‘Oxford introductions to Language Study
Series Editor H.G. Widdowson
Published in tis series: z
Guy Cook: Applied Linewistce
Rod Elis: Second Language Acquistion
Claire Kramsch Language and Culture
‘Tim McNamara: Language Testing
Peter Roach Phonetics
Thomas Scovel: Psycholinguistics
Bernard Spolsky: Sociolinguistics
Peter Verdonk: Stylistics 2
H.G.Widdowson: Linguistics
George Yule: Pragmatics
OXFORDOXFORD
Asin ck re A Gp Tor Chee
Nair ‘Sangha Tai Tokyo Toes
rh Ono Uni Pr ar
No wnenbaed pocopins
rents elon ceca
‘em age i he sae ao i
orem seen at gn ror
IUrechics iy caries,
Contents
Preface
suney
Language change as a matter of fact
Aatieudes to language change
Language sate and process
‘The aims and scope of historical linguistics
Reconstructing the past: data and evidence
“The data of historical linguistics
The writen evidence
Sources of evidence
Comparing and reconstructing languages
Correspondences between languages
Laws ofchange
Internal reconstruction
Vocabulary change
Coining new words
‘Changes of meaning
‘Why do word meanings change?
Grammatical change
Morphological change
Syntactic change
Sound change
low sounds are produced
Phonetic change
Phonemic change
5
5
26
3S
38
43
46
se8 Language contact
Borrowing from other languages
Convergence and linguistic areas
Language birth: pidgins and ereoles
anguage death
7 How and why do languages change?
Functional explanations
Psycholinguistic explanations: language acquisition
Sociolinguistic explanations
‘The origin and spread of changes
8 Postscript farther developments
Socio-historical linguistics and historical pragmatics
volutionary linguistics
Standardization and language planaing
Conclusion
Readings
References
Glossary
ss
3s
59
6
68
na
76
Br
sr
S
83
8
3
Preface
Purpose
‘What justification might there be fora series of introductions ro
language study? After all, linguistic is already wel served with
introductory texts: expositions and explanations which are
comprehensive, authoritative, and excellentin cheir way. Generally
Speaking, however, their way isthe essentially academic one of|
providing a detailed initiation into the discipline of linguistics, and
they tend tobe lengthy and technical: appropriately so, given their
purpose, But they can be quite daunting to the novice. There is
nso a need for a more general and gradual introduction «0
language: transitional texts which will ease people into an
understanding of complex ideas. This series of introductions is
designed to serve this need
“Their purpose, therefore isnot to supplant but to support the
‘more academically oriented introductions to ingsties: to prepare
the conceptual ground. They are based on the belief thar itis an
advantage t have a broad map of the erain sketched out before
fone considers its more specific features on a smaller scale, a
‘general context in reference co which the derail makes sense. Itis
Sometimes the case that students are introduced to detail without
it being made clear what itis a detail of. Clearly, a general
understanding of ideas is not sufficient: there needs to be closer
scrutiny. But equally close scrutiny can be myopic and mean-
ingless unless itis related tothe larger view. Indeed it can be said
thar the precondition of more particular enquiry isan awareness
fof whan, in general, the particulars are about. This series isdesigned to provide this large-scale view of different areas of
language seudy. As such it can serve as preliminary to (and
precondition for) the more specific and specialized enquiry which
Students of linguistics are required ro undertake
But the serics is not only intended to be helpful to such
students There are many people who take an interest in language
without being academically engaged in linguistics per se. Such
people may recognize che importance of understanding language
for their own lines of enquiry or for their own practical purposes,
‘or quite simply for making them aware of something, which
figures so centrally in their everyday lives. If linguistics has
revealing and relevant things to say about language, this should
presumably not be a privileged revelation, but one accessible to
people other shan linguists. These books have been so designed as|
toaccommodate these broader interests too: they are meant to be
introductions to language more generally 2s well as linguistics
asa discipline.
Design
The books inthe serie
sare four parts: Survey,
allcutto the same basic pattern. There
eadings, References, and Glossar.
Survey
‘This isa summary overview of the main features ofthe area of
language study concerned: its scope and principles of enquiry, its
basic concerns and key concepts. These are expressed and ex
plained in ways which ae intended to make them as accessible as
possible ro people who have no prior knowledge or expertise in
the subject. The Survey is written to be readable and is un
clutered by the customary scholarly references. In this sense itis
simple. But ies aot simplistic, Lack of specialist expertise des
‘ot imply an inability to understand or evaluate ideas. Ignorance
means lack of knowledge, not lack of intelligence. The Survey,
therefore, is meant to be challenging. It draws a map of the
subject afea in such a way as‘ stimulate thought and to invite a
Critical participation in the exploration of ideas. This kind of
‘conceptual cartography hasits dangers ofcourse: the selection of
‘whats significant and the manner of is representation, will not
be to the liking of everybody, particularly not, perhaps, ta some
‘of those inside the discipline. But these surveys are written inthe
belie tae there must be an alternative toa echnical account on
the one hand, and an idiots guide on the other if linguistics sto
bbe made relevant to people in the wider world
Readings
Some people will be content to read, and perhaps re-read, the
summary Survey. Others will want ro pursue the subject and so
willuse the Survey as the preliminary for more detailed study. The
Readings provide the necesary transition. For here the reader is
presented with txts extracted from the specialist literature. The
‘purpose ofthese Readings is quite different from the Survey. [eis
{ogetreadersto focus onthe specific of what issaid, and how itis
sai, in these source texts. Questions are provided to further this
purpose: they are designed to direct attention to points in each
text, how they compare across rexts, and how they deal withthe
issues discussed in the Survey. Te idea iso give readers an intial
familiarity with the mote specialist idiom of the linguistics
literature, where the issues might not be so eadily accesible, and
to encourage them into close critica reading
Refer
One way of moving into more detailed study is through the
Readings. Another is through the annorated References in the
thied section of each book. Here there is a selection of works
{books and articles) for farther reading. Accompanying comments
indicate how these deal in more detail with the issues discussed in
the different chaprers ofthe Survey.
Glossary
(Certain terms inthe Survey appear in bold. These are terms used
ina special or technical sense in the discipline. Their meanings
acemade clear inthe discussion, bu they are also explained inthe
Glossary at the end of each book, The Glossary is cross
referenced tothe Survey, and therefore serves a the same time as
an index. This enables readers to locate the term and what it
signifies inthe more general discussion, thereby, in effect, using
the Survey asa summary work of reference.Use
“The series has been designed so as to be flexible in use. Each ttle
ie separate and self-contained, with only the basic format in|
‘common. The four sections ofthe formar, as described here, can
he drawn upon and combined in different ways, as required by
the needs, o interests, of different readers. Some may be content
with the Sarvey and the Glossary and may not want to follow up
the suggested References, Some may not wish to venture into the
Readings. Again, the Survey might be considered as appropriate
preliminary reading for a course in applied linguistics or teacher
fducation, and the Readings more appropriate for seminar
discussion during. the course. In short, the notion of an
ineroduction will mean different things to different people, but in
all cases the concern isto provide access to specialist knowledge
tnd stimilate an awareness ofits significance. This series as a
‘whole has been designed to provide this access and promote this
{wareness in respect to different areas of language study.
Author's acknowledgements
The fact that the Oxford Introductions 10 Language Study
include a volume on historical linguistics bears wimess othe fact
fat this time-honoured linguistic discipline has taken up a
tenteal place within the field again. This book tries to make the
Subject accessible to the uninitiated reader and to show how
closely historical linguistics is inked to the other linguistic areas
Covered in the series, My thanks go to Oxford University Press
And the series editor, H.G, Widdowson, fr including this volume
in the series,
Quite a: number of people have provided me with valuable
input and have helped to make this a better and more readable
book.
First and foremost, I owe a very special debt of gratitude to
HG, Widdowson for his continuous support and invaluable advice
through all the stages of my writing; he has made numerous
proposals for improvement, both in regard 10 the overall
structure of the book and to countless details, and has
painstakingly read through the various drafts ofthe manuscript.
‘The following fiends and colleagues have read the whole or
substantial parts of the manuseriptin various stages and have made
valuable suggestions: Clausdirk Polines, Hans Plazes, Angelika
Hirsch, Ute Smit, Barbara Seidlhofe, Gunther Kaltenbéck
Nikolaus Rite provided me with valuable information on
‘nco-Darwinian evolutionary theory. My heartfele thanks to all of
them. Iam much indebted to the people at Oxford University
Press forall their support. Las, but not least, my special thanks
0 £0 my wife, who has not only read hoth the manuscript and
the proofs, but has also been a constant source of
encouragement.
London, October 2000SECTION 1
Survey4
Language change as a matter
of fact
All physical aspects ofthe universe and all aspects of human life
re subject to change and languages reno excepion-Individual
Changes can be que abrupe and obvious, as when new words
| ake an appearance and become popular. Normally, however,
language change is gradual, almost imperceptible, a withthe
slow alterations in pronunciation when one generation speaks
Slghly diferey from anothe. Linguistic changes tend to he
the esl of two equivalent forms coensting a variants for some
| ime, and one giving way tothe oter Two words, for example,
{— Grtwo ways of pronouncing the same word, may coexist inthe
Same specch community for some time, hut may be used. hy
diferent subgroups oF on different ocssions. However, for
reasons tobe discused Inter, such variant forms may begin 0
onypete an finally one wll dominate andthe othe deci.
Sill ingusde changes may be evident in everyday exper
ace, and people may notice (and sometimes disapprove) when
‘words are ied of pronounced in different ways; but language
Change is mos obvious ona large scale when we lok at older
texts ofa particular language, and the Further back we go i
history, the more obvious the changes become. Here is an
example of Old English, eaken from the time of King Allred the
Late noth century aD}, for which a translation ia modem
English given below
(1) Alfred kyning hated gretan Werfer® biscep his wordum
Iufice ond freondlice ond Qe cyan hate, fet mecom wide
oft on gemynd, hwelce wiotan iu waron giond Angeleyan
aegier ge godcundra hada ge woruldcundra, ond hu
sgesliglicatida 0a waron giond Angeleynn.
LANGUAGE CHANGE 3,[King Alfred sends greetings t0 Bishop Warferth with his
loving and friendly words, and I would declae to you that it
has very often come to my mind what wise men there were
formerly throughout the English people, both in sacred and in
secular orders, and how there were happy. times. then
throughout England.)
Here the language has changed almost out of all recognition.
[A linguistic discussion ofthe passage would go beyond the scope
ofthis ingroduction, but itis evident that only a small number of
‘words ofthe modern language sill carry obvious traces oftheir
heredity, and even these have changed in various ways, e4
freondlice,lflice > friendly, lovely. Some ofthe eters used have
disappeared from modern English, such as Ofor modern th, or
forthe vowel in modern standard English hat
‘Aux here is an example from Middle English, almost half
‘of millennium later, taken from the Prologue of Chaucer's
Canterbury Tales
(2) Ye goon to Caunterbury—God yow speede,
‘The blisul mastir quite yow youre moede!
‘And wel woot, as ye goon by the weye,
Ye shapen yow to talen and o pleye;
For tewely, confort ne myrthe is noon
‘Taide by the weye doumb as a stoon;
The language here i less remote. There are obviously several
differences in spelling, for example, the endings of verbs (goon,
talen, both with plural ending (et, and in some word forms
{woot know", ye"you'). Ifyou heard the passage read aloud, the
distance from mosen English would be somewhat greater But
most of the features in (2} we can recognize as related to the
English language as we use it today.
‘And finaly, heres passage from Shakespeare's A Midsummer
Night’ Dream iti. 335) writen about two centuries after
(Chaucer, at the end ofthe sixteenth century
(5) Lysander Now she holds me not.
Now follow ifthou dart, to try whose eight,
(Of thine or mine, is most in Helena
Demetrius Follow! Nay, I'l go wih thee, cheek by jow.
Hermia You, mises, all this coils long of you: Nay, go not
back.
Helena Lill no erase you,
No longer stay in your curst company
Your hands than mine are quicker fora frays
My legs are longer, though, to run ay.
‘This isin many ways, including che pronunciation, very close
to modetn English, But there ace still obvious differences on all
linguistic levels, especially with regard to grammar and vocabu:
lary. In the second person singular of verbs, we notice che ending
«tn dart and the singular personal pronouns thoulthe besides
Jot. The negative sentence she holds me not would be expressed
inmodern English as she does wot hold me, and the word order of
Your hands than mine are quicker fr a fray sounds definitely
peculiar today: None of the words looks particularly unusual,
though in some cases their meanings have changed. The word
ti for example, wll be familia to many because it also occurs
in Hamers famous phrase shuffled off cis moral coil’. But they
twill almost certainly not know its sixtenth-century meaning of
arm’
Tn looking at a particular language over a longer period of
time i becomes apparent that language change does noc esult in
iifferent distinct stages of a language but in a historical con-
tinuum, so that speakers easly understand the language of the
generations immediately before and after them, but meet
increasing diffculties in understanding chronologically remote
stages oftheir language. Ths closely resembles the well-known
phenomenon of the dialect continaum: adjacent geogeaphical
Yarieties of a language are mutually intelligible, but dialece
Speakers may have problems with geographically remote varieties
oftheir language. The close relationship beeween temporal and
spatial linguistic differences may also be noticeable in another
respect: thus, revelling through rural Britain from south to north
for from east go west can in many ways resemble a journey
through the history of English, since rural dialects have often
preserved older forms of language.
Linguistic change, then, is not restricted to particular lan-
guages or generations, bu is a universal fact. This does not mean |
that people wil always be happy to accept the inevitable.Attitudes to language change
Languageis so closely associated with social identity thatitis not
surprising that people have steong felings about it. Language
change can be unsettling anda widespread attitude istosee tas.a
‘change forthe worse. Speakers of cifferent periods and cultures
have often tended to think that their own language is inferior ro
thae of their forebears. For them, language change is a matter of
decline or decay. In some societies this attitude can be traced back
to the biblical account of the Tower of Babel. Here the change
from one common language toa diversity of different languages
is presented as divine punishment for sinful behaviour.
Ta the history of European languages, negative statements on
language change as corruption and decay clearly predominate
‘over neutral ones, while positive views sem ro be conspicuously
absent. In mos of the emerging European national languages we
Find increasing atcempts to “purify” sind coliy the language, ie
tox prescriptive rules of correct usage and thus stop language
change. This task was partly undertaken by offical institutions,
so-called ‘language academies, such as the Accademia della
Ceusea in Florence (founded in 1382), che French Académie
francaise (1635), and various German ‘language societies’ ofthe
seventeenth and eighteenth centuries. In England the eighteenth
century, with its striving for regulariey and order, was especially
hostile to the idea of linguistic change, but here eoditetion was
‘mainly carried out by influential individuals. Many of the leading
intellectual and literary figures of the period, such as Jonathan
Swift and Samuel Johnson, vehemently opposed the idea of
language change. For Johnson, author of the famous Dictionary
ofthe English Language (1755) for instance al linguistic change
was ‘of tself an evil In the Preface o his Dictionary, he states:
(4) tongues [i.e languages) like goveenments, have a natural
tendency 0 degeneration; we have long preserved our
‘constitution, ler us make some struggles for our language.
‘This comparison of language with human institutions like
governments was in no way restricted to England. For the
‘American politician Benjamin Franklin, language reflected social
reality and the supposed ‘degeneration’ of language directly
tmicrored the degeneration of contemporary society.
4
‘This‘complaine ration’ (as ithas been called) has continued
tp t0 the present century. Here is a recent example from a
popular book on linguistic correctness by an American art critic,
expressing a view echoed by educated laypersons in many
‘countries. (See also Section 2, Readings, Text 1.
(s) [Bly and large, linguistic changes are caused by the
ignorance of speakers and writers, and in the last few
centuries—given ou school, dictionaries, and books on
grammar—such ignorance could have been, like the live
netle or poison iy itis, uprooted
{John Simon, Paradignss Lost, 1976)
‘Notonly individuals, But also governmental instirtions have
shown emotional or ideologically motivated attudes towards
certain changes. In Navi Germany attempts were made to assert
the integrity of the language by promoting words of German
character even for established foreign words (Femsprecher
‘telephone’, literally “distant speaker’, instead of Telephon). But
even democratic governments are not immune from nationalistic
tendencies in tying t0 stop excessive borrowing of foreign
"words. A recent case in point has been the French government’
(unsuccessful) measures against the use of “ranglais', English
borrowings such as le weekend and le shopping, by means of
governmental deroes and attempts to replace chem with French
tvords coined by an official committe, e.g le baladeur for le
twalleman. Though reversion to a real or imagined earlier (and
‘purer) state of language also involves change, this type of
change is seen to be ‘inthe sight diretion’ and acceptable for
political reasons.
Een professional linguists have in the past heen prone to a
conservative attitude 10 language change. Early nineteenth-
century scholars regarded language asa growing organism witha
stage of grout, a brief moment of evolutionary perfection, and
subsequent decay. Thus, for example, the disappearance of case
inflections between Old English and Modern English or from
Latin eo French, and their partial replacement by prepositional
phrases (for example, Old English freondes vs. English of the
[Fronds Latin amici vs, French de ami“of the friend) was seen 8
indicating decineContemporary linguists in general hold a neutral ot ev
positive atitude towards change. On the positive side it has been
aimed that changes are a necessary development to make
languages more communicatively effective as they become attuned
to changing socal needs. This also applies to che promotion of
conscious linguistic changes to achieve this goal, suchas language
planning and measures to make langage “politically correct (cf.
‘Chapter 8). Furthermore, changes have been viewed as necessary
therapeutic measures to restore the halance and symmetry of the
Tinguistc system, o a8 moves towards the simplification of the
grammar. In such a view, the change of language over periods of
time ia function oF influences operating at any given time. In this
respect, the study of history (of language or of anything else)
depends on an understanding ofthe present, just as the present is
tobe understood by reference tothe past.
Language state and process
[Nevertheless in much modern linguisties past and present have
been separated into different areas of enguity. It has been a
‘common assumption that synchronic linguistics, which concerns
itself withthe state of languages ata given time, in particular the
present, is most conveniently carried out in distegard of the
Findings of historical or lachrone linguistics about the processes
of language development overtime, However, this strict division
is based on a misunderstanding ofthe relationship between these
two aspects of the study of language. On the one hand, the
synchronie study of linguistic systems can provide insights chat
can be used in reconstructing thee past. On the other hand, we
Should also recognize that the implied assumption that syn
ron linguistic systems are completely systematic, static and
hhomogeneous, is a fiction. All of them are in some respects
tunsystematc: the numerous irregular relics of eaier systems (he
‘exceptions’ to the rule}, which are simply inexplicable in
synchronic terms, can only be explained by reference to past
States and developments. The unstable state ofa language at any
tiven point ofimeis the consequence of historical processes, and
its very instability is evidence that these processes continue to
‘operat in the present.
Fqually, there is a close inerrlationship between synchronic
linguistic variation, ic. the coexistence of mote or less equivalent
variants at given time, and diachronic linguistic change. The
[growing awareness of these facts over the past thirty yearshas led
to a major reorientation in the discipline, wit historical linguist
ies again taking its rightful place in the eld of language stu.
The aims and scope of historical linguistics
“The beginning of historical linguistic studies inthe modern sense
‘of the word dates back more than two hundred years, though
there isa much older tradition of language study in some cultures.
Soi isnot surprising that there have been very differen scholarly
traditions and approaches to historical linguistics, each of which
sets particular research objectives. and. calls’ for different
rethodologis. Wecan identify these broad areas of enquiry
1 The study ofthe history of particular languages on the basis of
existing writen data,
2 The study of the prehistory of languages by means of
‘comparative reconstruction, whereby the unrecorded past is
inferred on the evidence ofthe data that are available from
late period,
5 The study of ongoing changes in a language, ie. changes
happening at the present cme.
However fascinating these issues may be in their own right
(especially for historical linguists themselves) they should be
linked to other and more abstract aims, namely the discovery of
‘mote general, possibly universal, aspects of language change. By
relating che descriptive fats about a particular language to what
is common across all (or most) other languages, the historical
Tinguis seks to explain wy languages change, and how these
changes spread in space and time, The most promising area for
finding answers to these questions is the study of ongoing,
changes, especially when caried out within a framework which
femphasizes the interrelationship between socal factory sj
chronic linguistic variation at any one time, and diachronic
linguistic change overtime.iid
&
Reconstructing the past:
‘The data of historical linguistics
“The synchronic description of living languages in the present can
by based on a wide range of data suchas the introspection ofthe
linguise (asa native speaker), the guided elicitation of data from
ative informants, and observation including the use of coxpora}
Different linguistic schools have placed different emphases on
these types of data, The data of historical linguistics, on the other
hand, are much mote restricted. Obviously, much ofthe pasts
not accessible through introspection or elicitation. We only have
its observable traces to go on, most importantly the limited
orpos of written texts as a record of actual Ianguage use in
former times.
Fortunately, many languages have along recorded past, which
provides evidence for the development of the individual lan
{guages and also for more general properties of linguistic change
But there aze clear limitations with regard to both the quantity
and the quality of the data. In general, the further back we go in
time, the more sparse and unreliable the data become, while at
the same time the language becomes more and more remote
Frequently we lack suficientextralinguistic information on old
texts, such as theie author, purpose, or audience. Equally the
range of text types is limited, and authentic spoken daca are
altogether lacking before the twentieth cencury. The recon:
Struction of older written langage is difficult enough, but its
‘ren more difficult to reconstruct older speech from written data
‘We must beat in mind too that this reconstruction is not a
straightforwaed matter of facts, The interpretation—and eventhe selection—of the available data is always informed by
undeelying general assumprions about language or a specific
theory to which the historical linguist subscribes. This means
that, as with other aspects of history, we will find competing
explanations ofthe past.
‘Whenever che beginning of human language may have heen —
and most modern estimates vary between $0,000 and 100,000
years ago, though some extend it substantially 10 one milion
years ago or even more—it is obvious that the evolution of
language is documented only for a very small percentage of is
total history, and even this applies only to relatively few ofthe
5,000 fo. 6,000 human languages said to exist in-the-world at
present. The lack of data from these unrecorded stages, i. from
the prehistory of languages, can however be partly overcome by
systematic comparison of the oldest writen records of related
languages even f they are now extinct. These canbe as fascinating,
a8 dinosaus eges or Foss of extinct species and, just like these
winesses ofthe pas, they can help us to extend our knowledge of
linguistic evolution further hack into prehistoric sind wnrecorded
time. The reconstruction of the prehistory of languages isthe
domain of comparative reconstruction discussed later in this
chaptet
The written evidence
Written texts provide the most important data for historical
linguistics, and it is a crucial matter how these data are
interpreted as evidence for earlier spoken language as well as for
linguistic systems. Such interpretation may present almost
insurmounable dlfculkies in the case of extinct languages with
unknown writing systems. The deciphering of extinct languages
‘writen in such systems often depends on the existence of bi- or
trilingual texts, in which a east one ofthe languages is known. A
famous casein point isthe Rosetta Stone found inthe course of
Napoleon's expedition to Egypt and now in the
the trilingual inscription on tis stone in ancient
two different versions ofthe ancient Payptian script provided the
key to the decipherment of the Egyptian hirogyphs.
Bat interpreting written data as evidence of language systems
and speech is not unproblematic, even for languages whose
‘weiting systems are known and. well established. Consider
alphabetic writing as used for English and the other European
languages. Whereas in other weting systems symbols might
stand for syllables or words, here they represent sounds, 4.
vowels and consonants, more or less directly, whether in the
Latin, Greek, or Cyrillic alphabet. These alphabers stand in a
long tradition and their basie writing conventions have been
handed down for centuries. But even so, the nature of the
correspondences berwen ‘letters’ and ‘sounds’ ig by no means
‘easy {0 determine, Speech and writing are wo distinct though
clearly relaced systems, but the nature of this relationship may
change over time. It is widely agreed that alphabetic writing
inically aimed ac rendering the distinctive sound units of a
language and tended to neglect non- distinctive differences: chus
the letter in Latin, Old English, Old French, et. represents the
distinctive unit! as opposed 0 l/l, et), but does not ell ws
anything about its specific realization as atriled,retofle, oF
‘guttural sound (cf che different quality of /e! in English, Seoss,
French or German),
Furthermore it safe to assume thatthe scribes who ist used
the Latin alphabet to write down languages such as Old English
‘or Old High German adhered to the Latin spelling conventions,
since they had been fist trained in writing Latin. Forunately,
these conventions are quite well known as a result of the
unbroken Latin tradition in Western civilization. However, the
Latin writing system had to be adapted in a number of ways to
these newly written languages, Fr example, Latin did nothavea
sound corresponding tothe iia sound in English thing, while
‘Old English did. Here, the Old English scribes adopted two
alternative solutions: they either used a letter fom the old runic
alphabet, namely 6 ,orthey slightly changed the Latin eter dinco
the form 3
‘The originally more-or-less direct relationship between letters
and sounds, or more generally, between writing and speech
became, however, blurred as a esult of language change, since
spelling tends to be conservative and either does not record
changes at all or docs 0 only after a considerable time-lag. An
additional complicating factor may be the mixing of different
Bregional or national spelling conventions due to cultural contact.
‘Thus Anglo-Norman conventions appear widespread in English
texts after the Norman invasion of England in 2066. To give a
simple example from the history of English: the Old and Middle
English forms of house were both pronounced with long i (the
‘vowel in modern English goose) and the Old English spelling hus
for fhu ellets the original Latin-based one-to-one relation
between specch and writing, The new Middle English spelling
bous(e) fr the same pronunciation /husis due tothe adoption
of the Anglo-Norman convention of writing ow for fa! (et
modern French jour ‘day’, pour ‘for). This spelling is stil
Tetained today, though “tl was diphthongized in the Karly
“Modern period, yielding modern English ‘haus. Weean say that
English writing started asa system which was secondary 0, i
dependent on, specch, but became increasingly autonomous and
unrelated to actual pronunciation, (For the relation beoween|
‘writing and speech see also Readings, Text 3.)
Sources of evidence
“The hypotheses of th historical linguist depend crucially on the
interpretation ofthe daa. Iris not justa matter ofthe amount of
data available but primarily of their quality, To evaluate the
‘quality of old texts, we have to find out as much as possible abou
theie extealingustc context (such as the author, scribe, purpose,
and location of a text, et.), and about the textual tradition,
including the original form and date of composition and copying,
‘Thisis the task of che philologist, for whom auxiliary disciplines
such as history and palaeography, the study of ancient waiting,
are of major importance.
‘Only very few old texts are inthe authors own hand, and even
these may shaw various kinds of textual errors. Mostly they are
the result of mulple copying by diferent scribes in diferent
regions and over long period of time. Some textsare compilations
hy a specific author from linguistically divergent, possibly orally
transmitted original sources, as with Homer’ Iliad and Odyssey,
or the Rigveda, the oldest collection of religious texts written in
Sanskrit, Such texteal history may resultin linguistically composite
texts witha mised language fill of srbal errors due to negligence
‘or insffcient competence in the language(s) or varieties of the
original. These diferent linguistic layers, whether dialectal or
tiachronic, must be disentangled and scribal errors detected
before the textcan he used as data fr forming hypotheses about
specific stages of a language, Furthermore, old texts are often
translations, e from Latin into Old English, or from Greek into
carly Gothic or Old Church Slavic, so that we have fo reckon
With linguistic influence fom the original language.
“Though written texts constitute the maj source fr historical
linguistics, other types of data may provide imporeant sop-
plementary evidence. For example, archaeological findings such
as pieces of pottery, grave findings, and ther historical material
have contributed substantially to our knowledge of the setle-
ment history of early Anglo-Saxon England, which again may
help in reconstructing te dialectal distribution of Old English,
‘OF particular interest as daca are diocr descriptions of and
‘explicit comments on a language by contemporary speakers.
Such ‘metalinguistic’ evidence is, however, raze for the early
sages of most languages and not always reliable, though there
are excellent early grammars of Sanskrit, Greek, and Latin. The
bulk of such information for European languages only dates
from the modem period. Some detiled desriptions of English
sounds and even phonetic transcriptions of texts date hack tothe
sixteenth century and we have numerous surviving glossaries,
‘word lists and translations of Latin and other languages, which
provide information about word meanings in medieval Europe.
ast but not least, modem dialeets and related languages can
provide valuable information to help construct oF test our
bbypotheses.
Let ws now look at how data are wed to reconstruct linguistic
history.
Comparing and reconstructing languages
[A basic hypothesis in historical linguists is that for all their
‘curren differences languages may originate from one common
source (or prototanguage), to which they are this genetically
felated. We have unambiguous linguistic and extalinguistic
historical evidence for such genetic relationships in the case ofthe6
modern Romance languages (French, Italian, Spanish, Catalan,
Portuguese, Romanian, et.) These are the direc descendants oF
laughter languages of Latin (more precisely of its different
spoken forms, Vulgar Latin), from which they have evolved in
the course of centuries as 4 result of geographical distance and
isolation, social factors and political developments, and chrough
contact with oher languages. Genetically related languages form
language families and they show systematic and recurrent formal
correspondences, i. similarities and differences which are too
‘regular and frequent to be mere chance or the result of borrow.
ing. These correspondences become more evident and regular the
farther back we go in language history. They are; for example,
stronger between Old French and Old Spanish than beeween
Modern Prench and Modern Spanish
"The most famous and bestresearched language family isthe
Indo-European (TF) one, witha long extual tradition in a wide
range of geographically divergent daughter languages. Indo-
European languages have long been spoken from India to the
western borders of the European continent, and have in more
recent times been exported all over the world. They ate grouped
inco-a number of subfamilies (or branches) such as Germanic,
Italic (including Latin and che Romance languages), Balto-Slavie
(including the Slavie languages Russian, Czech, Bulgarian, ete),
Geki, Greek, and Indo-Iranian (with Persian, Kuedish, Sanskrie
and a mumber of modern Indian languages). The reconstruction
‘ofthe common ancestor of these languages, Proto-Indo-European,
was one of the outstanding achievements of nineteenth-century
comparative linguists.
“The most widely used way of showing genetic relationships
graphically sche fam tree model. The diagram on the next page
tives a simplified family tee ofthe Germanic languages, which
disregard lite stages such as Old Norse a the common,
ancestor of lelandic and Norwegian, and the older stages of the
‘modern languages such as Old English, Old High German, ec
Linguists vary in their interpretation of the informational
value of such family tres. For some this is only a convenient way
of visualizing the degree of genetic eelationship between lan-
‘guages, and labels such as North Germanic are seen as a cover
term fora group of languages showing closer similarities beeween
Family tree of Germanic languages
roto-Germanic
West Germanic NorthGermanic East Germanic
Duh Danish Gothic
English Ieelandic
Fesian Norwegian
High German ‘seednh
ow German
cach other than with th other languages of the family. For others
{and this was how it was fist conceived) this is a model of
linguistic change which directly shows the way in wich proto-
languages diversify into daughter languages. In such a view a
label such as North Germanic sands for an actual language, an
ancestor which acted as an intermediate proto-language. (Foran
alternative tothe family tree model, see Readings, Text 4.)
Correspondences between languages
“The comparative reconstruction tha enables us to establish such
language families is based on identifying correspondences between
related languages. These correspondences are most evident on the
levels of phonology and morphology, i. sounds and inflections,
and are accessible through the systematic comparison of so-alled
‘cognates. These are words which are similar both in form and
meaning and which go back to a common source. Cognates are
particularly frequent in the basic vocabulary of daughter
languages, since words which relate ro basic aspects of lfe orto
‘common human experience (such as time, place, food, or socal
relations) tend to be les readily replaced by borcowings from
other languages.
‘The following simple example demonstrates the basic
principles of comparative reconstruction. French champ, Italian
‘compo, Spanish and Portuguese campo all derive from Latin
‘campus ‘eld and are thus cognates. Fven if Latin had noe been
"preserved in a wealth of writen records, we could partly recon
Struct it by comparing such cognate forms of its daughter
languages, as illustrated inthe table below (the meanings ofthe
respective words in English re (x) ‘dea’ (2) ‘eld (3) house)
Correspondences between Latin and Romance languages
“Tata French alan Spanish Porragnese
T carnaikh—cher(fer|caro(k]—caro(k]caro[k]
2 campue(k) champ) campo) campo tk] campo tk)
3 cack] cheelfel'a caua{k] caualk} casa
The table above shows three sets of cognates from four
different Romance languages, as well asthe Latin form, Having
established the sets on the basis of ther similarities of form and
meaning, we now proceed to recoastruct the original sounds of
the proto-form of each set. For this purpose we establish
systematic sound correspondences within and between the sets of
cognates: in all three sets, for example, the French initial
‘eonsonant[f] (asin English shoe) corresponds o [ke] inthe cree
‘other Romance languages. There are evidently three possible
sources for these intial consonants in the common proto-
language: (i) it was (k} a8 preserved in Italian, Spanish, and
Portuguese; in this case, the original sound would have
undergone a change only in French; or (i it was original [f] asin
French, in which ease it would have changed ico [k] in the other
three languages or (i) it was neither [k) nor {s}, but another
consonant from which therwo attested ones developed as result
of diffeent sound changes. Asis evident from these hypotheses,
the reconstruction of the proto-sound also involves. recon
seructing the sound changes which occurred in the individual
daughter languages.
There are certain general methodological principles that we
«an bring to bear on deciding on the proto-sound. The most
important of these are:
(i) Any reconstruction should involve sound changes tha are
phonetically plausible, The phonetic plausibility of a change is
evaluated on the basis of general phonetic considerations as to
how sounds ae formed as wellas onthe extensive documentation
‘of sound changes in other languages. On tis evidence, a change
from {ki to [f] is more frequent and plausible (and hus mote
‘natura’ than a change inthe opposite direction, [j] > [x leis
even more natural i ie proceeds via an intermediate stage [Uf],
‘whose existences well documented in the history of French, This
stage is also reflected in carly English borrowings from older
French such as Charles, chief, which have preserved this earlier
French pronunciation with [tf]. On the basis ofthis principle of
phonetic plausibility or ‘naturalness’ we reconstruct “[k] a the
initial consonant ofthe above sets 1-3 for Proto-Romance, in
which che asterisk * indicates a reconstructed form, without
written evidence
Ui) A second, though lest rciable principle, isthe ‘majority
principle. Any reconstruction should involve af few changes
berween the proto-language and its davghter languages a5
possible. Thus, the sound which is more frequently met in the
related forms is more likely to be the original one. In sets 1-3,
three languages have (k} while only one has [f]. The secon:
structed proto-sound *[k] isnot only more frequently found in
the corresponding sets, but also involves only one sound change
inthe history of the daughter languages for French), while inthe
case of reconstructed "Ij three languages would have undergone
the same change *[] > [k.
Continuing. with our reconsteucion in. this manner will
eventually resale for sets 2-3 inthe Proto-Romance forms "caro,
“campo, *casa, although this reconstruction is not always a¢
casy and straightforward as with *[k]. These reconstructed
Proto-Romance forms are quite close tothe attested Latin forms
given in the table (as an inflected language, Latin actually had the
ase forms caro, cmp)
In the case ofthe Romance languages we are obviously in the
fortunate positon of being able to verify our reconstructions to a
lange extent, and thus 10 test our methodology and basic
assumptions, In general, however, we can only reconstruct those
fearures ofthe proto language which have left atleast some trace
in one of the daughter languages. Thus the quality of our
reconstruction crucially depends on the quality of the survivi
evidence
Ac the same time, the individual reconstructed sounds must
also forma plausible complete sytem, which should furthermore
9conform to more general principles of sound systems. Languages
tend to have symmetrically struceured sound systems and there
‘would have to be compelling evidence to disregard this general
tendency for a reconstructed proco-language. To illustrate this
with an example or two: a language having a set of so-called
“voiceless stops’, ie. [p. . KJ and the corresponding ‘voiced
‘stops [b, 4, ] i moze likely (or natural) than one with [p, 1, &]
butonly b, gh ie- without (]—chough gapsin asystem do occur
in natural languages. Similarly there are no known languages
‘which have only nasal vowels (asin French champ, cf. above) and
‘no 'pure’,e-non-nasal vowels, and only very few which have no
nasal consonants Sach general "typological considerations mast
influence the final shape of any reconstructo
‘We should aote here thar there ate certain reservations about
this method of analysis. In the first place, the reconstructed
roto-language i (misjrepresented as an idealized homogeneous
system, whereas in face natural languages are necessarily hetero
iencous and variable. Secondly, sound changes ace presented as
being regular and occurring without exception in all identical
‘contexts. As we shall se later, both these views have come under
attack in recent decades, though the main esuls of comparative
reconstruction, ie. most ofits hypotheses about the shape of a
number of proto-languages, have stood the rst of time. (For a
discussion of the status of reconstructed forms, see Readings,
Text.)
Laws of change
In the reconstruction of linguistic relaionships and develop-
iments we identify certain processes of change that ate so regular
a tbe considered las. One ofthese is evident in the develop
ment ofthe Germanic branch of Indo-European,
The Germanic languages show a series of distinctive and
related sound changes involving certain Indo-European conson-
ants. These are accounted for by Grimm's Law, named aftr the
jerman linguist who discovered it It tates that Indo-European
stops, ie: consonants produced with a bref closure such as [p,b,
1d, k, gh regularly changed into different consonants (the stops
Ip. Kh, for example, becoming the fricatives [f, 8, ah, i.
consonants produced with an audible friction due to the
narrowing of two speech organs; see Chapter 5 for phonetic
decals). More specially, Grimms Law staes the following,
stands for ‘develop into’ ‘voicedvoiceess refers tothe presence
br absence of vibration of the vocal folds ‘aspirated refers to an
atadile breath)
voiceless stops [pt k]> voiceless fricatives [fx]
voiced stops [b, dd] > voiceless stops [p, tk]
voiced aspirated stops [b8, db, gh > voiced plain stops, do]
This law applies to all Germanic languages. Other Indo
European languages, such as Greek or Sanskrit, basically show
the original sounds of Proto-Indo-European. This may be seen in
the following table, which gives (i) one example each of two
changes of voiceless stops to voiceless frcatives (lp >My [t> 8)
and of voiced stops to voiceless stops (fd > tp fg > k) inthe old
German
languages (Gothic, Old English "OE", Old High
2 ‘OHG'}, (i) examples of their preservation in two non-
manic Indo-European languages (Sanskrit and Latin), and
(i) the reconstructed Proto-Iado-Furopean (PeE") forms.
Reflexes ofthe working of Grimm's Law
Sanskrit Latin Gothic OE ONG Pile
pad: fot fot fox *p “Toot
trdyas rial brie det tthe"
dau). duo tai tw usd dew!
fras’ gems tant cmd haat “g “ace
[As shown above, comparative reconstruction is based on
consideration of related forms in genetically related languages,
Its basic methodology ca, however, also be used to reconstruct
unattested ealir stages or gaps in the tradition of a language by
‘comparing related forms from within a single language. This is
the domain of intemal reconstruction, which relies on the
linguistic races leftin a language from its earlier stages.Intermal reconstruction
Alllanguages show patterned alternations in diferent realizations
‘of morphemes, the smallest meaningful units of language. Thus the
English regular plural morpheme is cealized in three different
variants (or allomosphs), namely 2! and Fl, cf. ats (kets),
dogs [dog borses [hss]. Internal reconstrction starts fom
‘the assumption that such synchronic alernation is in general the
result of regular sound changes and chat the different forms have
developed from a single non-altenating form. A simple example
toillustrate this German words endingin voiceless tops, While
some of these, such as Rat [e] ‘advice’ and Lack [k] ‘varnish’,
"etain the voiceless stop in inflected forms, others sch as Rad [t}
“bicycle” and Tag [k] ‘day’ show alternation between a voiceless
stop in word-fial postion and 2 voiced one in the inflected forms
(asin Rades, Tages, with the genitive singular ending ines)
frat) fect)
flak} aks}
‘The mon-altemating ancestor of the variants [racvrasd] and
[taskta:g-] ofthe moephemes meaning “bicycle” and “day” could
have been ether (i) frat} fac], (i) red, [ag or fii) an
lunatested different form. Since internal reconstruction should
follow the same basic principles applied in comparative econ
struction, the hypotherical changes leading to the alternations
should fall the criteria of economy and naturalnes, and should
‘not lead to contradictory results in other forms. Looking at the
sound system of German, we notice that voiced stops never occur in
\wort+-final position, so that iis plausible to reconstruct the earlier
non-alterating single forms with the voiced stops" [aa] ‘ay’ and
* [sc ‘bicycle’ and a subsequent sound change of devoting voiced
stops in word-fnal position, while che voiced stops remain in non
final positions. We have thus reconstructed both the orginal non
altenating forms *[tagfasgas] and the sound change of final
devoicing, which hasled othe modern alterations [ak/tagos] OF
‘course, intemal reconstruction is no aways a straightforward as
this. There are alternations which are mich more complex and
where the original sate has been obscured by multiple sound
‘changes. In such cases, the reconstruction also has to establish the
‘advice’ frat) fracdos) “bicycle”
‘varnish’ [eck] [tagos} “dy
‘elatve chronology of the various changes, i. the order in which
they have aken place. However ifaspecifcchange has eft no trace
in the language, internal reconstruction may lead to a grossly
simplified description ofthe intermediate sags. Furthermore, we
‘eed to note that not all alterations can be traced hack toa single
‘on-alkemating form,
Tntcenalreconstrcion is most feutflly applied in cases where
we have insufficient material for comparative reconstruction, as
in the case of isolates, ie. languages with no know eelatves, or
‘when only very distant, posibly controversial relatives exist. In
such cases as well as with languages without any textual
tradition, internal reconstruction may be virtually che only way
tolearn about the unactested linguistic past. Ocherwise it should
ideally be used in combination with other methods such as
comparative reconstruction.
‘The present chapter has shovsn how the historical linguist—
much ike an archacologist—can piece together information even
fon unrecorded stages of languages and processes of change.
‘These methodshave been most secessully applied to phonology
and morphology, while syntactic reconstruction is more con-
troversial. ‘The reconstruction of che proto-vocabulary in core
areas such as kinship, plants, animals and metals has also
increased our knowledge ofthe societal structures, the economic
‘organization, and the original homeland ofthe Indo-Europeans.
‘Let us now turn to a more detailed consideration of linguistic
change at the different levels of language.
3Thus sound change can lead
‘morphological change, and this in tun may have consequences
a aa on
dlifferentlevelsof language, We will begin with the changes which
are most obvious, namely ths
Speakers constantly have
The unparalele cl increase in the
number of words and word meanings from Old English to the
present day makes English particularly wel suited for a discus
son of lexieal change
‘There are ewo main strategies for the introduction of new
words, nam
Languages may
ad abou hE REEGHES EBON aN, ie 7 pr26
«ent of modeen English is sad to consis ofloan words from more
‘an So diferent languages primarily Latin and French.
“There are obvious historical reasons for this extensive Borrow:
ing such as thee COR UESTO ENG|ARAIALND «and chehigh
prestig@fLatinastheinremational language of communication
‘GEO the eighteenth cer. Buc ic is ultimately che attitude of
speakers towards foreign influence in general and towacds
specific languages in particular which determines the acceptance
andl@eGree OF BGFFOWIng. Since borrowing is the result of
Jaguage cant, wil be discussed in more detail in Chapter
"rte folfowing secions, we will focus on the other strategies
‘which speakers gto festa newer changed concepts, nary
‘he coining of new words and semantie change, .c. change in
‘word meaning.
Coining new words
AAs the smallest meaningful units of a lnguag@/MORSReMee
constitute the basic building Blocks of words. New words are
‘normally formed by combining existing words and morphemes
into new, complex words. Frclsh reach consists oF a
:morplieme, while fzhcer can be analysed into two meaning
_units:The suffi -ay, which signifies “agent™In Teacher, and
“inst ORT TF VOI, Ws SO-TAMEN BOUHTOFPEATE, hic
When discusing historical word-formation, we must dif
feremate between the appearance of new words. the outpot
‘of word-formation rules, and these abstract rules themselves.
(Word-formation rules are bot languagespesific and subject to
clachronic chanye, especially in rezard to their productivity i.e.
the frequency and flexibility with which they are used to coin new
words.
“Two of the most important word-formation processes are
‘compounding and afxation. Compounds are the combination of
io independent words, i.¢ fee morphemes, ike guesthouse
kc quest + hous), wiiliafaixationa Bound morphemeisadded
10a base, asa prefix (eg wnlike) ora suffix (likeness)
Compounding. hss beet Righly productive througHGWnc
(try oF Rngish, and innumerable compounds were coined
cover the centuries. Many ofthese have survived since Old English
‘ims, ARE Kise others ave HONE OUD ws,
such a Old English which was replaced bythe Trench
loan fan, though thas survived aaa aR
and as ssa) the basis of the semantic lations Denyse
thee diferent parts we cd distinguish ilferene subtypes of com
pounds, such s guesthouse house fr guest. glfiond rend
Who isa gifs Not al of these subtypes have boo equally
productive in the history of English and iis act too dificul co
‘magine compounds which ae unaoeptable in English. However,
English, and German likewise, han Fewer SOSA
compounding than some other languages sucha French,
‘Ava result of sound changes compounds may lose their
teansparency and develop into unanalsabl simple words such
a= Old English qgodapell 9008 dings RaRpel, Few ‘non-
specialists are aware that English fore hy go back to the
.
Ina similar way,many feums refering to women have undergone
pejoration, while she corresponding terms for men have remained
hettral oF have improved, ct. master vs. mistress, bachelor ys.
SPinsterwhick agin vftors che traditional lower status of
‘women in our socery.
Extensive shifts of meaning often cloud the relationship
between the original meaning and the modern one(s), as in
English dll) (< Old English (g)salig‘Happyeblesied’, 2s sill
preserved inits German cognate sli). look atthe incermediate
‘ages, howevenmleesthe development more tinspatent. From
‘ts original meaning attested till the late 15th century silly passed
through the fol nM SERRE NRE” (ate x3thc—r8the.)
> (@eSeRNgBIy”(c_1300-19th <-) SAW FeRbe” (x3ch
roth c), Simplegiaibfane (r6th cc. 1800), feeblesninded)
(16th c- coday) MoishleRipepeaUED late 6th c.— today)
‘These dates illustrate that typically an old meaning is not
replaced immediately by a new one, but that both coexist for
some time, each in specific contexts. A semantic change very
similar to thavOFily happened to Freich erin 'stipid person’,
which, ike French chretien “Christian', derives from Latin
bristianus ‘Christian. (See also Readings, Texts.)
3