically have
monophthongal realisations.

Ongoing Sound Changes

The list of sound changes in English is of course not complete, and there are ongo-
ing sound changes that may make the list in the future. We have seen some of these
in the previous chapters on sound change: s-retraction in /str/ clusters, th-fronting,
the Northern Cities Shift, and the Low-Back Merger Shift. Two other changes
that receive a lot of attention in the sociolinguistic literature are t-glottalling (the
pronunciation of /t/ as a glottal stop [P] in certain phonetic contexts) and goose
84 5 Track Changes

fronting (the fronting of /u / to a more front position like [ʉ] or even [y]). Many
▼
▲

of these occur to a greater or lesser extent across varieties of English.

Changes That Are Not Time-Bound

In addition to these changes that applied in a strictly chronological order, albeit

with some overlap in the periods they occurred, there are also changes that applied
seemingly haphazardly at different times.

Metathesis
Metathesis is a change where two sounds change position. These sounds are often
adjacent, but this is not necessary. In the history of English, metathesis often
involves a vowel and r, e.g., Old English þridde θrid e ‘third’ → θf d. Although
▼
▲
▼
▲

metathesis can happen at any point in time, we can use logic to determine when
it must have happened. In the case of third, the r and i must have changed places
before the loss of rhoticity, because the r can only disappear from non-prevocalic
position.
In addition to a vowel and r, metathesis also sometimes happens with consonant
clusters, as in wæps wæps ‘wasp’ → wAsp.

Analogy
Finally, some changes were not in fact sound changes, but change in morphology.
This applies in particular to changes in plurals of nouns and past tenses of verbs.
These may be analogical, that is, they may be modelled on a pattern in another
form of the word. The Old English cȳ ‘cows’ ky cannot be directly related to
▼
▲

its modern equivalent ka*z. Regular sound change would result in kaI (in fact,
kye and another form kine are still dialectal forms for ‘cows’). In order to explain
ka*z, we would have to go back to the singular cū ku → ka* and add the plural
▼
▲

marker z separately.

PRACTICE

What phonological and/or morphological changes have resulted in the present-

day Standard Southern British English (SSBE) forms of the Old English words
given? Give changes in order, with intermediate stages of the form of the words.
Use IPA for approximate phonetic transcription.
The last five are Old English words that no longer exist in present-day SSBE.
Using your knowledge of sound change in the history of English, try to predict
what form these words would have, had they survived. Again, give changes in
order, with intermediate stages and IPA transcriptions.

1. ċēse (cheese)
2. gāt (goat)
3. wrı̄tan (write, inf.)
References 85

4. dūst (dust)
5. cniht (knight)
6. scacan (shake, pl.)
7. þrēo (three)
8. hlāfas (loaves, pl.)
9. clēofan (cleave, pl.)
10. stānas (stones, pl.)
11. blōd (blood)
12. tēþ (teeth, pl.)
13. ġēar (year)
14. lǣfan (leave, inf.)
15. womb (womb)
16. sēoþan (seethe, inf.)
17. bātas (boats, pl.)
18. brōþor (brother)
19. hlūd (loud)
20. blæc (black)
21. bēċ (books, pl.)
22. sprecan (speak, inf.)
23. scyttan (shut, inf.)
24. þrote (throat)
25. þus (thus)
26. cȳþiġ ‘knowledgeable’
27. hrān ‘reindeer’
28. grēosan ‘frighten’ (inf.)
29. wı̄ġ ‘war’
30. mand ‘basket’

References

Jones, Charles. 1989. A history of English phonology. London: Longman.

Minkova, Donka. 2014. A historical phonology of English. Edinburgh: Edinburgh University Press.
Natvig, David. 2020. Rhotic underspecification: Deriving variability and arbitrariness through
phonological representations. Glossa: A Journal of General Linguistics 5 (1): 48. https://doi.
org/10.5334/gjgl.1172.
Natvig, David, and Joseph Salmons. 2021. Connecting structure and variation in sound change.
Cadernos de Linguística 2 (1): 1–20. https://doi.org/10.25189/2675-4916.2021.V2.N1.ID314.
Sebregts, Koen. 2014. The sociophonetics and phonology of Dutch r. Utrecht: Landelijke Onder-
zoekschool Taalwetenschap.
Smith, Jeremy J. 2007. Sound change and the history of English. Oxford: Oxford University Press.
van Loey, A. 1970. Schönfelds historische grammatica van het Nederlands: Klankleer, vormleer,
woordvorming, 8th ed. Zutphen: Thieme.
Watts, Richard J. 2011. Language myths and the history of English. Oxford: Oxford University
Press.
Wells, J.C. 1982. Accents of English. Cambridge: Cambridge University Press.
 Part III
Contact-Induced Change
Language Contact
How English and Other Languages Influence Each Other
6

Throughout its history, English has been in contact with many different languages.
Especially in its earlier history, up to and including the Middle English period,
there have been a number of contact situations that left pervasive traces in the
language. The French influence after the Norman Conquest of 1066 is probably
the most famous of these, but contact with speakers of Latin, Celtic, and Old Norse
has also left its marks. These contact situations differed in intensity, but also in
the types of social relations between speakers—and in the linguistic outcomes of
contact.
In this part of the book, we deal with the effects of contact on English. In
this chapter, we set up a theoretical framework through which we can understand
language contact, and discuss influence from Latin, Celtic, Old Norse, and French.
The next chapter is focused on contact within English itself, dialect contact, and
discusses the similarities and differences between these two types of contact.

A Theory of Contact-Induced Change

Borrowing and Imposition

In the theory of contact-induced change developed by van Coetsem (1988, 2002),

contact-induced change is broadly defined as the transfer of linguistic material
from one language to another. Linguistic material could be at any level of the
language: sounds, words, grammatical constructions, or even pragmatic patterns.
The two languages involved are the source language, the language that the mate-
rial originally comes from, and the recipient language, the language in which the
contact-induced change takes place.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 89

R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_6
90 6 Language Contact

As already argued by Weinreich (1953) in his foundational descriptive work on

language contact, the agent of contact-induced change is the bilingual speaker.1
What ‘bilingual’ means here, however, is flexible: the bilingual speaker’s fluency
in either language can range from fully fluent to just a few words, virtually zero.
One of the two languages is generally dominant for the speaker, that is, they feel
more fluent or more comfortable speaking that language than the other. Which
language is dominant can change from one situation to the next, though: a Welsh-
English bilingual, for example, might be Welsh-dominant in the home situation,
but English-dominant in an educational setting.
Based on dominance, we can define two types of contact-induced change:

1. a receiving language-dominant bilingual uses source language material in the

receiving language: borrowing
2. a source language-dominant bilingual uses source language material in the
receiving language: imposition

If we take Norman French influence on English as an example, we would consider

as ‘borrowing’ a situation in which an English-dominant speaker uses a French
word when they speak English. They are taking something that is not really part
of their language, but use it anyway. If it were a French-dominant speaker using
a French word when they speak English, we consider that ‘imposition’. They are
speaking another language than their own, but impose some of their own words
on that other language while they are doing that.
We know from other studies that certain domains of language (e.g., phonology)
are more stable and more resistant to change, and other domains (e.g., lexicon)
are less stable and less resistant to change. You can test this for yourself: it does
not happen frequently that people learn a new phoneme in their native language
after puberty—acquiring a new allophone might occasionally happen, but learning
new distinctions is much less common—while almost everyone continues to learn
new words throughout their lifetime. Learning new grammatical constructions is
somewhere in between: it is much less frequent than learning new words, but
it does happen a bit more often than learning new sounds. Assuming that this
(in)stability of linguistic domains translates to our bilingual speaker, the agent
in change will have a natural tendency to preserve the more stable domains of
their language, but not necessarily the less stable domains. From this follows that
borrowing primarily involves lexicon, and imposition primarily involves structural
aspects of the language.
Stability and instability of linguistic domains in a bilingual speaker are the
answer to the actuation problem in contact-induced language change. Once lin-
guistic material has been transferred into the language (or more precisely, into

1 For the sake of simplicity, the discussion here will focus on just two languages in contact. There
are very many cases worldwide where speakers may have competence in three, four, or even more
languages. The underlying dynamic of contact-induced changes in these multilingual situations
remains the same as in the bilingual speaker.
A Theory of Contact-Induced Change 91

an utterance) of one speaker; it needs to spread to other speakers in the commu-

nity if the innovation is to become a successful change. The degree of societal
bilingualism influences how easily a contact-induced innovation will spread: if
the interlocutor of our French-English bilingual is also bilingual in English and
French, they will have fewer problems processing the odd French word in English
conversation, and it also increases the chance that they will reproduce that French
word (rather than the English equivalent) in some future conversation.
When some item is transferred from one language or another, it may be changed
in the process. This is most clearly seen in the pronunciation of loanwords. A
French-English bilingual who wants to use the word croissant, for example, may
simply adopt the French pronunciation /kʁwa.sã/, or they may adapt the word to
English phonology and say something like /kwAsn̩ t/. There could be many reasons
for choosing one or the other pronunciation. The speaker may simply have an
English accent in their French, imposing their L1 phonology on an L2 word they
are borrowing back into their L1. They may adapt the pronunciation because they
are trying to make the word more understandable to an English speaker who does
not speak French, or because they feel it is more socially acceptable to do so. (Try
ordering a /kʁwa.sã/ at a food kiosk in a UK train station.) More subtle nuances
of use are even more likely to change as an item moves from one language to
another, in a process called ‘transformation under transfer’ (Meyerhoff 2009).

The Borrowing Scale

The more contact there is between two groups of speakers, and the more bilingual-
ism there is in two speaker groups in contact, the more intense contact-induced
change we can expect. Again, Weinreich (1953, pp. 83–110) acknowledged the
role of socio-cultural contact settings in the outcome of language contact. An early
and famous attempt to categorise types of contact situation and link them to the
type of language contact-induced change that may happen and is the ‘borrowing
scale’ by Thomason and Kaufman (1991, pp. 74–75) (Table 6.1). The summary
table contains only very generic terms to the point of being almost non-descript,
but their discussion lists a lot more specific types of changes that may happen
at the various points on the scale. (Unfortunately, there is no such specification
for what they mean by the different intensities of contact.) Note that while the
borrowing scale is a descriptive tool, it is often incorrectly interpreted as if it has
predictive value.
Lexical borrowing at the lower intensities of contact begins with content words
and with non-basic vocabulary. This is not really a linguistic constraint, but more
of a cultural and functional constraint: with low-intensity contact and little societal
bilingualism, it is likely that the only words that are picked up from the other
language are words that refer to concepts related to the other language group’s
culture. As contact increases, we see that function words can be borrowed too:
conjunctions, adverbial particles, and discourse markers. From Level 3, borrowing
extends to affixes, pronouns, and lower (i.e., basic) numerals. With even more
intense contact than that, anything is possible in lexical transfer.
92 6 Language Contact

Table 6.1 Thomason and Kaufman’s borrowing scale

Contact Transfer
1 Casual contact Lexical borrowing only
2 Slightly more intense contact Slight structural borrowing
3 More intense contact Slightly more structural borrowing
4 Strong cultural pressure Moderate structural borrowing
5 Very strong cultural pressure Heavy structural borrowing

Transfer of structural material is unlikely to happen in casual contact. At Level

2, the first types of structural material that can be borrowed are phonemes and
morphemes. At this point, structural transfer can best be seen as collateral damage
from lexical borrowing: a loan word might for example contain a sound that is
new to the language or even just a sound that already exists as an allophone but
that is found in a different position in the loan word, making it a phoneme.2 At
Levels 3 and 4, structural borrowings spread to native lexical material. An example
here could be the suffix -able, which was borrowed along with Romance verb
roots (legible, edible) but which can now also be used with native English verbs
(readable, eatable). Only at Level 5, so under very strong cultural pressure from
the other language group, do we find major structural changes such as word order
change, phonetic change, or even complete typological restructuring.
The borrowing scale was designed to work in situations of language mainte-
nance, where there is a stable speaker population that passes on their language
to their children. Beyond Level 5 on the scale, we are looking at such intense
cultural pressure that a society might give up their old language and shift to the
other language. When children do not fully acquire the now-dying ‘old’ language
because parents are not teaching it to them anymore, but they still pick up some of
the language, the changes that we find in such moribund languages are extremely
diverse and very few generalisations can be made (Campbell and Muntzel 1989).
The borrowing scale hinges on the intensity of contact alone, but Thomason and
Kaufman (1991, p. 72) acknowledge that linguistic factors can also play a role in
what the outcomes of contact-induced change can be:

[T]ypological distance does not appear to have an effect on the linguistic results of the
most intense borrowing situations, i.e., those involving heavy to extreme borrowing; but
in slight to moderate borrowing, source-language features that fit well typologically with
functionally analogous features in the borrowing language tend to be borrowed first.

In other words, in really intense contact situations, the similarities between the
languages in contact are irrelevant because we expect a lot of change to happen

2For example, Dutch does not have a phoneme /l/ in what is described as ‘native’ vocabulary. It
does have an allophone [l] of the phoneme /k/ that appears as the result of assimilation, e.g., in
zakdoek ‘handkerchief’ / zAkduk/ [ zAlduk]. Loanwords from French and English that contain /l/
have led to minimal pairs like goal /lo l/ vs. kool ‘cabbage’ /ko l/, which means that /l/ is now
▼
▲
▼
▲

properly analysed as a separate phoneme in Dutch as well.

A Theory of Contact-Induced Change 93

anyway. But in less intense contact, similarities between languages might actually
help contact-induced change at a level on the borrowing scale where such change
would be less likely to happen if the languages were more different. We will see
some examples of this when we talk about contact between closely related Old
English and Old Norse.

Pivot Matching

When two languages are active in a speaker’s brain, lexical material and construc-
tions from both languages are available to fulfil a communicative function. At the
same time, speakers also want to use the language that is most appropriate in
a communicative situation. If someone is speaking Language A but a word from
Language B seems more appropriate to use, they might simply use that word as a(n
incidental) loanword. The same happens with constructions: sometimes a speaker
can use a construction from another language because it seems more suitable for
the purpose. If they also use the words from that language, we are again dealing
with a loanword (or perhaps a code-switch)—but they can also decide to use the
most appropriate construction from Language B while still speaking, and using
words from, the most appropriate language, Language A. When this happens, we
use a process called pivot matching (Matras 2009, pp. 240–243).
Using a construction in a different language involves not simply word-for-word
translation, but rather the deconstruction of the construction into its constitutive
elements (pivots), which get translated individually before the construction gets
reconstructed in the other language. The use of the English construction I am
hot in Dutch as ik ben heet (idiomatically, ik heb het heet ‘I have it hot’) may
look like a word-for-word translation, but when we consider other languages, we
can see that it is actually translated at a more abstract level. To be precise, we
have a copula construction with a first-person pronoun and a subject predicate
that gets translated, and how that translation ends up depends on the grammar of
the recipient language. Scottish Gaelic is verb-initial, so we get tha mi teth, not
a word-for-word translation *mi tha teth with English word order; Italian allows
pronoun subjects to be left out, so we get sono caldo ‘am hot’; and Arabic does
not use a verb in such constructions, just a subject and the predicate, so we get
‘ana sāhin ‘I hot’.
This¯ simple example shows that it does not have to be individual words, but can
also be the abstract construction that is transferred from one language to another. It
works the same with more complicated multi-pivot constructions such as idioms:
the grammars of both languages are always active in a bilingual speaker. Given
enough pressure, however, it is easy to imagine how such constructions become
more aligned, so that, for example, bilingual Italian speakers will no longer leave
94 6 Language Contact

out the subject in this construction, or bilingual Arabic speakers will use an explicit
verb.3

In other languages
LANGUAGE MIXING AS THE RESULT OF EXTREME CONTACT
In some situations of extreme language contact, two languages may ‘merge’
in specific ways to form a new variety. Examples of such “mixed languages”
are Media Lengua (spoken in Ecuador), Ma’á (Tanzania), Mednyj Aleut
(Bering Island, now extinct), and Michif (Canada). In some of these cases,
the lexicon is taken from one of the two languages in contact, and the gram-
mar is taken from another: Media Lengua has Spanish lexicon and Quechua
grammar, while Ma’á has mostly Cushitic and Maasai lexicon and Bantu
grammar. On other cases, the split is even more remarkable: everything to
do with nouns comes from one language, and everything to do with verbs
from another. This is the case for Mednyj Aleut (Aleut nouns and Russian
verbs) and Michif (French nouns and Cree verbs). The following example
from Michif shows how this works in practice (from Meakins, 2013; Cree
material in bold):

(1) êkwa pâstin-am sa bouche ôhi le loup ê-wî-otin-át and open-he.it his mouth this
the wolf COM-want-take-he.him ‘And when the wolf came to him, he opened his
mouth.’

Although they are rare, it seems like mixed languages always develop
in situations of community bilingualism. As such they are unlike pidgins
and creoles (see Chapter 7) where adult second-language learning is nec-
essary for communication. Rather, mixed languages may develop as an
identity marker: speakers of Michif, originally the descendents of Franco-
phone fur traders and Cree women, are both Cree and French, and not
exclusively either. Mixed languages are separate varieties that exist indepen-
dent from the contributor languages, so they are not the result of on-the-spot
codeswitching. Codeswitching is one of the mechanisms through which
mixed languages are thought to arise, however. The small number of known
mixed languages and the large degree of variation between them means that
there are still many questions about how they develop, but such languages
give a wealth of information about the possible outcomes of high-intensity
language contact.

3Research on Italian heritage language speakers in Toronto, Canada, however, shows that this is
not necessarily the case: they use and leave out pronouns in roughly the same proportions as Italian
speakers in Italy (Nagy et al. 2011).
Influence from Latin 95

Influence from Latin

The earliest contact situation for English was with Latin. The famous textbook by
Baugh and Cable (2013, pp. 73–87) identifies three periods of Latin influence on
Old English, which for some reason are numbered from 0 to 2. This periodisation
has become standard in many histories of the English language.
The zero period involves contact between Romans and Germanic tribes on the
continent. This is before the settlement of Angles, Saxons, and Jutes in Britain, so
before we can really speak of ‘English’ as a concept. Still, the material that was
borrowed in this period was part of the language of the Anglo-Saxon settlers and
ended up in English. It is important to remember that during the Roman occupation
of Britain (43–c. 410 ad), the Germanic tribes that would later become the Anglo-
Saxons were still in Germania. The Anglo-Saxon invasion of Britain is dated to
the mid-fifth century, and leftover Romans may have encountered the Anglo-Saxon
vanguard, but direct Roman–English contact in Britain is negligible.
Contact between the Romans and Germanic tribes on the continent was pre-
dominantly by means of trade and warfare. The Germanic tribes fought against
the Romans, but also with them as mercenaries in Roman legions. The loan words
borrowed from Latin in the zero period reflect this contact situation: the majority
has to do with money, foodstuffs, and military terminology.
The first period involves contact between Romans and Celtic tribes during
the Roman occupation of Britain. The words borrowed from Latin in this period
ended up in Old English not directly from Latin, but through contact with Celtic,
and so in a sense are second-hand loan words. Because of the social situation
surrounding contact with Celtic languages (see below), very few words survive
that can be traced back to this period. They are mostly restricted to (elements of)
place names.
The second period involves contact between Christian missionaries, who used
Latin, and Old English. Christianity was not entirely unknown in Britain due to
the earlier efforts from Irish missionaries, but the Christianisation took on a more
serious form with the Augustinian mission from 597 ad. It, therefore, comes as
no surprise that Latin loanwords from this period are in the domain of religion,
but there are also words to do with science and botany, as these were activities
common to monasteries. Not all religious terminology was borrowed from Latin;
some existing Old English words were re-purposed for Christian use.
There is an additional period of Latin influence on English later: the Renais-
sance saw a lot of Latin loanwords being introduced in English, too (see Baugh
and Cable 2013, pp. 212–231). The emancipation of vernaculars in this period—
people using their own language rather than Latin in all domains—meant that
English needed new words to be able to fulfil its new functions. Many of these
were simply borrowed from Latin, but sometimes the source was neo- or pseudo-
Latin, and sometimes the words may have entered English via French. There was
a deliberate movement to ‘enrich’ English with Latin loanwords, also because this
would give English more prestige, but at the same time there was also a counter-
movement that opposed such pretentious ‘inkhorn terms’. Many Latin loanwords
96 6 Language Contact

from this period were linked to new domains such as science, where English
simply did not have existing vocabulary; these loanwords are often called ‘need
borrowings’. The loanwords in higher registers for concepts for which English
already had a word (a native English word and/or a French loanword) are ‘prestige
borrowings’.
All in all, the contact with Latin was by far the most casual of the four con-
tact situations described in this chapter. There may be many Latin loanwords in
English, but that is really all there is. Some English grammarians in the eighteenth
century did try to fashion formal English grammar on a Latin model, banning com-
mon constructions like preposition stranding and split infinitives, and describing
English grammar using Latin concepts like case, but this has very little to do with
English–Latin language contact.

Influence from Celtic Languages

When the Angles, Saxons, and Jutes invaded Britain in the fifth century, they came
into contact with Celtic tribes. These spoke a variety of Celtic languages, probably
from the P-Celtic group—the ancestors of Welsh, Cornish, and Breton. (The Q-
Celtic group, the ancestors of Irish, Manx, and Scottish Gaelic, was spoken in
Ireland at the time.) We will gloss over any differences that may have existed
between Celtic varieties, and will just use ‘Celtic’ as an umbrella term. Contact
between Old English and Celtic was of a completely different kind than contact
with Latin and therefore had completely different results.
The most traditional view of Celtic–English contact was that Celtic tribes were
either enslaved, exterminated, or driven to the fringes of Britain (Cornwall, Wales,
and the Scottish Highlands). In such a situation, of course, there is little reason
to expect Celtic influence on Old English. Historical evidence suggests, however,
that this was not the case. The Celts may have been ‘subjugated’ in the areas with
the most Germanic settlers, but overall the two communities appear to have lived
side by side, and the Celts even outnumbered the English twenty to one (Hickey
2012, p. 499). In the areas where both population groups lived, the Celts do appear
to have shifted to English.
Celtic loanwords in English are restricted to place names and words for natural
phenomena, as well as some religious terminology. (Christianity was common in
Celtic areas before it reached Anglophone Britain.) Overall, this leads Baugh and
Cable (2013, p. 72) to conclude that ‘the Celtic influence remains the least of the
early influences that affected the English language’. But even if Celtic influence
on the lexicon was negligible, more recent evaluations of Celtic–English contact
suggest that there may be quite a few cases of Celtic influence on the structure of
English. This is especially seen where English is different from other Germanic
languages but similar to Celtic. The idea that Celtic had substantial structural influ-
ence on English is known as the ‘Celtic Hypothesis’, and it is based on a variety
of constructions.
Influence from Celtic Languages 97

Table 6.2 Old English

wesan bēon
paradigms for to be:
wesan and bēon iċ eom iċ bēo I am
þū eart þū biest you are (sing.)
hē/hēo is hē/hēo biþ he/she is
wē sind wē bēoþ we are
ġē sind ġē bēoþ you are (pl.)
hı̄e sind hı̄e bēoþ they are

The first of these constructions is the progressive construction, also known

as the -ing form. The origins of this construction are disputed, with also Latin
influence and independent development being proposed, but Celtic influence is a
common explanation. Both English and Celtic use a verbal noun and a preposi-
tion: iċ wæs on huntunge ‘I was (on) hunting’—the preposition was later lost in
English. The form is very similar to that in other Germanic languages (Dutch ik
was aan’t jagen or German ich war (am) jagen), so the origins of the construction
can probably not be attributed to Celtic influence alone, but the exact meaning
and the frequency of use in English are more similar to Celtic than to Germanic
(Hickey 2012, p. 502).
A second feature is that Old English had two paradigms for the verb to be. One
came with the infinitive wesan and one with the infinitive bēon (Table 6.2). These
verbs have different origins in Proto-Indo-European. In other Germanic languages,
like Dutch or German, and in present-day English, the paradigms have merged into
a single paradigm—in fact, wesan is already a merger of two paradigms, which
explains why the forms of this verb look so diverse. But in Old English, especially
West Saxon, the two paradigms appear to have been functionally different: wesan
was used for the existential present (things that are the case right now, such as the
cat is ill) while bēon was used for the habitual present (things that are generally the
case, such as the cat is fluffy). This mirrors the use of two different verbs for to be
in Celtic at the time. A possible explanation is that Celts learning English latched
on to variation in Old English paradigms of to be, and reallocated the forms to a
functional distinction from their own language (Hickey 2012, p. 501; Lutz 2009).4
Thirdly, Germanic languages have so-called external possessor constructions,
where inalienable possession is expressed not with a possessive pronoun, but
with a (dative) personal pronoun. This construction is found occasionally in Old
English—þæt him þæt heafod wand forð on ða flore ‘that him the head went forth
on the floor’ = ‘that his head rolled onto the floor’—but this is no longer pos-
sible in English. Celtic languages do not have external possessor constructions
either. The hypothesis is that they disappeared from English due to contact with

4 The present-day plural form are derives from an Old English variant earon, part of the paradigm
for wesan, which was found predominantly in the North of England. It spread at the expense of
sind throughout the Middle English period, probably as a result of English–Norse contact: the Old
Norse plural forms all started with er- (OED, q.v. be).
98 6 Language Contact

Celtic (Hickey 2012, p. 500) because Celtic learners of English did not use these
constructions in their English.
There is also an argument that do-support may have Celtic origins. The evi-
dence for this (McWhorter 2009) is based on both English and Cornish having
do-support. If we assume that it was transferred from one of these two languages
into the other through language contact, the question is which of the languages the
construction originated in. McWhorter argues that if do-support had been an origi-
nally English construction that was borrowed into Cornish, then Breton—which is
essentially Cornish with French rather than English influence—should not have do-
support. Seeing that Breton does have do-support as well, it is more likely that the
feature was originally Celtic and was borrowed from Cornish to English. However,
do-support also exists dialectally in Dutch and German, which have nothing to do
with Celtic, and the fairly generic semantics of do make it an obvious candidate to
develop into an auxiliary verb, so this argument is not entirely convincing: it might
have developed independently in both languages (Hickey 2012, pp. 502–504).
The final feature that is supposedly due to Celtic influence is that even though
English does have the words yes and no, it is much more common to (also) answer
questions by using mini-sentences with auxiliary verbs (I do, I’m not). Celtic
languages tend to answer yes/no questions with mini-sentences as well. This is
another case where English runs parallel with Celtic while it differs from other
Germanic languages (Vennemann 2009).
As we have very little evidence for the earliest Old English, let alone for
P-Celtic of the same period, the Celtic hypothesis cannot be conclusively proved.
Still, it appears that there was relatively intense contact with possibly considerable
structural borrowing, but without much lexical influence. The Celtic influence on
English is a case of substrate influence, where a less powerful group influences the
language of a more powerful group. In the terms of van Coetsem (1988, 2002),
it is a case of imposition or source language agentivity: speakers of Celtic lan-
guages used Celtic features in their second-language version of English. Because
the Celts outnumbered the English so enormously, the English would have heard a
lot of Celtic-influenced English, and even if it may not have been very prestigious,
some of the Celtic English features may have been acquired by the English as
well. Over time, the Celtic-influenced English may have lost the association with
a Celtic accent and have become simply English without ethnic associations.
It is important to note that the discussion in this section has focused on con-
tact between English and Celtic in the very earliest stages of the Anglo-Saxon
settlement of Britain. In later centuries, we find renewed contact between English
and Celtic languages, as English spreads into territories inhabited by speakers of
Welsh, Irish, Cornish, and Scottish Gaelic. The changes resulting from those con-
tact situations will not be reviewed in depth, but the different types, intensities,
and durations of contact have led to different outcomes in these situations from
what we saw in Old English (Hickey 2020).
Influence from Old Norse 99

Influence from Old Norse

The third contact situation that we discuss here is that with Old Norse. This is
an interesting case study because the effects of Old English–Old Norse language
contact are both lexical and structural. At the same time, the similarities between
Old English and Old Norse make it difficult to tease apart in retrospect exactly
how this language contact took shape.
This more complex situation of language contact necessitates a slightly longer
section than the other contact situations receive. First, we will look at the his-
torical context of English–Norse contact and the questions surrounding the extent
of Viking settlement in England. We will then discuss the questions of how sim-
ilar Old English and Old Norse really were and what that means for the ease
of linguistic transfer, before focusing on some of the more notable outcomes of
English–Norse contact. The section concludes with a discussion of a provocative
theory claiming that English–Norse contact was in fact even more intense than
conventionally thought.

Historical Context

Contacts between speakers of Old English and Old Norse started with Viking
raids of the North East English coast in the late eighth century, with the 793
raid of the monastery at Lindisfarne as a famous and symbolic starting point.
Within a century, the Vikings stayed and settled in England as the records in the
Anglo-Saxon Chronicle show (Townend 2002, pp. 1–2). There is no question that
Scandinavians settled in England. The question is, however, how large-scale this
migration was. There are various kinds of evidence, but there is nothing that really
gives a conclusive answer.
First of all, we can look at place name evidence. There are many (originally)
Scandinavian place names in England, which can often be recognised by suffixes
like -thorp (< þorp ‘village’), -by (< býr ‘town’) or -thwait (< þveiti ‘meadow’).
As can be seen from Fig. 6.1, these place names are concentrated in Yorkshire,
the East Midlands, and East Anglia. The sheer number of Scandinavian place
names suggests that the Vikings settled in large numbers, because they were able
to swamp the original English place names. There is also a great variety of name
types, and there is Scandinavian influence in so-called microtyponomy—names
of fields, fens, becks, etc., that were too insignificant for the later waves of elite
settlers like the Normans to bother with.
A second type of evidence comes from genetics. This type of analysis is not
straightforward, because it involves comparing genetic material from the present-
day UK with that from present-day Iceland—they use Icelandic genetic material
to minimise the influence from later changes in continental Scandinavia due to
migration, something that cannot be done for the British material—in order to
say something about population structures in the tenth century. The large time
depth makes this kind of evidence difficult to interpret, also because there are
100 6 Language Contact

Fig. 6.1 Distribution of Scandinavian place names in England (Adapted from Fellows-Jensen
2011, p. 70)

not that many significant genetic differences between the population groups to
focus on in the first place. We must also take into account that genes are not the
same thing as culture. With all these caveats, ‘there is genetic evidence which can
plausibly be interpreted as reflecting Scandinavian influx, albeit in regions […]
that were relatively sparsely populated prior to settlement. In other regions any
minor difference in gene frequencies may simply have been ‘swamped’ by the
indigenous population’ (Evison 2000, p. 289).
Influence from Old Norse 101

Finally, there is archaeological evidence for Viking settlement, but this is simi-
larly problematic. There are very few burial sites in England that, based on rites or
burial gifts, can be assigned undisputed Scandinavian provenance. But the small
number would suggest that there were only a handful of Norse settlers, while the
documentary evidence talks about Viking invasions as a micel here ‘great army’,
so the burial sites do not give conclusive evidence (Halsall 2000, p. 269). Sim-
ilarly, there is little clear evidence for Scandinavian settlements from buildings,
both because possible Scandinavian settlement sites have not been excavated—they
were generally successful and are therefore still inhabited—and because there is
little particularly Scandinavian about them. All of this might also indicate cultural
mixture between English and Vikings (Richards 2000).

Mutual Intelligibility

An important question in Norse–English contact is to what extent Old Norse and

Old English were mutually intelligible. If they were mutually intelligible, and
speakers of both groups could continue to speak their own language, with perhaps a
bit of accommodation, we would expect different outcomes of contact than if they
were not, and speakers would have to learn the other language in order to com-
municate. The degree of mutual intelligibility determines whether Norse–English
contact was more like language contact, with the processes and outcomes that
we know from that, or like dialect contact, with different processes and different
outcomes.
Over a thousand years after the fact it is difficult to determine how well speak-
ers of Old Norse and Old English could understand each other, but there are a few
possible avenues of research that could shed light on this (Townend 2002, p. 13).
Firstly, a test the informant approach is of course no longer possible, but Townend
says that the treatment of Scandinavian loan words and place names in Old English
boils down to essentially the same thing: if these were easily integrated and under-
stood, judging from documentary evidence, this would suggest the languages were
relatively mutually intelligible.
Present-day mutual intelligibility research could also ask the informant; this
translates into historical research as searching for anecdotal evidence in historical
sources about the mutual intelligibility of Old Norse and Old English. Fortunately,
there is quite a lot of this. The Icelandic sagas comment regularly on language
differences and language learning in other countries, like Ireland, Bjarmaland
(Russia), and Germany; but for England (and Friesland) such accounts are entirely
absent. This already suggests communication without second-language learning
was possible, but there is even positive evidence about this, such as this comment
from Gunnlaugs saga ormstungu:

The language in England then was the same as in Norway and Denmark. But the languages
changed in England when William the Bastard conquered it; from then on French became
current in England, because he was from France. (Townend 2002, p. 150)
102 6 Language Contact

Of course what we can do after the fact is do a post-hoc linguistic comparison.

It helps intelligibility if the two languages use the same words (cognates) for the
same concepts, but more importantly, if there are not too many phonological dif-
ferences between the languages, and if the differences that do exist are predictable.
Furthermore, we can attempt to analyse the social relations between the two popu-
lation groups. Social factors such as power relations, the amount of exposure to the
other language, and language attitudes are known from present-day research into
mutual intelligibility to influence how well people understand related languages
(Schüppert and Gooskens 2011; Gooskens and Heeringa 2014; Schüppert et al.
2015; Gooskens et al. 2017).
Based on the available evidence for all four methods, Townend (2002, p. 183)
concludes that although Old Norse and Old English were not fully mutually intel-
ligible in the strictest sense, they were ‘adequately’ and ‘pragmatically’ so: there
was enough mutual understanding of each other’s language to ‘permit face-to-face
and day-to-day transactions’, and there was no need for interpreters. This means
that ‘Viking Age England was a bilingual society, but not a society comprised of
bilingual individuals’ (p. 185), and that we are perhaps more likely to see typical
dialect contact changes in this contact scenario.

Changes Under English–Norse Contact

A considerable part of Norse influence on English was lexical. Townend (2002,

p. 201) distinguishes two periods of lexical borrowing from Old Norse, with
different mechanisms and outcomes. During the Old English period, loans were
need-based borrowings, denoting new concepts. These were often adapted in form
to Old English, so rather than direct borrowings they were loan translations. These
words ended up in English under recipient-language agentivity. Borrowings in
Middle English were clearly not need-based, as they included a range of basic
lexical items. Of course, basic vocabulary can be borrowed, but that would assume
much more prestige for Old Norse during the later periods (whereas the Vikings
were more powerful during the earlier periods). This is the period, though, that
Old Norse died out in England—so we are talking about source language agentiv-
ity, and (originally) Old Norse speakers retaining some of their vocabulary as they
underwent a language shift to English. Most notably, some of the basic lexical
items borrowed from Norse were words that English already had, and where one
cognate replaced another. We can tell that these words have a Norse origin because
the initial consonants did not undergo a typically early Old English sound change:
sky with /sk/ (OE sċēo ‘cloud’ with /∫/), give with /l/ (OE ġiefan with /j/), and
kettle with /k/ (OE ċietel with /t∫/), among others (Dance 2017, pp. 207–215).
In contrast to the French loan words we will see later on, the Norse loans
also included some function words. These form so-called closed classes, meaning
that they do not easily take in new members, which means that the borrowing of
function words requires intense contact according to Thomason and Kaufman’s
Influence from Old Norse 103

borrowing scale. The prepositions till, until, and fro (as in to and fro) are borrow-
ings from Old Norse, and they exist alongside the original English forms to and
from. Even more spectacularly, Norse influence may have been the driving force
behind the development of the pronouns she and they (including them and their),
although as we will see below, both these accounts are disputed. Pronouns are
borrowed at Level 3 or higher in the borrowing scale, so this is usually seen as
evidence that Norse–English contact was very intense.
The Old English third-person plural pronouns were hı̄e (nom./acc.), heora (gen.)
and him (dat.), all with initial h-. Confusingly, these forms were very similar to
hēo (‘she’, nom.), hı̄e (‘she’, acc.), and him (‘him, it’, dat.). The Old Norse third-
person plural pronouns (masc.nom. þeir, gen. þeirra, dat. þeim, acc. þá) were not
too different in the rhymes from their Old English counterpart, but conveniently
all started with þ-, which could help disambiguate between singular and plural
pronouns. If children hear both þ- and h- forms at roughly equal frequencies, they
will probably focus their acquisition on the þ- forms that unambiguously refer to
a plural antecedent. However, Cole (2018) has shown that Old Northumbrian, the
dialect of Old English spoken in the north of England, had started to use demon-
strative pronouns þā, þāra, þām ‘those’ with the function of personal pronouns
independently from any Old Norse influence. The Old Norse forms with þ- are
still relevant, however, as they would reinforce a development that was already
going on (p. 201).
Scandinavian influence has also been used to explain the development of the
pronoun hēo ‘she’. With regular sound change, this would regularly develop into
he (which would be confusing) or hoo (which has in fact happened, and this form
survives dialectally according to the OED). Spelling evidence shows a gradual
spread of forms with initial /∫/, via intermediate /ç/ (as in huge). This change
is extremely rare in English but much more frequently attested in Scandinavian
(e.g., Shetland < Hjaltland), so Old Norse influence seems an obvious explanation.
However, Britton (1991) and Laing and Lass (2014) investigate all the historical
evidence from manuscripts and conclude that Old Norse is not involved in this
change.
Apart from the lexical and morphological influence, there is also evidence that
points to structural syntactic influence from Old Norse on Old English. Often, this
concerns syntactic constructions that are shared between Old Norse and Middle
English, but not Old English (where Old English, on the other hand, patterns more
often with Dutch and German). We do see the beginnings of many of these changes
before we can really expect extensive Scandinavian influence, and some of these
features have later developed independently in Dutch and/or German, without any
Old Norse contact (Kortmann 2016). It is therefore likely that again, similarities
with Old Norse reinforced ongoing changes in Old English. However, there is a
fringe theory that Old Norse influence was much more important, which we will
briefly touch upon below.
The consequence of Townend’s (2002) conclusion—that the English and the
Vikings communicated with each other in their own mutually more-or-less intel-
ligible languages—is that concepts from dialect contact, such as accommodation,
104 6 Language Contact

become relevant. This is on top of the fact that adults are not very good at language
learning. There is reasonable consensus, therefore, that contact with Old Norse
did not necessarily initiate language change, but that it confirmed and accelerated
changes already in progress, where such changes went in the direction of (a) Old
Norse itself, or (b) something that would be easy for speakers of Old Norse to
understand and acquire.

Too Far? The Viking Hypothesis

When we compare the syntactic features of Old English, Old Norse, and Middle
English, there are of course features that are shared by all three languages. But it
is striking that there are many syntactic features that are shared by Middle English
and Old Norse, but not Old English, while there are no features that are shared by
Middle English and Old English, but not Old Norse. In other words, the syntactic
changes from Old to Middle English all bring the language in line with Old Norse
syntax, while nothing syntactic survives from Old English that is not supported by
Old Norse syntax. The combined evidence from all these changes in syntax has
prompted Emonds and Faarlund (2014) to propose a radically different account of
the history of English, which they term the Viking Hypothesis.5
In what Emonds and Faarlund call the ‘traditional’ scenario, Middle English
developed fairly straightforwardly from Old English. Old English underwent many
fundamental grammatical changes, it incorporated a lot of Old Norse vocabulary,
and became Middle English. In the alternative scenario of the Viking Hypothe-
sis, however, Middle English developed straightforwardly not from Old English,
but from Old Norse. This scenario requires essentially no grammatical changes,
except some changes that also happened in Mainland Scandinavia. Old Norse
incorporated even more Old English vocabulary and became Middle English. The
evidence that Emonds and Faarlund give in favour of their argument is predom-
inantly syntactic. The lexical evidence is not decisive, they claim, because both
scenarios require a substantial amount of lexical borrowing, and because Old Norse
and Old English were mutually intelligible—or at least had many cognates with
clear regular correspondences—it is difficult to establish the direction of borrowing
anyway.
As can be expected with a theory that deviates so clearly from the established
narrative, and perhaps especially in the case of the history of English (Watts 2011,
pp. 298–301), the Viking Hypothesis has received substantial criticism from many
different scholars, criticising Emonds and Faarlund’s use of theory, data, and socio-
historical evidence; the harshest criticism is that they simply do not ‘know [their]
stuff’ (Los 2016). Concretely, critics argue that syntax is not a good place to

5 Quite a few of the relevant syntactic changes require a bit more understanding of theoretical syn-
tax to discuss in detail. Faarlund and Emonds (2016) give a summary of the argument, which briefly
explains the changes they believe to be evidence of substantial Old Norse influence.
Influence from French 105

look for language relationships. More than phonology and morphology, syntax is
determined by human cognition and conceptualisation; and there are relatively few
options anyway. As we saw with pivot matching, the abstract patterns from syntax
are also more easily borrowed in intense contact situations than the concrete words
or endings from morphology are, and the morphology of Middle English derives
overwhelmingly from Old English (Thomason 2016). There are also claims that
Emonds and Faarlund overstate the disruption that took place between Old and
Middle English. Many features actually remained stable throughout the develop-
ment of the language. And many features that supposedly derive from Old Norse
have also developed in other West Germanic dialects (varieties of German, Dutch,
and Frisian) without influence from Old Norse (Kortmann 2016). Although this
argument—if a change can happen in a situation without language contact, contact
must not be the cause of change even in situations with language contact—betrays
a bias against contact explanations that is prevalent in historical linguistics (Farrar
and Jones 2002, pp. 4–7), it is true, conversely, that not everything that could have
come from Old Norse must have come from Old Norse.6

Influence from French

The final contact situation that discusses here is the contact with (Norman) French.
The Norman Conquest of England in 1066 and the subsequent introduction of
French are generally seen as the seminal event in the history of English. And it is
definitely the case that the lexicon of Middle English looks a lot more Romance
than the clearly Germanic Old English lexicon: around half of the 1000 most fre-
quent words in English are of Romance origin (Durkin 2018; this includes words of
French as well as Latin origin). But how intense was this French–English contact
really and is there more than just lexical influence?
In so far as this can be ascertained from historical sources of questionable
reliability, such as the Domesday Book, Norman migration was relatively limited
in numbers: a few tens of thousands, amounting to 2% of the total population
of Britain at the time. The incoming Normans were a military aristocracy that
replaced English leadership much like a hostile business take-over; the English
minions were simply left in place. There was some intermarriage between Nor-
mans and English, possibly as a way for the new aristocracy to access English
lands (Härke 2002, pp. 162–163). But we should definitely not see the Norman
Conquest as an instigator of large-scale demographic change in Britain.
Accounts of French–English language contact have traditionally held that the
Norman Conquest led to a situation of diglossia (Ferguson 1959): one language is

6 There are several other points of criticism against the Viking Hypothesis, and also some argu-

ments that support it. It would go too far to discuss these all here at length. The issue of the journal
Language Dynamics and Change where Faarlund and Emonds present a summary of their argument
also contains responses from a range of historical linguists, and as a whole gives a good overview
of the discussion around the Viking Hypothesis.
106 6 Language Contact

Fig. 6.2 Uptake of French loan words in English from the eleventh to nineteenth centuries, based
on data by Jespersen and Baugh (Baugh 1935)

used in a set of formal, official domains (the H or ‘high’ language) and one lan-
guage is used in informal, less official domains (the L or ‘low’ language). French
established itself as the H language, while English was the L language.7 Bilingual
skills were not unheard of, the story goes, but generally acquired on a need-to-
know basis (Baugh and Cable 2013, pp. 116–120). English speakers who had to
deal with the Normans professionally would learn a bit of French, and Normans
that had to interact with the English would learn some English, but most people
were essentially monolingual. A catchy example to illustrate such a situation is that
the Francophone elite would eat beef , pork, mutton, and venison, while the Anglo-
phone peasants would be rearing cows, pigs, sheep, and hunting deer. But such
an account where French loanwords trickle down from the elite has come under
scrutiny in more recent research, as it is not always supported by social or linguistic
evidence: while there are many loanwords from specialist areas, a vast majority
are extremely normal words. There is nothing particularly elitist about large or
very, or about emotions such as joy or anger. Bilingualism in English and Anglo-
Norman was also considerably more widespread lower down the social scale than
previously thought (Timofeeva and Ingham 2018). There are many examples of
multilingual, so-called macaronic texts from this period in different genres that
mix English, French, and also Latin in ways that suggest multilingual competence
(Wright 1992; Schendl 1996, 1997). We therefore need to reconsider the story.
Borrowing from French did not start directly after 1066. Around the turn of
the twentieth century, the Danish linguist Otto Jespersen charted when French
loan words first appeared in English texts; this was later re-done with a slightly
different methodology by Baugh (1935). They built a corpus of 1000 French loan
words and counted how many of them first appeared in documents from different
half-centuries; their data is shown in Fig. 6.2. French borrowings only take off in
the thirteenth century, peaking in the late 14th followed by an immediate sharp
decline.

7 What is meant by ‘French’ here is a collection of related varieties: Parisian-style French, which
later would become the basis for Standard French; Norman, as spoken in Normandy; and Anglo-
Norman, the variety of Norman that developed in England with language change different from
what happened back in Normandy.
Influence from French 107

Jespersen’s and Baugh’s analysis was re-done by Dekeyser (1986) with a dif-
ferent methodology. By focusing on raw numbers of earliest attestations in the
OED and Middle English Dictionary (MED), Dekeyser argued, the methodology
used by Jespersen and Baugh created an artificial peak in the late fourteenth cen-
tury. Many of the first attestations were in the works of Chaucer, which must have
coloured the results. In order to bypass this problem, Dekeyser looked at relative
frequencies of English- and French-origin words in Middle English texts as well
as the MED. This slightly different analysis also sees the beginning of massive
French borrowing from c. 1200, and a decline, though not as steep as in Jes-
persen and Baugh, from the late fourteenth century. Dekeyser found that the ratio
of Romance to English vocabulary is higher in type frequency than it is in token
frequency. This means that individual French-origin words were, on average, used
less frequently than English-origin words. While Chaucer may have used a lot of
different French-origin words (types), they make up only 7% of the words (tokens)
in his works. This should not come as a surprise, knowing that up to half of the
1000 most frequent words in English are of Romance origin, but only four of the
100 most frequent words (Durkin 2018).
Historically it has been shown that the peak in French loans, c. 1250–1350,
coincides with a peak in French–English bilingualism (rather than monolingual
French and English speakers). Specifically, this was spoken bilingualism learned
interactionally in the classroom (Ingham 2014). The agents of change in the
massive borrowing of French lexicon therefore appear to have been a few very
specific generations of relatively balanced bilingual speakers, and not (as previ-
ously thought) an imposition by the new Norman overlords on a more or less
welcoming Anglophone population. In particular, the bilingual clergy, who were
educated in French but preached in their first language English, were instrumen-
tal in this lexical borrowing. An increased frequency of preaching and itinerant
preachers as a result of church reform in the early thirteenth century helped spread
lexical change both socially and geographically. It was particularly the less fre-
quent religious lexical items that were replaced by French words, while the more
frequent words from Old English tended to remain (Timofeeva 2018).
The bubonic plague pandemic in the mid-fourteenth century, the ‘Black Death’,
caused death disproportionally in the French-speaking upper classes, especially
among monasteries and secular clergy, who attended to those dying. As these
were the people who kept the acquisition of French going, the Black Death was
instrumental in the demise of Anglo-Norman. In this period, we see a rise of
particular types of errors in Anglo-Norman, which suggest that it was becoming a
taught rather than a naturalistically learned language. With the balance in French–
English bilinguals tilting towards English dominance, this can also explain the
sharp drop in the number of new French loan words in the late fourteenth century:
there were simply fewer bilingual brains to act as the locus of language change.
The contact between French and English, overall, was probably of medium
intensity, as a significant proportion of the population was in contact with speakers
from both languages. The French influence on the language was overwhelmingly
lexical, and although the deepest core of the lexicon remained unaffected, a lot
108 6 Language Contact

of basic vocabulary was borrowed. There is also some structural change due to
French influence, although this is mostly secondary to the lexical influence. Most
importantly, the voiced fricatives in English, which were previously allophones
of voiceless fricative phonemes, now became phonemes in their own right when
French loan words introduced minimal pairs. Contact with French may also have
been a catalyst for the loss of case and was probably behind a pragmatic change
in the use of personal pronouns (the replacement of singular thou by originally
plural you).

Research Highlight
FROM THOU AND YE TO YOU AND Y’ALL
French is known for making a pragmatic difference in second-person singu-
lar pronouns: there is tu for informal contexts, and vous (which matches the
plural form) for formal contexts. In fact, vous is used for situations of dis-
tance and height, also metaphorically. It can be used to signal social status,
respect, non-intimacy and/or estrangement (Wales, 1983, p. 110).
This tu/vous or T/V distinction was taken over in Middle English, where
the originally plural form ye began to be used in “vous contexts” in addi-
tion to the original thou in “tu contexts”. This can be seen from analyses of
pronoun use in the works of Thomas Malory (14th century; Wiśniewska-
Przymusińska, 2020) and William Shakespeare (late 16th and early 17th
centuries; Busse, 2002). Over time, though, forms of thou were used less
and less, and ye took over as the unmarked second-person (singular or plu-
ral) pronoun. At the same time, the original object form you started to appear
in subject position as well, giving us the Standard English system we use
today. (Non-standard varieties of English have introduced new plural forms
of you such as youse, y’all,yins and you guys.)
The loss of the original tu pronoun is cross-linguistically very rare. It
has traditionally been argued that societies in late medieval Europe favoured
the marking of respect over solidarity, which would explain the rise of you.
But why, then, did the original tu pronoun only disappear from English (and
Dutch), and not from a large number of other European languages as well?
Aalberse & Stoop (2015) suggest that this was because losing thou sim-
plified the inflectional system. The verb ending -st only occurred in the
second person singular, and cross-linguistic evidence shows that second-
person inflection is acquired relatively late. Only in English, Dutch and
German would dropping the second-person singular inflection result in a
loss of person contrast in singular verb inflections. What sets English and
Dutch apart from German (which still has its original pronoun du and -st
inflection) is that the late medieval and early modern periods were charac-
terised by a lot of migration into cities, which caused large-scale language
and dialect contact. High-contact situations are conducive to morphological
simplification, and moreover studies show that people would rather replace
Preliminary Conclusions 109

pronouns than make agreement errors. Aalberse & Stoop (2015) conclude,
therefore, that the disappearance of thou is a result of politeness strategies
coupled with inflectional simplification in acontact scenario.

Language Contact in Late Modern English

The focus on this chapter has been on four contact situations in Old and Middle
English, with the borrowing of Latin vocabulary in the Renaissance as the most
recent case of language contact discussed. Although this follows the typical dis-
cussion of language contact in textbooks on the history of English, it may give the
impression that no language contact of any importance took place in the history
of English after the Renaissance—mirroring an incorrect overall impression that
there has been little language change in the Late Modern English period (Beal
2012, pp. 12–15). But there has, of course, been quite significant language con-
tact in this period, if only because of the enormous colonialist endeavour that was
the British Empire. It may be exactly because of Empire that post-Renaissance
language contact receives so little attention. The dynamics of contact changed:
English as the colonial language is now the prestigious language, one with a
strong standard language ideology at that, and there has been an unwillingness
to acknowledge that English has been influenced by languages seen as inferior
(Dollinger 2019, pp. 199–200). Much of the influence of the languages of the
colonised is moreover restricted to specific varieties of English, and a discussion
would not fit a narrative of the inevitable emergence of Standard Southern British
English.
The influence of other languages on (Standard) English in the Late Modern
English period has overwhelmingly been lexical. This is what may be expected
from a low-intensity contact situation and recipient-language agentivity. But this
is not to say that particular varieties of English did not undergo contact-induced
change in areas such as phonology or syntax. Salmons and Purnell (2020),
for example, discuss phonological and syntactic properties of ‘Native American
English’ and ‘Jewish [American] English’ as being contact-induced. The processes
through which such colonial varieties of English arise are discussed further in the
next chapter.

Preliminary Conclusions

But first, we can draw some preliminary conclusions about language contact in
(earlier stages of) English. The intensity of contact and the types of social relations
between the speakers of these languages and English speakers varied considerably
for each of the contact situations discussed, sometimes even between different
periods of contact with the same language. As a consequence of who the agents
110 6 Language Contact

Table 6.3 Overview of contact situations in English

Contact language Intensity Material Agentivity
Latin Low Lexicon Recipient language
Celtic High Structure Source language
Old Norse High Both Both (?)
Norman French Medium Lexicon Recipient language
(Late Modern English) Low Lexicon Recipient language

of change were in these situations—whether they were English-dominant bilin-

guals using elements from their second language in English (recipient-language
agentivity) or were Other-dominant bilinguals using elements from their first lan-
guage in their second-language English (source language agentivity)—we saw
different outcomes, with recipient-language agentivity predominantly causing lex-
ical borrowing, and source language agentivity mainly causing the imposition of
grammatical structures. This is visualised schematically in Table 6.3.
Our theoretical understanding of contact linguistics is still growing, as is our
socio-historical knowledge of early English societies. This means that we must
continue to revise and critically assess established and novel theories about how
language contact changed English especially in pre-modern times. In the past two
decades or so, such reassessment has led to new insights into the possible influ-
ence of Celtic on English grammar, and to a clearer view of how French loanwords
entered the English lexicon. As exemplified by the Viking Hypothesis, new theo-
ries may sometimes be provocative or even iconoclastic, but even if a hypothesis
is unsupported, it can spur on new research and help us better understand how
language contact proceeded in early English.

References

Aalberse, Suzanne, and Wessel Stoop. 2015. The exceptional loss of the pronoun T. Journal of
Pragmatics 88: 190–201. https://doi.org/10.1016/j.pragma.2015.07.003.
Baugh, Albert C. 1935. The chronology of French loan-words in English. Modern Language Notes
50 (2): 90–93.
Baugh, Albert C., and Thomas Cable. 2013. A history of the English language, 6th ed. London:
Routledge.
Beal, Joan. 2012. Can’t see the wood for the trees? Corpora and the study of Late Modern English.
In Middle and Modern English corpus linguistics: A multi-dimensional approach, ed. Manfred
Markus, Yoko Iyeiri, Reinhard Heuberger, and Emil Chamson, 13–29. Amsterdam: Benjamins.
Britton, Derek. 1991. On Middle English she, sho: A Scots solution to an English problem. NOW-
ELE 17: 3–51. https://doi.org/10.1075/nowele.17.01bri.
Busse, Ulrich. 2002. Linguistic variation in the Shakespeare corpus: Morpho-syntactic variability
of second person pronouns. Amsterdam: Benjamins.
Campbell, Lyle, and Martha C. Muntzel. 1989. The structural consequences of language death. In
Investigating obsolescence: Studies in language contraction and death, ed. Nancy C. Dorian,
181–196. Cambridge: Cambridge University Press.
References 111

Cole, Marcelle. 2018. A native origin for Present-Day English they, their, them. Diachronica 35
(2): 165–209. https://doi.org/10.1075/dia.16026.col.
Dance, Richard. 2017. English contact: Norse. In The history of English. Volume 2: Old English,
ed. Laurel J. Brinton and Alexander Bergs, 202–219. De Gruyter. https://doi.org/10.1515/978
3110525304-011.
Dekeyser, Xavier. 1986. Romance loans in Middle English: A re-assessment. In Linguistics across
historical and geographical boundaries: In honour of Jacek Fisiak on the occasion of his fiftieth
birthday, vol. 1, ed. Dieter Kastovsky and Aleksander Szwedek, 253–265. Berlin: Mouton de
Gruyter.
Dollinger, Stefan. 2019. Creating Canadian English: The professor, the mountaineer, and a national
variety of English. Cambridge: Cambridge University Press. https://doi.org/10.1017/978110859
6862.
Durkin, Philip. 2018. Exploring the penetration of loanwords in the core vocabulary of Middle
English: Carry as a test case. English Language and Linguistics 22 (2): 265–282. https://doi.
org/10.1017/S1360674318000138.
Emonds, Joseph E., and Jan Terje Faarlund. 2014. English: The language of the Vikings. Olomouc:
Palacký University.
Evison, Martin Paul. 2000. All in the genes? Evaluating the biological evidence of contact and
migration. In Cultures in contact: Scandinavian settlement in England in the ninth and tenth
centuries, ed. Dawn M. Hadley and Julian D. Richards, 277–294. Turnhout: Brepols.
Faarlund, Jan Terje, and Joseph E. Emonds. 2016. English as North Germanic. Language Dynamics
and Change 6 (1): 1–17. https://doi.org/10.1163/22105832-00601002.
Farrar, Kimberly, and Mari C. Jones. 2002. Introduction. In Language contact: The interplay of
internal, external and extra-linguistic factors, ed. Mari C. Jones and Edith Esch, 1–16. Berlin,
New York: Mouton de Gruyter.
Fellows-Jensen, Gillian. 2011. Light thrown by Scandinavian place-names on the Anglo-Saxon
landscape. In Place-names, language and the Anglo-Saxon landscape, ed. Nicholas J. Higham
and Martin J. Ryan, 69–83. Woodbridge: Boydell.
Ferguson, Charles A. 1959. Diglossia. Word 15 (2): 325–340. https://doi.org/10.1080/00437956.
1959.11659702.
Gooskens, Charlotte, and Wilbert Heeringa. 2014. The role of dialect exposure in receptive multi-
lingualism. Applied Linguistics Review 5 (1): 247–271. https://doi.org/10.1515/applirev-2014-
0011.
Gooskens, Charlotte, Vincent J. van Heuven, Jelena Golubović, Anja Schüppert, Femke Swarte,
and Stefanie Voigt. 2017. Mutual intelligibility between closely related languages in Europe.
International Journal of Multilingualism 15 (2): 169–193. https://doi.org/10.1080/14790718.
2017.1350185.
Halsall, Guy. 2000. The Viking presence in England? The burial evidence reconsidered. In Cultures
in contact: Scandinavian settlement in England in the ninth and tenth centuries, ed. Dawn M.
Hadley and Julian D. Richards, 259–276. Turnhout: Brepols.
Härke, Heinrich. 2002. Kings and warriors: Population and landscape from post-Roman to Norman
Britain. In The peopling of Britain: The shaping of a human landscape, ed. Paul Slack and Ward
Ryk, 145–175. Oxford: Oxford University Press.
Hickey, Raymond. 2012. Early English and the Celtic hypothesis. In The Oxford handbook of
the history of English, ed. Terttu Nevalainen and Elizabeth Closs Traugott, 497–507. Oxford:
Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199922765.013.0042.
Hickey, Raymond. 2020. English and Celtic: Contact-induced change in history. Linguistics Van-
guard 6 (s2): 20190037. https://doi.org/10.1515/lingvan-2019-0037.
Ingham, Richard. 2014. The maintenance of French in later medieval England. Neuphilologische
Mitteilungen 115 (4): 425–448.
Kortmann, Bernd. 2016. The Viking Hypothesis from a dialectologist’s perspective. Language
Dynamics and Change 6 (1): 27–30. https://doi.org/10.1163/22105832-00601006.
Laing, Margaret, and Roger Lass. 2014. On Middle English she, sho: A refurbished narrative. Folia
Linguistica Historica 35: 201–240. https://doi.org/10.1515/flih.2014.006.
112 6 Language Contact

Los, Bettelou. 2016. Rule 1: Know your stuff. Language Dynamics and Change 6 (1): 31–33.
https://doi.org/10.1163/22105832-00601007.
Lutz, Angelika. 2009. Celtic influence on Old English and West Germanic. English Language and
Linguistics 13 (2): 227–249. https://doi.org/10.1017/S1360674309003001.
Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press.
McWhorter, John H. 2009. What else happened to English? A brief for the Celtic hypothesis.
English Language and Linguistics 13 (2): 163–191. https://doi.org/10.1017/S13606743090
02974.
Meakins, Felicity. 2013. Mixed languages. In Contact languages: A comprehensive guide, ed. Peter
Bakker and Yaron Matras, 159–228. Mouton de Gruyter. https://doi.org/10.1515/978161451
3711.159.
Meyerhoff, Miriam. 2009. Replication, transfer, and calquing: Using variation as a tool in the study
of language contact. Language Variation and Change 21 (3): 297–317. https://doi.org/10.1017/
S0954394509990196.
Nagy, Naomi, Nina Aghdasi, Derek Denis, and Alexandra Motut. 2011. Null subjects in heritage
languages: Contact effects in a cross-linguistic context. University of Pennsylvania Working
Papers in Linguistics 17: 135–144. https://repository.upenn.edu/pwpl/vol17/iss2/16.
Richards, Julian D. 2000. Identifying Anglo-Scandinavian settlements. In Cultures in contact:
Scandinavian settlement in England in the ninth and tenth centuries, ed. Dawn M. Hadley and
Julian D. Richards, 295–309. Turnhout: Brepols.
Salmons, Joseph C., and Thomas Purnell. 2020. Contact and the development of American English.
In The handbook of language contact, 2nd ed., ed. Raymond Hickey, 361–383. Hoboken, NJ:
Wiley-Blackwell.
Schendl, Herbert. 1996. Text types and code-switching in medieval and Early Modern English.
VIEWS: Vienna English Working Papers 5 (1–2): 50–62.
Schendl, Herbert. 1997. ‘To London fro Kent/Sunt predia depopulantes’: Code-switching and
medieval English macaronic poems. VIEWS: Vienna English Working Papers 6 (1): 52–66.
Schüppert, Anja, and Charlotte Gooskens. 2011. The role of extra-linguistic factors in recep-
tive bilingualism: Evidence from Danish and Swedish pre-schoolers. International Journal of
Bilingualism 16 (3): 332–347. https://doi.org/10.1177/1367006911426389.
Schüppert, Anja, Nanna Haug Hilton, and Charlotte Gooskens. 2015. Swedish is beautiful, Dan-
ish is ugly? Investigating the link between language attitudes and spoken word recognition.
Linguistics 53 (2): 375–403. https://doi.org/10.1515/ling-2015-0003.
Thomason, Sarah G. 2016. Middle English: English, not Norse. Language Dynamics and Change
6 (1): 42–45. https://doi.org/10.1163/22105832-00601010.
Thomason, Sarah G., and Terence Kaufman. 1991. Language contact, creolization, and genetic
linguistics. Berkeley and Oxford: University of California Press.
Timofeeva, Olga. 2018. Survival and loss of Old English religious vocabulary between 1150 and
1350. English Language and Linguistics 22 (2): 225–247. https://doi.org/10.1017/S13606743
18000114.
Timofeeva, Olga, and Richard Ingham. 2018. Special issue on mechanisms of French contact influ-
ence in Middle English: Diffusion and maintenance. English Language and Linguistics 22 (2):
197–205. https://doi.org/10.1017/S160674318000096.
Townend, Matthew. 2002. Language and history in Viking Age England: Linguistic relations
between speakers of Old Norse and Old English. Turnhout: Brepols.
van Coetsem, Frans. 1988. Loan phonology and the two transfer types in language contact.
Dordrecht: Foris.
van Coetsem, Frans. 2002. A general and unified theory of the transmission process in language
contact. Heidelberg: Winter.
Vennemann, Theo. 2009. Celtic influence in English? Yes and no. English Language and Linguis-
tics 13 (2): 309–334. https://doi.org/10.1017/S1360674309003049.
Wales, Kathleen M. 1983. Thou and you in Early Modern English: Brown and Gilman re-appraised.
Studia Linguistica 37 (2): 107–125. https://doi.org/10.1111/j.1467-9582.1983.tb00316.x.
References 113

Watts, Richard J. 2011. Language myths and the history of English. Oxford: Oxford University
Press.
Weinreich, Uriel. 1953. Languages in contact: Findings and problems. Den Haag: Mouton. https://
doi.org/10.1515/9783110802177.
Wiśniewska-Przymusińska, Malwina. 2020. T/V pronouns and FTAs in The Works of Sir Thomas
Malory: Medieval politeness and impoliteness in directives, expressives, and commissives.
Studia Anglica Posnaniensia 55 (1): 139–176. https://doi.org/10.2478/stap-2020-0006.
Wright, Laura. 1992. Macaronic writing in a London archive, 1380–1480. In History of Englishes:
New methods and interpretations in historical linguistics, ed. Matti Rissanen, Ossi Ihalainen,
Terttu Nevalainen, and Irma Taavitsainen, 762–770. Berlin, New York: Mouton de Gruyter.
https://doi.org/10.1515/9783110877007.762.
Dialect Contact
The Power of Accommodation
7

We will now focus on situations of contact between speakers of different dialects

of English, rather than contact with speakers of English with those of different lan-
guages. At the most basic level, any change that occurs due to dialect contact can
be traced back to processes of accommodation, but the effects of accommodation
depend on the social setting, such as the range of diversity among the dialects in
contact. Following the work of Peter Trudgill, a pioneer in this area of research, we
will look at dialect contact within English-speaking communities (Trudgill 1986),
but also at more extensive cases of contact known as ‘new-dialect formation’ that
were the result of contact between speakers of many different dialects, predomi-
nantly in (British) colonies in the nineteenth century (Trudgill 2004). We will also
look at other varieties that are linked to colonisation and the subsequent commod-
ification of English in hyperdiverse settings: English-based pidgins and creoles,
Multicultural London English, as well as English as a Lingua Franca. Some of
these developments may point towards the idea that a strict split between language
and dialect contact is theoretically untenable, and we may begin to see all instances
of contact-induced change as fundamentally the same process.

Accommodation

The main purpose of communication is to get meaning across. As a speaker, it

is important that you package your message as efficiently as possible: you want
to provide the maximum necessary specificity about the content of your message,
and you also want to maximise the chances that your interlocutor will understand
your message correctly. One of the ways of achieving this last goal is by tailoring
the language of your message to your interlocutor. This is what is usually known
as accommodation.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 115
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_7
116 7 Dialect Contact

Accommodation is known by many names, and it is a part of many different the-

ories, most famously Howard Giles’s ‘Communication Accommodation Theory’
and Allan Bell’s ‘audience design’ (Bell 2001). For our purposes, an appropriate
definition of the process is given by Coupland (1984, p. 49):

People will attempt to converge linguistically towards the speech patterns believed to be
characteristic of their recipients when they (i) desire their social approval and the per-
ceived costs of so acting are proportionally lower (identity maintenance function) than the
rewards anticipated; and/or (ii) desire a high level of communication efficiency (cognitive
organization function).

This definition focuses on linguistic convergence between the interlocutors—their

speech patterns become more similar—but, although convergence is probably the
most common outcome, it does also allow for other outcomes. If the desire for
social approval is small, then speakers may not converge; they may even diverge
if they have a negative social stance towards each other. Social solidarity can
also be indicated by matching your interlocutor in non-standardness, which may
objectively result in linguistic divergence and less efficient communication. And
of course, if there is no desire for efficient communication at all, accommodation
may also be absent.
In most instances, though, accommodation is understood to be equivalent to lin-
guistic convergence. A famous example is Coupland’s (1984) study of Sue, a travel
agent in Cardiff. Sue has an incentive to gain the social approval of her customers,
and to want communication to go smoothly: she wants to sell her customers a
holiday. For a range of Welsh accent features, Coupland was able to show that
Sue used them more with customers who had a stronger Welsh accent, and less
with customers who had a more SSBE-like accent. This was a tendency and not
a perfect match: Sue was likely accommodating to generalised social types rather
than individual interlocutors (the italicised ‘believed’ in the definition). Moreover,
the range of variation that an individual is capable of is always smaller than the
range that is present in the community at large (Bell 2001, p. 145), which limits
the extent of accommodation somewhat.
There are also studies that show that there may be individual, person-specific,
differences in how much people accommodate (Van Hofwegen, 2015), and that
people may accommodate to some accent features but not to others (Watt et al.
2010). With those caveats though, and although a lot more research is needed to
get to the bottom of this, it is believed that accommodation is a universal and
automatic process.

Research Highlight
WHO ACCOMODATES?
A study on accommodation in interaction between African-American chil-
dren (Van Hofwegen, 2015) showed some small differences between boys
Accommodation 117

and girls. Although all children in the study overwhelmingly showed accom-
modation to their interlocutor, boys typically accommodated less than girls.
When the interlocutor was someone they knew well, however, girls accom-
modated less, presumably because they need to work less hard to gain social
approval. With interlocutors they did not know well, boys started to accom-
modate more, but only on features that are associated with African American
English.
An experimental study with data collected at a music festival in the
Netherlands (Wieling et al., 2020) showed the opposite pattern: men became
more similar to their interlocutor than women in the pronunciation of the
vowel /a / after a short game where they had to repeat nonsense syllables.
▼
▲

Future work will have to investigate what the personal characteristics are
of typical accommodators and non-accommodators. Another open question
is whether those who tend to accommodate more are also the leaders in
language change, as they may be able to produce a broader range of variation
in a wider set of contexts than people who stick to their own way of speaking.

Accommodation and Dialect Contact

Much of the work on accommodation focuses on the timespan of one or at most

a few conversations, but we accommodate to other speakers all the time. If all
your interlocutors speak roughly the same variety as you do, then the necessary
accommodation will be minimal and possibly not noticeable as it falls within your
normal range of variation. If you end up in a more linguistically diverse situation,
however—and this need not be many different languages, it can just be different
dialects of the same language—continued accommodation may actually move your
range of variation, and in the process cause potentially lasting changes in your
speech.
We will exclude here clear cases of second-dialect acquisition, where a small
number of speakers move to a different area. Their new speech community speaks
differently from them, but is fairly homogeneous in itself. As such it presents a
clear target for acquisition, which may or may not be successful. Research on
Canadian children in the UK (Chambers 1992), a Canadian family in the UK
(Tagliamonte and Molfenter 2007), and Canadians in New York City (Nycz 2013a;
2013b; 2015) shows that there is a lot of inter-individual variability, and that suc-
cess rates may also differ depending on what features are investigated. Siegel
(2010) shows that this holds also for situations not involving Canadians.
In more linguistically diverse situations, we would expect a degree of mutual
convergence between all members of the speech community, subject to any indi-
vidual, social, or linguistic constraints that may exist on accommodation. It would
be practically and ethically impossible to set up an experimental setting to test
such situations, but there are some observable real-life situations of longer-term
118 7 Dialect Contact

dialect contact between people otherwise not in contact with the rest of the speech
community. Sonderegger et al. (2017) analysed the speech of contestants on a sea-
son of reality tv show Big Brother UK, which is built on the premise of locking
people into a house together, and Harrington et al. (2019) analysed repeated (self-)
recordings of members of the British Antarctic Survey during a five-month stay in
Antarctica. These studies show the same individual and feature-based differences
in the extent of accommodation, but both point at general convergence between
the speakers in the isolated speech community. The Antarctic study even pointed
at a shared innovation: all speakers had a more fronted pronunciation of /o*/ in
Antarctica compared to before.

Dialect Levelling and Supraregionalisation

The Big Brother UK and Antarctic studies are nice small-scale petri dish examples
of historical situations that we can no longer investigate in real time, such as the
development of colonial new dialects of English in the nineteenth century as a
result of large-scale migration of people with different dialect backgrounds. But
we begin our discussion of dialect contact phenomena with situations that involve
much less diversity: dialect levelling and supraregionalisation.
Since World War II, there has been immense social change. People’s spatial
mobility has vastly increased through the increased use of a denser public and
especially private transportation network. Where people live has also changed,
with people moving to cities (urbanisation) as well as from cities to the countryside
(counter-urbanisation) and with increased migration both within and across coun-
tries. A growing proportion of the population is university-educated and works
in the tertiary sector, which is less locally bound than traditional occupations.
This means that people have increasingly come in contact with speakers of other
dialects than their own (Britain 2010), contact with loose social network ties that
are characterised by accommodation and the potential for language change.
As a result of these developments, social dialectologists in the UK have
described what they call ‘regional dialect levelling’, also known as regiolectisation
or supralocalisation. This is a process by which highly localised dialect features
are lost and are replaced by features that are found over a wider geographic area
(Kerswill 2003). Ultimately, this means that dialects become more similar to each
other.
One of the prime sites for regional dialect levelling research is the north of
England, where it has been claimed that local dialects have been giving way to a
General Northern English. For example, apparent-time studies in Newcastle-upon-
Tyne and Durham show that the local variant of the face vowel [I∂] is on its way
out and is being replaced by a long monophthong [e ], which is found across the
▼
▲

north of England. In this particular case, Kerswill (2003) argues that [e ] is being
▼
▲

adopted through a process of levelling, because both variants were present in the
community already and it is the more local variant [I∂] that is being avoided. An
Dialect Levelling and Supraregionalisation 119

alternative explanation could be that [e ] is spreading from nearby larger towns in

▼
▲

Yorkshire through geographical diffusion.

The majority of our evidence for regional dialect levelling comes from studies
of individual features in individual communities. Regional dialect levelling is a
likely mechanism for the language changes found in these studies, but as evidence
for a process that takes place over a large geographical area, it is rather weak.
As Britain (2011, p. 48) argues, ‘A robust demonstration of it would require real
or apparent-time analysis of data from a number of locations all within the same
apparent dialect region, with all demonstrating convergence away from locally
restricted dialect forms and towards the common adoption of some feature already
enjoying a wide geographical currency’. Such large-scale studies would be practi-
cally very difficult to organise. Some further evidence for regional dialect levelling
in Northern English, however, comes from a machine learning study on dialect
classification (Strycharczuk et al. 2020). A computer model trained on labelling
speech samples as coming from one of five urban centres in Northern England
struggled to tell the difference between speakers from Manchester, Leeds, and
Sheffield. This suggests that these speakers could be classified as speakers of
General Northern English, with no further localisation possible—although differ-
ences between these three cities remained. Speakers from Liverpool and Newcastle
were consistently identified as coming from those cities, and in fact, these are
the more peripherally located cities in the region with strong local identities
(Kerswill 2003; Watson 2006). More, and more sociolinguistically informed, work
on regional dialect levelling is clearly needed to truly understand the processes
involved (Britain 2011).
Regional dialect levelling is a slow process, because although there has been
increased contact between communities, this contact is restricted to only part of
the population: the highly mobile, often highly educated professionals. For other
speakers in the community, the social changes mentioned above may apply to a
far lesser extent, and their speech is still influenced by the conservative pressures
of locally bound strong social network ties.
A setting where we can see increased dialect contact without such conserva-
tive pressures is the ‘new towns’ in the UK. These were planned new housing
developments at the location of existing small villages to solve the housing cri-
sis in post-World War II London, such as Milton Keynes (Kerswill & Williams
2000), or to house workers at industrial sites, such as Corby (Dyer 2002). Because
the new arrivals numerically swamped the existing population, there was no local
dialect to take into account, and the newcomers, who all came with their own
traditional dialects, were free to negotiate the ‘new dialect’ for these new towns
through accommodation. Kerswill and Williams (2000) show that with each gen-
eration in Milton Keynes, the range of variation decreases, and variants that
index a local identity from elsewhere are avoided in favour of majority or more
‘neutral’-sounding variants that are not linked to one specific location.
120 7 Dialect Contact

In other languages
TUSSENTAAL BETWEEN DIALECT AND STANDARD
Regional dialect levelling is obviously not limited to English-speaking
settings. It has also been a well-researched topic in Flanders, the Dutch-
speaking part of Belgium, where a levelled dialect has developed that is
called Tussentaal, ‘in-between language’: in between disappearing dialects
and a Netherlands-influenced Belgian Dutch standard language.
Tussentaal shows the characteristics of regional dialect levelling as highly
local features are avoided in favour of features that have a wider geograph-
ical currency. We also see the spread of features outside the area where
they were originally used, with particularly features from the Brabant dialect
area being taken up in other dialect areas (De Caluwe & Van Renterghem,
2011). In West-Flanders, where the traditional dialect enjoys greater vital-
ity and Tussentaal is viewed more negatively, Tussentaal has made limited
inroads, but even there supra-regional features can be found (Ghyselen & De
Vogelaer, 2013).
The future of Tussentaal as a new Flanders-internal standard is still uncer-
tain. It has widely replaced traditional dialect as people’s primary vernacular,
which means that ‘in-between language’ is no longer an accurate description
(De Caluwe, 2009). On the other hand, it remains difficult to define what
features exactly constitute Tussentaal. Analyses show multiple clusters of
features that are used in different contexts, and it has not yet focused into a
coherent variety (Ghyselen, 2015).

New-Dialect Formation in Colonial Englishes

Milton Keynes and Corby may have been new towns without a traditional dialect,
but they were still located in England, and the inhabitants of these new towns
would still have been in frequent contact with traditional dialect speakers from
surrounding towns and villages, as well as with speakers from their region of
origin which they could easily travel to. In order to see the formation of new
dialects in its purest form, we need to travel further away and look at the history
of colonial varieties of English.
Peter Trudgill has long been interested in the outcomes of dialect contact (e.g.,
Trudgill 1986). Based on his research, Trudgill (2004) sets out his theory of new-
dialect formation, also known as koinéisation. This theory focuses on the linguistic
(sub-) processes involved and sees the development of colonial varieties of English
as a deterministic development that can be predicted on the basis of the proportions
of speakers of different dialects in the colonial population. We will look at this
model in more detail below and also ask the question of whether the model is
purely deterministic or whether identity also plays a role (cf. Schneider 2003).
New-Dialect Formation in Colonial Englishes 121

Before we dive into Trudgill’s model, it is important to make a sociolinguistic

qualification. Although we often talk about ‘colonial new dialects’ as a relatively
coherent group of varieties with similar linguistic histories, we should, in fact,
recognise that there are two separate groups: ‘settler colonial Englishes’ on the one
hand, and ‘postcolonial Englishes’ on the other (Denis and D’Arcy 2018). Settler
colonial Englishes are the dialects of the areas where colonisation went hand in
hand with a mass migration of people from the British Isles, whose initial wide
range of dialectal diversity has made place for what are now clearly recognised
native-speaker varieties. Clear examples are Australian English, New Zealand
English, and Canadian English. Postcolonial Englishes are the varieties from (for-
mer) colonies that were not targeted for white settlement, such as Indian English
and Singapore English. Settlement in these latter colonies was limited to a ruling
managerial class, who assimilated a local elite to British customs and speech. The
main difference between postcolonial and settler colonial English, therefore, is that
postcolonial Englishes clearly show the influence of second-language varieties of
English (differentiated in Indian English as Marathi English, Hindustani English,
Kannada English, etc.; Denis and D’Arcy 2018, p. 21). Second-language varieties
play no role in settler colonial Englishes, because the indigenous population of
these colonies was at best swamped in numbers by white settlers, but more often
systematically killed.
Not all varieties of English are easily classified, however. For example, although
it is clear that Irish English is a colonial variety with a history of language and
dialect contact (Hickey 2007), there are differences within Irish English(es) as
well. While Northern Irish English is clearly a settler colonial variety, (Southern)
Irish English has many characteristics of a postcolonial variety after the invasion
of the English in the twelfth century (Corrigan 2022)—despite speakers of Irish
English also contributing to settler colonial Englishes in the eighteenth and nine-
teenth centuries. General (White) American English is also clearly a settler colonial
variety, but how to account for African American English in this model is much
less clear. And while South African English also has the characteristics of a set-
tler colonial variety, there are ethnically defined subvarieties as is typical for a
postcolonial variety, including that of later (forced as well as voluntary) migrants
such as Indian South African English (Mesthrie 2008); there is also the issue of
a parallel colonial variety of Dutch, Afrikaans, which influenced the development
of South African English(es). Going further back in time, we can see the original
development of Old English from the mutually intelligible dialects of the Anglish,
Saxon, and Jutish settlers also as a settler colonial variety (Schreier 2015, p. 620).
The contact between Old Norse and Old English has also been described in terms
of new-dialect formation (Fischer 2013, p. 33).

Linguistic Processes in New-Dialect Formation

Although Peter Trudgill’s work on dialect contact involves a great number of vari-
eties of English, as well as examples from many other languages, the empirical
122 7 Dialect Contact

part of his theory of new-dialect formation builds heavily on research into the
development of New Zealand English in particular (Trudgill 2004). This is no
coincidence: the University of Canterbury in Christchurch, New Zealand, has a
collection of audio recordings of (sometimes very) elderly speakers, recorded for
an oral history project by New Zealand radio in the 1940s. With European migra-
tion to New Zealand starting only after 1840, this means there is an apparent-time
audio record of almost the entire history of New Zealand English. Some of the
speakers in this Origins of New Zealand English (ONZE) project migrated to New
Zealand at a young age, while others were among the first Europeans to be born
in New Zealand. This proved an ideal testing ground for Trudgill’s deterministic
model.
Trudgill’s model of new-dialect formation consists of three stages (cited in sum-
mary from Meyerhoff 2006, p. 178 below). Each stage represents the speech from
successive generations.

Stage 1. Characterized by rudimentary levelling of variants towards an interdialect; takes

place among adult speakers. Interspeaker accommodation shapes the interdialect, and the
relative salience of variants, including social salience, may be relevant.
Stage 2. Characterized by extreme intra- and interspeaker variability; takes place among the
first generation of children; adult speakers are not relevant. This may result in quite idiosyn-
cratic (and ultimately unstable) original combinations of forms with the loss of linguistically
marked forms and the survival of unmarked ones. This may produce results that are similar
to levelling (Stage 1) but differs from levelling because linguistic processes are involved,
not face-to-face accommodation. [Trudgill calls this process apparent levelling.]
Stage 3. Characterized by survival of majority forms […]; takes place in the second gen-
eration of children; again adult speakers are not relevant. As in Stage 1, interspeaker
accommodation is the process by which the most common variants are focused.

Trudgill describes a number of linguistic processes that are relevant to new-dialect

formation:

● mixing
● levelling
● simplification (unmarking, interdialect development)
● reallocation
● focusing

Any form of dialect contact obviously starts with mixing. Speakers of different
dialects of a language come together, in proportions that ensure that no one dialect
becomes an obvious target for (second) dialect acquisition. In small-scale migra-
tion, such as when a family migrates to a different dialect area, these speakers
may acquire (features of) the dialect of that area but features from their original
dialect do not spread to their new neighbours (Chambers 1992; Tagliamonte and
Molfenter 2007). In order to create a new dialect, none of the original dialects in
the mixture should have an overwhelming majority of speakers.
New-Dialect Formation in Colonial Englishes 123

When speakers of different dialects accommodate to each other, they will avoid
using features that may impede communication and will choose variants that are
more widely shared in the speech community. This leads to levelling, i.e., every-
one’s speech becoming more similar as people avoid uncommon features. The
result of this is that demographically minority variant features disappear. Trudgill’s
view is that this is entirely a numbers game, decided by the proportions of speakers
from different dialects in the population. A famous example comes from the towns
of Odda and Tyssedal in Norway, two towns about 10 kilometres apart, each with
their own smelting works. In the early twentieth century, the two factories recruited
workers in different areas of Norway, so that in Odda there was a clear majority
of speakers of Western Norwegian dialects (81%), whereas the Tyssedal factory
had no such majority. The dialect of Odda contains many specifically Western
Norwegian features, and that of Tyssedal is much less clearly regionally coloured
(Kerswill 2002).
Another result of large-scale and long-term accommodation can be the sim-
plification of the new dialect. This can take two forms, the first of which is
unmarking. Unmarking means that features are chosen not because they are in
a numerical demographic majority, but because they are somehow linguistically
more natural. Trudgill (1986, p. 102) discusses examples from Fiji Hindi, a new
dialect of Hindi developed from dialect contact between Indian immigrants to Fiji,
where demographically minority features ended up in the new dialect, presumably
because they were shorter or did not contain nasal vowels. The longer and nasal
variants were apparently disfavoured and disappeared from the mix.
The second form of simplification is interdialect development, when speak-
ers settle on a compromise between features from different input dialects. Such
features are newly developed and cannot be traced back to any one of the input
dialects. An example here can be found in Early Modern Dutch, which is the result
of dialect contact in Amsterdam between migrants from Flanders, Brabant, the
Eastern Netherlands, and even Germany, as well as the local Hollandic-speaking
population. The sounds now spelled ei and ij in Dutch were pronounced identi-
cally in some of these dialects, but differently in others. Dialect contact resulted
in a merger between these two sounds, because that was a simpler system than
having two different sounds, with the merged sound pronounced as [EI]; this pro-
nunciation occurred for either ei or ij in some of the dialects, so in that sense, it
was geographically neutral. But in none of the input dialects were ei and ij both
pronounced as [EI]. This was a new development that cannot be traced back to a
single individual input dialect (Howell 2006).
If more than one variant survives after levelling, reallocation may happen,
where variants can be assigned different functions in the new dialect. These func-
tions can be social or allophonic. Trudgill’s (1974) study of Norwich suggests that
variants of the goose vowel from different areas of Norfolk have been reassigned
sociolinguistic functions as markers of social class in the city. Trudgill (1986,
p. 159) also believes that ‘Canadian Raising’, an allophonic rule for the price and
mouth vowels that gives us night time [n2It taIm] and out loud [2*t la*d], is the
result of a reallocation of raised Scottish and lowered English variants to contexts
124 7 Dialect Contact

before voiceless and voiced consonants, respectively. (This account is disputed by,
e.g., Moreton and Thomas 2007, who wrote that such a system is likely to arise
as a result of articulatory differences between these diphthongs before different
consonants.)
The final process, focusing, is crucial to new-dialect formation.1 Focusing is the
process by which the new variety acquires norms and stability. Speakers recognise
a particular set of features as belonging to the community, and a variety that is
considered to ‘belong’ to a community tends to be more stable. Focusing implies
levelling, but levelling does not need to imply focusing: less variation in a variety
does not necessarily mean stability. (We will look at some levelled but unfocused
varieties below.)

Testing Trudgill’s Model

Trudgill’s model of new-dialect formation is meant to be deterministic. Based on

any given input (a mix of dialect, and therefore a mix of features, in particular
proportions), it should be possible to predict what the outcome of new-dialect
formation would be. Social factors matter little to Trudgill in this scenario; he
even goes as far as to say that (national) identity is ‘irrelevant’ in new-dialect
formation (Trudgill 2008). We will return to the question of identity below.
Because Trudgill’s model is deterministic, it can be tested by means of mathe-
matical modelling. This was done by Baxter et al. (2009) for the development of
New Zealand English. They built four different models that simulated the interac-
tion between speakers of different varieties of English who would accommodate to
each other, with each model representing a different reason for why change might
occur. The model with the outcome that most resembled New Zealand English was
one that included social factors as well as simply the frequency of variants. This
would mean that the sociolinguistic associations of a feature also play a role in
new-dialect formation.

Identity in New-Dialect Formation

Trudgill’s model is (implicitly) aimed at new dialects in ‘white’ settler colonies,

where there is no previous English variety; in his theory, new-dialect formation
takes place in a so-called tabula rasa, a blank slate where the three-generational
process can take place undisturbed. It is, therefore, important that there is no fre-
quent (continued) contact with other varieties of English, as that would disturb the
process with constantly changing input from outside. However, Dollinger (2008)

1 In a narrow reading of terminology, focusing is what sets new-dialect formation apart from
koinéisation. Koinéisation then is an umbrella term for all the linguistic changes that take place
in dialect contact, but we need a more social process of focusing for all these changes to settle into
a new dialect.
New-Dialect Formation in Colonial Englishes 125

has successfully applied the model to early Canadian English with only a few
small modifications. This was a semi-tabula rasa context in which early settlers in
Ontario maintained contact with the newly formed United States.
Tabula rasa also means that, just as there are no other varieties of English that
matter, there is no other language that really plays a role either. This is not to say
that other languages are not compatible with the model: second-language varieties
of English, with local substrate influence, could easily be added to the mix (see
Millar 2008; 2010; Knooihuizen 2009, on the development of Shetland Scots, or
Blanc 1968 on new-dialect formation in Modern Hebrew with only L2 varieties).
Such non-tabula rasa situations typically appear to lead to postcolonial varieties
of English, where issues of identity are perhaps more important in the process.
Edgar W. Schneider (2003) has proposed a five-stage ‘Dynamic Model’ of the
development of postcolonial Englishes, where he looks at the role of both settler
and indigenous populations for different parameters: sociopolitical background,
identity construction, sociolinguistic conditions, and linguistic effects (Britain and
Matsumoto 2015, p. 307). The summary here is from Schneider (2011, p. 35):

1. Foundation. English is brought to a new territory, which leads to incipient bilingualism,

the borrowing of toponyms, and other minor processes.

2. Exonormative stabilization. During a stable colonial situation, the politically domi-

nant ‘mother country’ determines the norms of linguistic behaviour, and elite bilingualism
spreads amongst some representatives of the indigenous population, with lexical borrowing
continuing.

3. Nativization. The most vibrant and interesting of all the phases. With ties with the set-
tlers’ country of origin weakening, and interethnic contacts increasing, bilingual speakers
forge a new variety of English, shaped strongly by phonological and structural transfer—
though conservative speakers resent such innovative usage.

4. Endonormative stabilization. Implies that, after independence and inspired by the need
for nation-building, a new linguistic norm is increasingly recognised (and commonly per-
ceived as remarkably homogeneous), is beginning to be codified and to be accepted in
society, and is employed culturally in literary representations.

5. Differentiation. May follow in the end, i.e., in a stable young nation, internal social group
identities become more important and get reflected in the growth of dialectal differences.

In contrast to Trudgill, Schneider focuses not so much on the purely linguistic

processes, but more on the sociolinguistic situation in which new dialects develop.
Throughout the different stages, we see an increasing independence from the nor-
mative yoke of the original dialects and a shift of focus to the new community as
its own entity. In this process, there is a role for settlers as well as for the indige-
nous population, both as a source of loanwords (in the ‘Foundation’ stage) and
as a source of second-language influence on the new dialect (in the ‘Nativisation’
and ‘Endonormative stabilisation’ phases).
126 7 Dialect Contact

Schneider’s model explicitly applies to ‘non-white’ resource extraction

colonies, as it very clearly discusses different roles for the settler and the indige-
nous populations, and assumes that the indigenous population will shift to English,
or at least learn English as a second language. But it may apply equally to
white settler colonies, in the sense that, e.g., New Zealand English today is also
endonormative and does not look to the UK for its linguistic norms.
The issue of identity in new-dialect formation is still very much an open ques-
tion. Trudgill excludes it from the process, or at least from the initial three-stage
model. The fact that New Zealand English was described as ‘emerging endo-
normative’ (Hundt 1998, p. 140; emphasis mine) a full century and a half after
new-dialect formation was set in motion might suggest that identity is not the
important driver of the process that Schneider claims it to be. On the other hand,
developing (national) identities may play a more important role in the develop-
ment of local norms in postcolonial situations. Trudgill’s (2008) argument against
the role for identity is part of a special issue of the journal Language in Society,
where it is followed by short papers arguing in favour of identity, and a reply by
Trudgill. Schneider is one of the discussants.

Atypical Cases

The theory of new-dialect formation was developed and verified on the basis of
evidence mainly from fairly straightforward cases like New Zealand English. The
real test for the theories is to investigate whether they also hold for more atypical
cases, where some factor may disturb the regular development of the new dialect.
In this section, we briefly discuss three of such cases: the English varieties spoken
in the Falkland Islands, Palau, and Tristan da Cunha.

The Falkland Islands

The Falkland Islands are situated in the South Atlantic Ocean, close to the South
American mainland. Settlement, and therefore English speech, in the islands goes
back to 1833. Falkland Island English is interesting for two reasons: firstly, there
was no previous population in the islands, so no other language was involved in
the dialect mix, and secondly, the population history of the Falkland Islands is
atypical for a colony because it is mostly transient. The presentation of Falkland
Island English here comes from Sudbury (2004).
Despite the recency and the limited extent of settlement, the population history
is poorly documented. The main areas for nineteenth-century settlers seem to be
Scotland and the South West of England (Devon and Somerset). Many labourers
and civil servants came to the islands on short-term contracts and did not stay
longer than five years. Sudbury (2004, p. 404) notes that many Scottish settlers
came from the Highlands and Islands, so they might have spoken Scottish Gaelic
(too). She quotes a historical document stating that ‘the Scotch language was fairly
well represented’—but by this time ‘Scots’ is unlikely to refer to Scottish Gaelic,
New-Dialect Formation in Colonial Englishes 127

and anyway there is nothing in Falkland Island English that suggests Scottish
Gaelic influence.
Because of the high turnover of the relatively small population (c. 500 by 1860,
c. 2000 by 1900, and c. 3500 now), the founder principle—the idea that a rela-
tively small number of early settlers have a more lasting influence on the language
than larger numbers of later settlers, unless they are vastly outnumbered (Mufwene
2001, pp. 25–80)—does not hold, and more recent settlers are also relevant. Set-
tlers in the twentieth century also came from Scotland and the South West, but
later also from Northern Ireland (1950s) and from St. Helena, Chile, Australia,
and New Zealand (from the 1970s). Anecdotally, these settlers appear to have
influenced the speech of the locals.
The key characteristic of Falkland Island English is that it has not been focused.
Although the most marked regional features have been levelled out of the input
dialects, there is a lot of variation along different dimensions: between social
groups, between individuals within social groups, and even within individuals.
Sudbury (2004) lists a number of possible reasons for this. Firstly, there could be
linguistic reasons. Focusing can be expected to take longer if the dialects in contact
are more different. However, Australian and New Zealand English also have sub-
stantial settlement from Scotland and Southern England, and they have focused, so
this cannot be the (only) reason. Secondly, there are social reasons. There is rela-
tively little contact between speakers outside the capital, Stanley. Combined with
the transient nature of settlement, that restricts the forming of social networks in
which norms can develop. This lack of social networks and focusing has also been
described for Norwegian in Spitsbergen (Mæhlum 1996). In addition, small com-
munities tend to show a lot of inter- and intra-speaker variation without any social
meaning necessarily being attached to it (Dorian 1994). Finally, there are psycho-
logical reasons. There is no real separate Falkland Islands identity, and people are
happy just to be British. Especially after the Falklands War in 1982, the focus on
being British (i.e., not Argentinian) has become stronger and there is no real need
for a specific Falkland Islands identity.

Palau
The English of Palau, a Pacific island nation between the Philippines and Guam,
is interesting because English is a relatively recent introduction, from after World
War II. Another thing that makes Palauan English interesting is that this is one
of the few cases where we are looking at colonial American English rather than
British English. The presentation of Palauan English is taken from Britain and
Matsumoto (2015).
Palau was colonised relatively late, but saw different colonisers. It was a colony
of Spain (1885–1899), Germany (1899–1914), Japan (1914–1945), and finally the
United States (1945–1994) before becoming a (sort of) independent nation. The
colonisers had vastly different language policies. The Spanish only set up missions
in the island, leading to Spanish loanwords in Palauan in the religious domain. The
Germans were even less involved, but they did try to set up an education system
and eradicate Micronesian Pidgin English. Japan set up a settler colony in Palau,
128 7 Dialect Contact

building up to a four-to-one ratio of Japanese to Palauans in 1941. The Japanese

were integrated into Palauan society and intermarried with Palauans. Together with
education, this supported the diffusion of the Japanese language. The United States
captured Palau from Japan in World War II and continued to run it as a United
Nations trust after the war.
Formal colonisation is against US ideology, but ‘having no economic develop-
ment plan was, in fact, the intentional US policy, making the Palauan economy
totally dependent on the US’ (Britain and Matsumoto 2015, p. 314). Another
important incentive for the United States was to provide access for the military
close to East Asia.
There was no systematic Anglophone immigration. American military person-
nel were in Palau on short contracts and did not engage with the local population
much. From the 1960s onwards, military and Peace Corps volunteers were a bit
more involved. There were important demographic changes, however: all earlier
Japanese settlers (80% of the population) were expatriated shortly after the war,
and Filipino labour was imported from the 1980s onwards.
In terms of identity, the Americans knew they would not be staying in Palau.
Palauans understood that speaking English—or rather, collaborating with the
colonisers—would be beneficial. Increased education (including college education
in the United States) and increased contact with Americans meant that English was
not at odds with Palauan identity when Palau became a separate administrative unit
in 1974. But compared to American service personnel, Filipino contract workers
are much more important for the spread of English: they are in daily contact with
Palauans, also with the elderly, and with small children. A linguistic analysis of
Palauan English shows many features that are typical of postcolonial Englishes, as
well as many characteristics shared with Filipino English. As Palauan and Tagalog
are related languages, the question is whether this is Palauan substrate or Tagalog
(Filipino English) adstrate.
Britain and Matsumoto (2015, p. 339) claim that ‘Palauan English seems
currently to be at the second exonormative stabilisation phase in [Schneider’s]
Dynamic Model, but the potential for nativisation seems great given the continued
lack of American (English) input to everyday life in Palau’.

Tristan da Cunha
The last and possibly one of the most atypical cases of new-dialect formation is
that of Tristan da Cunha English. Tristan da Cunha is the world’s most remote
island, located in the middle of the South Atlantic, 2000 km from the nearest pop-
ulated place (St Helena, another remote island in the Mid-Atlantic), and 3000 km
from Cape Town. It was settled by about a dozen mostly British English speakers
beginning in 1817 and now has about 300 inhabitants. The demographic history
of Tristan da Cunha and its present-day variety of English have been described
very extensively, which makes it an excellent testing ground for Trudgill’s theory
(Schreier 2002; 2003; 2004; Schreier and Trudgill 2006).
The first settlers on Tristan da Cunha were all single men, with the excep-
tion of a Scot who brought his wife and daughter. Five women from St Helena
Dialect Contact in Hyperdiversity 129

arrived on the island in 1827, followed by (individual) settlers from the United
States, Denmark, and the Netherlands in the 1830s. Despite this great variety in
settler backgrounds, Schreier and Trudgill (2006) have traced back many features
of Tristan da Cunha English to individual settlers.
The first children were born on Tristan da Cunha around 1830; the 1830s, there-
fore, mark Stage 2 in Trudgill’s model, where the first generation of locally born
children starts acquiring language in a hyperdiverse situation. However, in the
mid-1850s, the death of the community leader led to two-thirds of the islanders
emigrating to South Africa. The timing of the mass emigration is significant: the
1850s should be when Stage 3, the focusing of the variety by the next generation
of children, occurs. Instead, we see a possibly major disruption in the process and
a re-balancing of the proportions of input varieties.
Another mass emigration followed in the 1880s, after most of the able-bodied
male population perished in a boating accident in 1885. This emigration was less
abrupt than the one in the mid-1850s, but still halved the population over a span
of six years. Once again we see disruption in the process and a change to the
(proportion of) input varieties in the mix at a very conspicuous moment: we are
now roughly a generation on from the first emigration event.
The development of Tristan da Cunha English has therefore not been as smooth
a process as we have seen for New Zealand English: the chain of intergenerational
transmission that is presupposed in Trudgill’s model was simply not continuous
in the first eighty years or so of the colony’s existence, and the incredibly small
scale of the community would suggest that individual speakers could have a more
significant influence on the dialect than others, e.g., if they had a more important
function in the community or if for some reason people spoke more to them than
to others. Despite these problems, though, Schreier and Trudgill (2006) argue that
the model is still applicable to Tristan da Cunha as well.2

Dialect Contact in Hyperdiversity

Theories of new-dialect formation and koinéisation are ubiquitous when we look

at dialect contact, or contact between mutually intelligible varieties, but as we
have already seen with the contact between Old English and Old Norse, it is not
always clear cut where the boundary lies between ‘mutually intelligible’ and ‘not
mutually intelligible’, and we know that in those borderline cases, the distinction
between what counts as a language or a dialect is a social construct. Along these
same fuzzy boundaries, there are also attempts to apply theories of koinéisation to
what we would intuitively see as cases of language contact: the development of
pidgins and creoles, of Multicultural London English, and of English as a Lingua

2 A final note about some of the data from Tristan da Cunha: this is a volcanic island. The volcano

erupted in the early 1960s, and the entire population was evacuated to England, where they stayed
in a refugee camp for three years before returning. Some of the available data consists of record-
ings made during this period, when they also had increased dialect contact with speakers of other
varieties of English, as well as access to mass-mediated standard English. There is a chance that
this has influenced our understanding of what the dialect was like.
130 7 Dialect Contact

Franca. As we shall see in the next section, this may lead us to conclude that all
contact is ultimately the same.

Koinéisation in Creole Languages

The first of these cases of hyperdiverse contact situations that we shall look at from
the point of view of koinéisation is creolisation. This is done with the explicit
caveat that English-lexifier creoles are not varieties of English, but rather lan-
guages in their own right. The inclusion of creolisation in a book on the history of
English may therefore be seen as an anomaly or even as a re-colonisation of these
varieties. Rather, this section serves to show how creole languages likely origi-
nated through processes of language and dialect contact, arguing against ‘creole
exceptionalism’, the problematic notion that creole languages are a special type
of language in which linguistic processes occur differently than in non-creole lan-
guages (see DeGraff 2003; 2004; 2005; 2020 for a critique). In this light, it is also
good to note that the concept of the founder principle was introduced into linguis-
tics by Salikoko S. Mufwene (2001) in his work on the development of creole
languages before it was included in Trudgill’s model of new-dialect formation.
Creoles are languages that developed in a situation of language contact between
speakers of two or more different languages. Creoles are often mentioned along
with pidgins, but they differ in that creoles are spoken as a native language whereas
pidgins are not, and creoles are used for all social functions, whereas pidgins are
restricted to only a few domains (Meyerhoff 2019, pp. 279–283). They usually
originate in situations where access to a language is severely restricted, so that
successful acquisition of the language is not possible. The prototypical situation
here is the Atlantic slave trade, where enslaved African people had little access
to the slave traders’ and plantation owners’ language, and where they also—by
design—had a large variety of first languages, further hampering communication.
Many different creole languages exist worldwide, with many different combi-
nations of superstrate language (the prestige language that provides the bulk of
the lexicon) and substrate languages (the non-prestigious languages). Despite this
variation, however, there are a lot of grammatical similarities between creoles. A
list is given in e.g., Markey (1982), but some highlights are the lack of grammat-
ical gender, invariant subject-verb-object word order, a general lack of inflection,
and the use of pre-verbal particles to indicate verb tense, mood, and aspect. None
of these features are exclusive to creoles, and creoles do not necessarily have all
features from the list, but if a language ticks (almost) all the boxes, it should be
taken as indicative of creole status.
Theories about the origins of creole languages must not only give a socio-
historical account of the development creoles, but must also be able to account
for why creoles tend to share this particular combination of features that is much
less common in non-creole languages. Early theories on the development of creole
languages claimed that they are so similar because they all originate from a single
ancestor, a Portuguese-based pidgin spoken along the West African coast in the
Dialect Contact in Hyperdiversity 131

fifteenth century. This pidgin would give the structure for all creole languages,
with the lexicon coming from various (European) superstrate languages.
Others have claimed that creoles are natural continuations either of the super-
strate (lexifier) language, or of the substrate language(s). These would involve a
lot of language change resulting in simplification due to high-intensity and high-
diversity contact. The similarity between creole languages would be a result of
the similarities between the different substrate languages, perhaps in addition to
universal strategies in language learning.
A final theory is based on Universal Grammar, the innate language learning
device from Noam Chomsky’s theory of syntax. Because the community input for
children learning a creole language at its earliest stages is extremely variable—
everyone speaks the (superstrate or lexifier) language in their own way—the input
does not give enough evidence for how the language works. Child learners would
then settle for the ‘default’ settings. This explains the similarities between cre-
ole languages, and, so the theory holds, can give more information about how
Universal Grammar works.
An elegant compromise position was taken by Lefebvre (2004). It seems clear
that contact and language learning play a role, both in the acquisition of the super-
strate language based on very little input, the acquisition of the creole based on
however much of the superstrate was acquired by others in the community, and the
levelling of second-language varieties into a more or less focused language. More
concretely, in the case of an English-based creole, every individual would do their
best to learn English from what they heard around them. This was not much, and
we are dealing with an adult second-language learner, who therefore is doubly dis-
advantaged; there will therefore most likely be a lot of imposition of first-language
structures (from lots of different first languages!) and a lot of simplification as a
result of learner strategies. The community ‘language’ (maybe ‘medium of com-
munication’ is a better term) then becomes relatively uniform in terms of lexicon,
but diverse in terms of structure—a mutual intelligibility situation that lends itself
to koinéisation.

Multicultural London English

A phenomenon that forms a middle ground between new-dialect formation and

creolisation is the formation of urban multiethnolects. From the 1960s onwards,
there was a migration to European countries, initially from former colonies, but
later also from elsewhere. This has led to demographic changes in urban centres,
which in turn has led to the formation of new varieties of language. In London,
this is Multicultural London English, or MLE (e.g., Cheshire et al. 2008), which
developed in East London boroughs previously the domain of Cockney; similar
processes have taken place in other major European cities.
MLE is an urban youth style that is characterised by a range of features; not
all speakers use all features, however, so there is discussion about whether MLE
constitutes a coherent ‘variety’. Characteristic features of MLE are, among many
others, a monophthongal face and goat, the backing of /k/ to [q], the lack of
allomorphy for a and the (the forms /∂n/ and /ði/ are not used before vowels),
132 7 Dialect Contact

the development of a new generic and first-person pronoun man (Cheshire 2013),
and the quotative construction this is X. In addition, MLE participates in ongoing
UK-wide changes such as th-fronting and goose-fronting (Cheshire et al. 2011).
MLE is used by speakers of all ethnicities, including White British, and it is not
the speaker’s own ethnicity but the ethnic diversity of their friendship network that
constrains the density of MLE features in their speech (Cheshire et al. 2008).
The origins of MLE are thought to lie in ‘group second-language acquisition’
(Cheshire et al. 2011, p. 153): speakers acquire English from their peers, who
themselves may not have English as their (only) home language either. Kerswill
and Torgersen (2020, p. 257) explicitly compare this to creolisation, with the main
difference being that in the case of urban multiethnolects, learners have much more
exposure to the language and access to appropriate models to learn from than in
the case of creole languages. This makes the process much more similar to new-
dialect formation, although with predominantly second-language varieties as the
input. As immigration rates remain high, the process remains ongoing as more
new features are added to the feature pool and become available for acquisition.
In a situation of new-dialect formation, it should be possible to trace back fea-
tures to original input dialects: the second-language varieties of subsequent groups
of immigrants from the Caribbean (from the 1960s), South Asia (from the 1970s),
and sub-Saharan Africa (from the 1980s), first-language varieties of English that
speakers may have had access to, as well as interdialect forms and new inno-
vations. Kerswill and Torgersen (2020) have attempted this with some success
and have also found some light evidence for a founder effect. Caribbean features,
from the initial group of ‘settlers’, dominate in MLE, but very much not exclu-
sively so, and despite the early migrants from the Caribbean assimilating to the
traditional Cockney dialect so that they were hardly distinguishable from other
Cockney speakers.

English as a Lingua Franca

The language contact situation that English finds itself in most frequently recently
is in its role as a lingua franca. The global spread of English, especially in recent
decades, is a complicated story of push and pull factors that we will not go into fur-
ther here. But it is being used as a language of wider communication by speakers
with different native languages has also changed what English is like. Contact-
induced change in the English of non-native speakers may eventually influence
first-language English as well.
The English that is spoken in international contexts may or may not adhere
closely to the norms of anyone’s standard native-speaker English variety of choice.
But whether this matters is another question. A study by van den Doel (2006)
found that native speakers of English do not actually seem to care much about non-
native speakers’ use of the language: they are much more forgiving than non-native
speakers themselves are (p. 217). English native speakers’ apparent indifference
towards ‘errors’ in the English of non-native speakers may simply be a result of
Dialect Contact in Hyperdiversity 133

increased exposure to non-native English. More than any other language, English
can be said to have a specific register for non-native speech, one that we may call
English as a Lingua Franca.
English as a Lingua Franca (ELF) is defined as ‘any use of English among
speakers of different first languages for whom English is the communicative
medium of choice, and often the only option’ (Seidlhofer 2011, p. 7). Note that
this definition is purely functional, and that it does not make any claims about
what ELF looks like. Note also that the definition includes interactions that may
involve native speakers of English, although in practice we tend not to see their
English as instantiations of ELF.
While a functional definition of ELF is relatively easy to give and is relatively
uncontroversial, it is less clear what ELF is in terms of form. Descriptions often
take the form of recommendations for easy communication between non-native
speakers (e.g., Jenkins 2002). But rather than trying to decide what ELF should
be, it is more useful to look at what people actually do when they speak English
in ELF contexts. Seidlhofer (2011) gives some examples of features that are quite
common in ELF conversation, but are in breach of the rules of Standard English
grammar, such as lack of plural agreement on demonstratives, lack of third-person
-s on verbs, non-idiomatic use of prepositions, as well as the use of innova-
tive word formation: conspirate, examinate, financiate, pronunciate, increasement,
increasive, approvement, etc. These are not real words in Standard English, and
will not normally be used by native speakers of the language, but they have been
created on the fly by ELF speakers who show at least some awareness of English
morphological processes. These speakers are actually fairly competent users of
English, because they have distilled abstract patterns from the language that they
are able to apply analogically to new situations.
A question that has recently started to be discussed is whether ELF can be com-
pared to new-dialect formation. If ELF usage differs from native-speaker usage,
and if this does not cause controversy, this may be a reason to believe that ELF
is undergoing new-dialect formation into Euro-English, or into specific non-Anglo
national Englishes. Edwards (2016) believes that this may indeed be the case for
the Netherlands. There is evidence of Dutch English (or ‘English in the Nether-
lands’, which is a slightly different concept) not sticking to the norms of Standard
English. This can be seen, e.g., in the use of verb tenses, the lack of -ly marking
on adverbs, the use of existential constructions, the use of grammatical gender,
etc. This could just be ‘poor English’, but anecdotally, translators and proofread-
ers have been overruled by Dutch English-speaking authors who insist that their
formulation is ‘better’, at least in that particular context and for that particular
audience of other Dutch English speakers.
We can try and link such examples to Schneider’s (2003) five-stage model of
norms in new-dialect formation: foundation, exonormative stabilisation, nativisa-
tion, endonormative stabilisation, and differentiation. The simple fact that Dutch
authors of English-language texts refuse to accept corrections by native-speaker
proofreaders already suggests that they have moved past a stage of exonormative
stabilisation. Native speakers are no longer seen as the ultimate authority on the
134 7 Dialect Contact

norms of the language. The insistence on forms that can be very clearly linked
to L1 Dutch interference takes this further in the direction of nativisation, or per-
haps even the beginnings of endonormative stabilisation. While this may work at
a national level, at a larger European level, however, there is very little evidence
as of yet for the emergence of endonormative, pan-European, non-native English
(van den Doel & Quené, 2013).

Are Language and Dialect Contact the Same Thing?

Given the similarities between Lefebvre’s theory of the development of creole lan-
guages, Cheshire et al.’s account of the origins of Multicultural London English,
and Trudgill’s model of new-dialect formation, as well as the applicability of
Schneider’s model of norms in colonial dialect formation to instances of English
as a Lingua Franca—in short, the similarities between models of dialect contact
and (apparent) language contact—it becomes fair to ask whether we must main-
tain the strict distinction between language and dialect contact that has been so
engrained in historical linguistic theory. The difficulty of deciding which of these
two categories the Old English–Old Norse contact falls under is another reason to
question the dichotomy.
The American-Australian linguist Jeff Siegel, who has a background in both
creole languages and second-dialect acquisition, clearly believes that differences
between contact varieties are a matter of degree, not principle:

The broad linguistic differences among contact varieties are due to the nature of the pool
of variants that undergoes mixing and levelling. In the case of koines, the pool is com-
prised mainly of already existing varieties of a language. But in other varieties, such as
pidgins and creoles, the pool contains a large proportion of newly modified versions of a
language resulting from transfer and restructuring in second-language learning and from
model simplification. The degree of modification depends on both linguistic factors, such
as the typological distance between the languages involved, and social factors, such as the
degree of access to the language being modified. Thus, with regard to input into the pool of
variants, there is a continuum ranging from very little restructuring to drastic restructuring.
This continuum is reflected in the contact varieties which emerge after levelling, ranging
from koines at one end and pidgins and creoles at the other. (Siegel 2001, p. 193)

Siegel’s idea is visualised in Fig. 7.1. In this view, all new varieties emerge by
levelling out the variation that exists in a community when people with different
source varieties mix. What ‘type’ of contact variety results from the levelling is a
function of the degree of restructuring that has happened already at the stage of the
source varieties as a result of (adult) second-language acquisition. Colonial new
dialects are based overwhelmingly on the speech of first-language speakers, with
hardly any adult second-language acquisition, while the input for pidgins and cre-
oles is almost exclusively that of second-language speakers who have had to learn
the language under extremely disadvantageous circumstances. It is also important
to note that this is a continuum: types of contact varieties are not clearly separable
References 135

Fig. 7.1 Restructuring along a spectrum of contact varieties (Adapted from Siegel 2001, p. 193)

categories. We have earlier seen that a variety such as South African English, with
many first-language speakers of English but also many second-language speak-
ers in the original population mix, is likely somewhere in between settler colonial
variety (or ‘koiné’, in Siegel’s terms) and postcolonial variety (or ‘indigenised new
dialect’).3 Siegel’s model further distinguishes a number of other points along the
continuum, which are less relevant to the history of English and which we shall
therefore not discuss here.
The similarities between language contact and dialect contact that are apparent
from Siegel’s model most clearly pertain to the mixing and levelling stages of
new-dialect formation (or rather new-variety formation). But ‘restructuring and
transfer’, too, applies to both language contact and dialect contact. Many of the
issues that second-language learners experience are the same as those that second-
dialect learners experience, such as the acquisition of new phonemic contrasts
(e.g., Nycz 2013b). It would appear, then, that regardless of the range of varieties
in the mix, and regardless of whether these varieties are ideologically constructed
as ‘languages’ or ‘dialects’, at the very core of it contact is just contact.

References

Baxter, Gareth J., Richard A. Blythe, William Croft, and Alan J. McKane. 2009. Modeling lan-
guage change: An evaluation of Trudgill’s theory of the emergence of New Zealand English.
Language Variation and Change 21 (2): 257–296. https://doi.org/10.1017/S09543945099
9010X.
Bell, Allan. 2001. Back in style: Reworking audience design. In Style and sociolinguistic variation,
ed. Penelope Eckert and John R. Rickford, 139–169. Cambridge: Cambridge University Press.

3 As discussed before as well, there is also variation within South African English, a lot of which
is ethnically based. Second-language varieties of English, spoken by people with many different
first languages, continue to co-exist with a range of South African English varieties spoken as a
first language, for which the second-language varieties served as input. All these varieties inhabit
different points on Siegel’s continuum.
136 7 Dialect Contact

Blanc, Haim. 1968. The Israeli koine as an emergent national standard. In Language problems of
developing nations, ed. Joshua A. Fishman, Charles A. Ferguson and Jyotirindra Das Gupta,
237–251. New York: Wiley.
Britain, David. 2010. Supralocal regional dialect levelling. In Language and identities, ed. Carmen
Llamas and Dominic Watt, 193–204. Edinburgh: Edinburgh University Press.
Britain, David. 2011. The heterogeneous homogenisation of dialects in England. Taal & Tongval
63 (1): 43–60. https://doi.org/10.5117/TET2011.1.BRIT.
Britain, David, and Kazuko Matsumoto. 2015. Palauan English. In Further studies in the lesser-
known varieties of English, ed. Jeffrey P. Williams, Edgar W. Schneider, Peter Trudgill, and
Daniel Schreier, 305–343. Cambridge: Cambridge University Press. https://doi.org/10.1017/
CBO9781139108652.014.
Chambers, J.K. 1992. Dialect acquisition. Language 68 (4): 673–705. https://doi.org/10.2307/
416850.
Cheshire, Jenny. 2013. Grammaticalisation in social context: The emergence of a new English
pronoun. Journal of Sociolinguistics 17 (5): 608–633. https://doi.org/10.1111/josl.12053.
Cheshire, Jenny, Paul Kerswill, Sue Fox, and Eivind Torgersen. 2011. Contact, the feature pool and
the speech community: The emergence of Multicultural London English. Journal of Sociolin-
guistics 15 (2): 151–196. https://doi.org/10.1111/j.1467-9841.2011.00478.x.
Cheshire, Jenny, Sue Fox, Paul Kerswill, and Eivind Torgersen. 2008. Ethnicity, friendship net-
work and social practices as the motor of dialect change: Linguistic innovation in London.
Sociolinguistica 22: 1–23. https://doi.org/10.1515/9783484605299.1.
Corrigan, Karen P. 2022. Language, migration and identity on the island of Ireland through time
and space. Paper presented at Methods in Dialectology XVII, Mainz, August 1.
Coupland, Nikolas. 1984. Accommodation at work: Some phonological data and their implica-
tions. International Journal of the Sociology of Language 46: 49–70. https://doi.org/10.1515/
ijsl.1984.46.49.
De Caluwe, Johan. 2009. Tussentaal wordt omgangstaal in Vlaanderen. Nederlandse Taalkunde 14
(1): 8–25.https://doi.org/10.5117/NEDTAA2009.1.TUSS339.
De Caluwe, Johan, and Evelien Van Renterghem. 2011. Regiolectisering en de opkomst van tussen-
taal in Vlaanderen. Taal & Tongval 63 (1): 61–77.https://doi.org/10.5117/TET2011.1.DECA.
DeGraff, Michel. 2003. Against creole exceptionalism. Language 79 (2): 391–410. https://doi.org/
10.1353/lan.2003.0114.
DeGraff, Michel. 2004. Against creole exceptionalism (redux). Language 80 (4): 834–839. https://
doi.org/10.1353/lan.2004.0178.
DeGraff, Michel. 2005. Linguists’ most dangerous myth: The fallacy of creole exceptionalism.
Language in Society 34 (4): 533–591. https://doi.org/10.1017/S0047404505050207.
DeGraff, Michel. 2020. Toward racial justice in linguistics: The case of Creole studies (Response
to Charity Hudley et al.). Language 96 (4): e292–e306. https://doi.org/10.1353/lan.2020.0080.
Denis, Derek, and Alexandra D’Arcy. 2018. Settler colonial Englishes are distinct from postcolo-
nial Englishes. American Speech 93 (1): 3–31. https://doi.org/10.1215/00031283-6904065.
Dollinger, Stefan. 2008. New-dialect formation in Canada: Evidence from the English modal
auxiliaries. Amsterdam: Benjamins. https://doi.org/10.1075/slcs.97.
Dorian, Nancy C. 1994. Varieties of variation in a very small place: Social homogeneity, prestige
norms, and linguistic variation. Language 70 (4): 631–696. https://doi.org/10.2307/416324.
Dyer, Judy. 2002. “We all speak the same round here”: Dialect levelling in a Scottish-English com-
munity. Journal of Sociolinguistics 6 (1): 99–116. https://doi.org/10.1111/1467-9481.00179.
Edwards, Alison. 2016. English in the Netherlands: Functions, forms and attitudes. Amsterdam:
Benjamins.
Fischer, Olga. 2013. Contact and syntactic change in Old and Middle English. In English as a
contact language, ed. Daniel Schreier and Marianne Hundt, 18–40. Cambridge: Cambridge
University Press.
Ghyselen, Anne-Sophie. 2015. ‘Stabilisering’ van tussentaal? Het taalrepertorium in de Westhoek
als casus. Taal & Tongval 67 (1): 43–95. https://doi.org/10.5117/TET2015.1.GHYS.
References 137

Ghyselen, Anne-Sophie, and Gunther De Vogelaer. 2013. The impact of dialect loss on the
acceptance of tussentaal: The special case of West-Flanders in Belgium. In Language
(de)standardisation in Late Modern Europe: Experimental studies, ed. Tore Kristiansen and
Stefan Grondelaers, 153–170. Oslo: Novus.
Harrington, Jonathan, Michele Gubian, Mary Stevens, and Florian Schiel. 2019. Phonetic change
in an Antarctic winter. Journal of the Acoustical Society of America 146 (5): 3327–3332. https://
doi.org/10.1121/1.5130709.
Hickey, Raymond. 2007. Irish English: History and present-day forms. Cambridge: Cambridge
University Press.
Howell, Robert B. 2006. Immigration and koineisation: The formation of Early Modern Dutch
urban vernaculars. Transactions of the Philological Society 104 (2): 207–227. https://doi.org/
10.1111/j.1467-968X.2006.00169.x.
Hundt, Marianne. 1998. New Zealand English grammar: Fact or fiction? A corpus-based study in
morphosyntactic variation. Amsterdam: Benjamins. https://doi.org/10.1075/veaw.g23.
Jenkins, Jennifer. 2002. A sociolinguistically based, empirically researched pronunciation syllabus
for English as an International Language. Applied Linguistics 23 (1): 83–103. https://doi.org/
10.1093/applin/23.1.83.
Kerswill, Paul. 2002. Koineization and accommodation. In The handbook of language variation
and change, ed. J. K. Chambers, Peter Trudgill, and N. Schilling-Estes, 669–702. Oxford:
Blackwell.
Kerswill, Paul. 2003. Dialect levelling and geographical diffusion in British English. In Social
dialectology: In honour of Peter Trudgill, ed. David Britain and Jenny Cheshire, 223–243.
Amsterdam: Benjamins.
Kerswill, Paul, and Ann Williams. 2000. Creating a new town koine: Children and language change
in Milton Keynes. Language in Society 29 (1): 65–115. https://doi.org/10.1017/S00474045000
01020.
Kerswill, Paul, and Eivind Torgersen. 2020. Tracing the origins of an urban youth vernacular:
Founder effects, frequency, and culture in the emergence of Multicultural London English. In
Advancing socio-grammatical variation and change: In honour of Jenny Cheshire, ed. Karen
V. Beaman, Isabelle Buchstaller, Sue Fox, and James A. Walker, 249–276. London: Routledge.
https://doi.org/10.4324/9780429282720-17.
Knooihuizen, Remco. 2009. Shetland Scots as a new dialect: Phonetic and phonological consider-
ations. English Language and Linguistics 13 (3): 483–501. https://doi.org/10.1017/S13606743
09990207.
Lefebvre, Claire. 2004. Issues in the study of pidgin and creole languages. Amsterdam: Benjamins.
Mæhlum, Brit. 1996. Semi-migration in the Arctic: A theoretical perspective on the dialect strate-
gies of children on Spitsbergen. In Language contact across the North Atlantic, ed. P. Sture
Ureland and Iain Clarkson, 313–331. Tübingen: Niemeyer.
Markey, Thomas L. 1982. Afrikaans: Creole or non-creole? Zeitschrift für Dialektologie und
Linguistik 49: 169–207.
Mesthrie, Rajend (ed.). 2008. Varieties of English 4: Africa, South and Southeast Asia. Berlin:
Mouton de Gruyter. https://doi.org/10.1515/9783110208429.
Meyerhoff, Miriam. 2006. Linguistic change, sociohistorical context, and theory-building in vari-
ationist linguistics: New-dialect formation in New Zealand. English Language and Linguistics
10 (1): 173–194. https://doi.org/10.1017/S1360674306001833.
Meyerhoff, Miriam. 2019. Introducing sociolinguistics, 3rd ed. London: Routledge.
Millar, Robert McColl. 2008. The origins and development of Shetland dialect in light of dialect
contact theories. English World-Wide 29 (3): 237–267. https://doi.org/10.1075/eww.29.3.02mil.
Millar, Robert McColl. 2010. The death of Orkney Norn and the genesis of Orkney Scots. Scottish
Language 29: 16–36.
Moreton, Elliott, and Erik R. Thomas. 2007. Origins of Canadian raising in voiceless-coda effects:
A case study in phonologization. In Laboratory phonology 9, ed. Jennifer Cole and Jose I.
Hualde, 37–64. Berlin: Mouton de Gruyter.
138 7 Dialect Contact

Mufwene, Salikoko S. 2001. The ecology of language evolution. Cambridge: Cambridge University
Press.
Nycz, Jennifer. 2013a. Changing words or changing rules? Second dialect acquisition and phono-
logical representation. Journal of Pragmatics 52: 49–62. https://doi.org/10.1016/j.pragma.
2012.12.014.
Nycz, Jennifer. 2013b. New contrast acquisition: Methodological issues and theoretical implica-
tions. English Language and Linguistics 17 (2): 325–357. https://doi.org/10.1017/S13606743
13000051.
Nycz, Jennifer. 2015. Second dialect acquisition: A sociophonetic perspective. Language and
Linguistics Compass 9 (11): 469–482. https://doi.org/10.1111/lnc3.12163.
Schneider, Edgar W. 2003. The dynamics of new Englishes: From identity construction to dialect
birth. Language 79 (2): 233–281.
Schneider, Edgar W. 2011. English around the world: An introduction. Cambridge: Cambridge
University Press.
Schreier, Daniel. 2002. Terra incognita in the anglophone world: Tristan da Cunha, South Atlantic
Ocean. English World-Wide 23 (1): 1–29. https://doi.org/10.1075/eww.23.1.02sch.
Schreier, Daniel. 2003. Isolation and language change: Contemporary and sociohistorical evidence
from Tristan da Cunha English. Basingstoke, New York: Palgrave Macmillan.
Schreier, Daniel. 2004. English transported to the South Atlantic Ocean: Tristan da Cunha. In Lega-
cies of colonial English, ed. Raymond Hickey, 387–401. Cambridge: Cambridge University
Press.
Schreier, Daniel, and Peter Trudgill. 2006. The segmental phonology of nineteenth-century Tristan
da Cunha English: Convergence and local innovation. English Language and Linguistics 10 (1):
119–141. https://doi.org/10.1017/S136067430600181X.
Schreier, Daniel. 2015. Historical phonology and koinéization. In The Oxford handbook of his-
torical phonology, ed. Patrick Honeybone and Joseph Salmons, 619–636. Oxford: Oxford
University Press. https://doi.org/10.1093/oxfordhb/9780199232819.013.037.
Seidlhofer, Barbara. 2011. Understanding English as a Lingua Franca. Oxford: Oxford University
Press.
Siegel, Jeff. 2001. Koine formation and creole genesis. In Creolization and contact, ed. Norval
Smith and Tonjes Veenstra, 175–197. Amsterdam: Benjamins.
Siegel, Jeff. 2010. Second dialect acquisition. Cambridge: Cambridge University Press.
Sonderegger, Morgan, Max Bane, and Peter Graff. 2017. The medium-term dynamics of accents
on reality television. Language 93 (3): 598–640. https://doi.org/10.1353/lan.2017.0038.
Strycharczuk, Patrycja, Manuel López-Ibáñez, Georgina Brown, and Adrian Leemann. 2020. Gen-
eral Northern English: Exploring regional variation in the north of England with machine
learning. Frontiers in Artificial Intelligence 3: 48. https://doi.org/10.3389/frai.2020.00048.
Sudbury, Angela. 2004. English on the Falklands. In Legacies of colonial English, ed. Raymond
Hickey, 402–417. Cambridge: Cambridge University Press.
Tagliamonte, Sali A., and Sonja Molfenter. 2007. How’d you get that accent? Acquiring a second
dialect of the same language. Language in Society 36 (5): 649–675. https://doi.org/10.1017/S00
47404507070911.
Trudgill, Peter. 1974. The social differentiation of English in Norwich. Cambridge: Cambridge
University Press.
Trudgill, Peter. 1986. Dialects in contact. Oxford: Blackwell.
Trudgill, Peter. 2004. New-dialect formation: The inevitability of colonial Englishes. Edinburgh:
Edinburgh University Press.
Trudgill, Peter. 2008. Colonial dialect contact in the history of European languages: On the irrele-
vance of identity to new-dialect formation. Language in Society 37 (2): 241–254.
van den Doel, Rias. 2006. How friendly are the natives? An evaluation of native-speaker judgements
of foreign-accented British and American English. Utrecht: LOT Publications.
van den Doel, Rias, and Hugo Quené. 2013. The endonormative standards of European English:
Emerging or elusive? English World-Wide 34 (1): 77–97. https://doi.org/10.1075/eww.34.1.
04van.
References 139

Van Hofwegen, Janneke. 2015. Dyadic analysis: Factors affecting African American English usage
and accommodation in adolescent peer dyads. Language & Communication 41: 28–45.https://
doi.org/10.1016/j.langcom.2014.10.004.
Watson, Kevin. 2006. Phonological resistance and innovation in the North-West of England.
English Today 22 (2): 55–61. https://doi.org/10.1017/S0266078406002100.
Watt, Dominic, Carmen Llamas, and Daniel Ezra Johnson. 2010. Levels of linguistic accommoda-
tion across a national border. Journal of English Linguistics 38 (3): 270–289. https://doi.org/
10.1177/0075424210373039.
Wieling, Martijn, Mark Tiede, Teja Rebernik, Lisanne de Jong, Anouck Braggaar, Martijn
Bartelds, Masha Medvedeva, Penny Heisterkamp, Tom Freire Offrede, Hedwig Sekeres, Anna
Pot, Mara van der Ploeg, Karin Volkers, and Gregory Mills. 2020. A novel paradigm to inves-
tigate phonetic convergence in interaction. Proceedings of the 12th International Seminar on
Speech Production: 1–4.
 Part IV
Structural Change
Analogy
Patterns and Morphological Change
8

Introduction

A lot of theory of language change is concerned with sound change or lexical

change. In this chapter, we will look at a different part of the language: mor-
phology. The study of morphological change is difficult for a number of reasons.
Firstly, unlike phonology and syntax, where there are (often multiple, competing,
and contradictory) overarching theories, there is no established theory of morphol-
ogy from which we can build a theory of morphological change. Secondly, more
than for example phonology and syntax, it is difficult to consider morphology as
just morphology. It interacts with phonology: the vowel alternation in plural forms
like foot ~ feet began purely as allophony, but through a series of sound changes
now carries the full weight of the plural distinction (Bybee 2015 calls this ‘mor-
phologisation’). Similarly, morphology interacts with syntax: when words occur
in the same construction regularly, they can also turn into morphological markers.
This is ‘grammaticalisation’, and we will talk about that more in the next chapter.
For now, we will stick with changes in morphology that we can explain just by
looking at morphology.

The Wug Test

In a well-known linguistic experiment, Berko (1958) asked children to provide

plural forms of made-up words. Children were given a drawing of a fictional ani-
mal and were told the name of the animal, e.g., ‘This is a wug’. Then they were
given a drawing with two of these animals: ‘Now there are two of them. There
are two…’ The name of the blue bird-like creature from the first stimulus is now
often more generally applied to experiments with nonce words: wug tests.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 143
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_8
144 8 Analogy

Berko’s experiment aimed to find at what age children acquire morphological

rules in English, and if some rules were acquired earlier than others. The chil-
dren who participated in her experiment were between 4.5 and 7 years old and
were asked to provide plural forms of nonce nouns ending in different conso-
nants. This way, Berko could test first of all whether children could apply the
pluralisation rule in English with an {-s} morpheme. The younger children strug-
gled with this sometimes, but the older children could all form plural forms of
words they had never heard before. But additionally, Berko could test whether the
children knew which allomorph to use: wug becomes wugs /w2lz/, but bik
becomes biks /bIks/ and niz becomes nizzes /nIz∂z/. Especially this final allomorph
proved to be difficult for the children.
Berko framed her study in terms of the children’s acquisition of morphological
rules in English. And indeed, plural formation with the allomorphs of {-s} is so
frequent that we could say it constitutes a rule. But imagine what happens if we
do a wug test with slightly more advanced users of English, and the stimulus
is a nonce word like talimus. A large part of respondents would probably offer
talimuses as the plural form, but a non-negligible part would answer talimi. That
plural form does not follow the normal pluralisation rule, but instead seems to be
modelled on a pattern where singular -us gives plural -i in words that look like
they may be of Latin origin. Speakers recognise this pattern from other words
( fungus, cactus, alumnus, radius, etc.) and extend it to a novel word. This is a
case of analogy.

Analogy

Proportional Analogy

There are two main types of analogy. The first of these is proportional analogy.
This is when we extend an existing pattern to a different word that the pattern did
not apply to before. This could be a nonce word as in a wug test: we know that
the plural of fungus is fungi; therefore, the plural of talimus must be talimi. But
we could also extend the pattern to existing words: based on fungus ~ fungi we can
decide that the plural of octopus is octopi. When we do this, we are making an
analogical change: octopus is not actually a word of Latin origin, and the form
octopi did not exist before we made it.1
Proportional analogy is also behind hypercorrection in dialectal accom-
modation, as when speakers of Northern English put on a Southern
English accent and say butcher /b2t∫∂/ for /b*t∫∂/, or understand /2nd∂stA nd/ ▼
▲

for /2nd∂stænd/. After all, they are aware that Southern English speakers say
duck /d2k/ where they themselves say /d*k/, and class /klA s/ where Northerners
▼
▲

´ pous was oktō´ podes. If you

1 Octopus is ultimately of Ancient Greek origin, and the plural of oktō

want to be a purist or a pedant, the form octopodes /ˌAk tAp∂di z/ would be the original plural
▼
▲

adapted to English phonology.

Analogy 145

say /klæs/. Similarly, it is behind overgeneralisations in child language, such as

sing ~ sang ~ sung → bring ~ brang ~ brung. Such overgeneralisations by language
learners generally do not stick, but when they do, we have language change.
Proportional analogy is the analogical extension of a paradigm to items it did
not apply to previously. Usually, this is the extension of ‘regular’ patterns at the
expense of ‘irregular’ patterns. We can see this throughout the morphological
system of English. For example, the regular plural morpheme {-s} has spread
to book, brother, and egg, replacing the original ‘irregular’ plurals *beech (OE
bēċ), brethren, and eyren. Similarly, comparatives and superlatives of adjectives
are regularly formed with the suffixes {-er} and {-est}. These are now applied
to old and late forming older ~ oldest, later ~ latest instead of the original forms
elder ~ eldest, latter ~ last. (Some of the original forms still exist with specialised
meanings, which we will return to briefly below.) And the past tense and past
participle of wash are now washed, rather than *woosh (OE wōsc) and *washen
(OE wascen). Analogical extension of ‘irregular’ patterns does occur, however,
especially in the verbal inflection system, and we will look at that in more detail
below.

Paradigmatic Levelling

The other type of analogy is paradigmatic levelling. Where proportional analogy

applies outside the original lexeme, paradigmatic levelling is the extension (usually
regularisation) of a pattern within the lexeme. Consider the different forms of
the verb ċēosan ‘to choose’ in Old English in Table 8.1: some of these started
with /t∫/, others with /k/. The stem-final consonant also differed between /s/, /z/,
and /r/. This apparent chaos is the result of completely regular sound change from
the Proto-Germanic forms, but it is not immediately obvious that ċēas and curon
are actually forms of the same word anymore. In modern English, the form of the
stem has been made consistent across the entire paradigm, analogically modelled
after one particular form (in this case a present-tense form). Something similar has
happened in modern Dutch and German, but with slightly different results.
Paradigmatic levelling is sometimes difficult to spot because it applies by defi-
nition to individual words. But there are cases of paradigmatic levelling also today.

Table 8.1 Paradigms for the verb ‘to choose’

Old English Modern English German Dutch
Present ċēosan t∫–z choose t∫–z küren k–r kiezen k–z
Past SG. ċēas t∫–s chose t∫–z kor k–r koos k–s
Past PL. curon k–r chose t∫–z koren k–r kozen k–z
Past PPL. coren k–r chosen t∫–z gekoren k–r gekozen k–z
Note Because voiced obstruents are devoiced in Dutch, the past singular form koos is underlyingly
k-z, so this paradigm is regular as well
146 8 Analogy

In varieties of English in and around London, the goat vowel (usually [2*])
has a different allophone before the ‘dark l’ [ë] allophone of /l/, where we get
[A*] instead. This means that whole is pronounced as [hA*ë], but holy as [h2*li].
According to this regular allophonic alternation, whole + -ly = wholly should also
be pronounced as [h2*li], because the appropriate syllable-initial allophone of /l/
here is ‘clear l’ [l]. But to maintain the link with the base word whole, speakers
pronounce wholly with [A*] as well. Paradigmatic levelling has thus created a new
minimal pair wholly [hA*li] /= holy [h2*li] and therefore even a new phonemic
opposition (Wells 1982, pp. 312–313).

Sound Change and Morphological Change

The example of ċēosan shows the interplay between sound change and morpho-
logical change. A completely regular process of sound change is responsible for
creating a highly irregular verb paradigm, which is tidied up by the analogical pro-
cess of paradigmatic levelling. This process applied fairly randomly, or at least, we
cannot set up a generalised rule for why paradigmatic levelling applied in this case.
There are a number of such examples in other languages as well, which Sturtevant
(1947, p. 109) has attempted to capture in a generalised rule (here in the words of
Hock 2003, p. 450):

Phonetic laws are regular but produce irregularities. Analogic creation is irregular but pro-
duces regularity.

This generalisation is known as ‘Sturtevant’s Paradox’. The paradox seems attrac-

tive, but upon further inspection lacks nuance. For one, the premises may not be
entirely accurate. We have already seen that sound change is not always regular,
for example, in the case of lexical diffusion. And as we shall see below in the
discussion of strong and weak verbs in English, analogy does not always apply
irregularly and definitely does not always cause regularity. Another issue is that
the sound change causing an irregularity and the analogical change ‘correcting’ it
may lack a clear relationship: the sound changes underlying the pattern in ċēosan
date back to Proto-Indo-European, but the analogical levelling took place in Mid-
dle English, multiple millennia later (Hock 2021, p. 196). These changes may
formally be related, but seem much less so than in the example of wholly–holy.
Sturtevant’s Paradox only applies to paradigmatic levelling and does not take
into account other forms of analogy. A change in plural formation such as
cactus ~ cacti → cactuses, which happens in response not to sound change but
to borrowing, also serves to increase morphological regularity. Similarly, sound
change plays no role in analogical re-analysis of ME pese /pe z/ ~ pesen /pe z∂n/
▼
▲
▼
▲

→ EModE pea ~ peas, where the /z/ in the original singular was reinterpreted as a
plural ending, again increasing morphological regularity.
Discussion of Sturtevant’s Paradox has raised questions about how exactly
sound change and morphological change go together, some even claiming that
Analogy and the English Verb System 147

sound change in lexical diffusion is itself an example of analogy (Hock 2003,

p. 452). Integrative accounts such as Ritt’s evolutionary approach (see section
“Language Change as Evolution” in Chapter 2) may be better able to explain
such sporadic changes as they simultaneously take into account phonological,
morphological, syntactic, and extra-linguistic pressures on competing forms.

Analogy and the English Verb System

Strong and Weak Verbs in Old English

Let us now turn to an example where analogical change is clearly visible: the
English verb system. English has two categories of verbs, based on how the past
tense and the past participle are formed. The largest group of verbs make both
these forms with an {-ed} suffix, e.g., walk ~ walked ~ walked. These verbs are
called ‘weak verbs’. There are also just over 100 ‘strong verbs’ in English; these
do not have a dental suffix, but instead (usually) employ vowel alternation, and
(usually) have a nasal suffix {-en} in the past participle. An example of a strong
verb is write ~ wrote ~ written.2
Both weak and strong verbs are typical innovations for Germanic languages;
they did not exist in Proto-Indo-European and did not develop in other Indo-
European language groups. The exact origin of weak verbs is controversial, but
a common explanation is that they come from a Proto-Germanic construction with
auxiliary do: roughly walk + did → walked (Hill 2010). The process by which this
happened, grammaticalisation, is something we will discuss in the next chapter.
Strong verbs are also typically Germanic, but the vowel alternation patterns they
are based on, also known as Ablaut, are known from Proto-Indo-European and
other Indo-European languages as well. In grammars of modern English, strong
verbs are often called ‘irregular’, but this is a misnomer. In fact, there was a lot of
regularity to strong verbs. Based on the vowel alternations in the verb stem, verbs
(mostly) fit into one of seven patterns called ‘classes’.
Table 8.2 shows the conjugations of Old English verbs from each of these
seven classes. The vowel alternations in these classes formed clear patterns that the
strong verbs in Old English adhered to with very few deviations. Where possible,
examples in the table have been chosen that preserve the original alternations in
present-day English (with regular sound change), but in many cases, there have
been additional changes so that the current verb inflections are no longer a faithful
reflection of the patterns in Old English.
From Old English to present-day English, the strong verb system has broken
down considerably: strong verbs have become weak, or they have changed their

2 There is another set of verbs in English that have both a dental suffix and a vowel alternation, e.g.,

keep ~ kept ~ kept. These verbs were weak in Old English (cēpan ~ cēpte ~ cēpt), but regular sound
change, in this case, pre-cluster shortening, caused a vowel alternation that was carried through to
present-day English (see Chapter 5).
148 8 Analogy

Table 8.2 Strong verb classes in Old English

Class Infinitive Past SG. Past PL. Past part.
1 wrı̄tan wrāt writon writen ‘write’
2 frēosan frēas fruron froren ‘freeze’
3a drincan dranc druncon druncen ‘drink’
3b helpan healp hulpon holpen ‘help’
4 stelan stæl stǣlon stolen ‘steal’
5 wefan wæf wǣfon wefen ‘weave’
6 scacan scōc scōcon scacen ‘shake’
7 healdan hēold hēoldon healden ‘hold’

vowel alternation patterns in different ways so that second-language learners of

English now need to memorise different patterns for individual verbs. Although
this development is not unique to English, the extent of change in English is much
greater than in most other Germanic languages. And for all of these changes,
analogy is the best explanation for what happened.

Analogy Causes… Irregularity?

As the vast majority of English verb is inflected according to the weak pat-
tern, we may assume that that weak pattern with {-ed} endings constitutes
the ‘default rule’ in English. Loanwords and neologisms overwhelmingly are
inflected according to the weak pattern: it is twerk ~ twerked ~ twerked, and not
twerk ~ twark ~ tworken (class 3b). Sturtevant’s Paradox, which states we should
expect analogy to apply irregularly but cause regularity, then means we would
expect strong verbs to occasionally move over to weak inflection. And this
does happen occasionally: the Old English verb rēocan ~ rēac ~ rocen ‘to smell’
(class 2) became reek ~ reeked ~ reeked, and the verb wasċan ~ wōsċ ~ wasċen (class
6) became wash ~ washed ~ washed. Some other Old English strong verbs have
become weak in standard English, but remain strong in some dialects, e.g.,
gnagan ~ gnōg ~ gnagen (class 6) → gnaw ~ gnew ~ gnawn → gnawed.
But there are many examples of the opposite happening. Some loanwords were
incorporated in English with strong inflection: strive, from Old French estriver,
is traditionally inflected according to the class 1 pattern: strove ~ striven. (The
weak form strived is a newer innovation.) Another French loanword, catch (Anglo-
Norman cachier), followed phonologically similar verbs such as teach and latch
in having a past tense and participle caught rather than the ‘rule’-based catched.3
In addition to loanwords, we also have originally weak verbs switching to a strong

3Teach, and latch before it regularised, were not in fact strong verbs in Old English, but are another
example of weak verbs where sound change introduced a vowel and consonant alternation. They
Analogy and the English Verb System 149

inflection: for many speakers of English, the past tense of dive is now dove, rather
than the original dived. Corpus data shows that this change started in American
English at the turn of the twentieth century, but really took off after 1980. It spread
to British English only after 2000, probably as a result of dialect contact rather than
independent innovation. It is also interesting to note here that only the past tense
form is innovated: no one has a strong past participle diven. (Yet.)
However, what has had a more devastating impact on the strong verb system
in English is a series of individual analogical changes in individual verbs—
analogical change that applied irregularly, but that did not cause regularity.
Take the verb slide, for example. The Old English slı̄dan ~ slād ~ sliden (class 1)
should have given present-day English slide ~ *slode ~ *slidden, but instead, we
have slide ~ slid ~ slid. What happened? One explanation for the change of the
vowel in the past tense by analogical extension from the participle (paradig-
matic levelling). The loss of the {-en} ending from the participle may be
simple sound change (the loss of unstressed endings that also made the infini-
tive ending {-an} disappear), or it could be proportional analogy with other
verbs where this ending was lost.4 We can make exactly the same argument for
sċēotan ~ sċēat ~ sċoten (class 2) → shoot ~ *sheet ~ *shotten → shoot ~ shot ~ shot,
for winnan ~ wan ~ wunnen (class 2) → win ~ *wan ~ *wunnen → win ~ won ~ won,5
and several others. We do often see that the past tense form adopts the vowel from
the participle, but this is not a hard and fast rule.6 The end result in modern English
is that the previously regular inflection classes have become a much more diverse
series of patterns. Some of these, in particular the old class 1, still apply to a fair
number of verbs, but others apply to only one or two verbs. The high number of
such idiosyncratic patterns makes that the strong verb system in English is now
widely regarded as ‘irregular’.

are therefore typically included in overviews of irregular verbs in English, and analogical change
from the default {-ed} to this new pattern is counter to expectations.
4 Impressionistically, the {-en} ending seems to be lost most frequently in verbs with a stem ending

in an alveolar stop (put, hit) or a nasal (swim, run). It may well be that it was felt that these final
consonants already signalled a participle, given that participles in English tend to end in an alveolar
stop or a nasal.
5 Do not be distracted by the spelling <o> in the past tense and participle. These are pronounced

/w2n/, which is a completely regular development from Old English short /u/. The spelling with
<o> was introduced by Norman French scribes to minimise chances of confusion in a sequence of
strokes like <uuunnen>.
6 One could posit an alternative purely phonological explanations for past tense forms like slid and

shot. Like we saw for cēpan earlier, these past tense forms could have been regularised to *slı̄d-de
and *sċēot-te at an early stage, after which pre-cluster shortening and the loss of unstressed vowels
could lead to the forms slid and shot. However, the OED does not give these forms at such an early
stage and instead gives evidence for the expected forms slode (slade in the north of England and
Scotland) and sheet in the Middle and Early Modern English periods.
150 8 Analogy

Research Highlight
NON-NATIVE SPEAKERS USE ANALOGY
The ‘wug test’ method is very powerful in its simplicity and has therefore
been re-used often. Cuskley et al. (2015) used it to investigate the verb sys-
tem in native and non-native speakers of English. They asked participants to
form the past tense of non-existing verbs and looked at whether they applied
the regular ending {-ed} or did something irregular.
Surprisingly, non-native speakers gave almost twice as many irregular
answers (41%) as native speakers (22%). The pattern held regardless of
whether the non-existing verb was more similar to an existing regular verb
or an existing irregular verb, although both native and non-native speakers
alike gave more irregular answers for the second category. Non-native speak-
ers particularly used vowel changes to form the past tense, or made past tense
forms that were identical to the present tense (as in cut).
An interesting finding was that among the non-native speaker participants,
those who gave more irregular answers had a lower self-reported proficiency
in English than those who gave more regular answers. The authors suggest
that this is because English teaching emphasises the learning of irregular
forms. The more proficient someone is in English, the less they have to
worry about lists of exceptions, and therefore they apply the regular ending
{-ed} more frequently. This once again suggests that native and non-native
speakers are not in a dichotomous opposition, but rather that there is a cline
in how accessible English-language material is to the individual speaker.

Generalisations About Analogy

Even though analogy applies irregularly, we can formulate tendencies of how anal-
ogy will apply when it does. There are two sets of tendencies stated by Polish
linguists, Jerzy Kuryłowicz and Witold Mańczak, which sometimes contradict each
other (McMahon 1994, pp. 76–80); a selection of the most important ones is given
below. There are also broader tendencies that take into account word frequency.

Kuryłowicz’s and Mańczak’s Generalisations

A summary of Kuryłowicz’s ‘laws’ of analogy is as follows:

1. Complex marking replaces simple marking.

• Old High German had gast ~ gesti ‘guest(s)’, with i-umlaut. This results in
Modern German Gast ~ Gäste. But in OHG boum ~ bouma ‘tree(s)’, with-
out i-umlaut, it has analogically been introduced, giving Modern German
Baum ~ Bäume with a doubly marked plural: both the umlaut and the ending.
Generalisations About Analogy 151

2. Analogical change proceeds from a basic, simple form to a derived form.

• English teach ~ taught has analogically led to catch ~ caught, based on sim-
ilarity between the (basic) present tense and/or infinitive forms. It does not
work the other way: the similarity between the past tenses brought and
sought does not lead to analogical change seek → *sing to be more similar
to bring.
3. When a form undergoes analogy, the new form takes over the primary function.
The old form may remain for secondary (specialised, marked) functions.
• The plural of brother ~ brethren was analogically changed to brothers. The
new form is now the basic plural, but the old form still exists in religious
contexts. Similarly, OE hūswı̄f regularly developed to hussy; the (analogi-
cally) reconstructed form housewife now has the basic meaning, and hussy
has a very specific meaning.
4. Contrasts of marginal significance will be sacrificed to strengthen more basic
distinctions.
• The Old English case system had different plural forms for the different
genders and cases. The analogical extension of plural {-s} across the board
goes at the expense of gender and case distinctions, but strengthens the more
basic plural meaning. It was more important to have a clear marking of the
plural than it was to maintain grammatical gender and case.
Mańczak’s ‘tendencies’ sometimes contradict these:

1. Longer words are more frequently remade on the model of shorter words than
vice versa.
2. Root alternation is more often abolished than introduced.
• The OE words bōc ‘book’, āc ‘oak’ and gāt ‘goat’ had plurals with i-umlaut
that would roughly be *beech, *eech and *geet in present-day English. Ana-
logical extension of plural {-s} means that the singular roots also remain in
the plural. Note that Kuryłowicz’s Baum ~ Bäume is a counterexample.
3. Longer inflectional forms are more often remade on the model of shorter forms
than vice versa.
4. Zero endings are more frequently replaced by full ones than vice versa.
• Many neuter words in OE had a zero plural ending: þæt word ~ þā word
‘word(s)’, þæt þing ~ þā þing ‘thing(s)’, þæt ġear ~ þā ġear ‘year(s)’ (here
with definite articles to show the singular/plural distinction). In Modern
English, these words now all have plural {-s}.
5. Monosyllabic endings are more frequently replaced by polysyllabic ones than
vice versa.
6. The forms of the indicative more frequently bring about the remaking of other
moods than vice versa.
7. The forms of the present more frequently bring about the remaking of other
tenses than vice versa.
152 8 Analogy

In general, Kuryłowicz’s laws apply mostly to proportional analogy, while

Mańczak’s tendencies work better for paradigmatic levelling. They do disagree
at points, but there are two main points that they do agree on: first, there are basic
forms that serve as a model for analogy, and second, grammatical marking should
be as clear as possible. The differences between Kuryłowicz and Mańczak mostly
hinge on different generalisations about what makes forms ‘basic’ or ‘clear’.

Frequency Effects

Apart from the more structural constraints on the application of analogy that
Kuryłowicz and Mańczak tried to capture, it is generally thought that also fre-
quency influences the outcome of analogical change. In order to pick this apart,
we need to distinguish between different kinds of frequency (Strik 2015, p. 22;
based on Dammel 2011, p. 57):

• type frequency is the number of lexical items that inflect according to a partic-
ular pattern. The overview on Wiktionary shows 150 Old English strong verbs
in class 1, but only 61 in class 4. The type frequency for class 1 is therefore
more than twice as high as the type frequency for Class 4.
• token frequency, which comes in a few flavours:
– lexical frequency is the number of occurrences of all forms from a particular
paradigm. It is safe to assume that the lexical item fall (including all its
forms: falls, falling, fell, fallen) has a higher lexical frequency than plummet
(including plummets, plummeting and plummeted).
– form frequency is the number of occurrences of one particular form in a
particular paradigm, so we would look at only the occurrences of falls or
falling, etc.
– categorical frequency is the number of occurrences of a particular gram-
matical form, so the occurrence of all past tense singular forms, or all past
participle forms, regardless of lexical item.
• type-token frequency is the total number of tokens in a given type. Here we
would be looking at how often you encounter a verb that inflects according to a
particular pattern. It might well be that even if there were fewer verbs that fall
into class 4 in Old English, these verbs had a much higher lexical frequency
than the many verbs from class 1, so overall your chances of encountering a
form from class 4 were higher than those of encountering a form from class 1.7

Forms with high token frequency appear to be relatively resistant to analogical

change. Many of the strong verbs have relatively high token frequencies, for exam-
ple. But for non-regularising analogical change (i.e., weak verbs that become

7 This is not the case here, though. Class 1 verbs were, and are, very common.
Generalisations About Analogy 153

strong) high token frequency may actually facilitate change, as the innovative
forms occur more often and therefore have a higher chance of sticking.
Type and token frequency seem to interact in which patterns are most likely to
spread by analogy. Bybee (1985, p. 133) discusses some data from French from
the 1920s (!) where children analogically extended verb inflection mostly from the
first of three conjugation classes in French (e.g., chanter ~ chantais ~ chanté ‘sing’),
not from the other two classes. She linked this to type and token frequency from
the same children’s own conversational data. The third conjugation class (e.g.,
vendre ~ vendais ~ vendu ‘sell’) has the highest type-token frequency of the three
classes with almost 60% of uses, but it has a relatively low type frequency: under
20% of verbs use this pattern. For a speaker of French, it is probably more efficient
to store the forms of third-class verbs as separate units in their brain rather than
assemble them from a base form and a conjugation every time they want to use
them. The first-conjugation class, on the other hand, is less frequent overall (36%
of uses), but has a lot more verbs (76%), so here it is more efficient to assemble
forms from scratch every time you want to use them. This means, Bybee argues,
that the first conjugation is more prominent as a tool than the third conjugation,
which makes it easier for first-conjugation forms to spread. Overall, ‘we can expect
most productivity from classes that have a high type frequency, but the members
of which on average have a medium to low token frequency’ (Strik 2015, p. 24).

In other languages
DUTCH VERBS STILL GOING STRONG
The strong verb system from Old English has disintegrated significantly
over centuries of language change. Other Germanic languages show differ-
ent developments: strong verbs have almost completely disappeared from
Afrikaans and Luxembourgish (two languages with a lot of language con-
tact) while Dutch still has c. 250 strong verbs. The original classes are still
clearly visible, because by chance, sound change has not interfered with
them as much as in English.
A wug test with verbs in Dutch (Knooihuizen & Strik, 2014) showed
that Class 1 was a very strong pattern for non-existing verbs that had the
relevant stem vowel ij /EI/ (past tense and participle ee/ eI/). For most other
non-existing verbs, participants defaulted to oo/ o*/ or o /O/, depending on
whether the verb had a long or a short vowel. These vowels appear in several
classes and therefore have a high type frequency regardless of what the root
vowel is. Participants even applied it to non-existing verbs with a vowel that
does not even occur in existing strong verbs.
Wug tests like this do not show language change in progress, but they
do uncover what patterns live in people’s minds. In a sense they show the
potential productivity of a pattern. We may not know if and when analogical
change happens, but our understanding of the process is such that we can
154 8 Analogy

make some pretty strong predictions about what form analogical change will
take when it does.

Conclusion

The look at English strong verbs and analogical change in this chapter forces us
to reconsider a few commonly used terms in morphology (Strik 2015, pp. 13–
22). First, the distinction between ‘regular’ and ‘irregular’ verbs (for ‘weak’ and
‘strong’, respectively) is false: strong verbs follow clear patterns, can be described
as following rules, and are therefore literally ‘regular’, too. Of course, the English
strong verb system has undergone so many analogical changes that the strong verb
rules have relatively little scope of applicability nowadays. Secondly, it makes little
sense to talk about forms or patterns, like weak inflection, as the ‘default’ or the
‘rule’. Attempts to define what pattern is the ‘default’ have failed: type frequency,
token frequency, and environmental diversity all play a role in what form is most
likely to spread in a particular context. This also means that we need to redefine
productivity: the ‘default’ must by definition be productive, but other patterns may
also be productive; it’s more a matter of relative rather than absolute productivity.

References

Berko, Jean. 1958. The child’s learning of English morphology. Word 14 (2–3): 150–177. https://
doi.org/10.1080/00437956.1958.11659661.
Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam:
Benjamins.
Bybee, Joan. 2015. Language change. Cambridge: Cambridge University Press.
Cuskley, Christine F., Francesca Colaiori, Claudio Castellano, Vittorio Loreto, Martina Pugliese,
and Francesca Tria. 2015. The adoption of linguistic rules in native and non-native speakers:
Evidence from a Wug task. Journal of Memory and Language 84: 205–223. https://doi.org/10.
1016/j.jml.2015.06.005.
Dammel, Antje. 2011. Konjugationsklassenwandel. Prinzipien des Ab-, Um-, und Ausbaus ver-
balflexivischer Allomorphie in germanischen Sprachen. Studia Linguistica Germanica 103.
Berlin: De Gruyter.
Hill, Eugen. 2010. A case study in grammaticalized inflectional morphology: Origin and devel-
opment of the Germanic weak preterite. Diachronica 27 (3): 411–458. https://doi.org/10.1075/
dia.27.3.02hil.
Hock, Hans Heinrich. 2003. Analogical change. In The handbook of historical linguistics, ed. Brian
D. Joseph and Richard D. Janda, 441–460. Oxford: Blackwell.
Hock, Hans Heinrich. 2021. Principles of historical linguistics, 3rd ed. Berlin: De Gruyter Mouton.
https://doi.org/10.1515/9783110746440
Knooihuizen, Remco, and Oscar Strik. 2014. Relative productivity potentials of Dutch verbal
inflection patterns. Folia Linguistica Historica 35: 173–200. https://doi.org/10.1515/flih.201
4.005.
McMahon, April. 1994. Understanding language change. Cambridge: Cambridge University Press.
References 155

Strik, Oscar. 2015. Modelling analogical change: A history of Swedish and Frisian verb inflection.
Dissertation. Rijksuniversiteit Groningen, Groningen.
Sturtevant, Edgar H. 1947. An introduction to linguistic science. New Haven: Yale University Press.
Wells, J.C. 1982. Accents of English. Cambridge: Cambridge University Press.
Grammaticalisation
From Words to Grammar (and Back?)
9

We have seen that analogy is a useful explanation for changes in morphology,

such as the spread of the {-ed} past tense ending on verbs or the {-s} plural end-
ing on verbs. These cases entail the extension of existing patterns to new cases.
But where do the patterns come from? What is the origin of these endings? In
this chapter, we will discuss a process by which morphemes can come into exis-
tence: grammaticalisation. We start with a theoretical overview before looking at
a number of case studies. At the end, we will also briefly investigate the claim that
grammaticalisation is a unidirectional process.

Grammaticalisation

The earliest definition of grammaticalisation is ‘the attribution of a grammatical

character to a formerly independent word’ (Meillet 1921). This definition fits some
high-profile examples, such as the development from the independent word did to
the grammatical item {-ed} in Proto-Germanic,1 but there are also many cases of
language change that qualify as grammaticalisation, but that do not start with an
independent word (such as a noun or a verb). It is more accurate to see gram-
maticalness as a cline and to see grammaticalisation as movement on that cline
(Hopper and Traugott 2003, p. 7).

1The precise word forms were of course different in Proto-Germanic, but the English forms do
show the general idea.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 157
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_9
158 9 Grammaticalisation

• content item → grammatical word → clitic → inflectional affix2

A better definition, therefore, would be one that takes this gradualness of

grammatical meaning into account, such as the following from Kuryłowicz (1965):

Grammaticalisation consists in the increase in the range of a morpheme advancing from a

lexical to a grammatical, or from a less grammatical to a more grammatical status.

As we will see, grammaticalisation is not a regular process like Neogrammarian

sound change that applies everywhere. There are quite well-defined (or well-
definable) restrictions on where a grammaticalisation change applies and where
it does not. Grammaticalisation always happens in the context of a particular con-
struction, and the same word does not grammaticalise outside that construction.
Bybee (2003, p. 602) writes that ‘it may be more accurate to say that a construc-
tion with particular lexical items becomes grammaticalized’, and some authors
even write about ‘constructionalisation’.

Processes in Grammaticalisation

Grammaticalisation is not a single process, but rather a frequently occurring com-

bination of processes. This also means that grammaticalisation theory is not a
theory as such, but rather a way of generalising over many instances in which these
processes together have turned lexical items or constructions into grammatical
markers (Heine 2003, p. 579; Traugott 2003, p. 638).
The processes that (can) make up grammaticalisation are explained below. The
example that is used is the development of future marker will in English, from Old
English willan ‘to want’.
desemanticisation The first process is a change to the semantics of the word
or construction. In most cases, the word loses (part of) its meaning content, which
explains the term de-semanticisation, or its alternative, ‘semantic bleaching’. In
the case of will, the ‘desire’ aspect of the meaning of the verb slowly fades to
the background and a more grammatical future meaning becomes foregrounded.

2 We can define a ‘content item’ as an independent word that has a real-world semantic meaning,
such as a noun (dog), verb (swim), or adjective (blue). A ‘grammatical word’ is also an independent
word, but its meaning is not so much concrete as it is abstract and grammatical: the auxiliary might
has a grammatical meaning that tells us something about the likelihood with which something will
happen in the future. ‘Clitics’ and ‘affixes’ are not independent words, but attach to (the end of)
other words. The difference between them is that affixes have clear restrictions on the word class of
the word they attach to—for example, {-ed} must attach to a verb, so that walk+ed is well-formed,
but *seagull+ed or *sharp+ed are not. Clitics have fewer restrictions: the possessive ending {-’s}
must attach to a noun phrase but the word class of the exact word it attaches to is not important:
the dog that I saw chasing a cat down the street yesterday when I almost stumbled+’s is entirely
grammatical (if a bit clunky). The grammaticalisation cline therefore tells us something about the
meaning as well as the syntactic freedom of a word.
 159

These meanings are clearly connected: if you want to do something, you probably
intend to do it, and eventually you will do it in future.
extension The next process is a change to the syntactic possibilities of the
word or construction. It becomes possible to use it in new contexts that were
previously not possible. In most cases, this means the construction can be used
in more contexts (‘extension’) with fewer constraints (‘context generalisation’, an
alternative term). In the case of will, we in fact probably have a context reduction,
as we lose the possibility to have a noun phrase as a complement and can only
have verb phrases: I will food is no longer correct in modern English, but it was
perfectly fine in Old English.
decategorialisation There is also a change to the morphology of the word
or construction: the item loses the morphosyntactic properties characteristic of the
word class that it belongs to. Normal verbs in English are negated with don’t and
have a past tense, for example. Will does not need do-support for negation, and
what used to be the past tense, would, has developed a meaning that is different
from simply ‘whatever will means, but at an earlier point in time’. This suggests
that will is not really a (normal) verb anymore, so it has left the category of verbs.
In extreme cases of decategorialisation, words can lose independent word status
and become clitics or affixes.
phonetic reduction As a phonological change in grammaticalisation, gram-
maticalised items often lose phonetic substance, usually as a result of being
pronounced in unstressed position. It is typically content words that receive lexical
stress, not function words; the defining change in grammaticalisation from being
a content word to being a grammatical word means that an item will appear in
unstressed position much more often. Of course, it is still possible to pronounce
will with a full form [wIë], but it is much more common for it to be pronounced
as [w∂ë], [∂ë], or even just [ë].
reanalysis The key stage in any change is when children reinterpret a con-
struction to mean something else from what their parents’ generation took it to
mean. They latch on to the meaning shift from ‘desire’ to ‘intent’ to ‘future’, the
changes in frequency of the verb in different constructions, and learn that will is
now an auxiliary verb marking the future.
In contrast to previous lists like this, it is important to realise that these are not
five stages that a construction goes through sequentially when it grammaticalises.
Rather, these are changes to different aspects of the construction (semantics,
syntax, morphology, and phonology) that occur roughly simultaneously and that
feed into each other to drive the grammaticalisation further, as new users and
learners of the language constantly reanalyse and reinterpret the construction.
The interconnectedness between these different changes makes that we regard
grammaticalisation as a single multi-layered process of language change.
Two examples should clarify how this process works.
160 9 Grammaticalisation

Grammaticalisation of English Going to

The English future construction be going to have its origins (obviously) in the pro-
gressive construction of the verb to go and the directional preposition to. This
development movement > future is a cross-linguistically common pathway
of grammaticalisation (Kuteva et al. 2019, p. 214–217). The going to future in
English has been extensively described both theoretically and on the basis of his-
torical corpus studies, often in comparison to other future markers (will, shall) or
to similar constructions in other (Germanic) languages (e.g., Bybee and Pagliuca
1987; Bybee et al. 1991; Gries and Stefanowitsch 2004; Hilpert 2008; Budts and
Petré 2016). A descriptive diachronic overview of the development of the going to
future with textual examples, many from Shakespeare, is given by Pérez (1990);
the presentation here borrows heavily from her work.
The original meaning of going to in Old English was unambiguously lexical: it
refers to movement towards a certain place, as shown in these early examples:

1. þū oferfærest þone sǣ and bist gangende tō Rōme byrġ
‘You cross the sea and are going to Rome’
2. þā sume dæġe wes hē tō þām bæþe gangende
‘Then some day he was going to the bath…’

This same meaning was still primary in Early Modern English, as many examples
from the works of Shakespeare exemplify:

3. There are pilgrims going to Canterbury with rich offerings (1 Henry IV, 1.2)
4. Your partner, as I hear, must die tomorrow
and I am going with instruction to him.
Grace go with you! (Measure for Measure, 2.3)
5. I tell thee, man, ‘tis better with me now
Than when I met thee last where now we meet
Then was I going prisoner to the Tower
By the suggestion of the queen’s allies (Richard III, 3.2)

Most examples in Shakespeare are clearly directional like this, but we do see a
subtle change happen. Whereas previously, the complement of going to could only
be a noun (including pronouns and proper nouns), we also occasionally see verbs
as the complement, as in (6). This is an extension of the construction to allow
new contexts.
In addition, we see a semantic change in this construction from meaning purely
‘travelling to a place’ to meaning ‘travelling (to a place) to perform an action’.

6. Duke: Sir Valentine, whither away so fast?

Valentine: Please it you grace, there is a messenger
That stays to bear letters to my friends,
And I am going to deliver them (Two Gentlemen of Verona, 3.1)
Grammaticalisation of English Going to 161

From the context, it is clear that the meaning of movement to a certain place is
still very prominent. The duke asks Valentine where he is going, and Valentine
answers that he is going to some (unspecified) place in order to deliver letters.
This does involve a slight change in meaning, as the intended action is added as
the ‘destination’.
Already in Shakespeare’s time, there are instances where the meaning of inten-
tion and the meaning of physical motion are more difficult to distinguish. Because
they are so difficult to distinguish, it is hard to say when exactly going to start
to be used as a future construction. Some very early examples come from the fif-
teenth century (Tagliamonte et al. 2014, p. 78) (7), but ambiguous cases can also
be found in Shakespeare (8):

7. Therefore while this onhappy sowle by the vyctoryse pojmpys of her enmyes was
going to be broughte into helle for thesynne and onleful lustys of her body
‘Therefore, while this unhappy soul by the formidable victories of her enemies
was going to be sent to hell for the sin and wicked lusts of her body’
8. Sir, the Germans desire to have three of your horses:
The duke himself will be to-morrow at court,
and they are going to meet him (Merry Wives of Windsor, 4.3)

This ambiguous situation remains for a few more centuries, as the attestations
of the going to construction all retain a hint of movement as well as agency:
intention and physical motion continue to play a role in the background. Only
around the turn of the twentieth century do we find examples that unambiguously
have a future reading, and where a movement reading is clearly nonsensical, such
as in the examples from F. Scott Fitzgerald’s The Great Gatsby (1925) (9, 10). A
reading where the earth and the fruit (intend to) move is clearly nonsensical, so an
interpretation as a future marker is the only possibility here.

9. It seems that pretty soon the earth’s going to fall into the sun.
10. … as if all sorts of fruit were going to fall into your hands.

At this stage it is clear that there has been desemanticisation: going to no longer
implies movement, the subject is not (literally) going anywhere anymore. We can
therefore also be sure that reanalysis has taken place.
Since at least the beginning of the twentieth century, the going to future has
been undergoing phonological reduction in various ways, although not all
these reduced forms are available in all varieties of English. (Most examples from
Poplack and Tagliamonte 2000.)

11. I think it’s gonna get worse before it’ll get better.
12. Well dear, I’m gon tell you the difference.
13. Who you gonna call?
14. If you looking for good you’ll find good. You look for bad, you gon find bad.
162 9 Grammaticalisation

15. Yo Taylor, I’m really happy for you—I’ma let you finish—but Beyonce had one
of the best videos of all time. (Kanye West, MTV Video Music Awards 2009)

The phonological reduction also highlights that grammaticalisation happens in

a very specific construction, and that other instances of the same word are not
affected. English now has two going to constructions: the original construction
meaning ‘movement to a place’, and the new, grammaticalised construction mean-
ing ‘future’. The phonologically reduced pronunciations are only acceptable for
the grammaticalised construction: things like I gonna London or I’ma London are
all nonsense, because the ‘movement’ meaning still requires a more or less fully
pronounced going to.

The Development of Discourse Marker Anyway

Another category of words that often arises through a process of grammaticalisa-

tion is discourse markers: words that organise the structure of conversation and/
or indicate a speaker’s stance towards an utterance, such as indeed, sort of , kind
of , like, or anyway (Traugott 2003). The discourse marker anyway started as the
construction in any way, with the literal meaning ‘following any path’ (16), but
already before 1500, we find ambiguous readings and even cases where a more
metaphorical reading ‘in any manner’ is more likely (17).

16. He schall haue accusars aboue hym, wythyn hym, on aythyr syde hym, and vndyr
hym, þat he schall no way scape. (<1500)
‘He shall have accusers above him, within him, on either side of him, and
under him, that he shall no way escape’
17. Sche wolde yet excusyn hir yf sche myth in any wey, and þerfor sche seyd …
(1438)
‘She would yet excuse her if she might in any way, and therefore she said …’

By the late sixteenth century, these unambiguous uses as an adverbial of manner

can be found without the preposition in (18); this is a case of decategoriali-
sation, as this is not normally a sentence position that noun phrases can occur
in. Up to this point, anyway only occurs in negative, conditional, or other irre-
alis contexts—things that are not (yet) the case, which are generally the contexts
where you use any—but by the end of the Early Modern English period we see an
extension of this context, and examples can be found in positive, realis contexts
as well (19).

18. and moreover so, that they bee not, any way overloaded or discouraged, nor yet
indangered, by the overcharging of their wits and memories (1627)
19. The Generation of all things, and every Progression of changeable Natures, and
all things which are any way moved, receive their Causes, Order and Forms out
of the Stability or Constancy of the Divine Mind. (1695)
The Development of Discourse Marker Anyway 163

What follows is a range of small, incremental changes in meaning and syntactic

placement. A usage with the meaning ‘nonetheless’ appeared in the early eigh-
teenth century out of the idea that to do something (in) any way (at all) is to do it
despite normal expectation or reason (19). Over time, the anyway with this mean-
ing tended to occur at the end of a clause (20), but syntactic placement generally
became a lot freer, as it is for other discourse markers as well.

20. This is certain, that whereas we behold selfish Actions of others, with Indiffer-
ence at best, we see something amiable in every Action which flows from kind
Affections or Passions toward others; if they be conducted by Prudence, so as
any way to attain their End. (1726)
21. The tape shows Barry picking up the crack pipe and asking how it worked,
adding ‘I never done it before.’ But when he received no directions, he lit up
anyway and inhaled the drug. (1990)

By the middle of the nineteenth century, the function of clause-initial discourse

marker anyway shifts to elaborating and justifying what has been said (22). But
like other discourse markers, it may co-occur with earlier forms that have a slightly
different meaning: in (23), we have both the ‘moving on with the conversation’
and the ‘contrary to expectations’ meanings in the same sentence.

22. It’s queer, very queer; and he’s queer too; aye, take him fore and aft, he’s about
the queerest old man Stubb ever sailed with. How he flashed at me! — his eyes
like powder-pans! Is he mad? Anyway there’s something on his mind, as sure
as there must be something on a deck when it cracks. (1851, Moby Dick)
23. Anyway, and so we ended up sleeping under there anyway and I only scared
two people. (1997)

These are just two examples of grammaticalisation in English. Such diachronic

overviews where we can see the meanings and grammatical status of construc-
tions change through ambiguously interpretable usage can be set up for many more
words and constructions. In addition to the future markers will and going to, also
modal verbs such as can (knowledge → ability → possibility → permission) and
may (permission → possibility) are examples where a verbal construction shows a
semantic shift as well as decategoricalisation: these verbs can no longer take a ver-
bal complement. These examples are cross-linguistically rather common (Kuteva
et al. 2019).
In the area of discourse markers such as anyway, Brinton (2008) offers a gram-
maticalisation account of a range of clause-like ‘parenthetical pragmatic markers’:
I think, I mean, I’m afraid, you see, if you will, what’s more, and many more.
Although the exact details and the timing of these changes differ, they all share
a semantic change towards more pragmatic meaning (e.g., textual organisation or
stance-taking) as well as a common syntactic change. Although Brinton is crit-
ical of the idea that all these constructions originate from a main clause with
a complement clause (e.g., I’m afraid that …), the constructions can no longer
164 9 Grammaticalisation

take a complement clause and have developed towards a placement more or less
independent from the sentence structure of the main clause.
The more subjective interpretation of meaning that underlies the development
of pragmatic markers can also be found in the history of quotative be like (see
Romaine and Lange 1991; Meehan 1991, for an early grammaticalisation analysis);
in the grammaticalisation of intensifiers really, very (< Old French veray ‘true’),
hella, and -ass; and in many more examples of grammaticalisation.

In other languages
NEGATION IS THE ULTIMATE GRAMMATICALISATION
One of the earliest cases of grammaticalisation that was described is the
development of the French negative marker ne … pas. It is a special case
because we see parallels of this development in many other languages,
including English, German, Dutch, Arabic, and Welsh. It is also remark-
able because this is a case of cyclical grammaticalisation: the development
of one negative marker generates a grammaticalisation process that develops
another negative marker. This cyclical pattern was first discovered by Otto
Jespersen and is therefore called Jespersen’s Cycle.
The negation non from Vulgar Latin developed into an unstressed ne
in Old French. By the time we have written forms for Old French, the
negation is often intensified by a separate particle, pas ‘step’. This word
had undergone DESEMANTICISATION (it no longer means ‘STEP’) and
DECATEGORICALISATION (it does not require an article, so is no longer
a noun), so that je ne vais pas ‘I not go STEP’ is merely an emphasised
version of the negation: from ‘I’m not going one more step!’ to ‘I’m not
going at all!’ Initially, pas could only occur with motion verbs, but over
time it appeared with other verbs as well, so it had really lost its semantic
content. There are also many cases where pas does not seem very intensive.
This means that the reanalysis was complete. Different methods of negation
(simple ne,ne… pas with intensified meaning, and plain ne … pas) were used
side by side for many centuries.
In colloquial spoken French, ne is often left out, which leaves pas as the
only marker of negation. What used to be only an emphatic marker is now
the negative marker. Interestingly, Latin non also derived from an emphatic
version of the negation: ne ‘not’+unum ‘one’ → noenum ‘not one’→ ‘not’.
The origin of English not lies in nāht, an Old English emphatic form that is
made up of ne ‘not’ +ān ‘one’ + wiht ‘person’ (Wallage 2008, 2013), which
replaced the original pre-verbal negation marker ne just like we ended up
with French pas.
Grammaticalisation as a Unidirectional Process 165

Grammaticalisation as a Unidirectional Process

Grammaticalisation is often thought of as a unidirectional process (Hopper and

Traugott 2003, p. 99–139). That means that we expect to find only language change
with items going from a less grammatical to a more grammatical status (the gram-
maticalisation cline), but we should not find change where items go from a more
grammatical to a less grammatical status, going along the grammaticalisation cline
the wrong way.

• content item → grammatical word → clitic → inflectional affix

• not: inflectional affix → clitic → grammatical word → content item

Strong claims of unidirectionality are dangerous. We earlier posited it for lenition

and then found cases of fortition, so it should come as no surprise that we also
find the opposite of grammaticalisation: degrammaticalisation. It is much less
common than grammaticalisation, but there are a few clear examples in English.
For example, the possessive marker {-s} in present-day English is a clitic that
attaches to the end of a noun phrase, regardless of what word class that noun phrase
ends in. But in earlier forms of the language, {-s} could only attach to a noun,
even when other words followed in the same noun phrase (24). That means that
{-s} changed from an inflectional affix to a clitic (see Chapter 11). Another exam-
ple is the possibility of using pronouns (grammatical words) as nouns (content
items) that also show the grammatical properties of nouns, such as being able to
take articles (25).

24. the archbishop’s of Canterbury hat → the archbishop of Canterbury’s hat

25. Is that cat a he or a she?

These cases of degrammaticalisation happened in Early Modern English, but a

more recent example is the development of ish, which appears to have travelled
the entire degrammaticalisation cline (Traugott and Trousdale 2013, pp. 233–237)
from being an inflectional suffix in Old English to becoming a clitic, a grammatical
word, and perhaps even a content word in present-day English (Pierce 2014, 2015).
The original {-ish} was clearly an inflectional suffix. In Old English, it was
used mostly to indicate ethnic identity (English, Danish), but there was also a less
productive ‘associative {-ish}’ that had the meaning ‘characteristic of’ (childish,
foolish). The ethnic {-ish} is no longer really productive in English3 ; the associa-
tive {-ish} became more productive over time and is still productive, especially
in pejorative contexts. In the Middle English period, the meaning developed into
‘approximative {-ish}’. There was a morphological change: it did not only attach
to noun bases, but could also attach to other word classes like adjectives, ini-
tially only colour terms (yellowish, bluish) but later also other adjectives. Note

3 Tolkien of course had Elvish, but he was a specialist in Old English and it is not unlikely that
he used the Old English ‘ethnic {-ish}’ as a model. Nationalities associated with more recently
formed countries tend to have {-an} or {-ese} as the suffix.
166 9 Grammaticalisation

that there is also a slight semantic change here: the emphasis is no longer on
similarity (childish = ‘like a child’) but on dissimilarity (greenish = ‘not quite
green’).
In the nineteenth century, we find a more significant morphological change, as
there are now examples where {-ish} attaches not just to a single word, but to the
end of a compound noun or even a noun phrase (26, 27). This means that it has
become a clitic: a suffix that can attach to words regardless of word class, as long
as that word is part of a phrase of the correct type.

26. a clean cravatish formality of manner (Dickens, Sketches by Boz, 1836)

27. Now onto The End We Start From, a not-quite-prose-not-quite-poetry end of
the world-ish novel by Megan Hunter @picadorbooks. Different. (Twitter,
@jonmhowells, 22 April 2017 18:32)

Approximative {-ish} then developed into a stand-alone word. Initially, this could
only be used in the same kinds of slot as interjections like yes, no, maybe, etc.,
and it is probably accurate to call this a grammatical word (28). And from there, it
developed into an actual content word. There are a few examples where it seems
to function like a pronoun (29),4 but mostly it is an adjective that can be modified
in the same ways as other adjectives (30–32). The exact meaning appears to be
a bit variable still. More recently, there are also examples where ish is used as
a euphemism for shit, in a variety of grammatical constructions (33–35) (Pierce
2015).

28. — At least you’re alive

— Ish.
— I thought you were dead.
— We’ve said all that.
— Right.
29. You seem like a chill person with similar interests—just want to make sure there
is ish to do on the scene. (‘something’, Pierce 2014, p. 117)
30. Show starts at 10 pm-ish (very very ish because we’ll still have the Clucking-
Blossom fundraiser going on). (‘approximately’)
31. If you’re like me and feel a little ish about dirty dining, you’ll need more than a
couple drinks. (‘unsure’)
32. I have another one that’s ish to this. (‘similar’, Pierce 2014, p. 117)
33. Now I’m not excluding myself from the list of people who need to get their ish
together.
34. What the heck Florida. Get your ish together.

4Pierce (2014, p. 117) interprets this ish as an indefinite pronoun, roughly meaning ‘something’,
but it may in fact also be a euphemistic replacement for shit, as in examples (33–35).
References 167

35. Need an ass to twerk? Are you ishing me? As long as youve got toned legs n can
hula hoop, you can twerk. Course theres a bit more to it than that but ive come
to find that good equestrians can twerk the ish out of anything. Its all in the hips
really. But I mean, thats kinda a duh righ?

These last examples, where ish stands in for shit, are probably a different
phenomenon, and the actual case of degrammaticalisation at hand is that of
approximative {-ish} to an adjective that denotes something along the lines of
‘uncertainty’—this is also a meaning change that makes sense. Interestingly, the
processes that were at play in this degrammaticalisation are the same as we saw for
grammaticalisation: gradual shifts in meaning, changes in the (morpho)syntactic
placement possibilities, and a reanalysis of what ish really is. Our cognitive biases
usually cause these processes to produce one type of result, but do not exclude
the opposite, and so can underlie grammaticalisation and degrammaticalisation
simultaneously.

Conclusion

In summary, grammaticalisation is an umbrella term for a set of linguistic changes

that together allow language users to reanalyse and reinterpret lexical items within
a particular conclusion as having a grammatical (rather than purely lexical) mean-
ing. There is no single way in which these changes combine, but the outcome is
something that has a more grammatical meaning than the lexical input. Gram-
maticalisation is what is thought to underlie the development of nominal and
verbal inflection, modal verbs, discourse markers, prepositions … the list is end-
less. Degrammaticalisation, the reanalysis of an item to have a less grammatical
meaning, much less frequently attested, but it is possible and is likely based on
similar processes.

References

Brinton, Laurel J. 2008. The comment clause in English: Syntactic origins and pragmatic
development. Cambridge: Cambridge University Press.
Budts, Sara, and Peter Petré. 2016. Reading the intentions of be going to: On the subjectification of
future markers. Folia Linguistica Historica 37: 1–32. https://doi.org/10.1515/filh-2016-0001.
Bybee, Joan. 2003. Mechanisms of change in grammaticization: The role of frequency. In The
handbook of historical linguistics, ed. Brian D. Joseph and Richard D. Janda, 602–623. Oxford:
Blackwell.
Bybee, Joan, and William Pagliuca. 1987. The evolution of future meaning. In Papers from the
7th International Conference on Historical Linguistics, ed. Anna Giacalone Ramat, Onofrio
Carruba, and Giuliano Bernini, 109–122. Amsterdam: Benjamins.
Bybee, Joan, William Pagliuca, and Revere D. Perkins. 1991. Back to the future. In Approaches to
grammaticalization, ed. Elizabeth Closs Traugott and Bernd Heine, vol. 2, 17–58. Amsterdam:
Benjamins.
168 9 Grammaticalisation

Gries, Stefan Th, and Anatol Stefanowitsch. 2004. Extending collostructional analysis: A corpus-
based perspective on ‘alternations’. International Journal of Corpus Linguistics 9 (1): 97–129.
https://doi.org/10.1075/ijcl.9.1.06gri.
Heine, Bernd. 2003. Grammaticalization. In The handbook of historical linguistics, ed. Brian D.
Joseph and Richard D. Janda, 575–601. Oxford: Blackwell.
Hilpert, Martin. 2008. Germanic future constructions: A usage-based approach to language change.
Amsterdam: Benjamins. https://doi.org/10.1075/cal.7.
Hopper, Paul J., and Elizabeth Closs Traugott. 2003. Grammaticalization, 2nd ed. Cambridge:
Cambridge University Press.
Kuryłowicz, Jerzy. 1965. The evolution of grammatical categories. Diogenes 51: 55–71. https://
doi.org/10.1177/039219216501305105.
Kuteva, Tania, Bernd Heine, Bo Hong, Haiping Long, Heiko Narrog, and Seongha Rhee. 2019.
World lexicon of grammaticalization, 2nd ed. Cambridge: Cambridge University Press.
Meehan, Teresa. 1991. It’s like, “What’s happening in the evolution of like?”: A theory of gram-
maticalisation. Kansas Working Papers in Linguistics 16: 37–51. https://doi.org/10.17161/
KWPL.1808.423.
Meillet, Antoine. 1921. L’évolution des formes grammaticales. In Linguistique historique et lin-
guistique générale, vol. 2, 130–148. Paris: Klinksieck.
Pérez, Aveline. 1990. Time in motion: Grammaticalisation of the be going to construction in
English. La Trobe University Working Papers in Linguistics 3: 49–64.
Pierce, Marc. 2014. The further degrammaticalization of -ish. American Speech 89 (1): 115–118.
https://doi.org/10.1215/00031283-2726431.
Pierce, Marc. 2015. More on ish. American Speech 90 (3): 394–398. https://doi.org/10.1215/000
31283-3441047.
Poplack, Shana, and Sali A. Tagliamonte. 2000. The grammaticization of going to in (African
American) English. Language Variation and Change 11 (3): 315–342. https://doi.org/10.1017/
S0954394599113048.
Romaine, Suzanne, and Deborah Lange. 1991. The use of like as a marker of reported speech and
thought: A case of grammaticalization in progress. American Speech 66 (3): 227–279.
Tagliamonte, Sali A., Mercedes Durham, and Jennifer Smith. 2014. Grammaticalization at an early
stage: Future be going to in conservative British dialects. English Language and Linguistics 18
(1): 75–108. https://doi.org/10.1017/S1360674313000282.
Traugott, Elizabeth Closs, and Graeme Trousdale. 2013. Constructionalization and constructional
changes. Oxford: Oxford University Press.
Traugott, Elizabeth Closs. 2003. Constructions in grammaticalization. In The handbook of histor-
ical linguistics, ed. Brian D. Joseph and Richard D. Janda, 624–647. Oxford: Blackwell.
Wallage, Phillip. 2008. Jespersen’s Cycle in Middle English: Parametric variation and grammatical
competition. Lingua 118 (5): 643–674. https://doi.org/10.1016/j.lingua.2007.09.001.
Wallage, Phillip. 2013. Functional differentiation and grammatical competition in the English
Jespersen Cycle. Journal of Historical Syntax 2 (1): 1–25. https://doi.org/10.18148/hs/2013.
v2i1.4.
Syntactic Change
Acquiring Language from Scratch
10

Introduction

In the previous two chapters in this part about structural change, we have dealt first
with very small structural items (morphology) and then slightly larger structural
items (grammaticalised constructions). The logical next step is to zoom out even
further and look at change at the level of the sentence: changes in word order, or
more precisely, constituent order. The diachronic changes in English syntax, much
of the relevant evidence, and early explanations of syntactic change have been laid
out in older sources such as Mossé (1959) and most notably Visser (1963–1973).
The study of syntactic change has made great advances since Noam Chomsky
developed his paradigm of generative syntax. This view of linguistic structure lent
itself to a model of linguistic change to an extent that earlier theories did not.
Most of the work on syntactic change therefore takes place in this Chomskyan
paradigm, and this is also the view we take in this chapter. The theory on syntactic
change will be illustrated with three case studies on syntactic change in Middle
and Early Modern English. The presentation will gloss over some of the finer
theoretical detail; for an accessible overview that goes deeper into syntactic theory
with regard to the same case studies, see Los (2015).

Syntactic Change in the History of English

There are some marked differences between Old English and present-day English
constituent order. The four sentences below give some examples of constructions
that were possible, or even frequent, in Old English, but that are no longer possible
today. This is not to say that these were the only constituent orders possible in Old
English; some of the constructions we find today were also fine in Old English.
But the loss of options in variation also constitutes change, and in this case, the

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 169
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_10
170 10 Syntactic Change

outcome of change shows a clear and consistent pattern in how Middle English
preferred to order its constituents.
Let’s go back to some sentences from the passage from Luke 9: 12–17 in the
Wessex Gospels for an illustration of Old English constituent orders that are no
longer possible:

1. þā gewāt sē dæġ forþ

then went the day forth
‘Then the day went forth’
2. hı̄e twelf him ġenēalǣhton
they twelve him approached
‘The twelve of them approached him’
3. þæt hı̄e (…) him mete finden
that they (…) them food find
‘that they find food for themselves’
4. wē ne-habbaþ būtan fı̄f hlāfas and twēġen fiscas
we not-have but five loaves and two fishes
‘We don’t have anything except five loaves and two fish’

In (1), we see a main clause in which, from a present-day English perspective,

the subject sē dæġ ‘the day’ and the verb gewāt ‘went’ have changed places. Old
English, as most present-day Germanic languages, had an unmarked constituent
order with the inflected verb in the second position in the sentence (‘verb-second’,
or V2). In many cases, the first position in the sentence would be occupied by the
subject, and a sentence like sē dæġ gewāt forþ ‘the day went forth’ does not strike
us as remarkable. But if some other constituent is put in this first position, such
as a topicalised object or an adverb like then, present-day English would simply
move the subject into the second position, and the verb into the third position. In
Old English, the inflected verb remained in second position, giving the constituent
order we see in (1).
In (2) and (3), we see evidence that Old English was a so-called OV language,
that is, that the underlying constituent order was Object-Verb, not Verb-Object as
it is in present-day English. A very basic mini-sentence in English that consists of
just a verb and an object, like the sentences produced by very young children or
muppets on Sesame Street, will have VO order: Eat cookies! But Cookie Monster’s
Dutch colleague Koekiemonster, who speaks an OV language, would say Koekjes
eten! with the object preceding the verb. Because in main clauses, the verb typ-
ically moves to the second position because of the V2 rule, main clauses do not
give very good evidence for OV order in Old English, but subordinate clauses do:
in (3) we clearly see the object mete ‘food’ before the verb finden ‘find’. There is
Acquisition and Change in Generative Syntax 171

occasional evidence from main clauses as well: in (2) we have the (indirect) object
him ‘him’ before the verb ġenēalǣhton ‘approached’.1
Finally, in (4), note the position of the negator ne in front of the main verb
habbaþ ‘have’. Present-day English would require an auxiliary verb do to negate
a main verb: we do not have two loaves. The introduction of do as an auxiliary in
negations is related to a change in where sentence adverbials (including negation)
can go. Present-day English does not allow these to come between a verb and its
object, which is why we place them before the verb: I never eat breakfast. In Old
English, again as in present-day Dutch or German, this was possible; in Dutch, we
have ik eet nooit ontbijt (lit. ‘I eat never breakfast’), and that is the order we would
expect in Old English as well.
Over time, then, English has become less similar to Dutch and German, and
more similar to—well, to present-day English, which in these respects seems to
be the opposite of Old English. The challenge is now to provide a syntactically
informed account of these changes.

Acquisition and Change in Generative Syntax

One of the most important developments in the study of syntax in the twentieth
century is the generative syntactic paradigm by Noam Chomsky. The most basic
idea behind this is that sentences can be divided up into constituents, groups of
words that belong together more than they belong to other words, and that con-
stituenthood is recursive, so that within groups you have other groups, and so
on. The internal sentence structure is often visualised in trees, such as the one in
Fig. 10.1. You can clearly see the constituents—noun phrases (NP), verb phrases
(VP), preposition phrases (PP), and adjective phrases (AP)—and how they are
embedded in one another.
As the theory of generative syntax developed and expanded, such trees became
more complicated in order to be able to account for different structures in a
coherent theoretical framework. Among other things, new types of phrases were
hypothesised connected to complementisers (CP) but also more abstract notions
such as tense (TP) and inflection (IP or InflP). In the theory, we have an under-
lying representation of the structure of language in our minds, the deep structure,
and we can derive different structures in actual utterances (surface structures) by
moving constituents according to certain rules. Regardless of the different surface
structures that we may produce, the deep structure of the language is always the
same, and it is just different types of movement that do or do not take place.
Generative syntactic theory includes ideas about syntactic change; of the differ-
ent theories about how syntax works, it is by far the theory with the most coherent
ideas about language change. A key person in the development of generative work

1Pronouns appear to have been able to escape the V2 rule, both as subjects and objects (Roberts
2007, p. 60), so this is not the best example of OV order in a main clause. Non-pronoun objects
occur in OV order in main clauses considerably less frequently.
172 10 Syntactic Change

Fig. 10.1 Tree diagram for The black cat found cheese from Germany in the pantry

on syntactic change is David Lightfoot (1979, 1991, 2003). In the generative view
of syntactic change, language change happens at the point of language acquisition.
Children who learn a language are genetically endowed with Universal Grammar.
This grammar, however, is underspecified; there is a whole set of options known as
parameters (Roberts 2022). These are essential questions that children will need
to figure out the answer to. What word classes does my language have? Does my
language require an overtly expressed subject? Is the basic word order in my lan-
guage OV or VO? Does my language require the verb to move to the T position
in finite clauses? Many of the details of how a language works are decided by
whether a parameter is set one way or another (Roberts 2007). Language acquisi-
tion, or at least the syntax part of it, then equates to parameter setting on the basis
of the evidence of the language that children hear around them. It is useful to
remind yourself here that children do not copy language wholesale, but that each
person acquiring a language has to construct it from scratch.
In the majority of cases, a learner will get sufficient cues from the language
around them to set a parameter in the same way as they were set by the previous
generation, and the language does not change. However, if something changes in
the evidence that a child has (a knock-on effect of some other change, or simply
a change in frequency of one construction over another) the available cues that a
learner has to set a parameter change as well, and they may mis-set a parameter,
that is, set it to a different value from that of the generation the child is learning
The Change from OV to VO Order 173

from. This means that they will build a mental grammar that is different from
that of the previous generation. This mental grammar can explain the structures
that the child heard around them during language acquisition, but could potentially
also generate other utterances that their parents’ generation’s grammar could not
produce. The child’s output, based on the ‘new’ grammar, will then be additional
evidence to subsequent learners, who will also decide to set the parameter in the
‘new’ way, and this is how language change spreads through the community.
In the following sections, we will look at three case studies to exemplify the
generative view of syntactic change in English. The first is the change from under-
lying OV to VO constituent order; the second is the loss of verb movement to
the T position, which gave rise to do-support in negations; and the third is the
loss of verb movement to the C position, which caused the disappearance of V2
from English. Much of this discussion ultimately goes back to the work by David
Lightfoot. This work has been criticised, however, because the chronology of the
changes he posited is not supported by the documentary evidence (Roberts 2007,
p. 127). Nevertheless, the discussion illustrates how generative syntacticians think
about language change.

The Change from OV to VO Order

Old English is thought to have been underlyingly verb-final, with Object-Verb

order. As in German and Dutch, there was direct evidence for this in subordinate
clauses, as we saw in (3) above. Intuitively, you would think that this would be
enough for children to acquire the OV order; after all, there are enough subordinate
clauses to serve as evidence. But Lightfoot (1989) argues that children do not
actually pay attention to all the linguistic evidence around them and that they
instead only use main clauses as evidence. In Lightfoot’s terminology, they are
‘degree-zero’ learners, because they only look at evidence with zero degrees of
embedding. And that becomes a problem for learning underlying OV constituent
order, because the V2 rule meant that in main clauses, verbs moved from the final
position after the object to the second position in the sentence, before the object,
and as such main clauses do not usually give great evidence of OV constituent
order.
Despite this, learners could still figure out the underlying OV order in Old
English, because some main clauses did provide the relevant evidence. Occasion-
ally, there might be a main clause with an unmoved verb still in final position, but
more often this evidence came from sentences with verbs with a separable prefix,
as in (5). The prefix would be left behind in the verb’s original place when the
verb moved to second place.

5. þā sticode him man þā ēagan ūt

then stuck him someone the eyes out
‘Then someone stuck his eyes out.’

How this works syntactically can be shown with the help of the Dutch sentences
in (6–7) and the syntactic trees in Fig. 10.2. Basic constituent order is shown in
174 10 Syntactic Change

Fig. 10.2 Verb movement and the change from OV to VO order

the subordinate clause in (6) and in the left tree, which clearly shows OV order in
the VP. In order to turn this basic order into main clause order, we move the verb
from its original position first to just below I' , where it picks up tense inflection,
and then to just below C' , both times leaving a trace tj . The subject moves to
below CP, leaving a trace ti in its original position below TP. Crucially, though,
when the verb moves up the tree, the separable prefix uit is left behind in its
original position, so learners can work out that that is where the verb originated,
and that its current position must be due to movement. This means that they are
able to construct a grammar that contains underlying OV word order and a set of
movements to derive surface structures from that deep structure.

6. hij zei dat hij zijn geld uitgaf

he said that he his money out-gave
‘He said that he spent his money.’
7. hij gaf zijn geld uit
he gave his money out
‘He spent his money.’

Over time, says Lightfoot, such cues for underlying OV order declined in fre-
quency, for example, because these verbs with separable prefixes disappeared from
the language. Contact with Old Norse introduced (a higher frequency of) phrasal
verbs instead of these prefixed verbs. Somewhere around the late eleventh, early
twelfth century, then, these cues from left-behind prefixes were so infrequent that
children no longer set the parameter for headedness in the VP to OV, but to VO
instead. This change in headedness was very rapid (‘catastrophic’). Lightfoot also
The Loss of English V-to-C (Verb-Second) 175

argues that when this reanalysis happened, subordinate clauses were still predom-
inantly verb-final, but despite that, it seems to children overwhelmingly started
setting the parameter to VO. This is one of the case studies that prompted Light-
foot to propose degree-zero learnability. (There were other factors that played a
role too, and quite a bit of examples from other languages than English, so that
degree-zero learnability is not a circular argument.)

The Loss of English V-to-C (Verb-Second)

Also with the ‘new’ underlying VO order, English still exhibited verb-second. In
syntactic terms, this can be seen as a movement of the verb to the C position. We
have already seen this movement in Fig. 10.2, where the finite verb gaf ‘gave’
moved from its original place, via I' , to the position below C' which is normally
reserved for complementisers (C). The loss of the possibility of performing this
movement in English is dated to the late fourteenth and early fifteenth centuries
(Fischer et al. 2000, pp. 132–137).
One peculiarity of the V2 rule in southern dialects of Middle English was
that different types of subjects behaved differently. If the subject was a pronoun,
it could break through the V2 rule and still come before the finite verb if the
first position was already taken up by some topicalised element, as in (8). Non-
pronominal subjects could not do this, and they always appeared in third position
if the first position was already occupied (9). Northern dialects, possibly due to
contact with Old Norse, did not show this differential pattern; also pronominal
subjects behaved as in (9) in those dialects. The change in the fourteenth and
fifteenth centuries, then, is that both pronominal and non-pronominal subjects
increasingly appear before the verb, regardless of whether the first position is
already taken (10).

8. ðas þing we habbað be him gewritene

these things we have about him written
‘These things we have written about him.’
9. Zewiss hafð godd forworpen ðan ilche mann
certainly has God rejected that same man
‘Certainly God has rejected that same man.’
10. sothely þe ryghtwyse sekys þe ioye
truly the righteous seeks the joy
‘Truly the righteous seeks the joy.’

An early hypothesis linked the change to the difference between pronominal and
non-pronominal subjects disappearing. Following Lightfoot’s idea that learners
build their grammar on the basis of the available evidence, this would then mean
that for whatever reason, the utterances these learners heard did not give enough
176 10 Syntactic Change

evidence that different types of subjects behaved differently. For example, sen-
tences with non-pronominal subjects and a topicalised element in first position
could for whatever reason have been infrequent, so that learners simply did not
pick up on this distinction, and start treating all subjects as if they were pronom-
inal, that is, all subjects can now come before the verb. Attractive as such an
analysis may be, however, the surviving written data does not support such a drop
in frequency, so the explanation for the change must be found somewhere else.
An alternative hypothesis posited by Fischer et al. (2000) is that the change
may be linked to another change in roughly the same period, namely the loss
of V-to-I movement (see below). Remember that I is a position where the verb
receives inflection such as tense and person and number agreement. If there is no
evidence that the verb has even been moved to I, then it cannot have moved on to
C either. The loss of verbal inflection in the Late Middle English period could be
a necessary cue, although the link between the loss of V-to-I movement and the
loss of V-to-C movement still needs to be supported by documentary evidence.

The Loss of English V-to-I

This loss of V-to-I movement, then, is our last example of syntactic change. This
is a change that was completed in English around 1600 (Roberts 2007, p. 57)—
it may therefore have been contemporaneous with the loss of V-to-C movement,
as hypothesised above—and which is connected to the rise of modal verbs as a
category and to the rise of do-support in negations.
Let us again look at some sentences to illustrate the effect of V-to-I movement.
Of the sentences below, (11) and (12) are formed according to a grammar with the
possibility of V-to-I movement, while (13) and (14) are the equivalent sentences
formed according to a grammar without V-to-I movement. The syntactic tree for
(12) is given in Fig. 10.3; that for (14) is given in Fig. 10.4.

11. he eats always breakfast.

12. … that she wields not her sword.
13. He always eats breakfast.
14. … that she does not wield her sword.

A grammar that does not allow V-to-I movement will produce sentences in which
sentence adverbials (like always or not) end up to the left of the main verb, or
phrased differently, in which sentence adverbials cannot come between the main
verb and its object. As can be seen from Figs. 10.3 and 10.4, the main verb origi-
nates to the right of the adverbial, but V-to-I allows it to ‘hop over’ it to a position
The Loss of English V-to-I 177

Fig. 10.3 A negated

subordinate clause before the
loss of V-to-I movement

on the left. The question now is what change in the input caused learners to con-
struct a grammar without V-to-I movement, even if an earlier generation’s grammar
did have it.
As Fischer et al. (2000) hinted at, one of the changes may have been the loss of
inflection. Lightfoot (1979) specifically discusses this in connection with the devel-
opment of a special group of verbs called ‘modal verbs’, which behave differently
from most other verbs.2
Modal verbs (can, may, must, will, etc.) were different to begin with: they orig-
inate from a class of verbs called ‘preterite present’ verbs, which are distinct from
other verbs in that they do not have the third-person singular ending (modern
English -s). (There were a few more verbs in this class, such as dugan ‘be of
value’ and munan ‘think’, but they were lost in the Early Middle English period.)

2As with earlier analyses, this account has been criticised as documentary evidence does not
unambiguously support the chronology of these changes.
178 10 Syntactic Change

Fig. 10.4 A negated

subordinate clause after the
loss of V-to-I movement

These verbs did have the second-person singular ending -st but when (unrelatedly)
second-person singular thou gets replaced by the originally second-person plural
you, that ending disappears from the language as well. As a result, this class of
verbs now showed no agreement at all.
Another characteristic of the preterite-present verbs that survived was that their
meanings allowed them to grammaticalise into modals: they were to do with ability
(can), permission (may), obligation (must, shall), etc., which are related to the
more abstract meanings that modal verbs have. It is likely that the change would
not have been possible if these were verbs like eat or dance, which do not have
these abstract meanings.
Modals also differed from other verbs in what complements they could take.
For one, they lost the ability to take nominal complements. Where in Old English,
sċulan ‘to owe’ could have a noun phrase as an object, its modal descendants shall
The Loss of English V-to-I 179

and should (originally a past tense) can only be followed by a verb phrase. Sec-
ondly, unlike other verbs that take an infinitive verb phrase complement, modals
do not require the infinitive marker to: it is you should go, not *you should to
go.
With all these changes having taken place, learners no longer had evidence that
modals were in the same class as other verbs, and they analysed them as a sepa-
rate class that simply originates in a different position—I—rather than them being
verbs that originate in the V position and then move to I. You would need more,
and more complicated rules, Lightfoot argues, to account for the surface form if
modals were still part of the larger class of verbs than if they were reanalysed as
a separate class.
This development of a clearly separate class of auxiliaries that occur in I, and
especially the rise of do in this position, has the additional effect of removing
evidence for the movement of ordinary verbs from V to I. And if ordinary verbs
do not move to I, they cannot participate in processes that occur in I. These include
a set of characteristics known as nice properties (Los 2015, pp. 91–93): modals
and auxiliaries, but not other verbs, can be negated (n, 15), they show inversion of
subject and verbs in questions (i, 16), they show ellipsis of a repeated complement
(c for ‘code’, 17), and they can be stressed for emphasis (e, 18).

15. You may not enter.

16. Can you tell me the answer?
17. You may not enter, but she may .
18. You should eat more healthily.

For such structures, where we would previously have to move the verb to I, we
would now have an auxiliary verb do in I while the main verb remains in the
V position. For a while, do was used as an auxiliary even in simple affirmative
sentences. Children acquiring English on the basis of this new evidence set the V-
to-I parameter to ‘no’. The words that they interpret as verbs always occur in V and
do not move. The other side of the coin is that modals cannot do certain things
other verbs can do—they do not have a progressive -ing form, for example—
because these things happen in V and modals were never in V.

In other languages
AFFIRMATIVE DO AS A FAILED CHANGE
The loss of verb-second and V-to-I caused an increase of do-support in
English. The rise of do-support was charted already by Ellegård (1953),
but there is a major question that remains unanswered to this day. Present-
day English has obligatory do-support in questions, negative sentences and
emphatic sentences, but in a simple affirmative sentence as (1) it strikes us
as marked.
180 10 Syntactic Change

(1) You’re the cutest thing that I ever did see.

Ellegård’s data shows a small increase in do-supported affirmatives from

the late 15th century, peaking at about 10% of all affirmative declarative
sentences in the second half of the 16th century, before disappearing almost
completely again by the late 17th century. It seems like affirmative do started
the beginning of an S-curve of language change, but petered out again before
it reached the phase of rapid spread. That affirmative do is the type of do-
support that constitutes a ‘failed language change’ is all the more surprising
considering that at its peak, it constituted 90% of all cases of do-support
(Budts, 2022, p. 340).
So why did affirmative do fail? After all, Wischer (2008) claims that it
actually fulfilled some very useful functions. For example, it regularised past
tense marking to some extent by making this universally ‘did + infinitive’.
This construction occurred most with strong verbs, verbs ending in a dental
stop, or verbs of Romance origin.
Budts (2022) suggests that the semantic function of affirmative do may
ultimately have contributed to its decline. Initially occurring predominantly
in statements of universal truth, affirmative do increasingly became used to
mark the speaker’s commitment towards the truth of the statement, as a
rhetorical tool to convince the hearer. With such specialised functions for
affirmative do, unmarked contexts were left once again to the simple present
without do-support.

In other languages
THE LOSS OF V-TO-I IN FAROESE
A language where the loss of V-to-I is a currently ongoing change is Faroese,
spoken by some 50,000 people in the Faroe Islands. Faroese allows both
word orders with V-to-I (1) and without V-to-I (2). It is generally thought
that this change is nearing completion, and that sentences such as (1) are
rare.

(1) at hon sveipar ikki svørð sítt

that she wields not sword her
(2) at hon ikki sveipar svørð sítt
that she not wields sword her
‘…that she does not wield her sword’
Conclusion 181

As a language where the loss of V-to-I is underway, Faroese offers a rare

opportunity to investigate this particular syntactic change as it happens.
For example, Heycock et al. (2012) have shown that while most sentences
produced by Faroese speakers suggest no V-to-I movement, speakers do gen-
erally accept sentences that can only be produced by a grammar with V-to-I
movement. This suggests that they have a grammar that simultaneously does
and does not have V-to-I movement.
Faroese moreover gives more information about the hypothesis that the
loss of V-to-I is connected to the loss of verbal inflection morphology.
Faroese has less of this morphology than Icelandic (which has also retained
V-to-I) but more than mainland Scandinavian languages (which have lost
V-to-I). It may therefore help answer the question of to what extent these
morphological and syntactic changes are causally related (Heycock et al.,
2012, p. 591–596).
Finally, although V-to-I is on its way out of Faroese, young children para-
doxically produce more sentences with V-to-I and also have more positive
grammaticality judgments than older speakers (Heycock et al., 2013). Their
language acquisition seems to hypothesise V-to-I more strongly than the
input they get would require. It may be that in the absence of unambiguous
cues, language acquisition is biased in favour of V-to-I, and its disappear-
ance from Faroese may be due to something else. Language contact with
Danish (without V-to-I) from primary school age onwards then becomes a
likely explanation.

Conclusion

Syntactic change in the generative syntactic paradigm, then, is a question of chil-

dren acquiring a language by setting parameters on the basis of the available input.
If there is some change in this input, they can set a parameter differently from what
earlier generations had done and change the grammar that way. A problem with
this account, however, is that it is necessarily all or nothing: parameter setting is
black or white, not shades of grey. This may happen to an entire generation of
children simultaneously, but a change may also spread gradually through a com-
munity, starting with a few children who for some reason got more of specific
cues, and—with their output adding to the evidence for the new setting—spread
faster as time goes on. But even in this ‘non-catastrophic’ scenario, the theory
predicts that change is all or nothing for each individual. You can only set your
parameter one way, not both.
The historical record, however, as well as contemporary experiments on syn-
tactic variation in English and other languages, shows that this is not the case.
Syntactic changes, like other language changes, progress via succeeding genera-
tions having differing proportions of the competing variants in their individual use.
182 10 Syntactic Change

Kroch (e.g., 2001) proposes that in periods of change, individuals acquire com-
peting parameter settings (often called ‘competing grammars’): output is variable
because they access the different systems at their disposal with particular proba-
bilities. Research in this paradigm tells a slightly different story of the OV-to-VO
change in English (Pintzuk 1993, 1996, 2003; Kroch 2001). We have evidence
of the same speaker (or rather, the same scribe in the same manuscript) produc-
ing both phrasal verbs with VO order and prefixed verbs with OV order, which
suggests that they had both these grammars available to them.
Such questions about variability and language acquisition remain open for the
time being and do not take away from the underlying idea of how syntactic change
works. As we have seen for many other types of language change, language learn-
ers build their language from scratch, modelled on whatever they hear around
them. Even small changes in the input—in frequency or distribution—can cause
them to learn a different language from their parents’.

References

Budts, Sara. 2022. A connectionist approach to analogy: On the modal meaning of periphrastic do
in Early Modern English. Corpus Linguistics and Linguistic Theory 18 (2): 337–364. https://
doi.org/10.1515/cllt-2019-0080.
Ellegård, Alvar. 1953. The auxiliary do: The establishment and regulation of its use in English.
Stockholm: Almqvist & Wiksell.
Fischer, Olga, Ans van Kemenade, Willem Koopman, and Wim van der Wurff. 2000. The syntax
of Early English. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO978051
1612312.
Heycock, Caroline, Antonella Sorace, Zakaris Svabo Hansen, Frances Wilson, and Sten Vikner.
2012. Detecting the late stages of syntactic change: The loss of V-to-T in Faroese. Language
88 (3): 558–600. https://doi.org/10.1353/lan.2012.0053.
Heycock, Caroline, Antonella Sorace, Zakaris Svabo Hansen, and Frances Wilson. 2013. Acqui-
sition in variation (and vice versa): V-to-T in Faroese children. Language Acquisition 20 (1):
5–22. https://doi.org/10.1080/10489223.2012.738741.
Kroch, Anthony S. 2001. Syntactic change. In The handbook of contemporary syntactic theory, ed.
Mark Baltin and Chris Collins, 629–739. Oxford: Blackwell.
Lightfoot, David W. 1979. Principles of diachronic syntax. Cambridge: Cambridge University
Press.
Lightfoot, David W. 1989. The child’s trigger experience: Degree-0 learnability. Behavioral and
Brain Sciences 12 (2): 321–334. https://doi.org/10.1017/S0140525X00048883.
Lightfoot, David W. 1991. How to set parameters: Arguments from language change. Cambridge,
Mass.: MIT Press.
Lightfoot, David W. 2003. Grammatical approaches to syntactic change. In The handbook of
historical linguistics, ed. Brian D. Joseph and Richard D. Janda, 495–508. Oxford: Blackwell.
Los, Bettelou. 2015. A historical syntax of English. Edinburgh: Edinburgh University Press.
Mossé, Fernand. 1959. Manuel de l’anglais du moyen âge des origines au XIVe siècle. II: Moyen-
anglais, 3rd ed. Paris: Aubier-Montaigne.
Pintzuk, Susan. 1993. Verb seconding in Old English: Verb movement to Infl. The Linguistic
Review 10 (1): 5–35. https://doi.org/10.1515/tlir.1993.10.1.5.
Pintzuk, Susan. 1996. Old English verb-complement word order and the change from OV to VO.
York Papers in Linguistics 17: 241–264.
References 183

Pintzuk, Susan. 2003. Variationist approaches to syntactic change. In The handbook of historical
linguistics, ed. Brian D. Joseph and Richard D. Janda, 509–528. Oxford: Blackwell.
Roberts, Ian. 2007. Diachronic syntax. Oxford: Oxford University Press.
Roberts, Ian. 2022. Diachronic syntax, 3rd ed. Oxford: Oxford University Press.
Visser, F. Th. 1963–1973. An historical syntax of the English language (3 vols.). Leiden: Brill.
Wischer, Ilse. 2008. What makes a syntactic change stop? On the decline of periphrastic do in Early
Modern English affirmative declarative sentences. Studia Anglica Posnaniensia 44: 139–154.
The Loss of Case
A Perfect Storm
11

Introduction

One of the most difficult grammatical features of Old English to speakers of

present-day English was that the language had case. Case is a morphological cate-
gory that, in its most basic form, expresses the function a word has in the sentence.
A word that functions as the subject in a sentence may have a different form than
the same word when it is the direct object in a sentence. This is something that you
may recognise from some languages frequently taught in schools, such as German,
Latin, or Russian. The case system of German is illustrated in Table 11.1.
Different language families have different case systems and different numbers
of cases. Proto-Indo-European is thought to have had eight cases, present-day
Russian has seven, Latin had five, but for Germanic languages like German or
Icelandic, and also Old English, only the four cases given in the table are relevant.
Because case forms differ for words of different grammatical genders, and even
for groups of words with the same grammatical gender, there can be quite a few
different sets of case forms. We shall see below that this was also the case for Old
English.
In German, case is nowadays mostly expressed only on the article, but in other
languages, the form of the noun differs much more between the different cases. A
side effect of using case to express the function of a constituent in the sentence
is that, as the function is already expressed by case marking, word order becomes
less important. Der Hund beißt den Mann ‘The dog bites the man’ is semantically
equivalent to Den Mann beißt der Hund; the only difference is where the focus
in the sentence lies. (The second sentence sounds awkward to native speakers of
German, but it is definitely grammatical and correctly interpretable.) The meaning
of the sentence depends on the case marking on ‘man’ and ‘dog’, and less on
constituent order. This applied to Old English as well, although like German, there
was a clear preference for sentences starting with a subject. The word order was
not as free as we can see in some Latin poetry, for example.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 185
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_11
186 11 The Loss of Case

Table 11.1 Cases in German

case function masc. sg. fem. sg. neut. sg. plural
nominative subject der Hund die Sonne das Haus die Hund-e
genitive possessive des Hund-es der Sonne des Haus-es der Hund-e
dative indirect obj dem Hund(-e) der Sonne dem Haus(-e) den Hund-en
accusative direct obj den Hund die Sonne das Haus die Hund-e

In addition to basic functions in the sentence, certain cases are also required on
the complements of different verbs and prepositions. Especially for the preposi-
tions, there is often no semantic clue to which case is needed. Some prepositions
can even select multiple cases—but in these cases, such as with in, it is usually
the dative for ‘stative’ meanings and the accusative for ‘movement’ meanings:

1. Der Hund geht in der Schule (herum)

the dog.nom goes in the school.dat (around)
‘The dog walks in the school’
2. Der Hund geht in die Schule (hinein)
the dog.nom goes in the school.acc (in)
‘The dogs walks into the school’

In this chapter, we look at the demise of the case system in English from the late
Old English and throughout the Middle English period. This was a change that
involved many different parts of the language system: morphologically, phonolog-
ically, and syntactically driven change, as well as language and dialect contact all
worked together in a cyclical, self-reinforcing cycle that eventually led to English
being a caseless language.

Decaying Case

The loss of case in English took place over the entire Middle English period, a
bit earlier in the north of England than in the south. Allen (1997) has charted the
process in detail, and the overview here is heavily based on her work.

Starting Point

Nouns in Old English fit into a number of different classes. Unlike in modern
German, these were not separated by grammatical gender, and each class also had
a different pattern of case endings in singular and plural. To illustrate this, the case
endings for some of the noun classes are given in Table 11.2. It is not predictable
what pattern applies to what noun, so learners of the language would just have to
acquire this for each word separately.
Decaying Case 187

Table 11.2 Some examples of case patterns in Old English

‘boat’ ‘eye’ ‘son’ ‘goose’ ‘moon’
nom. sg bāt ēag-e sun-u gōs mōn-a
gen. sg bāt-es ēag-an sun-a gōs-e mōn-an
dat. sg bāt-e ēag-an sun-a gēs mōn-an
acc. sg bāt ēag-e sun-u gōs mōn-an
nom. pl bāt-as ēag-an sun-a gēs mōn-an
gen. pl bāt-a ēag-ena sun-a gōs-a mōn-ena
dat. pl bāt-um ēag-um sun-um gōs-um mōn-um
acc. pl bāt-as ēag-an sun-a gēs mōn-an

Already in Old English, there was a lot of syncretism of forms. This means that
the same form was being used for different functions. For example, in the German
definite article for feminine nouns, there is syncretism between the nominative
and accusative (both are die) and between the genitive and the dative (both are
der). It is important to conceptually separate form and function here: the fact that
some German nouns use the same form in the nominative and the accusative,
does not mean that speakers are unable to tell the two cases apart—within the
context of a sentence, they know full well whether die Sonne ‘the sun’ is in the
nominative or accusative case, and if they have to replace the feminine noun with
a masculine noun such as der Mond ‘the moon’, where the genitive and dative
forms are different, they will choose the correct form.
The syncretism of Old English case forms is already clear from Table 11.2.
With only some exceptions, the nominative and accusative forms were the same,
and there were other endings that recurred as well: especially -a and -an were used
in different functions. Conversely, it is difficult to tell from the endings what cases
they indicated. Only -um for the dative plural is a solid pattern, and the genitive
plural endings all have a final -a.
This overview sketches case as in a bit more of a hopeless situation than it
actually was. Although the nouns showed a high degree of syncretism, this was not
the case for modifiers such as adjectives and especially articles, and for personal
pronouns. In other words, there were clear clues in the language for the use of
different forms for different functions.

Changes in Form

The earliest evidence of change in the case system dates back to the mid-tenth cen-
tury in texts written in the Northumbrian dialect, e.g., in the Lindisfarne Gospels.
Multiple forces work together to cause this change, which we will look at sepa-
rately at first. Allen (1997) emphasises the effects of sound change and analogy,
but we can also consider syntactic change and language contact as catalysts for
the development.
188 11 The Loss of Case

Two regular sound changes occurred that both had consequences for the Old
English case system. The first was the reduction of unstressed vowels to [∂]. This
change first applied to back vowels, and later also to front vowels. The second
sound change was the loss of final -n in unstressed endings. This final change
has a parallel in present-day Dutch where -en endings are usually pronounced [∂].
These changes started in the north of England, but later spread to the south. In
southern dialects, there was an additional sound change /m/ → /n/ in unstressed
syllables, which occurred before the loss of final -n. The effect of these sound
changes was that more and more case endings ended up as [∂].
The increased syncretism was not just due to sound change, but also analog-
ical extension. The {-as} plural, which already applied to some 60% of nouns,
was extended to many other nouns as well. This means that the vast majority of
nouns now had the same plural ending (which became {-es} with the reduction of
unstressed vowels) and also shared many -e endings elsewhere in the paradigm.
The clues that there were distinct inflectional classes become less and less clear,
which in turn results in further regularisation (Allen 1997, p. 69).
In the articles, we see analogical levelling too: the articles that started with s-
changed to start with þ-, as the other articles already did. There was variable syn-
cretism also here, which contributed to the spread of þe as a determiner that could
be used almost across the board. We saw before that the articles were important
in maintaining case distinctions with widespread syncretism in nominal inflection.
If case distinctions disappear also in articles, a case system becomes very difficult
for a learner of the language to acquire.
It is sometimes difficult to tell whether a certain development is the result of
phonological or morphological change. The loss of -n from case endings matches
a sound change that is also found elsewhere in the language, for example in verb
endings, but at the same time, we can arrive at the same end result by posit-
ing analogical extension of the ubiquitous -e ending. As Allen writes, ‘it seems
clear enough that the phonological tendencies of English supported the analogical
tendencies and speeded up the loss of some category distinctions’ (1997, p. 76).
We have already noted that the constituent order in Old English was not as
free as a case system would theoretically allow it to be. But as the case distinc-
tions became less clear, the burden of indicating the function of constituents in a
sentence fell increasingly on constituent order. The infrequent non-canonical con-
stituent order patterns therefore became even less frequent. With constituent order
giving even clear indications of the function of words in a sentence, there was less
need for maintaining case distinctions. This pattern becomes cyclical. Because con-
stituent order in Old English was already relatively fixed, this interaction between
syntax and morphology was probably less important than it was in parallel cases
in other languages.
Finally, contact may have played a role in the propagation of the loss of case
(though not so much its actuation). This means dialect contact between dialects
with slight differences in how the case endings had developed, but also language
contact with speakers of Old Norse. Old Norse had a very similar case system to
Old English, although the endings were different. Some examples from Table 11.2
 189

Table 11.3 Comparison of case patterns in Old English and Old Norse
‘boat’ ‘eye’ ‘goose’
Old English Old Norse Old English Old Norse Old English Old Norse
nom. sg bāt bát-r ēag-e aug-a gōs gás
gen. sg bāt-es bát-s ēag-an aug-a gōs-e gás-ar
dat. sg bāt-e bát-i ēag-an aug-a gēs gás
acc. sg bāt bát ēag-e aug-a gōs gás
nom. pl bāt-as bát-ar ēag-an aug-u gēs gæs-s
gen. pl bāt-a bát-a ēag-e aug-na gōs-a gás-a
dat. pl bāt-um bát-um ēag-um aug-um gōs-um gás-um
acc. pl bāt-as bát-a ēag-an aug-u gēs gæs-s
Note The standardised Old Norse forms given here are based on 13th-century Icelandic and there-
fore differ from the exact forms used by Old Norse speakers in England in the 9th and 10th
centuries

are repeated in Table 11.3 with their Old Norse equivalents, to illustrate both
the similarities and differences between the languages. A speaker of one lan-
guage trying to make themselves understood to a speaker of the other language
through accommodation will probably end up simplifying the case system signifi-
cantly. Loss of morphological distinction is a common outcome of accommodation
processes.
There is evidence from short, early Northumbrian and Kentish texts that these
developments start before, and therefore independently from the invasions and
subsequent settlement of the Vikings. When the Vikings had arrived, though, the
fact that their language, too, showed loss of final -n probably did contribute to the
spread of the ongoing change throughout the area of Norse settlement.
The loss of case did not proceed through the English-speaking area evenly: it
happened earlier in the north than in the south. For example, by the twelfth century,
the First and Second Continuations of the Peterborough Chronicle (a northern text)
show only minimal evidence for a dative/accusative distinction. It seems like the
scribe used a ‘dative’ form {-e} after a preposition, regardless of what case the
preposition actually governed in earlier language. But the later, thirteenth-century,
southern text Vices and Virtues does seem to have the dative as a separate category
from the accusative. The geographical location of the Norse settlement may have
something to do with these differences.
Overall, the case system of English collapsed because through sound change
and analogy, the forms became more and more similar, so—except for pronouns—
there was not enough evidence anymore to support the idea that there should be
different forms for the different functions. The nominative/accusative distinction
disappeared first, then the dative, and eventually only the genitive was left as a
specifically marked category. While the actuation was a question of phonolog-
ical and morphological change, syntactic change and contact contributed to the
propagation of the loss of case.
190 11 The Loss of Case

The Possessive as a Degrammaticalised Genitive

Although the genitive was eventually left as the only case-based category that
was still explicitly marked, it underwent a slightly different development from the
other case markings. It retained a separate form, different from the nominative,
accusative, and dative, but over time it lost the properties that qualified it as a case.
In present-day English, what was once the genitive case has become a possessive
clitic through a process of degrammaticalisation. This development is sketched
by Allen (2003). Although she does not explicitly make a degrammaticalisation
argument, the narrative and examples are based on her work.
In Early Middle English, there is still plenty of evidence that the genitive is
really a case. It is very widely applicable, also in cases where present-day English
would prefer an of -possessive (3). It is recursive, so you can have a genitive ending
on a phrase that already has a genitive ending within it (4). This is maybe gram-
matical in present-day English, but it is at the very least highly unusual. There
are also still some words (mostly plurals and irregular nouns) that use a different
genitive ending than {-s} (5, 6). When the head noun in a noun phrase is followed
by modifiers, the genitive is still marked on the head noun, and not after the com-
plete NP (7; Allen calls this construction the ‘combined genitive’). This evidence
together points at a true case status for the genitive at this time; these things are
not possible in, e.g., modern Dutch, where possessive {-s} is clearly no longer a
case.

3. ðis ferðe dais nigt

this.gen fourth day.gen night
‘the night of this fourth day’
4. þat he Constantines sones fur ssolde fle
that he Constantine.gen son.gen fire should flee
‘that he should flee Constantine’s son’s fire’
5. for Zho iss allre shaffte cwen
for she is all.gen creations.gen queen
‘for she is queen of all creatures’
6. hiss aZhenn broþerr wif
his own brother.gen wife
‘his own brother’s wife’
7. þurh þæs arcebiscopes gearnunge of Cantwerbyrig
through the.gen archbishop.gen yearning of Canterbury
‘through the desire of the archbishop of Canterbury’

But already in the same period, we see evidence of change. Previously, when the
possessor consisted of multiple words, they would all be marked with the genitive
case, such as King David in (8). But increasingly, we see examples where case
agreement within the noun phrase is lacking, such as King Herod in (9). With a
true genitive, we would expect Herod to be marked with a genitive ending -es
as well. From a (de)grammaticalisation perspective, this would be an example of
The Possessive as a Degrammaticalised Genitive 191

decategorialisation as the requirements for something to be a genitive case marker

seem to be changing.

8. Daviþess kingess kinnessmann

David.gen king.gen kinsman
‘King David’s kinsman’
9. upponn Herode kingess daZZ
upon Herod king.gen day
‘in King Herod’s day’

Another development is that the genitive ending started to be written separately

from the noun (10), so-called separated genitives. Although we must be wary of
spelling evidence, because changes in spelling need not be indicative of a change
in language itself, in this case, it does seem like these separated genitives go on
a different trajectory of change. From the late fourteenth century onwards, sep-
arated genitives no longer behave as a true genitive. We can see this when we
look at the behaviour of separated genitives in combined genitives. While we
can still find examples of possessor marking on the head noun (11), more and
more the marker is written after the complete NP (12). The separated genitives
are leading this change, which corresponds to a context generalisation change in
(de)grammaticalisation: the restrictions on where a marker can appear become less
strict. This means that at least the separated genitive has ceased to be an inflec-
tional marker and has been reanalysed as a clitic—it was still not an independent
word, but it is no longer attached to words of a particular word class but rather to
a full phrase.

10. ðe was Adam is sune

‘who was Adam’s son’
11. Dame Phylyppe, the kyng ys doughter of Fraunce
12. my seid lorde Bysshop of Exeter is commaundement

It is likely that the separated genitive had the same grammatical status as the
attached genitive, as is evident from the fact that it did not agree in gender and
number with the possessor: feminine and plural possessors get is as well, and not
a form of her or their (13, 14). In other words, it was a natural development of
the {-s} genitive marker and not an alternative form of the possessive pronoun his.
It is only in the sixteenth century that we see agreement pop up, and feminine
or plural possessors are marked with her and their instead of (h)is (15, 16). By
this time, though, these separated genitives are already on their way out of the
language again. Where they do occur, they always show agreement.

13. and the queen ys moder

14. … to your hurt and othyr men ys greate avaylle
15. but in his absence one Curio a gentleman of Naples of lyttle wealth and lesse
witte haunted Lucilla hir company
16. Beauty & agilitie their fame, hath their delyte
192 11 The Loss of Case

The construction with a possessive pronoun after the possessor also exists in other
Germanic languages, such as Dutch, German, and Norwegian, where it is gen-
erally seen as a non-standard, colloquial feature. This was not the case in Early
Modern English: (16) is an example from a letter by Queen Elizabeth I, for exam-
ple. Separated genitives with feminine or plural agreement are rare, though, and
most examples are with (h)is and masculine singular possessors. Allen (1997,
p. 19) suggests that literate people did, in fact, reanalyse the clitic is as a form
of the possessive pronoun his, and substituted the feminine and plural forms as a
hypercorrection.
In summary, Allen (2003, p. 21) sketches the development of from genitive case
in Middle English to possessive clitic in Early Middle English in three phases. In
the first phase (c. 1150–c. 1380), the marker is always adjacent to the possessor
noun and usually written attached to the noun. In this phase {-s} still more or
less behaves like a case ending. In the second phase (c. 1380–c. 1550), the marker
moves from the head noun in a possessor noun phrase to the end of the noun
phrase. It is predominantly written attached to the last word of the possessor noun
phrase, although separately written markers become increasingly common. This
is a transitional phase in the development. In the final phase (c. 1550–c. 1700),
the {-s} is always attached to the last word of the noun phrase. Separately written
markers disappear again; where they do appear, they tend to agree in number and
gender with the possessor.

In Other Languages
SEPARATED GENITIVES IN OTHER GERMANIC LANGUAGES
While the separated genitive—as in the queen her letter—constitutes a ‘failed
change’ in the history of English, we find parallel constructions in a num-
ber of other Germanic languages, even if the details differ somewhat from
language to language.
In Dutch, we find the same construction as in English in the later stages
of the construction: de koningin haar brief . The possessive pronoun agrees
with the possessor. We see the same thing in modern German, but here,
in addition, the possessor is in the dative case: der Königin ihr Brief . It is
clear that these constructions originated in a process of grammaticalisation
although the exact trajectory is less clear. A number of proposals have been
made, but none is fully supported by the earlier examples of this construction
from the historical record (Hendriks, 2012).
Norwegian also has a separated genitive, but here the possessive pronoun
agrees with the possessum instead: dronninga sitt brev. This construction has
its origins in dialects that experienced a lot of language contact with Low
German in the late Middle Ages, and its development is certainly influenced
by German. The differences in the agreement pattern can be explained by an
ongoing change in Norwegian possessive constructions at the time. As such
this is a form of grammatical replication (Norde, 2012) or pivot matching.
References 193

In Afrikaans, finally, the construction has an invariant possessive pronoun:

die koningin se brief . Early Afrikaans (Cape Dutch) followed the Dutch
construction, and possessor agreement is still found throughout the nine-
teenth century, although masculine agreement regardless of possessor is on
the rise. One of the earliest texts to show consistent masculine agreement is
an Islamic religious text from the late nineteenth century, written in Arabic
script. It would appear that the change to invariant se is a change from below,
potentially influenced by language contact (Roberge, 1996; Hendriks, 2010).
Interestingly, with the exception of Afrikaans and Austrian German, the
separated genitive is associated with a colloquial register and has only limited
acceptance in all of these standard languages. Nevertheless, the construction
has been more successful than in English.

Conclusion

The collapse of the English case system is a good example of how changes in
different parts of the language can combine and drive each other on, or in the evo-
lutionary framework of Ritt (2004), an example of meme co-adaptation. Regular
sound change, analogical change in morphology, accelerated by stricter syntac-
tic preferences, and accommodation processes in language and dialect started a
self-reinforcing cycle in which the case system progressively weakened. The last
remaining vestiges of the case system, in the form of the genitive, were removed
through a process of degrammaticalisation, itself also a combination of changes in
different linguistic subsystems.
It is exactly the combination of all these changes why the loss of case was
chosen as the final topic in this book.

References

Allen, Cynthia. 1997. Middle English case loss and the ‘creolization’ hypothesis. English Lan-
guage and Linguistics 1 (1): 63–89. https://doi.org/10.1017/S1360674300000368.
Allen, Cynthia L. 2003. Deflexion and the development of the genitive in English. English Lan-
guage and Linguistics 7 (1): 1–28. https://doi.org/10.1017/S1360674303211023.
Hendriks, Jennifer. 2010. Prenominal possessor doubling constructions in (West) Germanic:
Reassessing the evidence for grammaticalisation. In Grammatical change: Theory and descrip-
tion, eds. Rachel Hendery and Jennifer Hendriks, 27–48. Canberra: Pacific Linguistics.
Hendriks, Jennifer. 2012. Re-examining the ‘origins’ of the prenominal periphrastic possessive
construction Jan z’n boek in Dutch: An empirical approach. Diachronica 29 (1): 28–71. https://
doi.org/10.1075/dia.29.1.02hen.
Norde, Muriel. 2012. On the origin(s) of the possessor doubling construction in Norwegian. In Lan-
guage for its own sake: Essays on language and literature offered to Harry Perridon, ed. Henk
van der Liet and Muriel Norde, 327–358. Universiteit van Amsterdam.
Roberge, Paul T. 1996. Multilevel syncretism and the evolution of Afrikaans periphrastic posses-
sives with se. Stellenbosch Papers in Linguistics 30 (1): 123–153. https://doi.org/10.10520/AJA
0000015_101.
Ritt, Nikolaus. 2004. Selfish sounds and linguistic evolution: A Darwinian approach to language
change. Cambridge: Cambridge University Press.
Concluding Remarks
12

In this book, we set off to investigate the reasons why English changed over its
1500-year history and to find answers for the actuation question of a selected set of
changes in the language. To do this, we have taken not a chronological or reverse-
chronological tour through the history of English, nor given extensive synchronic
descriptions of different stages of the language—Old English, Middle English,
Early Modern English, etc.—but instead have taken theoretical explanations of
change as a starting point and used textual evidence from English as supporting
examples. As such, the book has not given an overview of the history of English,
but of language change in the history of English.
The three parts of the book, on sound change, contact-induced change, and
structural change, have also followed a traditional historical linguistic set-up,
focusing on separate parts of the language such as phonetics and phonology, mor-
phology, and syntax, with language contact as a wild card that is an ‘external’
rather than ‘internal’ driver of language change. The chapters in these three parts
have given different explanations for language, and as we look back, we may also
be able to categorise those in an alternative way.
First of all, there were physiological reasons for language change. We have seen
these in the discussion of sound change (phonetic change), where the biological
limitations of our articulatory system could be seen as the actuation reason for
s-retraction in /str/ clusters. It was also given as a reason for Grimm’s Law, and
many of the changes described in Chapter 5 can also be explained in this way.
The second set of reasons for language change was cognitive. Cognition is
potentially one of the most important drivers of change, as language at all differ-
ent levels is organised in our minds. We have a phonology, a sound system, in
which phonetic change can have knock-on effects exactly because we are trying
to organise sounds in a systematic way. This can explain chain shifts, for example.
We have a morphology, inflectional systems for nouns and verbs, where we look

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 195
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7_12
196 12 Concluding Remarks

for patterns that we sometimes overextend. The constraints on pattern productivity

show that it is exactly this we do: grabbing the most accessible patterns off the
shelf to construct word forms on the fly. And we have a syntactic system, where
we use or do not use certain constructions based on the abstract principles we
think we have spotted in the language around us.
We also find cognitive explanations in our discussion of language change. All
our linguistic material is stored in our brains as an integrated whole, and we pick
the material that is contextually appropriate as well as easy to grab from the shelf.
Sometimes that material is not labelled as ‘English’, and if we use it anyway, we
have actuated contact-induced change.
Finally, we have seen that language change may also be actuated in language
acquisition. Assuming that every person builds their language from scratch on the
basis of the available evidence around them, it is almost inevitable that everybody’s
language is slightly different, able to produce a slightly different set of utterances,
with the accumulation of these small differences sometimes triggering a significant
reanalysis. We have also seen that people’s mental linguistic representations may
change if the linguistic evidence around them changes, for example, in lifespan
change or second-dialect acquisition.
Once new features enter the world, the social meaning they acquire decides
whether they will spread; some features are socially promoted over others. In par-
ticular cases, such promotion may be a purposeful strategy, such as in the case
of standardisation. However, non-standard features may still spread and eventu-
ally end up as part of a standard variety, or survive indefinitely in addition to the
standard.
Language changes because it is being used by human beings in interaction.
With more interaction than ever in English, the language will continue to change.
In recent centuries, standardisation has hidden this change and standard language
ideology has even tried to halt language change. Today’s many online interactions
are based on written orality, non-standard language, so change is once again more
visible. And in our networked societies, change may spread faster than ever before.
There are many users of the English language, first-language or second-language
speakers, using a great many varieties of English. It is guaranteed that there will
be language change for future generations to study.
At the same time, we should not only look forward, but also back in time. There
are many unanswered questions of change in older stages of English. Much of the
source materials have now been made accessible in large corpora, and we are
much better placed to investigate Old and Middle English than even only twenty
years ago. Computational power, statistics, and exciting new approaches such as
cognitive linguistics and evolutionary linguistics give complementary insights to
traditional philology and the comparative method.
The history of English has a bright future ahead. I hope this book has been able
to inspire students to be a part of it.
References

Aalberse, Suzanne, and Wessel Stoop. 2015. The exceptional loss of the pronoun T. Journal of
Pragmatics 88: 190–201. https://doi.org/10.1016/j.pragma.2015.07.003.
Adams, Michael. 2013. Introduction: DARE from now on. American Speech 88 (3): 257–260.
https://doi.org/10.1215/00031283-2413957.
Ahlers, Wiebke, and Philipp Meer. 2019. Sibilant variation in new Englishes: A comparative
sociophonetic study of Trinidadian and American English /s(tr)/-retraction. In Proceedings of
Interspeech 2019, 291–295. https://doi.org/10.21437/Interspeech.2019-1821.
Allen, Cynthia. 1997. Middle English case loss and the ‘creolization’ hypothesis. English Lan-
guage and Linguistics 1 (1): 63–89. https://doi.org/10.1017/S1360674300000368.
Allen, Cynthia L. 2003. Deflexion and the development of the genitive in English. English Lan-
guage and Linguistics 7 (1): 1–28. https://doi.org/10.1017/S1360674303211023.
Anderson, Benedict. 2006. Imagined communities: Reflections on the origin and spread of
nationalism, 2nd ed. London: Verso.
Baker, Adam, Diana Archangeli, and Jeff Mielke. 2011. Variability in American English s-
retraction suggests a solution to the actuation problem. Language Variation and Change 23
(3): 347–374. https://doi.org/10.1017/S0954394511000135.
Barras, Will. 2006. The square–nurse merger in Greater Manchester: The impact of social and
spatial identity on phonological variation. MA thesis, University of Edinburgh.
Bauer, Laurie. 1992. The second Great Vowel Shift revisited. English World-Wide 13 (2): 253–268.
https://doi.org/10.1075/eww.13.2.04bau.
Bauer, Laurie. 2008. Lenition revisited. Journal of Linguistics 44 (3): 605–624. https://doi.org/10.
1017/S0022226708005331.
Baugh, Albert C. 1935. The chronology of French loan-words in English. Modern Language Notes
50 (2): 90–93.
Baugh, Albert C., and Thomas Cable. 2013. A history of the English language, 6th ed. London:
Routledge.
Baxter, Gareth J., Richard A. Blythe, William Croft, and Alan J. McKane. 2009. Modeling lan-
guage change: An evaluation of Trudgill’s theory of the emergence of New Zealand English.
Language Variation and Change 21 (2): 257–296. https://doi.org/10.1017/S09543945099
9010X.
Beal, Joan. 2012. Can’t see the wood for the trees? Corpora and the study of Late Modern English.
In Middle and Modern English corpus linguistics: A multi-dimensional approach, ed. Manfred
Markus, Yoko Iyeiri, Reinhard Heuberger, and Emil Chamson, 13–29. Amsterdam: Benjamins.
Becker, Kara, ed. 2019. The Low-Back-Merger Shift: Uniting the Canadian Vowel Shift, the Califor-
nia Vowel Shift, and short front vowel shifts across North America (Publications of the American
Dialect Society 104). Durham, NC: American Dialect Society.
Bell, Allan. 2001. Back in style: Reworking audience design. In Style and sociolinguistic variation,
ed. Penelope Eckert and John R. Rickford, 139–169. Cambridge: Cambridge University Press.
Berko, Jean. 1958. The child’s learning of English morphology. Word 14 (2–3): 150–177. https://
doi.org/10.1080/00437956.1958.11659661.

© The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer 197
Nature Switzerland AG 2023
R. Knooihuizen, The Linguistics of the History of English,
https://doi.org/10.1007/978-3-031-41692-7
198 References

Blanc, Haim. 1968. The Israeli koine as an emergent national standard. In Language problems of
developing nations, ed. Joshua A. Fishman, Charles A. Ferguson and Jyotirindra Das Gupta,
237–251. New York: Wiley.
Blevins, Juliette. 2004. Evolutionary phonology: The emergence of sound patterns. Cambridge:
Cambridge University Press.
Boberg, Charles. 2008. Regional phonetic differentiation in Standard Canadian English. Journal
of English Linguistics 36 (2): 129–154. https://doi.org/10.1177/0075424208316648.
Boberg, Charles. 2019. A closer look at the short front vowel shift in Canada. Journal of English
Linguistics 47 (2): 91–119. https://doi.org/10.1177/0075424219831353.
Bosworth, Joseph, Thomas Northcote Toller, Christ Sean, and Ondřej Tichy, eds. 2014. An Anglo-
Saxon dictionary online. Prague: Faculty of Arts, Charles University. https://bosworthtoller.
com/.
Brand, James, Jen Hay, Lynn Clark, Kevin Watson, and Márton. Sóskuthy. 2021. Systematic co-
variation of monophthongs across speakers of New Zealand English. Journal of Phonetics 88:
101096. https://doi.org/10.1016/j.wocn.2021.101096.
Brinton, Laurel J. 2008. The comment clause in English: Syntactic origins and pragmatic
development. Cambridge: Cambridge University Press.
Britain, David. 2010. Supralocal regional dialect levelling. In Language and identities, ed. Carmen
Llamas and Dominic Watt, 193–204. Edinburgh: Edinburgh University Press.
Britain, David. 2011. The heterogeneous homogenisation of dialects in England. Taal & Tongval
63 (1): 43–60. https://doi.org/10.5117/TET2011.1.BRIT.
Britain, David, and Kazuko Matsumoto. 2015. Palauan English. In Further studies in the lesser-
known varieties of English, ed. Jeffrey P. Williams, Edgar W. Schneider, Peter Trudgill, and
Daniel Schreier, 305–343. Cambridge: Cambridge University Press. https://doi.org/10.1017/
CBO9781139108652.014.
Britton, Derek. 1991. On Middle English she, sho: A Scots solution to an English problem. NOW-
ELE 17: 3–51. https://doi.org/10.1075/nowele.17.01bri.
Budts, Sara. 2022. A connectionist approach to analogy: On the modal meaning of periphrastic do
in Early Modern English. Corpus Linguistics and Linguistic Theory 18 (2): 337–364. https://
doi.org/10.1515/cllt-2019-0080.
Budts, Sara, and Peter Petré. 2016. Reading the intentions of be going to: On the subjectification of
future markers. Folia Linguistica Historica 37: 1–32. https://doi.org/10.1515/filh-2016-0001.
Burke, Peter. 2004. Languages and communities in Early Modern Europe. Cambridge: Cambridge
University Press.
Busse, Ulrich. 2002. Linguistic variation in the Shakespeare corpus: Morpho-syntactic variability
of second person pronouns. Amsterdam: Benjamins.
Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam:
Benjamins.
Bybee, Joan. 2003. Mechanisms of change in grammaticization: The role of frequency. In The
handbook of historical linguistics, ed. Brian D. Joseph and Richard D. Janda, 602–623. Oxford:
Blackwell.
Bybee, Joan. 2015. Language change. Cambridge: Cambridge University Press.
Bybee, Joan, and William Pagliuca. 1987. The evolution of future meaning. In Papers from the
7th International Conference on Historical Linguistics, ed. Anna Giacalone Ramat, Onofrio
Carruba, and Giuliano Bernini, 109–122. Amsterdam: Benjamins.
Bybee, Joan, William Pagliuca, and Revere D. Perkins. 1991. Back to the future. In Approaches to
grammaticalization, vol. 2, ed. Elizabeth Closs Traugott and Bernd Heine, 17–58. Amsterdam:
Benjamins.
Campbell, Lyle. 2004. Historical linguistics: An introduction, 2nd ed. Edinburgh: Edinburgh
University Press.
Campbell, Lyle, and Martha C. Muntzel. 1989. The structural consequences of language death. In
Investigating obsolescence: Studies in language contraction and death, ed. Nancy C. Dorian,
181–196. Cambridge: Cambridge University Press.
References 199

Chambers, J.K. 1992. Dialect acquisition. Language 68 (4): 673–705. https://doi.org/10.2307/

416850.
Chambers, J.K., and Peter Trudgill. 1998. Dialectology, 2nd ed. Cambridge: Cambridge University.
Cheng, Chin-Chuan, and William S.-Y. Wang. 1977. Tone change in Chao-zhou Chinese: A study
in lexical diffusion. In The lexicon in phonological change, ed. William S.-Y. Wang, 86–100.
Den Haag: Mouton. https://doi.org/10.1515/9783110802399.86.
Cheshire, Jenny. 2013. Grammaticalisation in social context: The emergence of a new English
pronoun. Journal of Sociolinguistics 17 (5): 608–633. https://doi.org/10.1111/josl.12053.
Cheshire, Jenny, Sue Fox, Paul Kerswill, and Eivind Torgersen. 2008. Ethnicity, friendship net-
work and social practices as the motor of dialect change: Linguistic innovation in London.
Sociolinguistica 22: 1–23. https://doi.org/10.1515/9783484605299.1.
Cheshire, Jenny, Paul Kerswill, Sue Fox, and Eivind Torgersen. 2011. Contact, the feature pool and
the speech community: The emergence of Multicultural London English. Journal of Sociolin-
guistics 15 (2): 151–196. https://doi.org/10.1111/j.1467-9841.2011.00478.x.
Clarke, Sandra, Ford Elms, and Amani Youssef. 1995. The third dialect of English: Some Canadian
evidence. Language Variation and Change 7 (2): 209–228. https://doi.org/10.1017/S09543945
00000995.
Cole, Marcelle. 2018. A native origin for Present-Day English they, their, them. Diachronica 35
(2): 165–209. https://doi.org/10.1075/dia.16026.col.
Corrigan, Karen P. 2022. Language, migration and identity on the island of Ireland through time
and space. Paper presented at Methods in Dialectology XVII, Mainz, August 1.
Coupland, Nikolas. 1984. Accommodation at work: Some phonological data and their implica-
tions. International Journal of the Sociology of Language 46: 49–70. https://doi.org/10.1515/
ijsl.1984.46.49.
Croft, William. 2000. Explaining language change: An evolutionary approach. Harlow: Longman.
Croft, William. 2008. Evolutionary linguistics. Annual Review of Anthropology 37: 219–234.
https://doi.org/10.1146/annurev.anthro.37.081407.085156.
Curzan, Anne. 2014. Fixing English: Prescriptivism and language history. Cambridge: Cambridge
University Press. https://doi.org/10.1017/CBO9781139107327.
Cuskley, Christine F., Francesca Colaiori, Claudio Castellano, Vittorio Loreto, Martina Pugliese,
and Francesca Tria. 2015. The adoption of linguistic rules in native and non-native speakers:
Evidence from a Wug task. Journal of Memory and Language 84: 205–223. https://doi.org/10.
1016/j.jml.2015.06.005.
Dammel, Antje. 2011. Konjugationsklassenwandel. Prinzipien des Ab-, Um-, und Ausbaus ver-
balflexivischer Allomorphie in germanischen Sprachen (Studia Linguistica Germanica 103).
Berlin: De Gruyter.
Dance, Richard. 2017. English contact: Norse. In The history of English. Volume 2: Old English,
ed. Laurel J. Brinton and Alexander Bergs, 202–219. De Gruyter. https://doi.org/10.1515/978
3110525304-011.
Dawkins, Richard. 1976. The selfish gene. Oxford: Oxford University Press.
De Caluwe, Johan. 2009. Tussentaal wordt omgangstaal in Vlaanderen. Nederlandse Taalkunde 14
(1): 8–25.https://doi.org/10.5117/NEDTAA2009.1.TUSS339.
De Caluwe, Johan, and Evelien Van Renterghem. 2011. Regiolectisering en de opkomst van tussen-
taal in Vlaanderen. Taal & Tongval 63 (1): 61–77.https://doi.org/10.5117/TET2011.1.DECA.
DeGraff, Michel. 2003. Against creole exceptionalism. Language 79 (2): 391–410. https://doi.org/
10.1353/lan.2003.0114.
DeGraff, Michel. 2004. Against creole exceptionalism (redux). Language 80 (4): 834–839. https://
doi.org/10.1353/lan.2004.0178.
DeGraff, Michel. 2005. Linguists’ most dangerous myth: The fallacy of creole exceptionalism.
Language in Society 34 (4): 533–591. https://doi.org/10.1017/S0047404505050207.
DeGraff, Michel. 2020. Toward racial justice in linguistics: The case of Creole studies (Response
to Charity Hudley et al.). Language 96 (4): e292–e306. https://doi.org/10.1353/lan.2020.0080.
Dekeyser, Xavier. 1986. Romance loans in Middle English: A re-assessment. In Linguistics across
historical and geographical boundaries: In honour of Jacek Fisiak on the occasion of his fiftieth
200 References

birthday, vol. 1, ed. Dieter Kastovsky and Aleksander Szwedek, 253–265. Berlin: Mouton de
Gruyter.
Denis, Derek, and Alexandra D’Arcy. 2018. Settler colonial Englishes are distinct from postcolo-
nial Englishes. American Speech 93 (1): 3–31. https://doi.org/10.1215/00031283-6904065.
Dixon, R.M.W. 2008. Australian aboriginal words in dictionaries: A history. International Journal
of Lexicography 21 (2): 129–152. https://doi.org/10.1093/ijl/ecn008.
Dollinger, Stefan. 2008. New-dialect formation in Canada: Evidence from the English modal
auxiliaries. Amsterdam: Benjamins. https://doi.org/10.1075/slcs.97.
Dollinger, Stefan. 2019. Creating Canadian English: The professor, the mountaineer, and a national
variety of English. Cambridge: Cambridge University Press. https://doi.org/10.1017/978110859
6862.
Dollinger, Stefan. 2022. Canadian English lexis and semantics: A historical-comparative resource
in contrastive, real-time perspective, 1683–2016. In Earlier North American Englishes, ed.
Merja Kytö and Lucia Siebers, 205–230. Amsterdam: Benjamins. https://doi.org/10.1075/veaw.
g66.09dol.
D’Onofrio, Annette, and Jaime Benheim. 2020. Contextualizing reversal: Local dynamics of the
Northern Cities Shift in a Chicago community. Journal of Sociolinguistics 24 (4): 469–491.
https://doi.org/10.1111/josl.12398.
Dorian, Nancy C. 1994. Varieties of variation in a very small place: Social homogeneity, prestige
norms, and linguistic variation. Language 70 (4): 631–696. https://doi.org/10.2307/416324.
Durkin, Philip. 2018. Exploring the penetration of loanwords in the core vocabulary of Middle
English: Carry as a test case. English Language and Linguistics 22 (2): 265–282. https://doi.
org/10.1017/S1360674318000138.
Dyer, Judy. 2002. “We all speak the same round here”: Dialect levelling in a Scottish-English com-
munity. Journal of Sociolinguistics 6 (1): 99–116. https://doi.org/10.1111/1467-9481.00179.
Eckert, Penelope. 2001. Style and social meaning. In Style and sociolinguistic variation, ed.
Penelope Eckert and John R. Rickford, 119–126. Cambridge: Cambridge University Press.
Edwards, Alison. 2016. English in the Netherlands: Functions, forms and attitudes. Amsterdam:
Benjamins.
Ellegård, Alvar. 1953. The auxiliary do: The establishment and regulation of its use in English.
Stockholm: Almqvist & Wiksell.
Emonds, Joseph E., and Jan Terje Faarlund. 2014. English: The language of the Vikings. Olomouc:
Palacký University.
Evison, Martin Paul. 2000. All in the genes? Evaluating the biological evidence of contact and
migration. In Cultures in contact: Scandinavian settlement in England in the ninth and tenth
centuries, ed. Dawn M. Hadley and Julian D. Richards, 277–294. Turnhout: Brepols.
Faarlund, Jan Terje, and Joseph E. Emonds. 2016. English as North Germanic. Language Dynamics
and Change 6 (1): 1–17. https://doi.org/10.1163/22105832-00601002.
Farrar, Kimberly, and Mari C. Jones. 2002. Introduction. In Language contact: The interplay of
internal, external and extra-linguistic factors, ed. Mari C. Jones and Edith Esch, 1–16. Berlin,
New York: Mouton de Gruyter.
Faulkner, Mark. 2020. Quantifying the consistency of “standard” Old English spelling. Transac-
tions of the Philological Society 118 (1): 192–205. https://doi.org/10.1111/1467-968X.12182.
Fellows-Jensen, Gillian. 2011. Light thrown by Scandinavian place-names on the Anglo-Saxon
landscape. In Place-names, language and the Anglo-Saxon landscape, ed. Nicholas J. Higham
and Martin J. Ryan, 69–83. Woodbridge: Boydell.
Ferguson, Charles A. 1959. Diglossia. Word 15 (2): 325–340. https://doi.org/10.1080/00437956.
1959.11659702.
Fischer, Olga. 2013. Contact and syntactic change in Old and Middle English. In English as a
contact language, ed. Daniel Schreier and Marianne Hundt, 18–40. Cambridge: Cambridge
University Press.
Fischer, Olga, Ans van Kemenade, Willem Koopman, and Wim van der Wurff. 2000. The syntax
of Early English. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO978051
1612312.
References 201

Ghyselen, Anne-Sophie. 2015. ‘Stabilisering’ van tussentaal? Het taalrepertorium in de Westhoek

als casus. Taal & Tongval 67 (1): 43–95. https://doi.org/10.5117/TET2015.1.GHYS.
Ghyselen, Anne-Sophie, and Gunther De Vogelaer. 2013. The impact of dialect loss on the
acceptance of tussentaal: The special case of West-Flanders in Belgium. In Language
(de)standardisation in Late Modern Europe: Experimental studies, ed. Tore Kristiansen and
Stefan Grondelaers, 153–170. Oslo: Novus.
Gooskens, Charlotte. 2005. Travel time as a predictor of linguistic distance. Dialectologia et
Geolinguistica 13: 38–62. https://doi.org/10.1515/dig.2005.2005.13.38.
Gooskens, Charlotte, and Wilbert Heeringa. 2014. The role of dialect exposure in receptive multi-
lingualism. Applied Linguistics Review 5 (1): 247–271. https://doi.org/10.1515/applirev-2014-
0011.
Gooskens, Charlotte, Vincent J. van Heuven, Jelena Golubović, Anja Schüppert, Femke Swarte,
and Stefanie Voigt. 2017. Mutual intelligibility between closely related languages in Europe.
International Journal of Multilingualism 15 (2): 169–193. https://doi.org/10.1080/14790718.
2017.1350185.
Gries, Stefan Th., and Anatol Stefanowitsch. 2004. Extending collostructional analysis: A corpus-
based perspective on ‘alternations’. International Journal of Corpus Linguistics 9 (1): 97–
129https://doi.org/10.1075/ijcl.9.1.06gri.
Grieve, Jack, Chris Montgomery, Andrea Nini, Akira Murakami, and Diansheng Guo. 2019. Map-
ping lexical dialect variation in British English using Twitter. Frontiers in Artificial Intelligence
2: 11. https://doi.org/10.3389/frai.2019.00011.
Grieve, Jack, Andrea Nini, and Diansheng Guo. 2018. Mapping lexical innovation on American
social media. Journal of English Linguistics 46 (4): 293–319. https://doi.org/10.1177/007542
4218793191.
Grund, Peter J. 2021. The sociopragmatics of stance: Community, language, and the witness
depositions from the Salem witch trials. Amsterdam: Benjamins. https://doi.org/10.1075/pbn
s.329.
Hadley, Dawn M. 2000. The Northern Danelaw: Its social structure, c. 800–1100. London:
Leicester University Press.
Halsall, Guy. 2000. The Viking presence in England? The burial evidence reconsidered. In Cultures
in contact: Scandinavian settlement in England in the ninth and tenth centuries, ed. Dawn M.
Hadley and Julian D. Richards, 259–276. Turnhout: Brepols.
Härke, Heinrich. 2002. Kings and warriors: Population and landscape from post-Roman to Norman
Britain. In The peopling of Britain: The shaping of a human landscape, ed. Paul Slack and Ward
Ryk, 145–175. Oxford: Oxford University Press.
Harrington, Jonathan, Michele Gubian, Mary Stevens, and Florian Schiel. 2019. Phonetic change
in an Antarctic winter. Journal of the Acoustical Society of America 146 (5): 3327–3332. https://
doi.org/10.1121/1.5130709.
Harrington, Jonathan, Sallyanne Palethorpe, and Catherine Watson. 2000. Monophthongal vowel
changes in Received Pronunciation: An acoustic analysis of the Queen’s Christmas broadcasts.
Journal of the International Phonetic Association 30 (1–2): 63–78. https://doi.org/10.1017/S00
25100300006666.
Hay, Jennifer B., Janet B. Pierrehumbert, Abby J. Walker, and Patrick LaShell. 2015. Tracking
word frequency effects through 130 years of sound change. Cognition 139: 83–91. https://doi.
org/10.1016/j.cognition.2015.02.012.
Heine, Bernd. 2003. Grammaticalization. In The handbook of historical linguistics, ed. Brian D.
Joseph and Richard D. Janda, 575–601. Oxford: Blackwell.
Hendriks, Jennifer. 2010. Prenominal possessor doubling constructions in (West) Germanic:
Reassessing the evidence for grammaticalisation. In Grammatical change: Theory and descrip-
tion, ed. Rachel Hendery and Jennifer Hendriks, 27–48. Canberra: Pacific Linguistics.
Hendriks, Jennifer. 2012. Re-examining the ‘origins’ of the prenominal periphrastic possessive
construction Jan z’n boek in Dutch: An empirical approach. Diachronica 29 (1): 28–71. https://
doi.org/10.1075/dia.29.1.02hen.
202 References

Heycock, Caroline, Antonella Sorace, Zakaris Svabo Hansen, and Frances Wilson. 2013. Acqui-
sition in variation (and vice versa): V-to-T in Faroese children. Language Acquisition 20 (1):
5–22. https://doi.org/10.1080/10489223.2012.738741.
Heycock, Caroline, Antonella Sorace, Zakaris Svabo Hansen, Frances Wilson, and Sten Vikner.
2012. Detecting the late stages of syntactic change: The loss of V-to-T in Faroese. Language
88 (3): 558–600. https://doi.org/10.1353/lan.2012.0053.
Hickey, Raymond. 2007. Irish English: History and present-day forms. Cambridge: Cambridge
University Press.
Hickey, Raymond. 2012. Early English and the Celtic hypothesis. In The Oxford handbook of
the history of English, ed. Terttu Nevalainen and Elizabeth Closs Traugott, 497–507. Oxford:
Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199922765.013.0042.
Hickey, Raymond. 2020. English and Celtic: Contact-induced change in history. Linguistics Van-
guard 6 (s2): 20190037. https://doi.org/10.1515/lingvan-2019-0037.
Hill, Eugen. 2010. A case study in grammaticalized inflectional morphology: Origin and devel-
opment of the Germanic weak preterite. Diachronica 27 (3): 411–458. https://doi.org/10.1075/
dia.27.3.02hil.
Hilpert, Martin. 2008. Germanic future constructions: A usage-based approach to language change.
Amsterdam: Benjamins. https://doi.org/10.1075/cal.7.
Hock, Hans Heinrich. 2003. Analogical change. In The handbook of historical linguistics, ed. Brian
D. Joseph and Richard D. Janda, 441–460. Oxford: Blackwell.
Hock, Hans Heinrich. 2021. Principles of historical linguistics (3rd ed.). Berlin: Mouton de
Gruyter. https://doi.org/10.1515/9783110746440.
Hopper, Paul J., and Elizabeth Closs Traugott. 2003. Grammaticalization, 2nd ed. Cambridge:
Cambridge University Press.
Horvath, Barbara M., and Ronald J. Horvath. 2001. A geolinguistics of short A in Australian
English. In English in Australia, ed. David Blair and Peter Collins, 341–355. Amsterdam:
Benjamins.
Howell, Robert B. 2006. Immigration and koineisation: The formation of Early Modern Dutch
urban vernaculars. Transactions of the Philological Society 104 (2): 207–227. https://doi.org/
10.1111/j.1467-968X.2006.00169.x.
Huber, Magnus, Magnus Nissel, and Karin Puga. 2016. Old Bailey corpus 2.0. http://hdl.handle.
net/11858/00-246C-0000-0023-8CFB-2.
Hundt, Marianne. 1998. New Zealand English grammar: Fact or fiction? A corpus-based study in
morphosyntactic variation. Amsterdam: Benjamins. https://doi.org/10.1075/veaw.g23.
Hutton, Christopher M. 1999. Linguistics and the Third Reich: Mother-tongue fascism, race and
the science of language. London: Routledge.
Ingham, Richard. 2014. The maintenance of French in later medieval England. Neuphilologische
Mitteilungen 115 (4): 425–448.
Jenkins, Jennifer. 2002. A sociolinguistically based, empirically researched pronunciation syllabus
for English as an International Language. Applied Linguistics 23 (1): 83–103. https://doi.org/
10.1093/applin/23.1.83.
Johnson, Daniel Ezra. 2010. Stability and change along a dialect boundary: The low vowels of
Southeastern New England (Publications of the American Dialect Society 95). Durham, NC:
Duke University Press.
Jones, Charles. 1989. A history of English phonology. London: Longman.
Kaplan, Abby. 2010. Phonology shaped by phonetics: The case of intervocalic lenition. University
of California (UCSC) dissertation, Santa Cruz.
Kerswill, Paul. 2002. Koineization and accommodation. In The handbook of language variation
and change, ed. J. K. Chambers, Peter Trudgill, and N. Schilling-Estes, 669–702. Oxford:
Blackwell.
Kerswill, Paul. 2003. Dialect levelling and geographical diffusion in British English. In Social
dialectology: In honour of Peter Trudgill, ed. David Britain and Jenny Cheshire, 223–243.
Amsterdam: Benjamins.
References 203

Kerswill, Paul, and Eivind Torgersen. 2020. Tracing the origins of an urban youth vernacular:
Founder effects, frequency, and culture in the emergence of Multicultural London English. In
Advancing socio-grammatical variation and change: In honour of Jenny Cheshire, ed. Karen
V. Beaman, Isabelle Buchstaller, Sue Fox, and James A. Walker, 249–276. London: Routledge.
https://doi.org/10.4324/9780429282720-17.
Kerswill, Paul, and Ann Williams. 2000. Creating a new town koine: Children and language change
in Milton Keynes. Language in Society 29 (1): 65–115. https://doi.org/10.1017/S00474045000
01020.
Knooihuizen, Remco. 2009. Shetland Scots as a new dialect: Phonetic and phonological consider-
ations. English Language and Linguistics 13 (3): 483–501. https://doi.org/10.1017/S13606743
09990207.
Knooihuizen, Remco, and Oscar Strik. 2014. Relative productivity potentials of Dutch verbal
inflection patterns. Folia Linguistica Historica 35: 173–200. https://doi.org/10.1515/flih.201
4.005.
Kortmann, Bernd. 2016. The Viking Hypothesis from a dialectologist’s perspective. Language
Dynamics and Change 6 (1): 27–30. https://doi.org/10.1163/22105832-00601006.
Kretzschmar, William A., Jr. 2015. Language and complex systems. Cambridge: Cambridge Uni-
versity Press.
Kroch, Anthony S. 2001. Syntactic change. In The handbook of contemporary syntactic theory, ed.
Mark Baltin and Chris Collins, 629–739. Oxford: Blackwell.
Kuryłowicz, Jerzy. 1965. The evolution of grammatical categories. Diogenes 51: 55–71. https://
doi.org/10.1177/039219216501305105.
Kuteva, Tania, Bernd Heine, Bo Hong, Haiping Long, Heiko Narrog, and Seongha Rhee. 2019.
World lexicon of grammaticalization, 2nd ed. Cambridge: Cambridge University Press.
Kytö, Merja, Peter J. Grund, and Terry Walker. 2011. Testifying to language and life in Early
Modern England. Amsterdam: Benjamins.
Labov, William. 1994. Principles of linguistic change. Volume 1: Internal factors. Oxford:
Blackwell.
Labov, William. 2001. Principles of linguistic change. Volume 2: Social factors. Oxford: Blackwell.
Labov, William. 2007. Transmission and diffusion. Language 83 (3): 344–387. https://doi.org/10.
1353/lan.2007.0082.
Labov, William, Sharon Ash, and Charles Boberg. 2006. The Atlas of North American English:
Phonetics, phonology and sound change. Berlin: Mouton de Gruyter. https://doi.org/10.1515/
9783110167467.
Labov, William, Ingrid Rosenfelder, and Josef Fruehwald. 2013. One hundred years of sound
change in Philadelphia: Linear incrementation, reversal, and reanalysis. Language 89 (1): 30–
65.
Laing, Margaret, and Roger Lass. 2014. On Middle English she, sho: A refurbished narrative. Folia
Linguistica Historica 35: 201–240. https://doi.org/10.1515/flih.2014.006.
Lass, Roger. 1992. What, if anything, was the Great Vowel Shift? In History of Englishes: New
methods and interpretations in historical linguistics, ed. Matti Rissanen, Ossi Ihalainen, Terttu
Nevalainen, and Irma Taavitsainen, 145–155. Berlin: Mouton de Gruyter.
Lawrence, Wayne P. 2000. /str/ → /∫tr/: Assimilation at a distance? American Speech 75 (1): 82–87.
https://doi.org/10.1215/00031283-75-1-82.
Lawson, Eleanor, and Jane Stuart-Smith. 2021. Lenition and fortition of /r/ in utterance-final posi-
tion: An ultrasound tongue imaging study of lingual gesture timing in spontaneous speech.
Journal of Phonetics 86: 101053. https://doi.org/10.1016/j.wocn.2021.101053.
Lawson, Eleanor, Jane Stuart-Smith, and James M. Scobbie. 2018. The role of gesture delay in
coda /r/ weakening: An articulatory, auditory and acoustic study. Journal of the Acoustical
Society of America 143 (3): 1646–1657. https://doi.org/10.1121/1.5027833.
Lefebvre, Claire. 2004. Issues in the study of pidgin and creole languages. Amsterdam: Benjamins.
Lightfoot, David W. 1979. Principles of diachronic syntax. Cambridge: Cambridge University
Press.
204 References

Lightfoot, David W. 1989. The child’s trigger experience: Degree-0 learnability. Behavioral and
Brain Sciences 12 (2): 321–334. https://doi.org/10.1017/S0140525X00048883.
Lightfoot, David W. 1991. How to set parameters: Arguments from language change. Cambridge,
MA: MIT Press.
Lightfoot, David W. 2003. Grammatical approaches to syntactic change. In The handbook of
historical linguistics, ed. Brian D. Joseph and Richard D. Janda, 495–508. Oxford: Blackwell.
Lightfoot, David W. 2013. Types of explanation in history. Language 89 (4): e18–e38.
Lindqvist, Christer. 2003. Thesen zur Kausalität und Chronologie einiger färöischer Lautgesetze.
Arkiv för nordisk filologi 118: 89–178.
Lindsey, Geoff. 2019. English after RP: Standard British pronunciation today. London: Palgrave
Macmillan.
Los, Bettelou. 2015. A historical syntax of English. Edinburgh: Edinburgh University Press.
Los, Bettelou. 2016. Rule 1: Know your stuff. Language Dynamics and Change 6 (1): 31–33.
https://doi.org/10.1163/22105832-00601007.
Lutz, Angelika. 2009. Celtic influence on Old English and West Germanic. English Language and
Linguistics 13 (2): 227–249. https://doi.org/10.1017/S1360674309003001.
Mæhlum, Brit. 1996. Semi-migration in the Arctic: A theoretical perspective on the dialect strate-
gies of children on Spitsbergen. In Language contact across the North Atlantic, ed. P. Sture
Ureland and Iain Clarkson, 313–331. Tübingen: Niemeyer.
Maguire, Warren, Lynn Clark, and Kevin Watson. 2013. Introduction: What are mergers and can
they be reversed? English Language and Linguistics 17 (2): 229–239. https://doi.org/10.1017/
S1360674313000014.
Markey, Thomas L. 1982. Afrikaans: Creole or non-creole? Zeitschrift für Dialektologie nd Lin-
guistik 49: 169–207.
Martinet, André. 1952. Function, structure, and sound change. Word 8 (1): 1–32. https://doi.org/
10.1080/00437956.1952.11659416.
Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press.
McMahon, April. 1994. Understanding language change. Cambridge: Cambridge University Press.
McWhorter, John H. 2009. What else happened to English? A brief for the Celtic hypothesis.
English Language and Linguistics 13 (2): 163–191. https://doi.org/10.1017/S13606743090
02974.
Meakins, Felicity. 2013. Mixed languages. In Contact languages: A comprehensive guide, ed. Peter
Bakker and Yaron Matras, 159–228. Mouton de Gruyter. https://doi.org/10.1515/978161451
3711.159.
Meehan, Teresa. 1991. It’s like, “What’s happening in the evolution of like?”: A theory of gram-
maticalisation. Kansas Working Papers in Linguistics 16: 37–51. https://doi.org/10.17161/
KWPL.1808.423.
Meillet, Antoine. 1921. L’évolution des formes grammaticales. In Linguistique historique et lin-
guistique générale, vol. 2, 130–148. Paris: Klinksieck.
Mesthrie, Rajend, ed. 2008. Varieties of English 4: Africa, South and Southeast Asia. Berlin:
Mouton de Gruyter. https://doi.org/10.1515/9783110208429.
Meyer, Heinrich. 1901. Über den Ursprung der germanischen Lautverschiebung. Zeitschrift für
deutsches altertum und deutsches Litteratur 45: 101–128.
Meyerhoff, Miriam. 2006. Linguistic change, sociohistorical context, and theory-building in vari-
ationist linguistics: New-dialect formation in New Zealand. English Language and Linguistics
10 (1): 173–194. https://doi.org/10.1017/S1360674306001833.
Meyerhoff, Miriam. 2009. Replication, transfer, and calquing: Using variation as a tool in the study
of language contact. Language Variation and Change 21 (3): 297–317. https://doi.org/10.1017/
S0954394509990196.
Meyerhoff, Miriam. 2019. Introducing sociolinguistics, 3rd ed. London: Routledge.
Millar, Robert McColl. 2008. The origins and development of Shetland dialect in light of dialect
contact theories. English World-Wide 29 (3): 237–267. https://doi.org/10.1075/eww.29.3.02mil.
Millar, Robert McColl. 2010. The death of Orkney Norn and the genesis of Orkney Scots. Scottish
Language 29: 16–36.
References 205

Milroy, James, and Lesley Milroy. 1985. Linguistic change, social network and speaker innovation.
Journal of Linguistics 21 (2): 339–384. https://doi.org/10.1017/S0022226700010306.
Minkova, Donka. 2014. A historical phonology of English. Edinburgh: Edinburgh University Press.
Mitchell, Alexander G. 1958. Spoken English. London: Macmillan.
Montemagni, Simonetta, Martijn Wieling, Bob de Jonge, and John Nerbonne. 2013. Synchronic
patterns of Tuscan phonetic variation and diachronic change: Evidence from a dialectometric
study. Literary and Linguistic Computing 28 (1): 157–172. https://doi.org/10.1093/llc/fqs057.
Moreton, Elliott, and Erik R. Thomas. 2007. Origins of Canadian raising in voiceless-coda effects:
A case study in phonologization. In Laboratory phonology 9, ed. Jennifer Cole and Jose I.
Hualde, 37–64. Berlin: Mouton de Gruyter.
Mossé, Fernand. 1959. Manuel de l’anglais du moyen âge des origines au XIVe siècle. II: Moyen-
anglais, 3rd ed. Paris: Aubier-Montaigne.
Moulton, William G. 1962. Dialect geography and the concept of phonological space. Word 18
(1–3): 23–32. https://doi.org/10.1080/00437956.1962.11659763.
Mufwene, Salikoko S. 2001. The ecology of language evolution. Cambridge: Cambridge University
Press.
Mufwene, Salikoko S. 2008. Language evolution: Contact, competition and change. London:
Continuum.
Nagy, Naomi, Nina Aghdasi, Derek Denis, and Alexandra Motut. 2011. Null subjects in heritage
languages: Contact effects in a cross-linguistic context. University of Pennsylvania Working
Papers in Linguistics 17: 135–144. https://repository.upenn.edu/pwpl/vol17/iss2/16.
Natvig, David. 2020. Rhotic underspecification: Deriving variability and arbitrariness through
phonological representations. Glossa: A Journal of General Linguistics 5 (1): 48. https://doi.
org/10.5334/gjgl.1172.
Natvig, David, and Joseph Salmons. 2021. Connecting structure and variation in sound change.
Cadernos de Linguística 2 (1): 1–20. https://doi.org/10.25189/2675-4916.2021.V2.N1.ID314.
Nerbonne, John. 2009. Data-driven dialectology. Language and Linguistics Compass 3 (1): 175–
198. https://doi.org/10.1111/j.1749-818x.2008.00114.x.
Nerbonne, John, and Wilbert Heeringa. 2007. Geographic distributions of linguistic variation
reflect dynamics of differentiation. In Roots: Linguistics in search of its evidential base, ed.
Sam Featherston and Wolfgang Sternefeld, 267–298. Berlin: Mouton de Gruyter. https://doi.
org/10.1515/9783110198621.267.
Nevalainen, Terttu, and Helena Raumolin-Brunberg. 2003. Historical sociolinguistics: Language
change in Tudor and Stuart England. London: Longman.
Norde, Muriel. 2012. On the origin(s) of the possessor doubling construction in Norwegian. In Lan-
guage for its own sake: Essays on language and literature offered to Harry Perridon, ed. Henk
van der Liet and Muriel Norde, 327–358. Universiteit van Amsterdam.
Nycz, Jennifer. 2013a. Changing words or changing rules? Second dialect acquisition and phono-
logical representation. Journal of Pragmatics 52: 49–62. https://doi.org/10.1016/j.pragma.
2012.12.014.
Nycz, Jennifer. 2013b. New contrast acquisition: Methodological issues and theoretical implica-
tions. English Language and Linguistics 17 (2): 325–357. https://doi.org/10.1017/S13606743
13000051.
Nycz, Jennifer. 2015. Second dialect acquisition: A sociophonetic perspective. Language and
Linguistics Compass 9 (11): 469–482. https://doi.org/10.1111/lnc3.12163.
Ohala, John J. 2003. Phonetics and historical phonology. In The handbook of historical linguistics,
ed. Brian D. Joseph and Richard D. Janda, 669–686. Oxford: Blackwell.
Orton, Harold, Stewart Sanderson, and J.D.A. Widdowson, eds. 1978. The Linguistic Atlas of
England. London: Routledge.
Pérez, Aveline. 1990. Time in motion: Grammaticalisation of the be going to construction in
English. La Trobe University Working Papers in Linguistics 3: 49–64.
Pierce, Marc. 2014. The further degrammaticalization of -ish. American Speech 89 (1): 115–118.
https://doi.org/10.1215/00031283-2726431.
206 References

Pierce, Marc. 2015. More on ish. American Speech 90 (3): 394–398. https://doi.org/10.1215/000
31283-3441047.
Pintzuk, Susan. 1993. Verb seconding in Old English: Verb movement to Infl. The Linguistic
Review 10 (1): 5–35. https://doi.org/10.1515/tlir.1993.10.1.5.
Pintzuk, Susan. 1996. Old English verb-complement word order and the change from OV to VO.
York Papers in Linguistics 17: 241–264.
Pintzuk, Susan. 2003. Variationist approaches to syntactic change. In The handbook of historical
linguistics, ed. Brian D. Joseph and Richard D. Janda, 509–528. Oxford: Blackwell.
Podesva, Robert J. 2011. The California vowel shift and gay identity. American Speech 86 (1):
32–51. https://doi.org/10.1215/00031283-1277501.
Podesva, Robert J., Annette D’Onofrio, Janneke Van Hofwegen, and Seung Kyung Kim. 2015.
Country ideology and the California Vowel Shift. Language Variation and Change 27 (2):
157–186. https://doi.org/10.1017/S095439451500006X.
Poplack, Shana, and Sali A. Tagliamonte. 2000. The grammaticization of going to in (African
American) English. Language Variation and Change 11 (3): 315–342. https://doi.org/10.1017/
S0954394599113048.
Richards, Julian D. 2000. Identifying Anglo-Scandinavian settlements. In Cultures in contact:
Scandinavian settlement in England in the ninth and tenth centuries, ed. Dawn M. Hadley and
Julian D. Richards, 295–309. Turnhout: Brepols.
Rickford, John R., and Sharese King. 2016. Language and linguistics on trial: Hearing Rachel Jean-
tel (and other vernacular speakers) in the courtroom and beyond. Language 92 (4): 948–988.
https://doi.org/10.1353/lan.2016.0078.
Ritt, Nikolaus. 2004. Selfish sounds and linguistic evolution: A Darwinian approach to language
change. Cambridge: Cambridge University Press.
Roberge, Paul T. 1996. Multilevel syncretism and the evolution of Afrikaans periphrastic posses-
sives with se. Stellenbosch Papers in Linguistics 30 (1): 123–153. https://doi.org/10.10520/AJA
0000015_101.
Roberts, Ian. 2007. Diachronic syntax. Oxford: Oxford University Press.
Roberts, Ian. 2022. Diachronic syntax, 3rd ed. Oxford: Oxford University Press.
Roeder, Rebecca V., and Matt Hunt Gardner. 2013. The phonology of the Canadian Shift revisited:
Thunder Bay Cape Breton. University of Pennsylvania Working Papers in Linguistics 19 (2):
161–170. https://repository.upenn.edu/pwpl/vol19/iss2/18.
Roeder, Rebecca V., and Matt Hunt Gardner. 2022. A unified account of the low back merger shift.
Paper presented at Methods in Dialectology XVII, Mainz, August 1.
Romaine, Suzanne, and Deborah Lange. 1991. The use of like as a marker of reported speech and
thought: A case of grammaticalization in progress. American Speech 66 (3): 227–279.
Rosenthal, B., ed. 2009. Records of the Salem Witch-Hunt. Cambridge: Cambridge University
Press. https://doi.org/10.1017/9781107589766.
Rutter, Ben. 2011. Acoustic analysis of a sound change in progress: The consonant cluster /stô/ in
English. Journal of the International Phonetic Association 41 (1): 27–40. https://doi.org/10.
1017/S0025100310000307.
Salmons, Joseph C., and Thomas Purnell. 2020. Contact and the development of American English.
In The handbook of language contact, 2nd ed., ed. Raymond Hickey, 361–383. Hoboken, NJ:
Wiley-Blackwell.
Sankoff, Gillian. 2019. Language change across the lifespan: Three trajectory types. Language 95
(2): 197–229. https://doi.org/10.1353/lan.2019.0029.
Sankoff, Gillian, and Hélène Blondeau. 2007. Language change across the lifespan: /r/ in Montreal
French. Language 83 (3): 560–588. https://doi.org/10.1353/lan.2007.0106.
Schendl, Herbert. 1996. Text types and code-switching in medieval and Early Modern English.
VIEWS: Vienna English Working Papers 5 (1–2): 50–62.
Schendl, Herbert. 1997. ‘To London fro Kent/Sunt predia depopulantes’: Code-switching and
medieval English macaronic poems. VIEWS: Vienna English Working Papers 6 (1): 52–66.
References 207

von Schneidemesser, Luanne. 2002. Settlement history in the United States as reflected in DARE:
The example of German. American Speech 77 (4): 398–418. https://doi.org/10.1215/00031283-
77-4-398.
Schneider, Edgar W. 2003. The dynamics of new Englishes: From identity construction to dialect
birth. Language 79 (2): 233–281.
Schneider, Edgar W. 2011. English around the world: An introduction. Cambridge: Cambridge
University Press.
Schreier, Daniel. 2002. Terra incognita in the anglophone world: Tristan da Cunha, South Atlantic
Ocean. English World-Wide 23 (1): 1–29. https://doi.org/10.1075/eww.23.1.02sch.
Schreier, Daniel. 2003. Isolation and language change: Contemporary and sociohistorical evidence
from Tristan da Cunha English. Basingstoke, New York: Palgrave Macmillan.
Schreier, Daniel. 2004. English transported to the South Atlantic Ocean: Tristan da Cunha. In Lega-
cies of colonial English, ed. Raymond Hickey, 387–401. Cambridge: Cambridge University
Press.
Schreier, Daniel. 2015. Historical phonology and koinéization. In The Oxford handbook of his-
torical phonology, ed. Patrick Honeybone and Joseph Salmons, 619–636. Oxford: Oxford
University Press. https://doi.org/10.1093/oxfordhb/9780199232819.013.037.
Schreier, Daniel, and Peter Trudgill. 2006. The segmental phonology of nineteenth-century Tristan
da Cunha English: Convergence and local innovation. English Language and Linguistics 10 (1):
119–141. https://doi.org/10.1017/S136067430600181X.
Schrodt, Richard. 1974. Die germanische Lautverschiebung und ihre Stellung im Kreise der
indogermanischen Sprachen. Wien: Halosar.
Schüppert, Anja, and Charlotte Gooskens. 2011. The role of extra-linguistic factors in recep-
tive bilingualism: Evidence from Danish and Swedish pre-schoolers. International Journal of
Bilingualism 16 (3): 332–347. https://doi.org/10.1177/1367006911426389.
Schüppert, Anja, Nanna Haug Hilton, and Charlotte Gooskens. 2015. Swedish is beautiful, Dan-
ish is ugly? Investigating the link between language attitudes and spoken word recognition.
Linguistics 53 (2): 375–403. https://doi.org/10.1515/ling-2015-0003.
Sebregts, Koen. 2014. The sociophonetics and phonology of Dutch r. Utrecht: Landelijke Onder-
zoekschool Taalwetenschap.
Seidlhofer, Barbara. 2011. Understanding English as a Lingua Franca. Oxford: Oxford University
Press.
Sekeres, Hedwig. 2022. Out of sight, out of mind? A cognitive geographic approach to dialectology.
Rijksuniversiteit Groningen MA thesis, Groningen.
Siegel, Jeff. 2001. Koine formation and creole genesis. In Creolization and contact, ed. Norval
Smith and Tonjes Veenstra, 175–197. Amsterdam: Benjamins.
Siegel, Jeff. 2010. Second dialect acquisition. Cambridge: Cambridge University Press.
Smith, Jeremy J. 2007. Sound change and the history of English. Oxford: Oxford University Press.
Smyth, Alfred P. 1998. The emergence of English identity, 700–1000. In Medieval Europeans, ed.
Alfred P. Smyth, 24–52. London: Macmillan. https://doi.org/10.1007/978-1-349-26610-4_2.
Sonderegger, Morgan, Max Bane, and Peter Graff. 2017. The medium-term dynamics of accents
on reality television. Language 93 (3): 598–640. https://doi.org/10.1353/lan.2017.0038.
Sóskuthy, Márton. 2015. Understanding change through stability: A computational study of sound
change actuation. Lingua 163: 40–60. https://doi.org/10.1016/j.lingua.2015.05.010.
Stevens, Mary, and Jonathan Harrington. 2016. The phonetic origins of /s/-retraction: Acoustic and
perceptual evidence from Australian English. Journal of Phonetics 58: 118–134. https://doi.org/
10.1016/j.wocn.2016.08.003.
Stockwell, Robert P., and Donka Minkova. 1988. The English vowel shift: Problems of coherence
and explanation. In Luick revisited: Papers read at the Luick-Symposium at Schloß Liechtenstein,
15.–18.9.1985, ed. Dieter Kastovsky and Gero Bauer, 355–394. Tübingen: Narr.
Strik, Oscar. 2015. Modelling analogical change: A history of Swedish and Frisian verb inflection.
Rijksuniversiteit Groningen dissertation, Groningen.
208 References

Strycharczuk, Patrycja, Manuel López-Ibáñez, Georgina Brown, and Adrian Leemann. 2020. Gen-
eral Northern English: Exploring regional variation in the north of England with machine
learning. Frontiers in Artificial Intelligence 3: 48. https://doi.org/10.3389/frai.2020.00048.
Sturtevant, Edgar H. 1947. An introduction to linguistic science. New Haven: Yale University Press.
Sudbury, Angela. 2004. English on the Falklands. In Legacies of colonial English, ed. Raymond
Hickey, 402–417. Cambridge: Cambridge University Press.
Sweet, Henry. 1900. The history of language. London: Dent.
Tagliamonte, Sali A., Mercedes Durham, and Jennifer Smith. 2014. Grammaticalization at an early
stage: Future be going to in conservative British dialects. English Language and Linguistics 18
(1): 75–108. https://doi.org/10.1017/S1360674313000282.
Tagliamonte, Sali A., and Sonja Molfenter. 2007. How’d you get that accent? Acquiring a second
dialect of the same language. Language in Society 36 (5): 649–675. https://doi.org/10.1017/S00
47404507070911.
Tamminga, Meredith. 2021. Leaders of language change: Macro and micro perspectives. In Lan-
guage variation—European perspectives: Selected papers from the Tenth International Confer-
ence on Language Variation in Europe (ICLaVE 10), Leeuwarden, June 2019, ed. Hans Van de
Velde, Nanna Haug Hilton and Remco Knooihuizen, 269–289. Amsterdam: Benjamins. https://
doi.org/10.1075/silv.25.12tam.
Thomason, Sarah G. 2007. Language contact and deliberate change. Journal of Language Con-
tact—Thema 1: 41–62.
Thomason, Sarah G. 2016. Middle English: English, not Norse. Language Dynamics and Change
6 (1): 42–45. https://doi.org/10.1163/22105832-00601010.
Thomason, Sarah G., and Terence Kaufman. 1991. Language contact, creolization, and genetic
linguistics. Berkeley and Oxford: University of California Press.
Timofeeva, Olga. 2018. Survival and loss of Old English religious vocabulary between 1150 and
1350. English Language and Linguistics 22 (2): 225–247. https://doi.org/10.1017/S13606743
18000114.
Timofeeva, Olga. 2022. Sociolinguistic variation in Old English: Records of communities and
people. Amsterdam: Benjamins. https://doi.org/10.1075/ahs.13.
Timofeeva, Olga, and Richard Ingham. 2018. Special issue on mechanisms of French contact influ-
ence in Middle English: Diffusion and maintenance. English Language and Linguistics 22 (2):
197–205. https://doi.org/10.1017/S160674318000096.
Todd, Simon, Janet B. Pierrehumbert, and Jennifer Hay. 2019. Word frequency effects in sound
change as a consequence of perceptual asymmetries: An exemplar-based model. Cognition 185:
1–20. https://doi.org/10.1016/j.cognition.2019.01.004.
Torgersen, Eivind, and Paul Kerswill. 2004. Internal and external motivation in phonetic change:
Dialect levelling outcomes for an English vowel shift. Journal of Sociolinguistics 8 (1): 23–53.
https://doi.org/10.1111/j.1467-9841.2004.00250.x.
Townend, Matthew. 2002. Language and history in Viking Age England: Linguistic relations
between speakers of Old Norse and Old English. Turnhout: Brepols.
Trask, R.L. 1997. Historical linguistics. London: Arnold.
Traugott, Elizabeth Closs. 2003. Constructions in grammaticalization. In The handbook of histor-
ical linguistics, ed. Brian D. Joseph and Richard D. Janda, 624–647. Oxford: Blackwell.
Traugott, Elizabeth Closs, and Graeme Trousdale. 2013. Constructionalization and constructional
changes. Oxford: Oxford University Press.
Trudgill, Peter. 1974. The social differentiation of English in Norwich. Cambridge: Cambridge
University Press.
Trudgill, Peter. 1986. Dialects in contact. Oxford: Blackwell.
Trudgill, Peter. 2004. New-dialect formation: The inevitability of colonial Englishes. Edinburgh:
Edinburgh University Press.
Trudgill, Peter. 2008. Colonial dialect contact in the history of European languages: On the irrele-
vance of identity to new-dialect formation. Language in Society 37 (2): 241–254.
Trudgill, Peter. 2012. On the sociolinguistic typology of linguistic complexity loss. In Potentials of
language documentation: Methods, analyses, and utilization, ed. Frank Seifart, Geoffrey Haig,
References 209

Nikolaus P. Himmelmann, Dagmar Jung, Anna Margetts, and Paul Trilsbeek, 90–95. Honolulu:
University of Hawai‘i at Mānoa.
Trudgill, Peter. 2020. Sociolinguistic typology and the speed of linguistic change. Journal of
Historical Sociolinguistics 6 (2): 20190015. https://doi.org/10.1515/jhsl-2019-0015.
Trudgill, Peter, Elizabeth Gordon, Gillian Lewis, and Margaret MacLagan. 2000. The role of
drift in the formation of native-speaker Southern Hemisphere Englishes: Some New Zealand
evidence. Diachronica 17 (1): 111–138. https://doi.org/10.1075/dia.17.1.06tru.
Turton, Danielle, and Maciej Baranowski. 2021. Not quite the same: The social stratification and
phonetic conditioning of the foot–strut vowels in Manchester. Journal of Linguistics 57 (1):
163–201. https://doi.org/10.1017/S0222226720000122.
Upton, Clive, and J.D.A. Widdowson. 2006. An atlas of English dialects. London: Routledge.
van Coetsem, Frans. 1988. Loan phonology and the two transfer types in language contact.
Dordrecht: Foris.
van Coetsem, Frans. 2002. A general and unified theory of the transmission process in language
contact. Heidelberg: Winter.
van den Doel, Rias. 2006. How friendly are the natives? An evaluation of native-speaker judgements
of foreign-accented British and American English. Utrecht: LOT Publications.
van den Doel, Rias, and Hugo Quené. 2013. The endonormative standards of European English:
Emerging or elusive? English World-Wide 34 (1): 77–97. https://doi.org/10.1075/eww.34.1.
04van.
Van Hofwegen, Janneke. 2015. Dyadic analysis: Factors affecting African American English usage
and accommodation in adolescent peer dyads. Language & Communication 41: 28–45. https://
doi.org/10.1016/j.langcom.2014.10.004.
van Loey, A. 1970. Schönfelds historische grammatica van het Nederlands: Klankleer, vormleer,
woordvorming, 8th ed. Zutphen: Thieme.
Vennemann, Theo. 2009. Celtic influence in English? Yes and no. English Language and Linguis-
tics 13 (2): 309–334. https://doi.org/10.1017/S1360674309003049.
Visser, F. Th. 1963–1973. An historical syntax of the English language, 3 vols. Leiden: Brill.
VOICE. 2021. The Vienna-Oxford International Corpus of English (version VOICE 3.0 Online).
https://voice3.acdh.oeaw.ac.at.
Wales, Kathleen M. 1983. Thou and you in Early Modern English: Brown and Gilman re-appraised.
Studia Linguistica 37 (2): 107–125. https://doi.org/10.1111/j.1467-9582.1983.tb00316.x.
Wallage, Phillip. 2008. Jespersen’s Cycle in Middle English: Parametric variation and grammatical
competition. Lingua 118 (5): 643–674. https://doi.org/10.1016/j.lingua.2007.09.001.
Wallage, Phillip. 2013. Functional differentiation and grammatical competition in the English
Jespersen Cycle. Journal of Historical Syntax 2 (1): 1–25. https://doi.org/10.18148/hs/2013.
v2i1.4.
Wang, William S.-Y. 1969. Competing changes as a cause of residue. Language 45 (1): 9–25.
https://doi.org/10.2307/411748.
Ward-Perkins, Bryan. 2000. Why did the Anglo-Saxons not become more British? The English
Historical Review 115 (462): 513–533. https://doi.org/10.1093/ehr/115.462.513.
Watson, Kevin. 2006. Phonological resistance and innovation in the North-West of England.
English Today 22 (2): 55–61. https://doi.org/10.1017/S0266078406002100.
Watt, Dominic, Carmen Llamas, and Daniel Ezra Johnson. 2010. Levels of linguistic accommoda-
tion across a national border. Journal of English Linguistics 38 (3): 270–289. https://doi.org/
10.1177/0075424210373039.
Watts, Richard J. 2011. Language myths and the history of English. Oxford: Oxford University
Press.
Wedel, Andrew, Abby Kaplan, and Scott Jackson. 2013. High functional load inhibits phonologi-
cal contrast loss: A corpus study. Cognition 128: 179–186. https://doi.org/10.1016/j.cognition.
2013.03.002.
Weinreich, Uriel. 1953. Languages in contact: Findings and problems. Den Haag: Mouton. https://
doi.org/10.1515/9783110802177.
210 References

Weinreich, Uriel, William Labov, and Marvin I. Herzog. 1968. Empirical foundations for a the-
ory of language change. In Directions for historical linguistics: A symposium, ed. Winfred P.
Lehmann and Yakov Malkiel, 95–188. Austin: University of Texas Press.
Wells, J.C. 1982. Accents of English. Cambridge: Cambridge University Press.
Westerberg, Fabienne. 2019. Swedish “Viby-i”: Acoustics, articulation, and variation. In Proceed-
ings of the 19th International Congress of Phonetic Sciences, Melbourne, Australia 2019,
ed. Sasha Calhoun, Paola Escudero, Marija Tabain, and Paul Warren,3696–3700. Canberra:
Australasian Speech Science and Technology Association Inc.
Wieling, Martijn, Mark Tiede, Teja Rebernik, Lisanne de Jong, Anouck Braggaar, Martijn
Bartelds, Masha Medvedeva, Penny Heisterkamp, Tom Freire Offrede, Hedwig Sekeres, Anna
Pot, Mara van der Ploeg, Karin Volkers, and Gregory Mills. 2020. A novel paradigm to inves-
tigate phonetic convergence in interaction. Proceedings of the 12th International Seminar on
Speech Production: 1–4.
Wischer, Ilse. 2008. What makes a syntactic change stop? On the decline of periphrastic do in Early
Modern English affirmative declarative sentences. Studia Anglica Posnaniensia 44: 139–154.
Wiśniewska-Przymusińska, Malwina. 2020. T/V pronouns and FTAs in The Works of Sir Thomas
Malory: Medieval politeness and impoliteness in directives, expressives, and commissives.
Studia Anglica Posnaniensia 55 (1): 139–176. https://doi.org/10.2478/stap-2020-0006.
Wright, Laura. 1992. Macaronic writing in a London archive, 1380–1480. In History of Englishes:
New methods and interpretations in historical linguistics, ed. Matti Rissanen, Ossi Ihalainen,
Terttu Nevalainen, and Irma Taavitsainen, 762–770. Berlin, New York: Mouton de Gruyter.
https://doi.org/10.1515/9783110877007.762.
Yu, Alan C. L. 2023. The actuation problem. Annual Review of Linguistics 9: 215–231. https://doi.
org/10.1146/annurev-linguistics-031120-101336.