THE SOCIAL DIFFERENTIATION OF ENGLISH IN NORWICH

P. J. Trudgill

Thesis Presented for the Degree of Ph.D.

University of Edinburgh. 1971-

 I
A
cTj^ov !<?. tents

X aia very prcutly in .ubte . to ny .vipervisors,

-
«L« Jones und J«Y» Pathcrt £or tho h©lpf advic© and

tneauraj.-©went I receive! fro.3 then while I was woT*<.±n{*

on this thesis* I an most grateful to both of then*

I an also very yratefwl to I tofeasor J* Lyons lor

his one© urn. ;cn©nt$ particularly in tho Initial sta es

of this work, and for his help it overconixt;; certain of

tho practical difficulties I experienced at that stfi'.e.

I would also lid© to t mnk . *..» Brown for r.roviein^

son© of the initial inpeiuo to this worf;» ami rin -*3

Coliof-e* Cmthrive, for ^Ivintf «ie son© nuch ne* VP

financial help*

X nust t in addition* express i ;y ©special thanka to

d.A# GRiuia, who road the entire thesis and nrw© uor o

oxtrcucly valuable coi.nents un<• aud.pestions. 1 have

also profited from discussions with A*.I* Foldvih*

b»C# fuclr©» . «L* f'od^nrt, », h rnowles, and p.-rtic-

ul' rly J.s ..1. Lawr, who provided invaluable assistance

in the preparation of one ol t a chapter^*

the following people unvo boon v. ry helpful in

providing- mterlal which 1 have used In tit© preparation

of the thesis t '•# fills i Professor A»r.. t tiers on»

Professor a , .
llalliday, Professor •£• »c -*avid, to
whom I owe especial thanks# s.F* Sanderson# F.M* Tilling#

and Professor W*N« Francis# for whose comments on the

Norfolk dialect I am also very grateful*

Several people who helped with the dialect survey

In Norwich must be thanked* 1 am particularly grateful

to J«A« Hannah# who carried out ten of the Interviews#

and made my task that much easier* Z am also very

grateful to G.E. Carter# J*N, Iiewltson and F. Kapherr,

who helped me with the selection of younger Informants|

and to my father# for the use of his ear# which saved

me a great deal of time# and ray mother* for her board

and lodging# and helpful comments on Norwich speech*

I am also, of course# very grateful to all my sixty

anonymous informants# most of whom will never see this

thesis# and particularly to the gentleman who refused

to help me# because he did not believe in universities*

Finally# I want to thank Carol Geddes for typing

the thesis and coping with my handwriting) and my wife

Sandra for her help and encouragement# and for so

usefully acquiring a Norwich accent during the last

three years*
 Summary (ill)

This work la a study in urban dialectology,

sociological linguistics, and generative phonology, Xt

takes the form of an urban dialect survey of the city

of Norwich, England, and is particularly concerned with

the correlation between phonetic and phonological aspects

of English, as it is spoken in Norwich, and various

sociological parameters, Sociological linguistic

research, it is claimed, Is of value both in the ob¬

taining of new and accurate linguistic data concerning

■

the linguistic characteristics of the majority of the

population of modern Britain, and in the solution of

problems of linguistic theory, Xt may also be of value

to sociologists and sociology*

The city of Norwich, together with its associated

suburban areas, has a population of approximately

160,000, Xn relation to the sine of its population, the

city has a very large rural hinterland, and ranks high

in the English urban hierarchy. This has some linguistic

consequences. A random sample of sixty Informants was

drawn from this population, and a tape-recorded interview

conducted with each of them. The interviews were based

on a questionnaire structured so as to produce different

contextual styles of speech, and the informants have been

classified according to an index of social class

characteristics.
 Norwich English, although a dialect of the South-

East of England, differs phonologically and phonetically

in sons interesting trays from the English of the Hose

Counties* In this work, certain of the phonological

features of Norwich English are selected as phonological

variables* By means of the development of index scores

for these variables, a study can be made of the exact

nature of the eo-varlatlon of phonological features and

sociological phenomena, such as social class, social

context, age, and sex. Certain conclusions arc drawn

from these measurements concerning the nature of the

class structure, the Norwich phonological system, and

stylistic variation.

In spite ef the many differences that are to be

found within the Norwich speech community, the vast

majority of its members are distinctively speakers of

Norwich English. This overall similarity is expressed

by means of the development of a Norwich diasystem,

which consists of a set of abstract underlying phono¬

logical elements common to all speakers, and a set of

rules which generate the various types of Norwich

English and relate them to each other. For several

reasons, the diasystem is developed within the

theoretical framework of generative phonology, rather

than any other theoretical framework,

 The largo amount of accurate data that has boon

obtained during the course of the survey front Informants

of all age groups moans that a study can also be made

of linguistic changes that have occurred and are occurring

within the Norwich dlasystem. Certain conclusions are

drawn from this study concerning the nature, causos,

mechanisms and effects of phonetic and phonological

change*
 CONTESTS

Parre

Section At Introduction

Chapter I» Sociological Urban Dialectology 1

Section D» Goographf and Sociology

Chapter 2» Norwich 11
Chapter 3» The Sample b5
Chapter 4t Social Zndleeo 6b
Chapter 3* Social Context and the
Questionnaire 86

Section Ci Language» Social Class and

Social Context

Chapter 6i The Social Differentiation

of a Grammatical Feature 108
Chapter 7* The Phonological Variables 127
Chapter 8t The Co-Variation of the
Phonological Variables with
Sociological Parameters 178

Section Di The Dlaeysteia

Chapter $>t The Theory of the Diasystest 237

Chapter 10* The Norwich Dlasystem &99

Section Ei Linguistic Change

Chapter lit Linguistic Chang© in Norwich bX7

Kotos 328

Appendix i The Questionnaire 56b

Bibliography 368
bi.CTION* A ( INV;v>» yCTIOtf
 1

Chapter Qnet Sociological t?rban Dialectology

Liasnists have employed several different terras to

signify that area of study which is concerned with the

relationahipe between language and society. These

different terms, moreover, have often been defined in

different ways, for different purposes, by different

linguists. These different terms include the following!

1 2
soelellnguistles, institutional linguistics,

anthropological linguistics, ethnolingulstics, and,

perhaps, paycholinguistics. For the purposes of the

present study we will subsume all these terms under the

heading of soci of ingui atioit.

Sooiolinguistics can be defined as that branch of

linguistics which employs or is concerned with the

methods, findings or subject matter of the social

sciences. As, therefore, the social sciences are con¬

ventional ly divided into the two main branches of social

anthropology and sociology, so sociolinguisties can be

regarded as consisting of anthropological linguistics

and sociological linguistics. The work that is presented
here can be considered under the heading of sociological

linguistics, in this senset what follows is a study in

sociological urban dialectology. In so far as sociology

is concerned with the study of complex urban sociotles

and communities, we shall be dealing with the subject

matter of sociology. Further, in the application of

 2

sampling techniques to linguistic problems and in the

handling of problems connected with the concept of social

Glass we shall be making use of the methods and findings

of sociology. The sociological aspects of this work will

be most apparent in Chapters Two to Fire.

We shall also bo making some reference to the methods«

findings and subject matter of another of the social sciencest

human geography. Although linguists have long boon intor¬

es tod in what they* havo tormod "linguistic geography",

they have most often ignored the advances that have boon

made in the field of theoretical geography itself. The

4 5
concepts of central place and urban field,-' and the the-
6 *7
ories of location and diffusion' ore all relevant for

linguistic studies. In so far as an attempt will bo made

here to rectify this omission, the following can also be

considered to be a work in geographtoal linguistics.

sociological Dialectology

It should be pointed out that sociological dialectology

Is not, and has not been, nocossarily urban. For example,

Kurath d Mcliavid,3 in their Investigations into the

dialects of the United States, made some attempt to ob¬

tain informants from different social class backgrounds

(see chapter Throe). As, however, their methods

 3

of securing Informants cannot be considered sufficiently

rigorous to satisfy tho higher demands ra.ule by socio-

logical theory In this field, their work is perhaps

bettor described as social, rather than sociological,

dialectolo.jy.

Urban -'ialectolory

It la also true to say that urban dialectology is

by no means necessarily sociological, .'.'any linguists

have attempted to describe the &poooh forms of urban

areas without recourse to any of tho methodology of

sociology. SJome work of this type will be discussed in

Chapter Throe, and ayain in Chapter Nine, whore it will

be ar{juod tliat the failure to adopt t?ie sociological

approach ©agential for accurate work in urban areas has

unfortunate linguistic as veil as sociological con¬

sequences.

Lin-rnlgtic diversity

The inadequacies in the work of those linguists

stems from the fact that they have, generally speaking,

chosen to ignore the fact that most if not all speech

corluunlties are riore or less socially and linguistically

hetorogenoous* This heterogeneity is, moreover, much more

 k

marked in urban arena than it is in ether linguistic

communities• For this reason the inadequacies of non-

sociological urban dialectology are all the more serious*

The application of sociological techniques to

linguistic material represents, historically speaking*

an advance, in that it permits not only the recognition

of the fact of linguistic diversity but also the develop¬

ment of a methodology for handling this diversity* Many

nineteenth-century linguists, for instance, vers totally

uninterested in the heterogeneous nature of linguistic

communities, and especially so sinoe non-standard dialects

a
were often considered to be debased farms of the standard,'

In their historical reconstructions of * proto"-languages

by the comparative method, moreover, no recognition was

given to the fact that these languages must themselves

have been internally differentiated.

Since that time, the heterogeneity of speech-

communities has generally been acknowledged* Many

linguists, however, have contrived to avoid facing up to

the fact of this heterogeneity. This practice was

particularly characteristic of American linguistics in the

I95O*s* Hookett, for example, has stated that, in working

with the comparative method to reconstruct "a single

language free from dialect variation*» we are making "a

potentially false working assumption*•Uockett never-

 5

theless makes this assumption. One knows, in other words,

that languages are subject to internal differentiation,

but, for practical reasons, one pretends that they are

not* In the synchronic treatment of modern languages,

moreover, methods were devised to permit the inconvenient

fact of linguistic diversity to be ignored* Great stress

was laid, for example, on the fact that analyses were

only relevant for one idiolect in one style at one time**"*"

(in the last decade, on the other hand, it lias been

shown that, in some cases at least, the idiolect is the

least stable object for linguistic study, and that it can

only be viewed as coherent against the background of the

12 v

linguistic community as a whole* ) Much of linguistio

diversity was also dismissed as "free variation". Later

studies have shown that most of this variation is not at

all "free", but is on the contrary structured and socially

determined in sociologically and linguistically interesting

ways*

Since the last war, however, and particularly in the

United States, there has been a steady increase in the

number of works which have both recognised and attempted

to cope with the fact of the non-homogeneity of language.

From the point of view of urban dialectology, undoubtedly

the most important of these works is Labov's study of the

11
speech of the Lower East Side of New York City. Labov,

whose work will be mentioned frequently in the course of

the following chapters, has applied sociological methodology

 6

to a linguistically heterogeneous community with results

that have several important implications for linguistic

theory# The following work can perhaps*# therefore, be

seen as part of the trend towards studies dealing with

heterogeneous «porch connunitics ^nd atto ptirit: to deal

with en'! draw conclusions fron linguistic diversity#

The Value of Sociological Urban Pinlectolo.-ry

What is the value of sociological urban dialectology*

and wti.it aro its alias? On© of the nost obvious results

of this type of work is the accumulation of a whole new

body of linguistic data# Labov ha«i sold, however, that

the interpretation of the tern sociollryatisties to signify

a new intordlselplinlary field doalin.j with the description

of the relations of lanrrun.,® an society is "an unfortunate

notion, foreshadowing a Ion- series of purely descriptive

studies with little beariiii? on the central theoretical

tk
problems of lirvrnistics or of sociology"1. Labov's con-*

donnation of "purely descriptive studies" would appear,

however, to be somewhat over-severe. "Purely descriptive

studies" ©f r»tral dialects have lon^ boon r©,-urdod as

legitimate ana worthwhile linguistic pursuits, hhiie

they nay have contributor not .in ; directly to the solving

of problems of llu rutstie theory, they have added to the

linguist's knowled;*u e.T>»»ut lan :ua. e, and have provided

data which has subsequently been turned to solving various

 7

15
linguistic problems, ■* In exactly the saroo way, socio¬

logical urban dlalcctolofjical methods con make possible

descriptions of urban speec t that aro in themselves of

value and interest* They increase the body of linguistic

data available to linguists, an.' ior this reason should

not b© re?mired to have any direct boarlrv: on theoretical

probleru *

At the aano time, it wuat be recognised that one of

the main alna of research of this type is to shed llj<jht

on various problems and other as pacts of lincTuistic theory*

Studies which aro able to do this aro obviously of nore

value than those that are not. Sociological urban

dialectology provides the linguist with a body of data

that is not only lart:e but also accurate* Soiie of the

conclusions we shall arrive at on the basis of this kind

of data In tho present wor?e - such aa, perhaps, the con¬

clusion that wonon use f w>re "correct1* 1 Insula tic forms

than men — may appear to bo rather obvious* Vs can claim,

however, that w© have presented raato.ial in an exact and

ricTorous way, which prove3* perhaps for tho first time,

that (in this case) sex differentiation of this type

occurs in Prltish English, and which demonstrates th©

exact nature and decree of the differentiation* This, w©

bcliovc, is a useful advance.

The lar:;© amounts of data that are obtained in this

way can be applied to many theoretical problems, and in

 3

this present work wo shall attempt to make some comments

on phonological theory, and to make soio contribution to

the theory of the dlasystem. Th© fact that th© data is

drawn from a large, dynamic and complex community, which

is characterised by many different types of social inter¬

action and by various types of social change, also means

that it i® particularly useful in the study of linguistic

change*

Sociological urban dialectology can also have th©

function — particularly in Prit&in, where little attention

has ©o far been paid to this kind of work — of providing

a description of th© linguistic characteristics of th©

vast majority of the country's population. It would soera

that the considerable amount of rural dialcctologlcnl

work that has been carried out in liritain has left th©

linguist singularly ignorant about th© way in which most

of the people in britain speak# The aims of rural

dialectology have, of course, been of a different nature,

Hany dialectologists have, lcgitipnitely, been concerned

to record older dialect forms before fcboy are lost for

good. The result has been, however, a neglect of current

speech forms which could have provided an excellent

"laboratory1* for the testing of linguistic hypotheses *

ftural dinlcotologists, too, can bo accused of having

neglected the heterogeneity that is prcoent oven in

rural speech comnu si ties.

 9

Material from sociological urban dialect studies

oan also bo useful for practical purposes* especially- in

the field of applied linguistics. Because we can now

know* often for the first tir*o, the exact nature of the

linguistic characteristics of large sections of the

population* it is possiblo to point more accurately to

the sort of difficulties that can arise in the teaching

of Standard English to children who have some other

variety of English as thoir native tongue#It is also

possible that conclusions will emerge that will be of

some value in the teaching of English as a foreign

language. Urban dialectology demonstrates, for example*

that ft# P. is very nueh of a minority accent* even in

England, This suggests that the teacher of English

should perhaps not concentrate oo exclusively as is now

usual on the teaching of this particular accent.

Sociological urban dialectology can also be of some

value to sociology. It oan ahod light, for instance, on

problems concerning the discreteness and continuity of

social classes, on certain aspects of role and status,

and on the class structure of tho con ntnity generally#

It can also provide material that can bo used in tho

study ©f reference groups, normative pressures, and

prestige patterns. (¥o shall have something to say in

the following chapters, for example, on the nature of the

class structure in modern Britain#) This Kind of work

 10

can also bo of uso in pointing to son© of the barriers

that exist in the way of social aal educational advane©•

stent for siany member a of our society# It has been shown,

for example, that linguistic differences not only arise

17
from social inequality but also help to reinforce it#

Mainly, however, w© shall be concerned in this work

with problems of interest to linguista aid linguistics#

The following work, which makes use of both sociological

and dialectological techniques, is the first attempt that

has boon rand© to describe the speech of the urban area in

question. It is also one of tho first essays in British

[
sociological urban dialectology. In tho following chapters

we shall first of all describe how tho urban dialect survey

*

was carried out and then attempt to illustrate, in some

i
l

detail, the exact nature of tho heterogeneity of the

linguistic community in question. V© shall then attempt

to establish a theoretical framework for the description

of all typos of speech that occur in the urban area and,

finally, describe and discuss the theoretical implications

of the linguistic changes that have been and »ro taking

place in this particular area#

SBCTXOK 8 t OKOORAPHY AND SOCIOLOGY
 11

Chapter Twot Norwich

This work in urban dialectology and sociological

linguistics tok©3 the fom of" a study of the English

spoken in the city of Norwich, England, More exactly,

it is a study of the speech, forms of the urban area

associated with Norwich, since the newer suburbs, which

form part of an organic whole with the older city, remain

outside the administrative boundary of the County Horough.

The population of the built-up area of Norwich in

1P67 was appraxin&toly 160,000, with about 118,000 actually

1 2
within the city boundary* This means that, at the moment,

the nearest town to Norwich which has a population exceed¬

ing that of Norwich is Greater London, which is about 120

miles away by road. In fact, Norwich, although by no

means one of the largest towns in I nglaxvi, is, as the

above information may suggest, of considerable cultural

and commercial importance for the surrounding area of

Norfolk and indeed for East Anglia as a whole. Like many

urban contres, it has acted as a goal for in-nigra tion

from the surrounding rural areas for much of' its history,

This is still true of today, although much of the movement

into the city is now of the commuting type.

Those factors have Important linguistic consequences.

The speech of Norwich is clearly founded on East Norfolk

rural ©pooch, but has become increasingly ffiJToronfciatod

 12

from awro rural a peach forms over the years • Ons can

•peculate that this differentiation has taken place, at

least partly, as a result ef the spreading of linguistic

Innovations, sot gradually across country, hut along the

mere important lines of communication from one member of

the urban hierarchy3 to another. On the other hood,

Norwich aleo exerts considerable linguistic influence en

its rural hinterland,^ and cultural innovations of all

kinds tond to spread from Norwich outwards*

It is, for example, a well-known fact amongst Norfolk

people (see Chapter Eleven) that "Norwich people drop their

&»8*. This is a linguistic fact which is striking even to

the layman, as the rural accents of East Ahglia, possibly

alono in the South of England,3 consistently preserve J|

in their phonological systems in all styles of speech*

One can therefore assume, albeit tentatively, that "fc-

dropping" is a linguistic feature that has spread to

Norwich from the Heme Counties, leaving the surroanding

rural areas unaffected, at least for several generations.

It now appears that "l^xlropping" is spreading to country

areas around Norwich amongst younger people*

It is worth noting in passing that the linguistic

changes due to diffusion that we are referring to here

arc of the type called by Labor "linguistic changes from

below",** that is to say, from below the level of conscious

awareness* Linguistic changes "from above* are likely to

 13

be of a normative type tending to modification in the

direction of H„P.» due to the influence of n.P.-speaking

prestige-groups, the educational institutione, and

possibly the mass-media. (This point will be taken up

again later* Dialect mixture in Norwich, and the mech¬

anism of linguistic diffusion into and within East

Anglia, will be discussed in Chapter Eleven)*

The importance of Norwich to its surrounding areas

suggests that it is first of all necessary to discuss

the place of Norwich in its East Anglian background.

This will have several purposes. Xt villi

(l) give an illustration of the kind of area

East Anglia is, and of the kind of people

that live there!

(ii) clarify the oity*s strategic position

with respect to Influences spreading from

Metropolitan London)

(ill) explain its function as a source of

innovating linguistic pressure on its

hinterland)

(iv) explain the dialect mixture that occurs

in the city)

and (v) help to explain the attitudes people

have to their city and to their speech.

ften,.AWlW

As a geographical* or indeed cultural or linguistic

terra, "East Anglic" (EA) can only be very imprecise* The

counties of Norfolk (Nfk) and Suffolk are certainly EAn,

and in many usages the term refers only to these two

counties* There has always been doubt, however* ae to

whether or net the Fenland should be considered part of

EA, and consequently some definitions have included all

or part of the counties of Cambridgeshire* Huntingdon*

shire, Lincolnshire and Northamptonshire* Essex* which

is culturally and geographically closely linked with

Suffolk and Nfk* has also been included, as have even

Bedfordshire and Hertfordshire*

In setting up EA as an economic planning region,

the Department of Economic Affairs has recently redefined

the area for its own purposes* It incorporates under

this heading Nfk, Suffolk, Cambridgeshire, Huntingdon*

shire and the Soke of Peterborough* Much of the

statistical information in this chapter concerning 1966

or later years is taken from the first publication of
*7
the East Anglia Economic Planning Council* and must

therefore be taken to refer to EA in this sense*

Linguistically speaking, EA* as a relatively

homogeneous dialect area, has been shrinking gradually

under the Impact of the spread of Home Counties speech

 13

forms* It would seem* for instance* that It does not

now sake very good sense, except perhaps at the level of

the very old* very rural speaker*** to Include Nfk and* say,

Hertfordshire In the same region, as has been dons pre*

9
vlously. for this reason* "Bin", as a linguistic tern*

mast he employed in a rather restricted sense* To avoid

confusion, the following conventions will he adopted in

the linguistic sections of this worki

(a) nNfk* linguistic forms will refer to the

speech of control and eastern Nfk and

north-eastern Suffolk* This area is more

or less co—extensive with the Seh "city

region* proposed by the EABPC,*® but

stretches further west to the edge of the

Fenland, and probably includes King's Lynn*

(b) "EAn" linguistic forms will refer to the

speech of Nfk* B.Suffolk* an ' those parts

of tf *Suffolk, eastern Cambridgeshire and

north-eastern Essex which remain distinct¬

ively Bin* or relatively unaffected by the

speech forme of the Home Counties*

These definitions are necessarily imprecise in the

absence of any thorough dialect survey of the present-day

speech-habits of all sections of the communities In these

areas.
 16

Coaraurfteatlona

Since the beginning of the Induetrial Revolution*

the history of EA has been one of declining economic

importance, and geographical isolation, followed in sore

recent years by population pressure and increasing

cultural and economic influence from the South-East.

Both the geographical isolation and the influence of the

SB are partly caused by and partly reflected in the

communications networks that serve EA. EA, which lies

between 40 and 120 miles distant from Greater London,

is set apart from the main national centres of commerce

and industry, and from the chief north-south road and

rail routes leading from London. Throughout EA transport

communications are poor, and most of the main road and

rail routes lead to London. Nch itself, the most north¬

erly of the four main EAn centres, suffers most from this

Isolationi "Norwich has become virtually the capital of

Bast Anglla, the attractive, odd and curiously remote

corner of Britain, out off on three sides by the sea and

on the fourth by British Hail. Although this is an

ancient and not very funny Joke# there is more than a

grain of truth In its reference to the difficulties of

communication with the rest ef the country, and this

has had a profound effect on the way in which Norwich

11
developed Over the years and will do so in the future*1,

Nch is in fact served by a good rail link to London.

 17

Trains leave on the average once every hour and conplcto

the journey in alternately two and two and a half hours.

This route la, importantly, via Ipswich and Colchester*

Other rail links are not good, a fact which has recently

been recognised by the plannersi "There must be a better

service across country and to the North and Scotland - for

12
too long all lines have led to London"* It la partic¬

ularly important that rail links to Cambridge are

significantly less satisfactory than those to Ipswich,

although Cambridge itself has a good service to London*

Travel west mid north generally is more difficult, with

fewer trains, and connections at Peterborough and Lly,

In terras of road transport the picture is somewhat

different. Here again the main links arc only with

London, but even these are not adequate* From Neh it is

necessary to travel either 32 miles south—west or 103 miles

north"west before meeting a road of motorway standards

(the A1). Of the main trunk roads leading from London into

EA, the Al impinges only on the western edge of the region|

the AlO leads through Cambridge to King* s Lynn; and the

A13 through Ipswich to Great Yarmouth* Only the All,

which avoids Cambridge and leads through Newmarket, term¬

inates in Nch. Routes from Nch to London arei All — A303

and on to the Al for central and west London| AlkO to

Ipswich and on to the A12 to east London) and All direct

to north—east and east-central London* All those roads,

13
except the Al40, aro overloaded at peak times* Travel
 ia

to the W©st ami North, as with the railways, is more

difficult* Xn particular, the route to th© North, through

the bottleneck of King's Lynn ami along tho seriously in¬

adequate A17 to Newark and the Al, is poor. (Tho figures

shown on the traffic density map published by th© EANPC
14 . .
for this route (aoo Map 1) way be misleadlngly high, as

they refer to tho month of August, whon largo nunbors of

holiday-makers from th© , lidlanda and tho North arrive in

Nfk fron this direction.) Th© planners have again roco'g-

nisod tho problemst "One stratofry problem which Nch • •.

faces is tho difficult/ of road communications with London

and with tho Midlands. Th© distances involved ar© ex-

oggoratod by tho poor quality of tho road links, which

IK

greatly extends Journey times1'* J Thus road and rail

links from I,A to tho seat of tho country are poor and

inadequate, and those that do exist are heavily dominated

by routes to London. Tho position concerning road trans¬

port remains true in suite of the fact that car-ownership

in LA Is higher than in any other part of the eotmtry,

probably because of the scattered nature of tho population.

It is difficult to know, without further study,

exactly how figures concerning railway timetables and

road networks relate to cultural and linguistic diffusion,

and what implications they have for the influence of the

N01:10 Counties in PA. It has been shown, however, that

isogloss bundles correlate closely with traffic network

|; TrafUc itt

Pa>^sth^e.r Qit H&C£$

 19

t f.
patterns In parts of the U.S.4. In fact« it is tempting

to Identify the breads in traffic iknuity shown on the nap

for Bin trunk roads (August 1965) witiv possible dialect

boundaries (see Map 1). Certainly, on the All the area of

lowest density occurs in the neighbourhood of Thetford, near

the eounty boundary, where the border of the "Nflc" speech

area was tentatively set (soo above). Similarly* the break

on the Ala oeours south of Lowestoft, which supports the

decision to include the "Nfk" region of north-oast Suffolk

in this area. On the A10 there is a similar drop In flow

south of King's Lynn* which could again coincide with the

Nfk dialect boundary. Mo significant drop in density is

shown for the King's Lynn - Neh route, A47* but, as mention¬

ed above, these figure® may not be typical for any whole year.

The EAa urban hierarchy is dominated by the four

major centres of Neh, Ipswich, Cambridge and Peterborough,

an« the three minor centres of King's Lynn, Great Yarmouth

and Lowestoft, and Bury St. Edmunds. In one classification

17
of the urban hierarchy of England and Wales, Neh is

described as a "major city*, Ipswich and Cambridge as

"cities", and Peterborough, together with King's Lynn,

Great Yarmouth, Lowestoft and Bury St. Edmunds, as a

"major town or minor city". (In Vest European terms,

Norwich has been classified, together with, for example,
18
Aberdeen and Cardiff, as a 4th order Metropolitan centre.

The other EAn towns are not olassiflod.) Each of these

towns acts as a centre for shopping, social services,

 20

entertainment t and employment • (Their oentral position

ia clearly illustrated by the travel-to—work map shown in

Map 2, Acoordins to the urban hierarchy theory of

diffusion* which both human and Unguistio geographera

20
have recognised as valid, it can be hypothesised that

linguistic innovations are likely to spread to Nch from

Ipswich and Cambridge, rather than or as well as direct

from London, Thie is particularly so in view of tho fact

that the main lines of communication from London to Nch,

as outlined above, lie through these two towns. We oan

omit Peterborough as it lies outside linguistic EA and

is not on a direct route from London to Nch,

(There is no a priori reason, incidentally, why

linguistic innovations should not also spread to Neh

from, say, the Midlands via Peterborough, However, the

increasing linguistic dominance of London is a fact

which the casual listener can notice in the South of

England from Oxford to Dover, and even further afield.

There is no reason to suppose that EA will escape this

influence, particularly in view ef the system of

communications and the evidence of influence already

observed. There is in any case no sign of any linguistic

Influence from other directions. See also Chapter Eleven,)

It is difficult to decide whether Cambridge or Ipswich

is the more important innovation source for Nch, We have

already seen that rail links are better with Ipswich,

 r -c
O
r 75>

MAP 2 : Local authority from which 5

areas percent +

(Norwich, Ipswich, Peterborough, Cambridge) or 15 per¬

cent h- (Lynn, Bury, Yarmouth, Lowestoft) of economically
active population travel to work in the urban centre.
 21

vhl«h i« also Unguis tioaliy mors similar to Noh as veil

as geographically closer than Cambridge* (The relevant

read distances arei Cambridge 62 miles, Ipswich kj miles.)
Some attempt will be made to discuss this problem at

greater length in Chapter Eleven, but, for the moment,

traffic density figure* (Map 1) ean supplement the inform*,

at ion given above concerningthe rail and road networks*

The map suggests that, in terms of road transport, Cambridge

is probably mere closely linked with Itch than is Xpswloh*

(in any ease, the Nch » Ipswich road, the AlkO, is not a
21.
trunk road. ) This is indicated by a much heavier

traffic flow inte Neh along the All from the direction

of Cambridge than along the Al40. On the other hand, it

must be recognised that much of the All traffic is direct

frost London, and that although traffic along the Ak5 from
Cambridge to Newmarket towards Nch is also heavy, much of

this may be earning from the South, tlio S« Midlands, and

the Vest.

ftmsijgratipn

The Migration tables of the 1961 Census and the 1966

22
Sample Census give further evidence of the cultural in*

fluenoe of the Rome Counties on FA and of the way this

influence Is effected. Immigrants (defined as persons who

twelve months previous were llvinr elsewhere) into the

Eastern Region2** in the year to I960 totalled approximately

 22

152,130 or 4*1 per cent of th© population of the Ronton*

Something over one half of these had cone from London

and the South-fas tern Region (77#340)« The next largest

Influx was iron the it. Mldlan i.egion (8,500), A furtXier

break-down of figures shows that In the same period Nfk

received a total of 17»700 immigrants• Of those, 2,960

were in fact migrants from the County Borough of Kch,

Of those coming actually from outside Nfk, nearly half

were contributed by the authorities listed In Table !•

Table 1

I: 'migration to Kf!c. I960.

London bdO Cambridgeshire 390

illddlesex 670 Hertfordshire 330

Essex 660 Sussex 270

Lly 570 Buckinghamshire 200

Surrey 430 Hampshire 190

Kent 440 Bedfordsnir© ISO

Figures for Nch for the sane period show a similar trend,

with lore® numbers of immigrants from tho south-eastern

counties* Tho total number of immigrants into Nch was

4,160, of whom 1,370 wore fron Nfk* Of the rest, nearly

half came from the areas shovn in Table 2.

 23

Table 2

IraaA/tratjon to ffch* i960.

Middlesex 210 La&o&abir* 120

E* Suffolk 130 Essex 100

London 140 Hertfordshire 100

E* Sussex 130 Kent 100

This trend continues In ttu» 1966 Sample Census figures*

whieh Indicate s further influx from the Southeast* Of

the total 1966 population of Neb (116*350)* 94*360 were

born In El* Of the others* 9*340 were bora in the South*

Eastern Region* and only between 900 and 1*300 in each

24
Of the other six English regions*

Population

The population of EA is both the smallest and least

dense of the eight English planning regions* This is* of

course* one reason for the expansion of densely populated

London in this direction* The percentage increase 1931 m

1366 was the largest in the country* however* and this

trend is likely to accelerate in the next fifteon-year

period* The 1967 population of EA was 1*611*910, of whom

376|430 (over one third of the total) were living in the

north-east (Norwich) subdivision or eity-regioa* ualf
of this number were living in 5?ch (113*610)* Croat
Yarmouth (31*910), and Lowestoft (49*160)* The Nfk
 £4

Rural Districts which contain the Jfch suburbs also had

large populationsi Blofield and Flegg (41,130), Forehoe

and Henstoad (31,200), and St* Faiths and Aylsham (54,150),
The region had a somewhat higher than average proportion

of old people, particularly In the north-eastern sub¬

division* the 13^1 Census shows that the age distribution

of Nik is significantly older than that of England and

Vales as a whole* The population agei under five (7*4

per cent) was lower than that of England and Wales (7*8
per cent). The proportion aged over sixty-five in Nfk

(13*8 per cent) was considerably more than that of England

and Vales (11*7 per cent) and had risen since 1951 (12*6
per cent). The population of Nfk increased at a rate of

0*23 per cent a year between 1951 and 1961, compared with

0,44 per cent a year between 1931 and 1951* This places
Nfk forty-second in a list of sixty-two administrative

counties in terms of annual increase 1951 • 19&1* The

population of the largest EAn urban settlements (more or

less continuously built-up areas) in 1906 wasi

Norwich 160,000 Cambridge 105,000

Ipswich 129,000 Peterborough 81,000

An important factor In the future population is likely to

be the organised London over-spill schemes which have

already affected, for example, Peterborough* Ipswich and

Thetford. There has as yet been no over—spill to Nch*

 23

Social and Beonoralo struoture

It is basic to the employment structure of EA that

the region, has a ouch higher percentage of people employed

in agriculture than the country as a whole. It also has

a higher proportion involved in the manufacture of food

and drlnkf in eonstruotion) in professional and scientific

servicesi and in public administration and defence. In

all ether fields, notably heavy Industry, the proportion

is less than that for the country as a whole, ffla-

eapbyment has recently been quite close to the national

level, but the If eh and Peterborough regions have had

higher rates than the Ipswich and Cambridge regions. It

is also noteworthy that a below-average proportion of

children stay on at school beyond the statutory leaving

age.

In terras of earned income per income-tax case, EA

falls eight per cent below the UK average, and has the

lowest average for all regions of England and Wales,

Investment ineorae, however, is relatively high. Within

EA, both total net ineorae and earned income are on

average significantly lower in Nfk than elsewhere. For

total net ineorae per tax case, the relevant 1964-5

figures were i

UK £1,003 P.a.

EA £950 p. a.

fffk £913 P. a.
 27

Norwich

Nch la generally thought to have grown up out of

a number of Anglo-Saxon settlementa made around the

lover-most ford of the Hirer ¥ ensurn, near Its confluence

with the Hirer Yare, By 1066 it was one of the largest

towns in England, with an Anglo—Banish population of

25
about 3»5GO, and indeed at this time east Nfk was the

most populous part of the country. By 1036, because of

the effects of the Noman Conquest, the population had

fallen to around 5,000, a largo number of whom were

Norman, French, Breton or Flemish, Subsequently, large

numbers of dews also arrired, but during the next 200

years many migrants were attracted from the surrounding

countryside by the wealth of the city, and their Influx

reduced the alien proportion in the population. The

Cathedral was founded in 1094, and Nch then became the

administrative as well as ecclesiastical capital of the

region* It was also the commercial capital of a fertile

area producing wool and barley. At that time, trade

routes eastward across the North Sea were as easy as

those leading Inland, because of the great distance from

London and the swampy impassable nature of the Fens,

There was a substantial immigration of Flemish weavers

into tho district in the 14th century, and Nch subsequent¬

ly became the centre of a large cloth-making area. There

was considerable Flemish cultural Influence, and the

shape of parts of Nch even today is not unlike that of

 23

Flemish. towns, with its large open market , and the open

squares known as "plains*. (This is also true of Great

Yarmouth and Kins*» Lynn. )

From the middle of the 16th century, and for two

hundred years or so, Nch was probably the second largest

city in England, rivalled only by Bristol and York, At

the beginning of this period the weaving industry was in

decline, and Queen Elisabeth attempted to revive it by

encouraging new immigration from the Low Countries, This

coincided with a wave of religious persecution in that

area, and by 1379 there were 6,000 Flemings and Walloons,

known as *Strangers*, in the city's population of 16,000*

The impact of those people, and their cohesion and in¬

fluence as a group, can be gauged from the fact that an

annual church servioe was held in Butch in Neh until the

26
1390*s, and services in French continued until 1332,

By tho middle of the 13th century Neh was one of the

wealthiest cities in Britain, with the large population

of 33*000, During the Napoleonic Wars, however, decline

set la, and the city depended, throughout the 19th eentury,

on Ita function as a market town for the surrounding

agricultural districts, It lacked the fuel and power

resources of the North of England and was therefore unable

to participate fully in the Industrial Revolution, with

the result that its place in the textile industry wae

taken over by the towns of Yorkshire and Lancashire*

During tho Victorian period the city gradually emerged,

 29

with the help of the development of the railway*, as tho

mainly commercial and administrative oentre that It la

today*

The Isolation which has affected tfch particularly

in the last 130 years has had unfortunate economic con*

sequences* However, as a local writer has recently

observed, "it is possible for us who are living today to

reflect that from an esthetic point of view isolation had

27
Its advantages"* The centre of Heh constitutes a

considerable tourist attraotlon, and one estimate, based

on current trends, has put the likely number of tourists

to visit Hch in the mid-1970*s at about two million a

23
year* The University of last Anglia is now situated in

Noh, as ar© also ITV and BBC TV and radio stations, la

1961 Hoh was the thirty-fourth town in else in England

and Wales»2^

Population

Since the First World War the population of Hch has

maintained a fairly steady level, as Table 3 shows* The

decline lii population elnoo 1931 has coincided with a

large increase in the number of people living in tho

suburbs and suburban villages. Sinoo 1921, the population

of these areas has risen by ovor 300 per cent* (We have
 30

Population of Neh CB* 1921. * 1^67

Year Population

1321 120,661
1931 126,236
1939 121,700

1931 121,236

1961 120,096

1966 116,350

1967 118,610

already noted that the actual built—up area of Roh contains

about 160,000 people*) in 1961 an extra 21,500 people

(equivalent to 13 per cent of the resident population)

came into Neh to work each day* Tho increases in popul¬

ation of the fire suburban parishes around Nch between

1951 and 1961 are shown in Table 4* Curing this saao

TnbX* 4

Population of 5 suburban parishes* 1931 - 1961.

Parish 1951 1961

Thorpe St, Andrew 3,231 10,788

Costessey 6,995 7,031

Cation 2,112 2,392

Itollesdon 6,359 9,744

Sprowston 5,483 9,609
 31

period* the largest population Increases la Nfk Rural

Districts were In St, Faiths and Aylshaja RD (♦ 8,374)*

Blofield and Flegg RD (♦ 2*841)* and Forehoe and Hens tend
RD (+ 2*434)* all of which adjoin If oh County Borough*
This represents Increases of 26*5 per cent* 8,9 per cent*

and 10*00 per cent respectively. The number of private

dwellings in St* Faiths and Aylsham RD rose by 49*3 per

cent* The Importance of these areas to Nch* and their

social characteristics* as opposed to those of Mch* are

shown In the occupational statistics in Table 5* This

Table 5

Percentage of male work-force

in selected categories

Employers, Junior Non-

Managers * manual* Skilled
Professional Manual

England a Wales 14,1 5^.8

Norwich 10,7 62,6
Blofield a Flegg rd 17*0 45,9
Forehoe A Henstead rd 18,5 46,3

St, Faiths a Aylshant rd 15.7 52*3

demonstrates that a large number of the Nch professional

and raiddle-elasses in fact live outside the city boundary*

and that the social composition of the suburbe is biased

towards the * top* end of the occupational scale.

 32

Tho population of Heh la relatively immobile

compared to that of the rest of the country. la 1961,

75»^ per cent of the population of England and Males had

lived at their present residence less than fifteen years.

The comparable figure for Ncli was 71*3 por oent (Nfk
76.3 per oent).

Social and Economic Character

Hoser & Scott, in their study of British towns,*'0

have placed Keh, together with, for example, Bristol,

Reading, York, Great Yarmouth and Ipswich, in a group of

sixteen towns which are characterised as "mainly cots*

mercial centres with some industry". This group is added

to those towns which are "mainly seaside resorts" and

those which are "mainly spas, professional and admini¬

strative centres", and together these form Group A

"mainly resorts, administrative and commercial towns".

This group of thirty-six towns differs in its social and

population eharaeteristics from Group B, sixty-five

"mainly industrial towns" and Group C, fifty-four "sub¬

urban and suburban type towns"• This classification is

made on the basis of four main components, which are

associated mainly with, respectively! eocial classt popul¬

ation growth 1931 - 1951I population change after 1951, and

new housing rate) and housing conditions and overcrowding.

 33

for a total of 157 English and Welsh townsv Moser A

Scott have made a comparison based on many different vari¬

ables, mainly taken from 1951 Census data* As veil as

being allotted a percentage or Index score, each town Is

arranged In rank order, from 1 to 157, for each variable.

That Is to say that a low number (high position) Indicates

a relatively high number or amount of each variable con¬

cerned, Variables for which Itch was significantly above

or below average for the country are listed In Table 6,

together with a rank position out of 157* Compared to

Table 631

Variable Position

Population 43
Per cent population aged 65 or over 34
Illegitimate births 1930-52 26
Illegitimate births 1955-57 31
Per cent overcrowded households 123
Per cent one-person households 43
Per cent households in shared dwellings 192
New housing rate 1945-53 27
New Local Authority housing rate 1945-53 14
LA per oent of total houses built 1945-53 47
Per cent employed in finance 3
Job ratio 14
Per capita retail sales 1950 20

Per cent 1955 voting Labour or other left-wing 42

Per cent voting in contested local electlone 24

Infant mortality rate 1953-57 150
w iVw< e s s al i *5j£< M4 s t J> © o ©i

5 i

I
J

* I 8 cf «
f ch
f

'« *

i
.fi 81
3£9
w *rt£

1 ! iM N
BJ h• e 94 9 h •*

©£

»4 8
I

!
2*

SS I
O

tf ©
si g *

94 k

94 £S »4 © r
! o £ i £ ©S

at

© 94 £ 9 *J
s

n
S« S* • i a ** « 8 E *4 1 »4 i a 1 © »4 4i g © «% H >rH

I0% © u 0 «•* 4© 8 gi. w• BS g 3 g 94 s

K© © | >•«"< l« S © H 4% • i a1 »4 t £ n»— © a i|M tf

,4*

© H< © « 4» .H 1* O 81 I «O © •© 4 KJ g © !9 © §1 o *tf c k8 •

!
ii © g 94 k 9
.

s I »4 ® k • *«

i* - | w sa OOO itf H

! tf # • 9 tm »k I « 3< a1 3< G H »4 i J* ©© i © $

sail
b4 H S© %"f O9S 1 © © 3 »4 i >4 4i« o£ © II - 4lif 94 i« S

I* 94 s I 94 tf e s© )ef © Gj o 94 £ *4 94 8k o a

ak o a 1H »4 ti ©a &C 4>• £ 94 s* 8 w © £o a o t* «4 X
 35

stores have a very wide delivery area* and most deliver to

the whole of Iffk and parts of neighbouring counties*) tteh

also has a low number of overercwded households and shared

dwellings, and a low Infant mortality rate* The Job Ratio

is defined as the population working in the town per

hundred of the resident occupied population! the greater

the net outflow of workers the smaller the ratio, and vies

versa* If eh* s high position indicates a relatively large

daily Influx ef workers and a relatively small outflow*

Industry

In 1961, only one textile firm employing 275 workers

remained in Neh as a reminder of its former dominance in

3U
the oloth industry* Clothing, however* remains the

largest single employment category* but 90 per cent of

the workers in this field are employed in the shoe industry.

Approximately 9*000 people are engaged in this trade, and

Kelt is the fourth largest footwear manufacturing centre

in the oountry* The construction Industry onploys about

7*00O workers, while 5*700 are engaged In the pro motion

ef food and drink, including chocolate, soft drinks and

Oolman*s mustard. There is also a large brewing industry,

The largest single employer in Neh is the engineering firm

of Laurence, Scott and Electromotors Ltd., with about

3,000 employees, and engineering and allied trades account

 36

for 8|500 of the labour fore© altogether* There Is also

a considerable print Art, j industry in N'ch.

Ponsln,-

Generally speaking, as v® have seen above, housin:;

conditions in Nch are f;ood. In 1366, 53»dQO people wore

livinij in council bouses, 32,00O in owncr-occupiout houses,

and 23,000 in privately rente*! housln > The proL>ortton

of council houses is lur?;®, around kO per cent, whereas

about 30 per cent are privately rented, find over 25 per

<5
cent owner-occupied.

Internal r Ifforcntlation

The only figures available lor a discussion of the

internal social and econocsio characteristics of the city

of Nch are tho I3CI Census figures concerning electoral

w rrfst This is tiouevhat utifortunate, sinee ward boundaries

»ro not necessarily econa .ically or socially sl:mifleant,

and wards are usually larger in size fian £.o0t',rns; ileal

neighbourhoods. However, an analysis of these l'i.rures can

help to convoy so: 1© picture of the chaxvjln.: social and

1 ;eographical structure of fcho city, and of the character—

*36
istios of dilforent areas.

U® have already sten that the population of Nch County

 37

Borough declined in the period 1931 - 1961. This change

did not affect all areas of the city equally, as the map

of population change shows (Map 3). The largest drops,

of over 20 per cent, were in Mancroft (- 36.6 per cent)

and Wsstwiek (•» 26*1 per cent) wards* Both those wards

are in the centre of the city* Hancroft contains the city

centres the main shopping areas; the main business sections;

the Caotle and Cathedral and other tourist attractions!

and the *bus station* During this ten-year period, areas

of lSMbh-century working-class terraced housing of inferior

quality were cleared in the King Street and her Street

areas in the south of the ward, Westwick is a stilted area

which* in terms of neighbourhoods, includes two distinct

areas associated respectively with the two subsidiary

shopping Centres around Magdalen Street, and around St.

Benedicts Street and Dereham Pond, It contains decayed

inner residential areas of 19th-century terraced housing,

touch of it being cleared! warehouses} an Industrial area

around the City hallway Station (now closed) including

boot-and-ahoe factoriesj and other signs of deeay-^ such

as small workshops, builders *-merchants * yards, and cycle

repairers* There are also many small comer shops* There

are In addition some areas of modern and pre-war Council

flats, and the area as a whole is in a state of rapid

change and renovation* Both Mancroft and w'estwlok have

a high proportion of Old Age Pensioners, 2k par cent and

26 per oent respectively (see Map k).

 38

A group of wards with o small or drop in population

(between 10 par cent and 20 per cent) is found mainly in the

north of the city and adjoining the two central wards* They
38 38
are Holloadon (• 18.1 per cent), Thorpe (• 16,3 per

cent), St, Stephen (- 13,4 por sent), -household (-14,7 per

4g
oont), Coslany (— 12,0 per cent) and Cotton (— 10,3 per

sent), Hollosdon ami Catton are two of the four outer wards

which do not adjoin either Mancroft or Westwiek, Hellesdoi^

eonsiste almost entirely of an area of pre-war Council

housing, with soma recent additions (einco ip6l and therefore

not Included in these figures) of Council tower flats and

maisonettes, which is known locally as the Drayton Road

Estate and "Hlle Cross*. Ofercrowdira<; in ly6l was high Cor

the city, although the proportion of persons living at more

than one and a half persons per room is given as still only

around six per cent• Sat&n comprises porta of the areas

identified locally as New Catton, Mile Cross, and the

S prows ten Read area, and is a mixed area consisting at pre¬

war Council houses an flats, post—war Council houses, and

both pro- and post—war semi-detached middle-class houses with

sissuable gardens. The other four wards consist pre;lominan t ly

of the pro—First World sfnr working-class terraced housing

that lias in a more or less continuous bolt around the oity

oontre. They also have, however, sou© more expansive older

houses. Many Of these are detached houseo in largo grounds

on the main roads leading out of the city, such na Newmarket

Road in St, Stephen ward, Thorpe has a number of middle—

class housing arena, particularly in the dietriots off

Thorpe T-toad, ami Household contains part of the anolnvo of

 s

a © © C\ I
9 1 < © 9

ft ss ♦

■0
♦*
"Vj

£
»

4 ©© ? r ? 9
t Cv
*
©>
1 1

«►
!
i
w
© © p

<♦

El !?
M 4 ?
1 © S m © I 9 3 P §
«

VR *
•I

•o

'U

a
w
# 5
© © 1
?
©

e*
©

«+
SP
9 4 « a

«*•
2

e*
5?
© §

e» «a
;r
© 3

•S

Hs © © 9 9 e ® 9 ©

t¥ e* »5
»*

«4

E &
© 4 $ « a a 9 S ©
0 a 3 ©

»*• «»■ «(• ct-

vt
53* i*
«*
|M

H
© © i © a

S?
I*

I 8IE
9 I

H M H
t #
9 © H- © a P O 8
s. © n » Wj © P a a © 0
Sr

<*
(♦»<* '/mf
ct 5r
<$■ Sr

»*
*<
 40

ajround Nelson Street off the Dereham Road, which la a con¬

tinuation of the inner residential working-claaa holt of

terraced houses, Rowthorpe contains the fringe of the Earl-

ham Council estate in its western half together with a

little semi-detached middle-class housing, and that part

of the inner belt of terraces that lies between the borehaa

and Earlham Roads in Its eastern half, Kelson is bounded

by Unthank Road, Earlhaa Road and Christehurch Road, and

contains streets of fairly expensive large pre-First World

War two- and three-storey detached and terraced houses| some

working-class terraces $ some pre-war semi-detached houses|

and some post-war middle-class and upper middle-class housing.

Only four out of the sixteen wards in the city have

had an increase in population, They are Eaton (+66,4 per

cent), Town Close {*13,2 per cent) and Lakenham {*9,6 per

cent), which lie tofios^r in the south and south-west of

the city, and Crome (4101,2 per cent). Crone contains

the large new Heartsease Council housing estate, together

with a new development area of owner-occupied bungalows

at its eastern end, and in the west some streets of

terraced housing which form a continuation of the Thorpe

Hamlet urea, Town Close contains the post-war Tuoftswood

Council housing estate on its outer edge, and the drove

Road area of mainly large semi-detached middle—class

houses nearer the city centre, hnkonhan has the new

Lakenliara Council housing estate at its outer edge, the

 kl

pr»i»>r OoonelJL estate in the middle, and part of the older

holt of terraced houses near the city centre, off City Road

and Hall Road, it also has some large detached houses in

bi^ grounds along the main roads, such as bracondale, and

both pre— and post-war Council flats, Eaton is composed of

several fairly distinct areas* the Raton Rise estate of

owner-occupied post-war detached housings the old. Raton

Tillage area, with its associated area of recent middle—

olass serai-detaahed housest the Unthank Road - Newmarket

Road - Ohristchurch Road area of older large middle-class and

upper middle-class housing, with some pre-Flrst world War

terraces and newer semi-detached houses$ the South Park

Avenue and North Park Avenue post-war Council estates| and

the Colaan Road area of mainly pre-war Council houses with

some smaller owner—occupied pre-war semi-detached houses,

Neh thus falls into three main regions of population

change (see Map 3)t

(a) a northern, central and eastern region where

emigration has been at a high level|

(b) a western region, where emigration has been

low |

and (e) a southern and south-western region which has

had modernte-to-larg* increases in population.

Only Crorao ward seems to fall outside this pattern, appear¬

ing as ft does to drive a wedge of high immigration areas

 42

into the centre of the city. Thie is, in fact, & e&se of

ward boundaries disguising what hae actually oceurred.

Only the Hearteoase Council touiing estate, which did not

exist in 1951* has had a significant increase in population*

This estate occupies the rectangular area which Jute out

beyond tho rest of the city in the eoet, Xt seems certain

that the older area of Croms (left blank en the nap. Hap

3) has the some kind of emigration characteristics as the

two words on either side of it, Household and Thorpe,

(This is also likely to be true, of course, of the inner

areas of Town Close and Lokenhom), There is, in any case,

a definite drift in population from the north, east and
ho
west of the city to the south.

This trend is also partly illustrated by the figures

concerning the proportion of Old Ago Pensioners in each

ward (see Map 4), which show, as it were, that older

people ore being left behind in the centre and north of

the oity* These figures are also helpful in demonstrating

the character and ago of each ward. The central areas of

the city, particularly those with a predominance of pre-

first World War terraced housing, have the highest pre*

portion of Old Age Pensioners, the outer areas a lower

proportion, with the south-western areas of Eaton and

Town Close having the lowest proportion of all (except for

the two predominantly post-war Council estate wards,

E&rlhani and drome).

ftPouVnoM C.HfcN6r6 , — U». PedCS fTlA&g Oo> h&i 9&\%i^€?s .

%£C^NTfAftg VK«TH 4° Pe&OSvN-fA£rg ©MN^fc-

OVLOPiGh,

fkf££hf^S L*\jifib M Ki£$ ftiM SsCWU. ^NfcgX - (fc&UftfcliGH * v

l*S OcAlPWCH)
 h'3

Further characterisation of eaoh area can he provided

by fiprcs concerning housing and household facilities.

These ward* (see hap 5) with a low percentage of house*

holds possessing all four facilities (hot and cold water

tap, fixed bath and WO) are in the central areas which

also have a high proportion of pensioners* The south

and south-west arc well—provided with facilities,

A similar pattern occurs in the figures for owner-

occupied houses (see Map 6)t Eaten and Kelson, in the

i

south—vest, have the highest proportion of houses owned

by their occupiers, generally an index of high social

status, This typo of pattern also emerges for those

areas of the city which have a below-average index of

overcrowding (defined as the percentage of persons living

at more titan one and a half persons to a roots)* Those

wards (see Map 7) forts a broad band across the city* but

again Eaton, Town Close, St* Stephen and Nelson are in¬

cluded in this favoured group.

Further evidence can be obtained by an analysis of

the social class composition of the different wards.

Maps 8 and 9 give details of the socio-economic class***

42
ami social elaes composition of eaeh ward. They show

that the highest ranked four wards in each ease are

Eaton, Kelson, St* Stephen and Town Close, A more de¬

tailed analysis is that given by Maps 10 to 16, showing

Socio- Economic "Sko#* 'fgfUfrttft&S &P fkfrflBSfcifeNftl. Weg&g^

/l&fNfA&f OP &4ftfcVl£R* *Nfc %&cg*>t*&6' CP owa N6**-w*Nvifc«,

if&eCWf&i&f lMSK.JU.ffc Wt$&£g$. fecd»Hfcft$ ONi^^&Mefc

 V*

the wards with the highest proportions of, respectively)

professional workers{ employers and managers) junior non-

manual workersi skilled workers) semi-skilled workers)

unskilled workers) and unemployed. There Is a olear

progression, as we move down the social scale, from the

south to the north and from the west to the east of the

city. Particularly striking arc the first three maps

(Maps 10,11, 12), which show Baton, Kelson and Town Close

leading in their proportions of non-manual workers,

411 thsss figures combine to present a picture of

the increasing economic and social dominance of the

southern and south-western part of the city in general,

and Baton ward in particular. It is perhaps significant

that this area lies on and around the approach roads from

London, the All and the AlkO.^ Thio bias to the south

and aouth-weot is to a certain extent counteracted by

the more prosperous suburbs outside the city boundary to

the north and east, such as Thorpe St, Andrew and Catten.

It is worth noting, however, that the most statuaful of

all the city suburbs, Grlngleford, lies just over the

city boundary immediately adjacent to Eaton.

From this city a sample of the population was taken

for the purposes of the linguistic survey.

 *5

Chapter Threei The Sample

Many previous linguistic and dialectoleglcal surveys

have been based on work carried out vith informants who

were chosen either beoause they were elderly natives of

the area under survey and therefore likely to be •pure*

dialect speakers** or simply beoause they were easily

available• Other surveys* such as that carried out by

2
Kurath & HcDavld in the U,S,A., have been of a more

sophisticated type* but have relied basically on the same

relatively haphazard methods* There are obvious and

serious dangers in this typo of approach* particularly

when one is dealing with a large urban population,rather

than with a small rural community* Typically* the urban

population is heterogeneous* and both socially and geo¬

graphically mobile| sociological factors are more important

from the point of view of linguistic differentiation than

geographical factorsj and the social structure is of a

complexity that makes close individual knowledge of the

area impossible* and person-to-person contact as a means

of selecting informants useless*

It is* of course* neither practicable nor desirable

to Interview the entire population of a city* That is to

say that a survey of this kind must necessarily be ia-

conpleto* Obviously* a survey can only be complete if the

population to be covered is small, as in a restricted rural

 46

area* or if the resources to be employed are very large*

(Evea if it were possible to interview every single

individual In a survey, this would serve no useful

purpose, since sample survey methods make the effort

unnecessary# and the vast amount of detail obtained

could well lead to confusion and inaccuracies.)

Chi the other hand, esse does want to ensure that the

language one is describing is truly the language of the

city, rather than that of a few hand-picked informants*

(It is, of course, legitimate to rely on a few hand-

picked informants if the object is, for example, to

record older dialect forms before they become extinct*

It is doubtful, however, if even this can be totally

justified with any large, heterogeneous population*)

Informants selected solely because they are available and

willing to be interviewed are simply a part of the popul¬

ation of the city, not a representative sample * and no

valid statements concerning the language of the city as

a whole can bo based on evidence obtained from informants

selected in this wayi "It is entirely wrong to make an

arbitrary selection of cases, to rely on volunteers or

people who happen to be at hand, and then to claim that

they are a proper sample of son© particular population"•

There is only one way to ensure that the results

obtained in an inconplete survey of this kind can

 k'J

legitimately be said to apply to th© population aa a

wholet the section of the population which is to be

studied Bust be selected by "accepted statistical methods"•^

The informants* that is, must constitute a genuine re¬

presentative sample of the city*© population* This

particular work is based on a series of interviews carried

out with a sample of the population of the city of Neh,

which was obtained by methods which are statistically

acceptable* The sample is large enough and sufficiently

scientifically devised to permit reasonably confident

assertions to be made concerning the population as a wholo*

and small enough to persit accurate and intensive study

over a limited period of time*

The Benefits of Sampling

Social scientists* geographers, botanists and ottors

have developed a set of very sophisticated sampling teoh<*

nlqucs in the last thirty or forty years* Linguists*

however* have begun to realise only comparatively recently

that this set of techniques Is a potentially useful tool

in their own work* Hany examples can be cited of linguistic

work which could have benefited from this type of approach*

*4
In her study of Cockney phonology, for instance* Sivertsenr

mad© no attempt to obtain a representative sample of th©

population of London or even of hethnal »roen. Instead,

 ha

she eimply worked with people who to her seemed suitable*

and who were willing to help her. In faet, her four sain

informants were all women ogoi over sixty. She did* it is

true* investigate the speech of other age-groups* men and

women* but not in any systematic way. She does concede*

of course* that she is using the term "Cockney" in a

restricted sense* hut* in view of the fact that she

seleoted, by purely subjective criteria* four Informants

out of a potential eight million or so* this appears to bo

an understatement.^ Xt eou be argued* of course* that hsr

work is a valid and scholarly study of the speech of a

small group of mainly elderly* working-class people* from

a email area of Fastern London* as indeed it is* hut X

hope to show (see Chapter fine) that this has important

consequences for linguistic theory, in that the type of

phonology employed by Sivertson in her work is as unreal¬

istic as the type of "sample* she was working with* and

only appears to be valid because she was studying an

artificially homogeneous speech community*

Similarly* in his work on the speech of San Francisco *

D© Gmms? worked with a larger group of informants seleoted

in the same kind of veyt and he himself points out some of

H
the disadvantages of using this kind of group* A mere

recent study is that of Vlereek^ on the urban speech of

Gateshead. Viorock chose* by a means he does not specify*

twelve Informants* all men, of idiom ten were a.-o«i over

seventy, out of a total population of 113,000. Vi©rock*s

pur!MMi« was not* of coiurue, to obtain a representative

sample of the population of the town, and by this means

to produce a genuine description of the dialect of

Oateshead as it is spoken today. His concern was rather

to seleet speakers of what he considered to be "pure*

Oateahea t dialect, and to write a description of their

speech* He rveognisea that there is in fact no such

thing as "pure Gateshead dialect* nowi *&i© alls arose*

stadtdialokte . • • 1st auch dor *biale!ct* von Oateahead

10
koine in sieh gooChiosson® kinhoit mehr", but the

implication is that there woe formerly such a thing as

"pure Gateshead dialect", ami that Viereek is trying to

get as close to this as he can. 1 hope to show that he

has set himself an impossible task, and that such a thing

as a "Phonomatlsche Analyse" of the dialect of Gateshead,

or indeed of any other speech counsunity, is an unreal¬

istic and impossible concept* Many other oxanplos could

11
b© given of this typo of work*

In the last decade* however, some work has been

carried out, particularly in the U#3.A., with large,

usually urban populations* using statistical sampling

techniques* One of the first studies of this type was

J «>
the work of Labov "* in How fork City, in which he was

fortunate enough to be able to us© a sample corn* true ted

very scientifically for uso in a previous sociological

survey. Another t'-rgr—ocnl© work la that which has been

 50

11
carried out in Detroit, Hero a considerable amount of

time and effort had to be expended on obtaining a sample.

The only feasible mean® of access to potential informants

was through the schools* which meant that the sample,

although otherwise very scientifically designed, hod a

heavy bias towards adults who had cliildren of school ago,

A similar work is that of Levine & Crockett*** also in

the U.S.A.

There has as yet boon very little work of this

15
nature in Groat Britain. however, the difficulties

encountered by the research workers in Detroit in con¬

structing their sample are totally avoidable in this

country. This is because wo always havo available a

relatively up-to-date erjr.plin/; frame, for all parts of

the country, in tho shape of tho heelstor of Electors,

This lists every member of the population who Is eligible

to votei all people aged twenty-one or over who are not

aliens (or otherwise excluded). There is no comparable

list in the U,o.A,

The ease with which a sample can bo drawn from a

large urban population by using tho rtogister of v.lectors

16
seems to havo boen overlooked by Iioucfc in on© of the few

linguistic studies based on a statistically selected

sample so far carried out in Britain, In his small

methodological study of the speech of tho city of Leods,

 \SI H

X$ 1 31 ii 8( »«* 5 *K

1I ♦* 2 m *4 85 4>< h 3^ 2 55< t

11 O 4* »»■ »* 5 »4 O ? t 4* eI 1e ©i

HV 8© .> « ©H I S* i »4 s ex 'C *h © 1
'

8© 5 »5
< SI 0 Wtfi £■ © 3 © i©
-:s ©SMZ 9 *0 8

Iv SJ © | §© % Ie » ©0 !

ip

©
I

! •

« 4•
S£ 0J

| I 85 € © "0 *4 t© * et2i
S rf

i8
©a
&
111 -8
C
'
s©

»4 © 4* K!
H 0 t »4 »

:a © t ©

8 3« I © 4^ ©& I 8s

88 e »%" © 3 © !« *4 s c 1 4« 1> 8» 94 « « o• f* I I I w 0 *S X $i

i4*
I !
.»■© |
* *1 " 0© S 1 © 4% 94 £ © & 3 © P< *t © a

! 2 5 I2 §• S n 1 I 4« »4 1 8 © »4 © e s>
-'

1 s io © *< i ©© © ©u 0• £ 4"« *
 52

auna^-random sample method, which is described below, is

often employed where the population in question la a

large one, end when the sample la to be drawn from some

kind of pre-arrange*! list, rather than toy using socio form

of lottery method, Strictly spooking, this method is not

equivalent to simple random sampling, since the number of

possible samples is much smaller, but it is "generally

Justified by the argument that the list" (in this ease the

Register of Fleetore) "can to© regarded as arranged more or

less at random, or that the feature by which it is arranged"

(streets and house numbers) "is not related'to the subject

of the etiTToy",^ It has one advantage for a linguistic

survey, apart from the obvious advantages of speed and

facility, In that there is a better chance of obtaining

an overall geographical coverage of a particular area,

and thus being able to investigate ;possible geogra-tfiie

variation.

Th® method is to divide the number of the total

population of the area in question toy the number of in¬

formants desired, in order to obtain the asbaling fraction,

A number smaller than the sampling fraction la then

randomly selected, an) the person with that number on the

list becomes the firot member of the sample group. Then

the remaining member© of the sample group are obtained by

adding tho sampling fraction number to the number of the

first somber selected, and than to that of the second

 53

member, and uo on, so that the selection of tho first

member automatically determines the solvation of all tho

others, (All northers of the population, in other words,

uo not have an oo.nnl chance of selection,) Thus, if tho

total population is 90, and the desired nunher of In¬

formants 9» tho sampling fraction will ho 10, If tho

first member randomly selected la number k, then the

remainder of tho sample will consist' of numbers 1^, 2k,

3'#, kk, 5k, 6k, 7k, and 8k, giving a total of 9 in all,

Tho sample for tho h'eh survey was not drawn from the

Register of electors for tho city as a whole. It was

decided inst© .d to sample four of the city*s electoral

wards only. This procedar • has tho ap >roval oi exports

in tho field of survey methodst "Let U3 suppose that a

sample of individuals is to >e selected in a town \;it?\ a

population of 250,000, dost probably, one would decide

to concentrate tho intervi. wo in a few area- , so the

18
first step may be to nio1 son© of tho wards In t.io town

The advantages of this kind of approach In a survey of this

kind are that it opens up the possibility of investigating

googrophical variation within tho city, and that it makes

contacting informants and conducting Interviews a much less

laborious and time-consunin,; business, as the geographical

area to be covered is much s.>allor, Tho four '..arris were

not seloctn at random, but were chosen so that they bad,

between thou, social .and economic civaract oris tics that

wore, on avurn ;e» tho oa « ?vi those oi tht c ity a whole.

They wore* moreover, chosen so no to represent different

types of area from the point of view of social, geograph-

leal and housing characteristics (see Table ?)« The four

wards weroi Eaton (D) | Lakenlian (JU)j Hollosdon (M) \ and

Weatwick (d) (sec Chapter Two), The sampling: fraction

was adjusted so as to give an equal number of informants

20
from each area. In addition, a sample was drawn from

one of the suburban puriahos outai !o the city bosandary,

Thorpe St# Andrew (see Fig# 1)« This was because, as we

have already seen (Chapter Two), the suburban areas fora

an important part of tho social structure o# the urban

area of Nch#

Each person listed in the Register of i lectors also

ha® a number, which ia a useful aid in the selection of

the sample# In each of tho five areas, an initial sample

of twenty—five persons was randomly selected from the

Register# A number smaller than the adjusted sampling

21
fraction was taken from a table of random numbers# The

person whoso number this vas became tho first member of

the eanplo in each area, and his selection automatically

determined that of tho remaining twenty-four members# Tho

Initial selection thus comprised 125 names and ruldr©93ca,

from five different areas of the urban region# It was

decided that tho maximum number of infon urmts who could

be interviewed in the time available) was fifty, ten from

each ward# Introductory letters were therefore sent out

 55

Table 719

/i rcononio and Social Cliaractori.atica

of Hen o.'VJ fmiy seloctod "nn»3«

% Workers
Social socio- Per A/to —

Class i con- dent Index " on- Foremen Semi-

Zndex omio Pens- Annual and ami tin-
Index toners Skilled skilled

C. 71#50 62,00 14,00 57 50.00 33.00 17.00

L, 47.50 50.70 17.00 5-2 26,00 43,00 30.00
II, 40.00 43.00 10, 00 50 l4,oo 53.00 32.00
¥. 53,00 43,00 26,00 37 13.00 53.00 32.00

fseon h% 25 50,10 13.75 43 26,25 46.00 27.75

Hoh.50,00 50.00 17.10 50 27,40 45.70 26.50

£ Per Cent Workers

Profess- employ- Junior rorenen Service Un-

ional era end Hon- and an? secii- skilled
Uana^ora rinual skilled skilled

E. 11,00 15.00 24,00 33.00 10.00 7.00

L. 1.00 5.00 20,00 45.00 18,00 12.00

n. 0.00 4.00 10,00 53.00 12,00 20,00

¥. H • 00 4.00 10.00 53.00 18,00 14,00

pean 3.25 7.00 16.00 46,oo 14.50 14.25

HCll 2,00 7.oO 17.00 45.70 14,00 12.50

to ten people randomly selected from each croup of twenty-

fiv©» the purpose of the letter was to the

secure f'/ood—
 §* s I© s£ I ®4 2 «« £ € © S1 H« © tif

©£ i a »* ! *« © : i 34 »4 © : t£ § 4*< £ I

»4 © ! © tif i2©© © »v 4 ® sIS s I *4 e

S * ! © U0 © § tif 4*> o »•* e 3 »4 I £ & V ot ir © S< »4 *4 ®

§

S g- '* I 4*■ 4£ eif © 4• £ © »4 » ? © 1 C- »4 © DJ © 0 > © o9

f ©s o »4
 37

to help* and showed considerable interest at the prospect

of talking about "old Nfk words" ami other allied topics*

Stone* however* wanted to be assured about the complete

anonymity of the study) an i the tape-recordor had often

to be Justified as a time-saving device*

The sueeeas—rate in securing interviews was not

particularly high* for several reasons. Xt was decided*

in the first plaee * not to interview anybody who had

moved to Neh from outside EA in the last ten years*

Strictly speaking* to obtain a realistic picture of the

speech of the city* and particularly of attitudes to Neh

opesch* it would have boon neeessary to interview every¬

body* irrespective of their origin* Xt was* however* feit

that* in view ef the short time available and the small

sine of the sample* time could not be spent on informants

whose linguistic behaviour wan radically different from

other informants• Some of the initial selection* moreover*

had moved away from fteh since the compilation of the

Hogieter, and others were blind or too senile or infirm

to be interviewed* Difficulties of this kind cannot be

OVeroome in a linguistic survey as readily as they can in

an ordinary social survey* people who have left Net* are

no longer part of the Nch speech community) blind people

cannot talc© part in the reading tests) and so on. In

addition* ae with all social surveys, some people had

died) and others refused to participate, or could not bo

contacted.
 53

In all ouch easow of failure, names of replacement©

wore a elected from araoiv; the remainder of the tw<mt]r*flTe

in each ward, In allv to obtain the fifty inforraants , it

was necessary to tend letters to ninety-five of the orig¬

inal 125 people selected* Of these ninety-five, twelve

were not from *ich, five had moved, three had died, and

three were blind or infirm* (Those who had left their

original add res©, but were still in H.©h» were located and

interviewed if possible.) Of the remaining seventy-two,

fifteen refused to bo Interviewed, and seven were not

contacted. This refusal and no-oeatoet rate is high for

a social survey. The standard approach in a social sur¬

vey is to follow up a refusal with second, third and

further visits, in order to secure an interview, because

it is important to minimise the number of refusal©, in

order that the sample should remain truly representative*

This Is particularly so in ordinary sooial surveys# sinoe,

as can readily be imagined, the attitudes of people who

refuse to be interviewed on, say, race relations, may

woll be significantly different from those of the rest

of the population*

This approach, however, was not adopted in this sur¬

vey, as it was thought that linguistic behaviour was

unlikely to show significant differences of this type*

2b
Labov has shown, in fact, that those Informants who

refused bin interviews or could not be contacted, and

 59

those linguistic behaviour was later studied by other

means, wore in no way different from the other inform¬

ants in their language eharaoterlstioe * It was therefore

decided, for the purposes of the ffeh survey, that follow-'

ing up refusals la the usual way would bo a time-consuming,

unrewarding and unnecessary task. A further consideration

was that, in order for an interview to be successful and

as mooti conversation as possible to be recorded, the

good-wlllof the informant was very necessary, and

repeated attempts at prosunaion were liksly to Jeopardise

this* The number of refusals was probably also fairly

high because the nature of the interview, which involved

the use of a microphone and a tape-recorder, was rather

more intimidating than the normal type of social survey

interview*2* (Reasons for refusal inoludedt inability

to afford the tinej not feeling well enough to take

part) failure to see the point of the study) unstated

but obvious fear of the whole exercise) and one gentlemen

who did not "believe In universities* • ) "Ho contact*

means that several visits were made to the address in

question, and that nobody could be found at heme* This

was often because of holidays, since the interviews were

conducted during July, The policy of not following up

refusals meant that very little time was waste, and

the fifty informants were written to, visited and inter—

viewed In a little over throe weeks,

 60

Schoolchildren

Th© Hof;ister of Electors has on© important defect as

a sampling frame* only people aged twenty-one or over ore

listed. To too of any significance# however# a linguistic

survey of this nature needs to study the linguistic char¬

acteristics of all age-groups# This defect was therefore

put right toy drawing a sample of ton schoolchildren from

two of the schools in Neh. Tho children were all aged

between ten and twenty# as there was not sufficient time

available to obtain a sample of young©** children as well#

The schools were not selected at random* thoy were in

fact two out of tu© three sfcate-mintalned grammar schools

in the city# The sample was drawn from this type of school

since they are th© only local schools which have the entire

city (but no areas outside tho city) as their catchment

area* This particular sample could for this reason act as

a Kind of control for the main sample# in tho investigation

of geographical variation within the city, and could poss¬

ibly also shod light on the speech of those areas of the

city not covered in the main sample. It sight be supposed

that the fact that tooth schools were grammar schools would

bias tho sample towards the higher end of the social scale#
26
This# however# was not the ease (so© Chapter Four)•

Characteristics of the Sample

Thus, the final sample of informants who gave inter

 2 .
/a persons U d^crcAt Nek. fits
r.^e-cj rc-ops, tw
report©! uv $W> SoMple. Census
4&i kkd. u\(itv^iusHc Survey Sample,

&>■

20-

!©•

M za-SJi

% Pers&ns in d^fer&nt cou,>pa#o«.at. groups u\ Mek<

Os reported, (A *%£> S«u*.p(K. Ou\t^, ft,a/£> U\
£& L^otki'ic Sorw^ Ssuupte.

40-

SB

2d

Ib

© UIOUWK!
-#- -*.-i
I'tv,v£ if Ill & vi

kutp^dfew iA'u p
 63

atlons. Groups I ami II comprise profeo <ion »1 workers#

employers» and tmaa^ersi Group III othar non-manual

workers I Group IV foremen# skilled workers and self-employed

persons} Group V semi-skilled and parsonal—service workers}

and Group VI unskilled workers. (The Census figures also

contain 2.1 per cent listed as Armod Forces or *occupation

inadequately described"•) The two curves showing educ¬

ational status# her© measured by the terminal education

age of those aged over fifteen# ire oxtre- sely close. *h©

lact that t'u ro Is a tine-lag of oigut year:} between the

two sets of figures could well account for tho ap >erent

slight over—rcprosontattori of t .10 middle two groups# bo-

cause of the steady increase in t «© proportion of tho

population who have remained longer at school.

.Sociological indices o. j^loyed for o?w*ract ©rising each

informant iiidividnaliy aro described in thu t ollot.ing

chapter.
 % (%«&«& u* dj^tr&A'k 'iWfcu&A.C Giucfi^oiv A^5-5rao?^
iw Mtk. &% rtPofi^k U\ lU<
C&nSos, owt\A t*.
loo- UA^vi^it S«rv^ Scuvfle,
%

<&

76

% 4&

&

So

<ZO

16

IS"fcR tk-n SToltfNH

Mafc6
'fcTKtMft.l
 6k

Chapter Fourt Social Indices

One of the main purposes of this study is to in¬

vestigate the nature and extent of the correlation be¬

tween and co-variation of linguistic and sociological

parameters in the city of Nch. This investigation* it is

hoped9 will illustrate the way in which a large* hetero¬

geneous speech community functions , and produce some

answers to theoretical problems, such as the status of the

dlasystem* and the nature of linguist!© change# It will

also provide a more realistic description of the speech of

an urban area than some of the earlier studies briefly

described in Chapter Three* in that it will be ait attempt

to relate language to its social background, and to show

how social differences are reflected in, and how they

act upon* linguistic differences#

In this chapter* we shall discuss briefly those

sociological parameters which can be subsumed under the

heading of "social stratification"* together with the

relationship between these parameters and language#

Mainly* however# we shall be discussing the ways in which

they can bo associated with measurements of an individ¬

ual' s social status* in order that he can be ranked in

the social scale and his linguistic behaviour compared

with his social position* (The measurement of those

sociological parameters associated with social context

will be discussed In Chapter Five, and the measurement

of linguistic! parameters in Chapters Six and Seven#)

 65

Social Stratification

Social stratification Is a complicated subject* and

discussions and definitions associated with this topic in

sociological literature arc both varied and complex* At

a simple level* however, social stratification can be

described as one particular form of social differentiation**

and has been defined as "any hierarchical ordering of
2
social groups or strata in a society*. In Britain* as

in other Western societies* social stratification takes

the form of the division of society into different social

classes and status groups* rather than into* say* different

castes.

There is iittls point in attempting to evaluate or

even enumerate the different approaches adopted by socio*

logists to the topic of social olass and soclo-eoonomio

class. There are very many different explanations and

definitions of these terms in sociological literature.

And* as is the case with some linguistic terms* it is

often necessary to define* in each Individual study* ex*

aotly what is meant on this occasion by "social class".

A simple and not particularly controversial definition of

social class is the following! "a major social group*

members of which are of approximately (,thej same economic

•j
position". An alternative definition, which could also

be eald to apply to the term "socio-economic class", is

"a major social group, members of which are of approximately

 66

LtheJ earn® economic position* prestige* occupational rank*

power, value orientations* and characterised by inter-

t±
action and class consciousness"• Other sociologists

introduce the concept of social evaluation* in addition

to tho concept of social differentiation* in order to

define social class*^ The type of attitude that is

adopted to tho term "social class" in this study will

emerge during the course of this chapter* hut it is in

any case widely accepted that social class stratification

is based primarily on differences in wealth and income*

and it is also true that most members of our society

have some kind of idea* intuitive or otherwise* of what

social class is.

Social classes* then* are not organised or sharply

demarcated social groups* but are rattier aggregates of

people with similar eeonomio characteristics* The

differences between classes in income and wealth are ex¬

pressed in different types of consumption* education*

manners* dress* taste, speech and so on.^ The exact

nature of the mechanism leading to this differential

expression of wealth is not clear* Obviously* it is

partly due to the fact that more expensive clothes*

consumer durables and so on are available only to those

able to afford them* But it also seems certain that it

is partly of the some nature as the mechanism associated

with the spread of dialect features across geographical

areas* so that dialect boundaries often coincide with

 61

troughs in the communications network (see Chapter Two)*

In other words, social barriers are as effective as goo*

graphical barriers in halting or slowing down the diffusion

of fashions, ideas, values and speech forms which have

originated in a particular social group, from ono section

of the community to another* lionce different groups havo

different customs, and so on* These social barriers arise,

in the first place, through differential access to socially

desirable typos of objects and activities, such as housing

and education, because of differences in wealth and income*

These two aspects of tho mechanists of differentiation are,

of course, inter*rolated, in that objects and activities

initially available only to the wealthier members of the

community come to bo thought of as socially desirable by

lower social groups* This typo of downward diffusion, how¬

ever, is by no means the only form of diffusion process*

These differences in consumption, manners, dress

and so on give rise to different status groups, which are

social groups whose members see themselves as equals,

with common understandings, attitudes and behaviour, and

who regard outsiders as social superiors or inferiors*

There is thus a hierarchy of status groups, which is

related to, but not identical with, the class hierarchy*'

This moans that in a class society the stratifieationaX

structure is very complex, since objective factors, such

as income, and subjective factors, such as status evaluation

associated with, for example, consumption patterns, are

 68

lnter«»related and combined In a very complicated way.

Moreover* In a eoclety of this type* close boundaries and

barriers are relatively fluid and flexible* and a relatively

large amount of social mobility is possible* both up and

down the social ecale. There are also a relatively large

number of anomalous individuals* and others whom it is

difficult to place with any degree of assurance in a

particular social class. There are those who* while ob¬

jectively members of a particular class, assume, or attempt

to assume* the statue characteristics of another, usually

higher, class. This anticipatory socialisation (when an

individual takes as his reference group a status group to

which he decs not belong) is particularly important from

a linguistic point of view.

In this kind of society* therefore* a study of the

correlation of linguistic and sociological phenomena is

much more complex and difficult than in* say* a caste

society. The different social groups in a caste society

are* relatively speaking, discrete units* there la little

social mobility, and it Is a relatively small distortion

of the facte to discuss each caste dialect as a separate

entity*in which contextual factors are the only relevant

sociological parameters determining linguistic variation.

This is Just not possible in a class society, although

some linguists (see Chapter Three) have attempted it.

For a linguistic study of any large community within a

class society to be in any way significant the class

 69

continuum must bo objectively measured against the Unguis t-

8
io continuum, and vioo versa*

It li, of course* on open question to what extent the

"class continuum" (and correspondingly the "linguistic

continuura") la* in fact* a continuum* Me have already

stated above that* in a class society, there is a relatively

large amount of flexibility and social mobility* However,

at least on© fairly large barrier scons to remain* Hits

is tlie gap between what are usually referred to as the

"middle classes" end the "working olaesoa"* a distinction

which is usually but not entirely related to the distinction

between those working in respectively nen-mnnual and manual

occupations, it has been shown* for example* that* in

Britain* opportunities for upward social mobility by

working class people ie relatively restricted,^ auw that

10
their access to higher education is also limited* Xt

has* moreover* been shown that even the most affluent

manual workers retain the values, ideas* behaviour patterns

and general culture of the working class, am) that there

has been little "ombourgooiseraent" of the British working

class•** This suggests that occupation is in fact the

main stratifying factor in our society* end that differences

in wealth and income, although obviously very closely re¬

lated to occupation, my net be the factors on which social

class differentiation are primarily based* litis of course

relates to the subjective experience of nont members of

our socaity, who know that to enquire about a person*a

 70

occupation is to enquire about his social position* and

that a car*factcry worker who earns £2*000 a year has

less social status titan a bank-clork who earns much less*

This split between "working class" and "middle class* is

therefore likely to be a very important factor in a

description of the linguistic characteristics of any

12
British urban area, particularly in so far as it acts
as a barrier to the sprcaeSlng of linguistic changes. If

the social barrier between the middle and the working

classes is a large or significant one, tills will be

reflected in tho extent of the diffusion of various

linguistic innovations, and in differing norms of ling*

IT
uistio behaviour generally*

Social, class and Lln/mlstlo Studies

In linguistic work it has long been realised that

language is in many ways a forri of social behaviour, and

that social factors are as important as geographical

factors in determining linguistic variation, For this

reason linguists have attempted, in some previous studies,

to classify their informants sociologically, in order to

relate their speech to their social background. This has

been done, in more or less sophisticated ways, in an

attempt to avoid ank*.ng subjective am! therefore possibly

unreliable and circular Judgements of social status* In

only a few studies, however, have linguists actually been

 71

successful in achieving any kin a oi' aon«Hajal3tic and

sociologically satisfactory measurement of their inform¬

ants* social cIass position* Am^ those studies arc the

1&
surreys which have boon carried out in ffow York City*
ift 16
Detroit« ^ and the Worth Carolina Piedmont. habov* for

instance, classified his informants by means of a scient¬

ifically calculated index based en occupation, income and

education* Shuy, wolfram <£ ftiley based their calculations

of social status on occupation, education, and the type

of neighbourhood, objeotlvely seasarod, in which the in-

*: "

ferment was resident* ftrui Leviae d Crochett measured

their informants* status using occupation and education

as parameters*

There is a large amount of sociological literature

on this topic which is both readily available and extremely

17
useful for linguists engaged in this type of worts* Xn

spite of this, however* most attempted methods of measur¬

ing social class: in linguistic studies, apart from the

three Just mentioned, have been unsatisfactory* One con

cite, for example* the work of Kurath A Meuavid in the

]Q
tJ.s.A,* which was one of the first large-scale attempts

to measure social as well as regional variation* This

to
attempt, as described by autrath, relies on a view of

social class which regards class position as being more

or less identical with level of educational attainment*

This somewhat naive <>osit.ion (together with the "sampling"

 72

methods used in the survey) has Veen strongly criticised

20
Vy Piokford for ignoring the great importance of other

stratifying factors in American society* This means that

the classification of speech forme in the survey as

"cultivated"* "middle class" and "folk" can be neither

objective nor consistent and is therefore of doubtful

value* A similar study is that of Do Camp (of* Chapter

2i
Three) whose work is in many ways baaed on that of

Kur&th A Hebavid* Here again the social classification

of informants* which Is if anything somewhat lees sophist*

looted than that of fCurath A HeDavid * is based only on

education* he Camp divides his informants into three

educational groups* They consist of people who have*

respectively* only elementary educations only secondary

education* and some college education* {Informants are

also classified according to ethnic group*) Both these

studies represent a significant advance on* for example*

the work of Orton A Dietfc*22 and Siverteen*2^ who pay very

little or no attention to sooial differentiation. They

are* however* still seriously deficient*

Two smaller studies are alee worth mentioning at

this point for their attempts at assessing and employing

social differences* Flecher* in his interesting study of

2&
social influences on the choice of a linguistic variant*

although more concerned with intra-group variation associated

with social context* personality and sex* does include

 7«

in this study, rather than some other measurement of

social class, is that it provides an objective, stand*

ardised and easily calculated indicator of social class

which is at tho same time reliable and capable of ranking

27
informants in a seals, ' The advantage of using an index

eonstrueted specifically for this particular study lies

in the fact that * there can be no such thing as a single

index of socioeconomic status for all purposes of social

2S
research in a modern, complex society*, Xt is also

the ease that different geographical areas may require

different indices.

The index developed for the purpose of this portio*

ul&r work was a multiple-item index, rather than the

simpler and more unreliable type of single-item index

29
used, for instance, by he Camp* 7 A multiple-item index,

by increasing the number of indicators of social class

involved, is a much more refined and reliable means of

measuring social classt "several indicators in combination

TO
increase the validity of an index"* It Is also possible,

with a multiple-item index, for individual indicators to

be examined separately for correlations with linguistic

behaviour, in order to gauge the relative importance of

each one for linguistic variation* Many types of multiple-

item indices for measuring social class have been devised

by sociologists, and a large amount of work has been

done on comparing these indices and assessing their

 75

31
validity, The problems of index construction, for the

purpose* of non-sociological work are for this reason

not too complex* The six indicators used in compiling this

index* moreover* have all been employed previously by

sociologists in constructing social class indices* although

perhaps not in exactly this combination. It was felt that

a six-item index would give a finer stratification of the

informants than* for Instance* Labov'a three-item index*

This was thought to be especially necessary since social

differentiation is probably much greater in New York

City than in Nch* and a more sophisticated index is

therefore essential in order to bring out the less obvious

but equally important differences which are to be found

in Ncli,

The six indicators used in constructing the index

weret occupation! income| education! housing! locality!

and father's occupation. Me have already stated that

occupation is probably the most Important stratifying

element in brltish society* This assumption has the

support of sociologists! •occupational position is the

best single indicator of social stratifleational position

in contemporary American society •,« this is probably

32
also true in any industrial society* « It therefore

follows that *in contemporary Industrial society* the

single item most commonly used for social class indices

33
is occupational position"* There is a considerable body
 76

of work on the social. grading of occupations, which

simplifies the task of devising this kind of index,^ In

many multiple-item Indices the occupation indicator has

been weighted relative to the other indicators in the

calculation of the final index score, since it has been

35
thought to be of prime importance• This was not done,

however, in this particular work, since it was considered

that weighting was implicitly present in the choice of

the next two Indicators, income and education* These

two indicators are clearly closely correlated with

occupation. They are also fairly obviously Important

stratifying factors in our society, in their own right,

and have figured as such in many sociological indices.

Housing and locality aro less frequently used as in¬

dicators of social class, but are obviously of some

significance, and by no means rare in indices devised

by sociologists, ~*** They are perhaps not so useful in

wider studies, since they tend to be products of differ¬

ences in the local community, and are therefore likely

to vary in the nature and degree of their significance

from place to place. They are, however, entirely suitable

in a one-area study of this kind. Father's occupation

is also clearly of some importance in assessing social

class, since one is initially born into the particular

social class group which is that of one's parents. This

is particularly important from a linguistic point of

view, since social mobility is known to have an effect

 77

37
on linguistic behaviour.'" It is also known that "father's

occupation" and class of origin arc important factors

3d
from a purely sociological point of view. By including

these last three Indicators, housing, locality, and

father's occupation, wo hope to obtain a more precise

subdivision of informants than would otherwise have been

possible. We can now proceed to a description of exactly

how the index is calculated.

We have already stated that the social class index

is composed of six separate indicators. For each of

theso indicators a six-point scale was developed, and a

score ranging from O to 3 allotted to each informant.

Possible social class index scores therefore ranged from

0 to 30.

Occupation

The six-point scale developed for the calculation

of scores for occupation and father's occupation is shown

in Table 1. The groups used in the scale are those of the

Registrar General's Classification of Occupations, 1966,^

The arrangement of the groups is based on, but not

identical with, the arrangement used in certain Of the

itQ
tables In the 1966 Sample Census. The composition Of

the groups is as follows* Groups 3 and 4 consist of

 78

professional workers| Groups 1, 2 and 13 of employers

and managers) Groups 3 and 6 Of other non-manual workers t

Groups 8, 9, 12 and 1% of foremen, skilled manual workers,

and own account workers! Groups 7» 10 and 15 of personal

service, semi-skilled and agricultural workers) and

Group 11 of unskilled workers* More detailed explanation

of exactIf what these terms imply can he found in the

Classification.

Married women and widows were rated on their husbands*

occupation, and unmarried women on their fathers *• This

was done because "with the still limited employment

opportunities for women - that is, especially in pro¬

fessional and administrative roles - occupation is not a

satisfactory index of social status for women in our

4i
society"* It also seems clear that the social class

position of the bread-winner determines the social class

position of the family. However, in those cases where

working women had occupational status higher tlum that

of their husband or father, the informant * a own occupation

was used, since it seemed probable that her higher status

would in many ways be recognised as such by society as a

whole, and that her linguistic behaviour would according¬

ly differ from that of her husband or father* School¬

children, like unmarried women, were ranked according

to their father's occupation, both for occupation and for

father*a occupation. Father*s occupation was scored

 79

according to the father*s present occupation* or* if lie

kz
vm no longer working* hie last occupation*

I2222Z2S

The scale for ranking informants according to thoir

income is shown in Table 2* Women and schoolchildren

were ranked as for occupation* Information concerning

income was obtained by presenting the informant with a

card showing the scale illustrated in Table 2, and asking

him to state which of the groups A to 0 he came into*

This was done in order to overcome the reluctance which

moat people feel about discussing their financial circum¬

stances, and in fact no informant refused to give this

information. There was, on the other hand, no means of

knowing how true the information given was, In two cases,

married women did not know how much their husbands earned•

In these cases, and in the cases of the schoolchildren,

who could not bo expected to know the details of their

father's income, a reasonable assessment was made based

on knowledge of the husband or father's occupation,

I ducation

The education-index scale is shown in Table 3* The

 80

educational level was in all eases that of the informant»

Schoolchildren who had not yet taken any external exam¬

inations were Riven a score of 3, since, as grammar

school pupils, they could reasonably be expected to take

either 0-1eve1 or C. S. E. examinations in the future.

The social status of different looalities within a

41
city can to some extent be measured objectively. In

this work, however, no attempt was made to achieve this

kind of measurement* Instead, the different areas invest¬

igated in the survey were ranked subjectively - a much

quicker and simpler process* This ranking was based on

knowledge, acquired during many years* residence in the

City, of the status significance of different neighbourhoods.

It is likely that this ranking would, to a large extent,

44
bo agreed upon by most inhabitants of the elty. The

justification for the inclusion of this indicator in the

index io that the city neighbourhood in which a person

lives is an important component of his social status,

and that neighbourhood can sometimes be the only differ¬

entiating feature between two people of otherwise equal

status characteristies• The importance of locality is

underlined by the well-known fact that a house in a

Statusful neighbourhood can command a higher price than

 61

an equally good house elsewhere* Details of the ranking

of localities are shown in Table 4*

Bousing

The housing scale was based on three different

parameters» house ownership! age of house| and house type

Xt was consid©red, first* that* other things being equal*

an owner—oecupied house is associated with higher social

status than a rented house* and that a house rented from

a Local Authority carries less status than one rented

privately* Secondly* it was thought that* other things

again being equal* the newer the house, the higher the

status* The age of each house was measured as *pre~19l4*

"pre-1939* or *post-war*» since the two major wars

represent significant gaps in house-building programmes*

This means that* because of changes in building styles

and materials* houses fall into three main recognisable

and relatively discrete groups* with associated status

connotations* Thirdly* it was considered that a detached

house has higher status than a semi-detached house* which

in turn has higher status than a terraced house or a flat

at least in Nch. Each of these three features (ownership

age and type) can easily be ascertained and measured*

The matrix combining these three parameters * from which

the housing score was read, is shown in Table 5* Only

 82

those scores which ore underlined in the Table actually

46
occurred in the sample, partly for obvious reasons*

The informant*s scores for each of the six indicators

vers combined, and the total was taken as his social class

index*

The social class characteristics cf the sample as a

whole, as measured by the index, are shown in Fig* 1*

This shows the distribution of informants obtained in the

sample by social class lovsl, and demonstrates that the

sample adequately represent# the whole range cf class

differences, from the lowest to the highest* This re*

mains true in spite of the fact that both extremes, as

one would expect from their proportion in the notual

population, are not so well represented as the middle

ranges* The graph ehows some breaks in the social class

continuum, and is suggestive of how the eontinuum might

be broken down into more or lees discrete units or

classes* Natural breaks which occur around dusters of

informants may Justify the social elaes divisions used in

subsequent chapters* Figs* 2 and 3 show that an index

based on only two or three indicators would have given a

much more obscure picture of the social class differences

between Informants and of the clusters and breaks in the

olass continuum. Fig* 4, when compared to Fig* 1, shows

 Aa.i. e.ks<;
£&i*pk.

i* Ifc 1% Z& i-?

Seoa.1 tloSA u*fltat.

%2. Secsal ck^ t&&fi4.en;to ^

~kj outspatto'A Q.4ui^dcjt«:iion

S&cicd cks$

tlfcfi ctdMOeC. tri^CS 6* Satv.pb

IK®.Sorel b*j (xztjp»:ht?*f
feiutcck^ fUvtd k\C©&\€ ,

I© II iV SsV

Se£«tl zlfcVl WU#,

 83

that the inclusion of th« locality indicator has a

significant effect in further stratifying: the sample.

fig:. 5 shows the social class characteristics of

the sample by age-group, This demonstrates that the

i

sample of schoolchildren, although taken from grammar

schools, is not biased towards the higher end of the

scale (see Chapter Three), If anything, the reverse is

true, Xt also appears that older informants tend to be,

on average, of a lower social class than younger inform¬

ants, as one would expect from the general social and

educational characteristics of the population as a whole.

Some supplementary information is given in Fig, 6,

which shows the sex distribution of the adult sample, by

age—group* (Ago-characteristics of the sample as a whole

were given in Chapter Three).

Table 1

Index of Occupation and Father's Occupation

Index Score

X Groups 3, k. 5
XX Groups 1, 2, 13* k
XXX Groups 5, 6. 3
XV Groups 8, 9, 12, l*t. 2

V Groups 7, 10, 13# 1

VI Group 11. 0

(General Register Office, Classification

of Oeoupatlons, 1966).
 |e<
5L 4» Seciat CUs;. CUff&d<tA<rfics o$ ■sai^pl#
3 % fH£4&tf<Ki fey IDC^U-Cl&H IfeoUv PUHUt
1
w
Uc&tiiy WUeeeW,
o b'
T
4*

■£. 2

(jlsaa
i€ 20

S&tie {r d&$?& (Mtl&si „

Pv}. £ Sod&Ucl^ CWa.cten\-lt<& ef la^pU

111
fcvj d^"3ro©p.
14

I*
H
on fO'

11,
•"3
J *
4

2-

0
Z3-2-? 30-31 40-41 &>"& &£<><» 1©+

*s- Oreop-

fvj. (o. Sey of

a^a-^foop.
FWU
 8k

Table 2

Index of Incorae

Index Score

G £2*000 ♦ ) 5
F £1,000 • £1,993 { (annual salary) k
£ £999- ) 3
D £20+ > 3
C £15 - £19
J (usual weekly 2

D £10 — £1^ 1 wago) 1

A £9- ' 0

Tabid 3

Index of Education

Index Score

I Soma University or College Education 3

II A~1gv©1 or equivalent 4
III 0~level» G.S»«E# or equivalent 3
If 15*
j 2
V 1* Terminal Education
) Age
VI 13- ) 0
 85

Table k

Index of Locality

Index Score

X Eaton* exeept Council Estates*

and R»E. Lakenbam 5
IX Tborpe k
XXX S. bakenham* Eaton Council Eetateo 3
IV Central Lakenham 2

V Hellesdon 1

VI tfeatwiok 0

Table 5

Index of HonaIn/?. Index Scores

Council Privately Owner

Rented. Rented. Occupied
T/r 8-P P T/F S-D D T/F S-a P

0 0 1 0 1 2
1 2 3
pre-1939 2 I 2 1 £ 3 2 a a
Poat-war 1 £ 3 2 3 4 3 a a

T a Terrace
F a Flat
8-D « Setai-detached
D a Detached
 B6

Chanter FIve t Social Contest ami tho Quoatlonnalro

The co-variation of llnguistlo and sociological

phenomena can be thought of as taking place along two

main dimensions* The two dimensions ares (a) the

dimension of social differentiation# and the sooial class#

age and sex of the individual{ and (b) tho dimension of

social context# and tho social situation in which the

individual is involved in social interaction* In Chapter

Four we examined the first of the«o dimensions of ©o-

variation# and discussed the relationship between language

and social class* We also demonstrated the methods de¬

veloped for measuring social class position* In this

chapter we shall briefly discuss the relationship between

language and sooial context# and examine ways in which

social context can be measured and controlled*

Linguists have always been aware that different styles

of language occur in different social contexts* and some

theoretical work# such as that of italliday# Mcintosh &

Strevens,3" (see below) has appeared on the subject of

stylistic variation* It is still the ease# however# that

very little work has actually been carried out on comparing

the speech of individuals in different social contexts, or

on specifying tho particular social contexts in which

certain forms occur* In this work an attempt has been

2
made, following Labov, to control social context accur¬

ately and to correlate context with linguistic forms* The

 87

purpose of this 1st first» to secure accurate statements

about the way In which language varies with social context)

second, to discover the linguistic norms of the speech

community* as they are revealed in the direction of

stylistic variation* from informal to formal contexts)

and third* to permit an accurate study to be made of the

effects of stylistic variation on linguistic change* and

of the mechanism of linguistic change as it is revealed

In the interplay of stylistic variation with social class

and age variation*

Style and Register

The two main dimensions of the co-variation of

language and society are reflected in the linguistic

terms social dialect* and style and register* Register,

1
as defined by Jfalliday, Mcintosh A Strevens, is the term

used to cover varieties of a language distinguished

according to use* rather than according to user* Registers*

moreover* "may be distinguished according to field of

4
discourse* mode of discourse* and style of discourse*.

These three features are of course in many ways closely

Interrelated* Style of discourse represents the second

main dimension of linguistic variation* and refers to the

relations among participants In a discourse* which is what

we are primarily interested in here* If the term style

 aa

is employed in this sense, an<1 not, aa is often the case,

fj
in a wider sense equivalent to register,^ it is clear that

this dimension, like the first, is capable of a scalar

interpretation and treatment* The ran^e of possible

social contexts, from the most intimate to the most formal,

/

is a cljlrso, but one which, in the view of lialliday et al»

is "unlikely ever to yield clearly defined, discrete
7
registers'*, Labov, however, hao demonstrated that it is

In fact possible to divide up th© stylistic continuum in

a useful, if arbitrary and artificial u&y, and control

the social contexts in which different styles occur.

This is largely because the how York City study, like tho

present work, is concerned with phonolojy, which is more

susceptible to this kind of treatment, whereas nalliday,

Mcintosh & Strcvens arc interested in th© grammatical

and lexical properties of different styles and registers.

Registers, however, are differentiated primarily by

(jamar ami lexis phonology, according to those thro©

authors, is not properly a characteristic of register at

all, Labor's stylistic study ©f phonological features is

possible largely because in Lrv;lish, much nor© so than in

n
most other lan/tuapcs, choico of dialect and accent^ is

very closely bound up with choice of register,(and

phonolo,7y is, of course, n. characteristic of dialect and

accent). Phonology is there;ore involved to a considerable

extent in stylistic vnri ;.tlon,rm * tends to vary alyo with

 89

field and mode of discourse* This Intimate connection

between dialect and register is largely duo to the very

high prestige of Standard English and lis associated

(although of course by no neons necessarily concomitant)

prestige accents* and in particular in Britain to the

prestige of the B.P* accent* It should be noted * howover*

that this connection does net necessarily imply that a

speaker selects a particular dialect for a particular

situation* H©r® again the dialect differences* and in

particular the phonological differences* are likely to

represent a cllne, as will b© demonstrated in later

chapters*

This means* then* that the two main dimensions of the

co-variation between language and society are closely

interpolated, in that linguistic forms which are assoc¬

iated with higher social groups tend also to be associated

with form! social contexts» since Standard English and

S,P, forme arc felt to be more suitable for formal

registors* (This it also true to a certain extent of

lower social group speech and Informal social contexts*)

This is an obvious -"-at * but one which is not capable cf

a simple explanation* Any explanation* however, would

appear to have to take three main factors into consider¬

ation* The first of these* the social prestige ami dominance

of Standard English and P.P., has already been mentioned

 90

above* Second* part of the explanation would seem to lie

in the fact that forml situations are higher class

situations* Formal situations normally occur when in*

dlvlduals who are not well acquainted* or who are playing

roles of significantly different status* are involved in

some form of soeial interaction* Members of the higher

classes* because of their access to higher education* and

prestigious occupations* have a wider range of social and

occupational contacts* They therefore more frequently

experience formal situations* and for this reason* as

well as for reasons oonnooted with their actual social

prestige* tend to dominate them* The community as. a

whole regard* formal social interaction as typically in*

volvlng members of the higher Glasses* and reacts accord¬

ingly when in a formal situation*

This reaction* which will naturally vary greatly

with the class* age* sex* personality and beliefs of the

individual* will be manifested phonologically in some

kind of movement towards pronunciations considered stand¬

ard or correct* Lexically and grammatically it will be

signalled by the adoption of a style and register con¬

sidered appropriate to the occasion* Some individuals*

according to their experience* background and education*

will be more successful than others In this kind of

response* Bernstein has shown* for example* that the

elaborated code* which is more appropriate to formal

 91

situations, is not available to children from working-

class hones* and that they accordingly have difficulty in
||
adapting to formal (particularly educational) situations*
12 11
Two critics of Bernstein** work, Robinson and Regan,

have shown that working-class children do, after all,

have an elaborated code available* it la simply the case

that they do not choose to use it, because playing a role

In a formal situation is not a familiar experience for

them, nor an accepted part of their culture. If they are

directed to assutas a role in a formal situation, then they

can in fact employ an elaborated code.

This faet is Closely connected with the third partial

explanation for the association of formal contexts and

higher class linguistic forms. Formal situations, as we

have already stated, usually occur in confrontations

between individuals not well known to each other, or of

differing role status, Xn these social contexts, for

psychological as well as social reasons, the individual

will often choose to attempt to turn the confrontation to

his advantage, and to play an impressive, or dominant, or

at least non-dominat ed social role, "Thus, when an in¬

dividual appears In the presence of others, there will

4:
usually be some reason for him to mobilise his activity

so that it will convey an impression to others which it

14 *
is l*i his interests to convey*. (From a linguistic
 92

point of view this is ©specially true of formal contacts*

particularly* it seems* if the individual is not from a

working-class background*) for this reason* dominant

(that is to say standard, higher class) linguistic forms

are selected, (Higher class forms are normally also

standard forms, because of the social prestige of the

higher classes, ami the control they have over the educ¬

ational institutions and mass communications media*)

Later on in this work: (see Chapters Eight and Eleven)
there is a demonstration of the importance of this con¬

nection between formal content end higher class linguistic

forms* The ooraplexlty of the inter-relationship* however,

is also pointed out* and there is a discussion of the

extent to which this connection affects and brings about

linguistic change,

***** ViG

It is clear, then, that stylistic variation, includ¬

ing stylistic variation in phonology* is of considerable

linguistic importance* particularly from the point of view

of linguistic change. The problem is now one of obtaining

information on this type of variation* and on the connection

between social class and social context in language. Ob¬

viously* a fairly formal interview situation is unavoidable

 93

in securing Information from a large number of informants

in a study of this kind* Labov has shown* however (and

much of his methodology has been adopted in the Detroit

survey) that the Interview situation can be structured so

as to ensure that information concerning different con¬

textual styles of speech is obtained* and that all in¬

formants are placed in a series of "contexts" which are*

15
relatively speaking, the same for each of them* In

this way* accurate information can be obtained on

stylistic variation of the type discussed above*

Formal Speech

The structuring of the interview is reflected in the

structure of the questionnaire* The questionnaire used

in this survey was designed specifically for work in Nch*

but the assumptions that underlie its construction are the

same as those upon which Labov based his New York City

Questionnaire* It was assumed* for example* that the

contextual style of speech elicited from the informants in

the bulk of the interview would be that style of speech

most appropriate to a tape-recorded interview with a

stranger. This type of speech is defined as formal speech*

Some informants* of course* feel more constrained and in¬

hibited than others, and are therefore more "formal". But

the position of this particular style on the style continuum

 94

is the «am«, relative to the other styles* for all speakers*

The formality of the context, too, Is relative* There are

obviously contexts which are much more formal than this

particular type of interview situation* and an attempt Is

made in certain sections of the questionnaire to increase

the formality of the situation* in order to elielt forms

which are likely to occur in theso contexts (see below)*

formal speech is likely to be elicited in response to the

questions in sections 1* III* V and VIII of the questionn¬

aire (see Appendix)* Sections X* XXX and VIII are in¬

cluded * not only to elicit formal speech* but also to

obtain information which is necessary for the survey.

Section V is included mainly to elicit formal speech* but

also to provide a suitable context for question V*4 (ill)

(see below) which is designed to elicit casual speech*

This section is also useful in that it provides for a

comparison between attitudes to Nch English (section VIII)

and attitudes towards JJeh itself* Formal speech* ae it is

elicited in these eeotlons* is easily recognisable as such*

Since this work is concerned mainly with phonology* it is*

for instance* valuable to have the formality of style

signalled by the use of grammatical and lexical features

such as the pronoun "one** and the passive voice*

The Reading Passage

Once we have achieved the setting up of one contextual

 95

stylo within the Interview# that of formal aueeeh* It Is

a relatively easy task to extend th** spectrum of contexts

in a more formal direction* The object of this is to gain

insight into styles that are more formal than formal speech

as here defined| to obtain information on a wider range of

stylistic variation) and to discover more precisely what

are the norms of the speech coiamuniiy* Section If# which

consists of the Heading Passage, is designed to elicit the

next most formal style in the interview, which is defined

as reading style. Here we attempt# in the context of the

interview, to simulate a change along the continuum of

s^vle of discourse by switching the mode of discourso*

(Heading aloud from a written text "is a special case of
l£
written rather than of spokon language"* We have al«
/ ••

ready stated that choice of dialect in Britain is closely

bound up with choice of register) the expectation here is

that a change of mode will produce a more formal style of

pronunciation#) The informant was asked to read the

passage, which includes many examples of phonological

variables (see Chapter Seven)# as naturally as possible#

"just like you'd normally say it"# The effect of this

17
instruction is probably not very considerable# ' but the
result is to standardise the style for all speakers towards

th© informal end of the range of possible styles in tills

context* Th© actual material and nature of the Heading

Passage is also important in this respeoti the passage is

 96

written in a colloquial stylo on an informal topic, and

as If spoken by a young man, In order to produce a per-

fomnet that is as relaxed and natural as possible*

(Here we are making use of the interconnection between

style of dieoourse and field of discourse*) The advantage
of standardising in this direction is that it avoids

certain characteristics of formal, slow reading style,

which would make the results difficult to compare with

speech.

TfteVfeqd

The next stage in the extension of contextual style

in a more formal direction is to be found in the responses

elicited in Section ZI, the Word List, which are said to

be in word list style* The informants were coked to read

aloud, at a normal speed and as naturally as they could,

10
a list of 212 lexical items* ' This means that, for three

or four minutes, tho informant's attention is directed at

a single item at a time, and at his pronunciation of that

particular item* The effect of this is to produce n stylo

of pronunciation more formal than that of tHe Heading

Passage, and thereby to introduce a further degree of

sophistication into the examination of contextual styles*

In addition to this, the list is designed to provide

 97

information concerning all segments Of Nch English, and

the phuiologioal variables in particular* The list,

moreover* ineludee ail the items in the Heading Passage

which contain phonological variables* This makes comparison

of individual items in different styles possible, and weans

that overall scores for the two tests are reasonably com¬

parable, It is also worth noting that a tost of this form

ensures that a large amount of necessary phonological in¬

formation can be gathered which one could not otherwise be

certain of eliciting during an entirely conversettonal

interview, (This is also true of the Heading Passage*)

The Word Eiat teat is of course an artificial one*

in that informants are not likely to use their *normal

everyday pronunciation** when reading out a list of single

words. It should be stressed* however* that wo are con¬

cerned here to produce a formal (rather than a "normal")

style of pronunciation* in order to obtain information

about the community's linguistic norms* and to extend the

range of stylistic variation studied* and that we are not

specifically interested in the pronunciation of individual

items in this context* but rather in scores produced for

different phonological variables* The Word List method

Is therefor© perfectly adequate for our purposes and there

is no need to adopt the /tore time—consuming "sentence

20
frame" type of question employed, for instance, by llouok*
 93

Houck. believes that, by using this kind oi toohnique, he

21
will obtain • Isolated approximations of casual speech",

because the informant*s attention is concentrated on

obtaining the correct answer, rather than on pronunciation.

This would appear, in the absence of evidence to the con¬

trary, to be n rather vain hope* especially in an admittedly

linguistic interview. The response, however obtained, is

a single item pronounced in isolation (in the presence of

a tape-recorder) and is therefore likely to have at least

son® of the phonological features cliaraot eristic of this

typo Of unusual utterance. The only way to obtain examples

of casual speech is actually to record speech. It la much

more satisfactory to recognise that a list of single items

will provide a style of pronunciation much more formal

than casual speech and to take advantage of this fact,

houolc's Intention is also presumably to avoid "spelling

pronunciations". Xt was found in the Noh survey, however,

that the effects of this type of pronunciation arc minimal,

and at the same time interesting and informative. Xt must

be said in favour of houck * a method that in its original

22
use in the English Dialect Survey and other similar

surveys it serves a definite and useful purpose. The rural

dialect survey is concerned to discover what different

objects and activities are actually called in different

areas of the country, ami there is thereforo a need to

avoid prejudicing the informant towards any particular

variant. There is also likely to be a problem of illlteraoy

 99

In this type of survey* By extending this technique to a

large urban survey, however, and by his excessive pre¬

occupation with "minimal pairs", ilouck appears to have

expended a great deal of unnecessary effort, particularly

in the development of suitable sentence frames for items

like "pud" and "gill"* In Nch, on the other hand, a very

large amount of information was obtained in a very short

period of time, without Irritating the informant with such

questions as "A little flying creature with feathers is

called a *. A small number of mistakes occurred in

the reading of the Word List, naturally enough, but since

the total number of items elicited for this test in the

survey was almost 13,000, the effects of mistakes was

totally negligible*

The Rapid Word List

Xt was thought possible that differences in scores

between the Reading Passage and the Word List might be

partially due to speed of utterance, rather than to formal¬

ity of context* This possibility was checked by the mater¬

ial elicited in Section VI, the Rapid Word List* Informants

were asked to read a list of kk items, and to say aloud 27

others (numerals from 1 to 20, and the days of the week)

as quickly as possible, but without being incomprehensible*

It was found that scores for this test were not aignific-
 100

antly <2liferent from scores for the Word List test* It

was therefor© conalxlerott that differences between Pending

Passage and Nord List scores could be accounted for solely

in terms of stylistic variation, and rapid fe'ord List scores

havo therefore bom treated as word list stylo*

The Pairs Test

Tho final stage la the extension of formality of con¬

text is contained in Section VII, the Pairs Test* In this

test, pairs of ite s which are often horaoplionous in Nch

English, tout not in H*P«, (or vie© versa), havo to be

read together. In this way, a maximum amount of attention

is focussod on a particular phonological variable, and tho

expectation 1® t!wt pronunciations will bo produced that

are even more formal than those elicited by tho Word List

test* In a similar tost in New York City, Labov found

that by contrasting normally homop'vonous pairs such as

dock and dark* a higher number of post-voeallo r*s in dark

on
was elicited than in other contexts. (Post-vocalic r

is a prestige marker in Now York City English). The Nch

results, however, are often more complex than this* Pop

instance,tho Juxtaposition of a normally honophonoua pair

such as bootiboat (.buitj might produce a corrected (towards

Ti*P*) form of boat as (.boutJ* It might also, however,

particularly with lower class speakers, produce instead a

 101

"discorrected" form of boot as j^bMttJ (see Chapter

Tleven)• This test has also been used to investigate

the Nch vowel system more accurately, and to study the

nature of phonetic differentiation within the system#

Casual Speech

The spectrum of contests has now been extended as

far as possible in the formal direction# The next problem

is therefore one of extending the spectrum in the informal

direction# It is obvious that, during most of the inter¬

view, the informant will be using formal speech# and will

not speak to the interviewer as he would to his fatally or

to his friends# How then do we obtain examples of this

more natural and typical casual style of speech? Labov has

shown that» although it is almost by definition impossible

to elicit casual speech of this kind within the context

of tho interview, it JL3 possible to obtain examples of

spontaneous snosch. Spontaneous speech Is "the counter¬

part of casual speech which does occur in formal contexts,

not in response to the formal situation, but in spit© of

£>ii
it"# It is also possible to tapo-rocord examples of

casual spooch outsi>© the iornal context of the interview#

(Spontaneous and casual speech arc in effect Identical

from the point of view of stylistic variation, and will

be referred to simply as casual speech la the rest of this

 102

work)» habov has, in addition, sot up a schema for tho

identification Of casual & poochi ho defines a scries of

contexts where casual speech may occur, ami lists a series

of channel cues which arc characteristic of casual speech*

This schema has been adopted with slight modification hero*

Tho contexts whoro casual speech my occur arc as

follows t

1) Speech outside tho context of tho fomai

Interview* This can occur before tho inter¬

view starts, after it has finished, or during

tho interview in breaks* for instance, for a

cup of tea*

2) Speech addressed to a third person.

3) Speech not in direct response to questions*

limy speakers, particularly older informants,

digressed considerably from tho subject at

hand, and wore of course encouraged to con¬

tinue with tho.tr reminiscences, stories and

favourite topics* In many cases it was

tbsr©fore possible to dispen3o altogether with

Section V. (There is, incidentally, a slight

danger of a bias occurring in tho casual spooch

 103

material, tiaco older informants tended to

provide nor© ©xanplos of this typo of speech#

This neans that speech forris characteristic

of older people nay bo over-represented in

the results calculated tor ensoul ©reach)«

h) Speech in response to questions in Section

111# on FAn dialcot words# 'lost aponhors

found this section onusin humourous and

nostalgic reminiscences wore frequentt and

the constraints of the formal interview

situation v.re often forgotten.

5) Speech in response to tho question "Have

you ever boon in a situation, recently or

sot ;o time ar;o, where you had a 4;ood lau^jh,

or oomethtn;; funny or humourous happened

to you, or you ©aw it happen to someone

©Is©?" Tills question* which was uuoptod

25
from liottch*s hoods questionnaire, conos

at tho end of a ca ries of questions on Nch,

ano on whether it i© possible to enjoy

oneself in the city# In this hind of con¬

text, moat informants loan 1 the question

quite natural and acceptable, .>n-* responded

readily with an amusing incident# Tho in—

 10k

formant is under some compulsion to tjake

the story soeta amusltust usually becomes

involved in tho story-telling; and the coiaedy

of tho situation to an extont thai overrides

the forianl constraints of tho interview,

The result is casual (in fact spontaneous)

8oooch« This question was used rathor than
■^6
Lnbov*s of death* question,*" as

nost Nch people sec-nod to have lived rather

more peaceful arid uncvontml lives from this

point of view tJv :i the inhabitants of New

Yorh City, Labov*s question was used,

however, if the informant was unable to re¬

call any anusin,; incident, Tho interviewer

was able to us© tho "danger of death"

question raost often in connection with

war—tiiao experiences, a favourite topic for

di(jressions»

The channel cues involved in the identification of

coo tin 1 speech are "modulations* of the volco production

27
which aflect sjieoch as a whole", ' Tho cues aroi a chanc©

in the tempo of speechj a chnnt;e in tho pitch rmv;oI a

change in voluno{ an.; n chruv;o in the r:\tc of breathing

(including; the occurrence of laughter)• Those changes

"fores socially significa- t a Inns of a shift towards a

 105

28
mop® spontaneous or more casual atyl© of speech". When

on© or moro of these channel cues occur in conjunction

with on© of the five contests listed above, the utterance

which contains them Is considered to b© casual speech*

Interviewa

Of the sixty Interviews, fifty were carried out by

myself and ten by a second interviewer, Mr, Adrian Hannah,

Both he ami I have lived most of our lives in rich, wo arc

the Same aand attended the sane- school, He wore,

moreover, both likely to appear to informants as university

students, and v® both attempted to conduct all the inter¬

views in thi' sane fashion. There are slight phonological

differences in our speech, but this eeeraa to have boon of

no consequence. There appears to b© no si, ;nif leant

difference between the two acta o£ interviews with respect

to either th© dt^rreo or the nature of the stylistic vari¬

ation elicited. This demonstrates the usefulness of a

scientifically structured interview procedure which per¬

mits objective comparison between on© interview and

another. Interviews were kept as irifortml as possible,

and moat wore conducted sitting in armchairs in the

Informant*© 11vine rood, with the tape-recorder placed

 107

to ©licit information concerning knowledge and use of

local dialect words, follows tho Word List, and allows

the informant ticio to recover before tho ?J ©a : in» Passage,

Section IV* Then con©a another relatively relaxing

section, Section V, which contains the questions on Keh

itself, and gives tho informant a breathing-space before

tho final reading tests. Tho Pairs Teat cones last of

all the reading tests, sine© in this test attention is

drawn, to a certain extent, to tho phonological variables*

Tho questions on ?«'ch speech aro also placed at tho end,

since these too are likely to draw attention to particular

aspects of the infornant*s own speech* Finally, Sections

IX and X are designed to tost the extent to which inform¬

ants accurately perceive their own speech, and the extent

to which they hoar the..iselves usi ig tdo eon mnity»s

linguistic norms, ©von when this is not tho case* They

aro also desivmou to test the extent to which Informants

ar© insecure about their own speech, in t!iat they char¬

acterise their snooch as incorrect* This information will

be employed in tho cliapter m linguistic change (see

Chapter Eleven).
 soditi orvse

smasxnr 0 • nmomm*

YMO 899IV1
 108

Chapter Six* The Social Differentiation ef a

■ ,
. ,. ■■ , I^ure

In the last tec chapters we have been discussing

methods of measuring social class and social context

which will permit a study to be made of the relationships

that Obtain between these sociological parameters and

Unguistic material. ¥e shall now turn, in this and the

following chapter, tc a discussion of methods of measuring

the linguistic material Itself*

The main purpose of this work is to investigate and

draw conclusions from phonological material* (The reasons

for the selection of phonological rather than grammatical

or lexical material will bo discussed in the following

chapter*) However, the investigation of the co-variation

of phonological features with sociological parameters is

a complex matter, and some reasonably sophisticated tech¬

niques arc required for this purpose* Before we begin

this investigation, it is useful at this point to give a

simple, preliminary demonstration of the fact that it is

indeed possible to measure linguistic material and to

investigate the co—variation which it under, roes with

sociological features* This will be done by taking on©

non-phonological ami easily quantifiable linguistic

feature and examining the way 1*1 which it correlates

with social class, social context, an * age-group.

 109

The Present Tense Marker

The non-phonological feature which we will take

for this purpose is the form of the third-person singular#

non-past tense# non-eontinuous aspect of the verb la

Norwich English# as in "he loves"# "she has"# "it goes"*

It Is a characteristio of EAn dialects that this parties

ular form for all verbs other than the copula "be"* is

identical with the form for all other persons In the

2
same paradigm* That is to say that forms such as "he

love"# "she have"# "it go" are usual. It is likely that

this feature was at one time more wide-spread geographic¬

ally in the South of England than it is at present#*' but

it has recently gained some recognition from linguists

h
as a typically EAn feature* Nelson Francis has stated#

for instance# that the lack of the third-person singular

marker is standard even for the youngest speakers in

5
certain of the areas of Nfk which he investigated*

This feature is also said to be one of the character-

istics of American Negro speech*

In Norwich Itself this particular linguistic feature

is very common* It appears to have# however, all the

characteristics of a linguistic variable* it does not occur

consistently in the speech of most individualsj it is not

used equally by all speakers{ and its incidence varies

from verb to verb* This particular aspect of tne hetero-

 110

geneity of the speech community eon be ascribed to the

Influence of the Standard Language* through the presence

of Standard speakers in the eeramimlty* the noraativo in*

fluonoo of the educational institutions* and possibly

the mass media* and to the influence of the dialects of

those areas adjoining ISA whloh do not have this feature*

through diffusion and immigration preeesses« The hypo¬

thesis underlying the work described in this chapter Is

that the proportion of verb foms without the third-

person singular marker in a given amount of speech will

shew a direct relationship to the social class of the

speaker and to the social context* Because of the increas¬

ing influence of the standard Language it is also likely

that there will be son© kind of relationship with the

age-group membership of the speaker*

Method

To tost this hypothesis is a relatively siraplo task*

especially when compared te the problem of investigating

phonological variation* The method employed for invest¬

igating the co-variation of this linguistic variable with

sociological factors is as follows* For each informant*

in both fornal speech (FS) and casual speech (fl ) styles#

a count was taken of the total number of marker—loss forme

 112

introduelnc a bias into the flpres at tMi pr^lioinary

stage, jTigorea for the andor-twenties aorc not intro¬

duced until lator.

SSSS2&

The thiri'peraoR singular present to?*»© oarkcr aoores

obtained by the adult sample are shown In f ig» 1, for 0£>»

and Fig* 2, for F3» No informant with a social class index

of 19 or over had any ooourrenooo of marker-less forms, in

either contextual style* Thee© iafermnta do not appear

in Fige. 1 or 2. Note also that no relevant ferae wore

recorded in CS for these informants with social class in*

dex scores of id, 1% or 16* A oonparioon of Figs• 1 and 2

reveals immediately that the hypothesis is at least partly

confirmed* Scores for CS are on the whole significantly

higher than those for FS* with six groups of informants

Showing soores of over 90 per cent in CS, and only two in

FS* The percentage of marker-less forms, that is to say,

falls as the formality of social context Increases. This

linguistic variable is thsrofors a significant indicator

of social context. The other part of the hypothesis, that

this linguistic variable will show correlation with social

class, is not so readily demonstrated. There is, it is true,

a definite tendency for marker scores, in both stylos, to

c e ce a af © aX9 o I © © 4• % 4« ee 4«■ i3 *4 3J • 4« e iaf C u

»4 a af P( c( a 4< ie fi 1 f 0 4• 1 a0 0 I

a f sS *9 o> a g
.4• 4• I 3^ 4« 4< 0 .af■' 0 c »4 4< 3 »4 4• a c O laf « G■ tf 1* n 4• 0 #" 1

*« 0 *ag tA 3 4■ £a 4• :f■ e P• w ?T S"

e© 0< 1 4■ 0 * >+ X cf 4« u< taf s § *4: 0 o9

9 4« © 0 « © © a© af c a 4•. a 4• c im »> >o

65 O 9 4• 0 4« o 4* ©e0

X9 c» 4• 11 4• es 4• 9 »4 im S af © U © af 9 0 c9 © 9 *a 4« 4« 4• 4■ 3a »4 SJ 9
'

'

9 4• 0 9 9 c 0 S »4 4- C 33 a 1 4• © X *2 e Xe G o !X c 4« C 9 'X e 9 as a e taf 19
.

38 O 9 4< 4 9 4< 3 0 iff 4- ©• G « 4• c © H 9 af3* © af » a © af 9 9* 4- ® 9 af © »♦ 9 Saf © "»4 "i a© t o4

af © 4« & 34 4• > © 4i s! § 4■ saf 4• 4< 4 X 9 4■ af 4• 5 34■ lff 34 4■ 5J 9 X © > 4< § 3^

r
3< af © 9 C o .•« s af »
 114

which reflect the class structure of society as a whole*

There is no reason* incidentally* why a particular class

grouping should he relevant for all linguistio variables*

since different variable* will have differing degrees of

social significance* and will also he capable of reflect-*

lag break® in the class continuum at differing points*

For the purposes of the linguistic variable in question

here# however* the first step is to look for clusters of

scores* or for breaks in the continuum of scores which

may well reflect breaks in the social class continuum

itself* and to make tentative class divisions at these

points*

An obvious first move in this direction is to group

together all those informants with social class scores of

19 or over* who have 0 per cent roarker-1 ess forms in both

FS and OS* This* the highest social class group* will be

called Class X*

Xf wo now look at Fig* 1* the CS scores* the compos¬

ition of the second highest group also suggests itself*

The CS scores for all the groups shown in Fig* 1 range

between ?0 and 1GO per cent with the exception of the

anomalous group with the social class index score of 11

and those groups with index scores of 15 and over* Thla

series of groups has much lower scores, of below 40 per cent*

 &j. >, % Hftffe^r-iess •for*** fcvj etass
5 r^DpS - C% r 0 »\ lv.
tec

4 ts

4&

£0

•JEUHIWI TmllcMn -vrs-c 7UJW. «-F ^

3» 4- 4" 7 « <1 " '2 is 14 K lb r? if

Class. Ih.de«:

|eo 5g>.%^irh^te |offc& bq ck«s

~~

^rtsfcpS —■ fi Mu.(& &yd^.

j

4©

20

EX .i^^SSroy^W—

3 * € & 7 l i lb n d is IS l& I? It

Q&%% "£x-d,&X.

loo %%, % Mmrfcw- f»r*$ bij %d<

Ctass -AbLiXH e^Oj.
CS

cs

FS
20 FS
C.S

0 rvri
5 El 1
Seci&l Ckss
 115

There is, moreover, a break in tho eo tinuui.i between

tho 100 per coot at index 13 and 33 per cent at 15, Wo

will therefore set up Class XX as consisting of all the

informants from the four cells 15 to 18, The decision to

divide at 15, rather than at 1'*, for which no CS scores

wore obtained, is supported by the FS scores (Fiji 2)

which ©how a lar.,7© drop from CO per cent to 25 per cent

between 1'* and 15, FS figures also ©how that Class IX

groups Siav© the lowest ©cores of all the groups (with tho
exception of the anomalous colt tt 12), aid th'it two of

thoso groups Siave 0 per cent, including the coll at 16,

which did not 11 *i.tr© in the CS score®.

In the selection of further class croups, CS scores

are not ©specially helpful, since thoy are not very sij^nif-

icantly differentiated, 'J© therefore turn to tho FS scorea

(F1&. 2). ".ore, what stands out most clearly Is tho eon-

position of the lowest social class croup. Groups 3*6

liavo scores of above 80 per cent, whereas all tho other

groups have scores of around 70 per cent or below, There

is also a eijTiif icruit Pro-. •' la tho continuum of scores

between 83 P©* cent n,t social claw * index score 6 and

6h per ocnt at 7* e will therefore sot »»•» tho lowest

class as consisting of the Groups 3 "6*

There thus remains a do far undifforentl&tcd central

series of ei;:ht .'.troups, 7 - 14, Kit'tin this series, there

 116

la one really sl&nifleant break in scores, between 71 per

cent at 10 and 38 per cent at 11* If we divide here* we

have two classos consisting of four cells oaohi Clans 111,

11 - 14, with F* eoores or over 60 per cents and Class IV,

7 * 10, with PS scored of 60 per cent or less* This Is

In fast the only out which will produce such a satisfactory

division* 1 cut between 11 and 12, for instance* where

there is also a largo break in scores, produces two classes

of differing sice with considerable overlap In scores and

little significant clustering*

Ve have thus established live major social class

groups whoso composition with respoet to the social class

Index is as shown in Table I.

Table 1

figaag&i&ga M

Class Class Indox

X 19*
XX 13-18
XXX 11-14
XV 7-10
V >6

The relatively sophisticated and comprehensive nature

of the eoolal class index moans that occupation is by itself

 I » t § E
ft ft ft ft © ft K

5? ♦t i*

£ £
If I

«*
*%
ft H
3
*

s>
ft
a

«si
I

«3

w
# i I t 6
ft ft ft » 5 © ft » ft

*
t- «+
<*

I! • ? 8

fcS
til *S
o
i s
M
ft
I
ft ft

r?
a t a
&
ft a ft § ft e
ft C 1 © a

«f
jS I*

<s

9
* I
l

«*
« c I iE i
ft a
ft

ft § § s? o ft
i
!
» a
H

f
i
e

» o I ft
n
0
1
O ni*
a
«►
£ % S o $1S or H ©
4 £

15
§ s »
s

<5
3

cf
t-
MJ
J-
©

S3
«
11 2 £P
gr

l- •ft
M4
U
a* 1
 ii y

cent) the possible number of marker-loss forms, whereas

the MlfC score stays relatively high at 64 per cent*

Lower mlddle-olaaa (LIO) (Class XX) speakers use hardly

any marker—less forms in their FS, at 5 per cent, bat a

fairly considerable and much larger amount, 29 per cent,

in their OS,

We mm say, therefore, that there ie an overall norm

for the working-class (LtfC, iWC and UWC) of over 76 per

oent marker-lees forms in everyday speech, and that, while

each of the three Classes is clearly- differentiated by the

use of this variable in their everyday speech, the most

striking difference is due to the adjustment made by the

WC in FS. The middle-class (LMC and MMC), on the other

hand, use very few marker-less forms, or none at all, in

FS, and a relatively small amount, less than 30 per cent,

in OS* Use of marker-less forms in CS distinguishes the

LHC from the MJiC* Ths very large gap between the WC ami

LMC scores of respectively 75 per cent aiMl 29 per oent in

CS support® the hypothesis developed in Chapter Four that

the biggest class division in modern British society is

that whioh exists between the working and the middle

classes* Further support for the groat social importance

of this division le given by the fact that the two Classes

which show the greatest awareness of the social signif¬

icance of this linguistic variable, that t® to say the two

Classes which shov the greatest amount of stylistic vari¬

ation between rs and CS, are those two classes which Have
 120

greatest cause to be aware or this major social class

division, the two "border-lino" Classes, the UC and tho

UWC. One can propose, as one of the mechanisms behind

this phenomenon, the social need for the LMC to dissociate

themselves from the working-classes, and the desire of the

UWC to acquire some of the a tributes of osiddle-elasa status*

Fig* k now shows that, If we also include the scores

for those informants aged nineteen or loss, no significant

change in the social stratification of this variable occurs•

The only change that does occur has the effect of increasing

the differentiation between the UWC and the JflfC, ami slight¬

ly decreasing the differentiation between the HWO and i#C.

We can therefore conclude tivnfc the incorporation of the

sample of schoolchildren into the win sample Introduces

no serious bias into the figures for this linguistic

12
variable•

As we mentioned above, a further aspect of the social

differentiation of language which is worth exploring here

is the possible correlation of this variable with the age-

far each ten-year age-group of informants in FS only.

(There is not sufficient OS material to permit a similarly

differentiated analysis for that contextual style*) It is
 too
fig -4; % H&r&er-lgss |or^s ^ Social Cl&s$
full i:uv.,f{£.

c.s

40 PS

zc PS &S

o IE
5V X

S&ci&l Class»

10© %€, % Marbtf-U^ kij ^jt~3reup

to

40

20

ItrW 20-z4? 36-SFJ 40-41 10+

A^-^ooP.

loo %k % Marker- Utt $r&vi ftijt^ffcup

.j&rml ^pfetik, civil casual ifA,
-

cs
SO
c%
y (>0 <$ c$
PS
40
PS
PS PS
2S

0*
10*$ as-2)
 121

el©or, first of all, that the linguistic variable in

question is involved, in a considerable amount of age-

group variatloni scores for different groups range from

37 to lOG per cent* It is striking, moreover, that

speakora aged seventy or over use 100 per oont marker-

less forma even in PS« For people bora around or before

the turn of the century, the marker—less form is clearly

the norm* This suggests that for the great majority of

the population of Nch, the omission or inclusion of the

third-person singular present tense marker has become a

linguistic variable of social significance, to any great

14
extent, only since the First World War, when people now

aged less than seventy were still growing up* One eon-

elusion that oan be drawn from this, albeit tentatively,

is that tho general increase in population movement during

the war, and perhaps in particular the contact wfiieh men

from Neh experienced with speakers from "narker-ful" areas

in the armed forces, marked the first stage in the local

stlgmatisatlon of tmrker-lessneas.1** Speakers aged sixty

to sixty—nine shew the first sta ,tm of this stlgiaatlsation

in that they have a fairly considerable amount of styl¬

istic variation, falling from a high rate of 92 per cent

in CS to a FS score of 73 per cent* There is next a very

sharp drop to the scores obtained by younger speakers,

starting with the fifty to fifty-nine year-old group, who

were, at the most, nine years old at the end of the First
 122

World War* This group has a score of 38 par cent in PS,

The drop* however, Is considerably leas marked for C3,

This means that there has been a marked fall In the per¬

centage of marker-less forms characteristic of everyday

speech* hut there has boon an even more marked rise In

the stigmatiaation of raarker-lesanoas. The scores ob¬

tained by all groups of speakers aged between twenty and

fifty-nine are very constant, around hQ per cent for FS(

and around 70 per cent for OS (see Fig* 6)* This suggests

that this variable is no longer involved in linguistic

change in Nch and that marteer-leasness, after undergoing

a marked set-back around the time of the First World War*

is now holding its own in Nch, and since that time has

been functioning as a marker of social class and social

contest at a reasonably constant level for all age-groups*

The continued survival of this linguistic feature is

further indicated by the figures for the under-twenty

age—groupt 38 per oont in FS, and 30 per cent in OS* This

rise in scores, however* should not be interpreted as In¬

dicating that marker-lessness is again increasing* It

rather confirms its status as a stigmatised feature acting*

for those members of the population aged undor seventy*

as a marker of social class and context* The higher

scores for the under-tweatiss can be ascribed to their

relative immunity from the normative and stigmatising

influences of the speech community, and to the greater

amount of group cohesion that exists amongst this age-

 123

group* XI is significant, too* that the aaoant el styl*

istle variation amongst this a^e-group is smaller than

that saongsi the twenty to fifty-nine groups, which

farther stresses the relative isolation of the under-

Id
twenties to normative influences*

Added confirmation of the status of the variable

as a stigmatised feature which is not in fact increasing

among younger speakers can be obtained from a further

breakdown of the figures. Table 2 shows the OS scores

for younger speakers (forty-nine or younger) end older

speakers (fifty or older) in the five Glass groups.

Table 2

Per Cent Marker-less Forms by

Two Ago-areups and Social Class

Class 10-39 40+

X MMC 0 0

XX LMC 33 25
XXX OkC 77 43
XV Mire 87 89
V IMC 100 95

This shows that older speakers in the two groups most

sensitive to this feature, the LMC and the OtfG, (and

particularly the latter), have lower scores than the

younger speakers, but that there are no significant

 12 k

<liffemo«« in the other Class groups. This variable

therefore has all the characteristics vhioh Labov*^

associates with a "stlpstited language feature11 with

no associated Unguistio change in progress. According

to labor* the different class and age groups in the

population will show the following amounts of thie type

of feature«

lower Working lower tipper

Glass Glass Middle Middle
Glass Class.

Younger high higher higher low

Older high lower lower low

Xn this particular ease the configuration is in faot as

follows (of. Table 2)i

tire Mire uwc img mmc

Younger high high higher higher low

Older high high lower lower low

The position, in other words, is identical with labor1s

schema, if we interpret the Lire and MUG as being equival¬

ent to his "Working Class", Labov*e explanation of this

configuration is that "the middle-ranking members of

lower status groups, such as the upper sections of the
 12 £

entiatlon occurs in FS, bse&uss of the large amount of

stylistic variation exhibited by the twenty to fifty—nine

year-old groups* It would appear that* Just as the 8*0

and the UfC are the classes most aware of the sooial

significance of this variable, the middle age-group, as

to
Labov has suggested, ' are the most aware, beeanss of

their experience and occupational contacts, of the

inappropriateness of marker-leas forms to formal con¬

texts and of their general sooial stigmatlsation,

Xt has thus been demonstrated that it la possible

to measure linguistic material sufficiently well to

permit an examination to be carried out of the sooial

olass and age variation that occurs in language. The

particular linguistic variable on which an examination

of this type has been carried out in this chapter has

emerged as a significant marker of sooial class. It is

also involved in stylistic variation, and is, at least

subconsciously, recognised by the speech oomrrmnlty as a

stigmatised feature. The linguistic variable was also

involved in linguistic change, at about the time of the

first World War, when stigmatisatlen and the influence of

the Standard Language first became apparent, but appears

to be no longer involved in this type of change.

 127

Chapter Severn The Phonological Variable^

V© have already stated In preceding chaptera that

the main purpose of this work is to investigate and draw

conclusions from tho co-variation of sociological para*

meters with phonological material# It could# however# he

argued that lexical and grammatical material might have

provided more profitable objects of study for tills kind

of work# There are two main counter—argumenta to this

point of view# The first is that the speech of Nch is an

urban variety of English spoken less than 110 miles from

London in what is# at least linguistically, the South of

England# This means that lexical and grammatical devi¬

ations from Standard English are not particularly numerous

or especially groat# This in turn moans that the amotmt

of grammatical and lexical variation found in Nch# although

in some way© considerable, is not em ©specially worth¬

while target for investigation# and presumably of much

less linguistic interest than corresponding variation in

cities, say,in the North of England# This is not to claim

that a largo amount of work of son© interest could not

bo carried out in this field in Nch, but merely to ©tress

that phonology itself offer© a more useful sphere for

investigation# The second argument is that phonological

features cannot bo consciously suppressed by informants

nearly so readily as nrara*.tical ami, particularly#

lexical features# Naturally, informant© can attempt to

alter thoir pronunciation during the course of a linguistic

 128

interview* and usually do* (This alteration or die-

tortion is itself of considerable Unguistio interest*)

They cannot, however, refrain from using a particular

phonological unit altogether* A stigmatised

item* on the other hand* eon quite easily be suppressed

during the course of the interview,1 Phonological units*

moreover* are of maoh hi^i^r frequency of occurrence in

Speech than most grammatical and lexical features* and

are for this reason alone a more suitable field of study*

There is also the possibility with phonological material

of relating variation in one unit to that in other units

in the same system* and of measuring the degree of vari¬

ation along some form of (phonetic) linear scale* Those

possibilities do not present themselves in the ease of

non-phonological material. Phonological material* in

other words* is more readily quantifiable*

The linguistic variables dealt with in the rest of

this work* therefore, will be Phonological variables* A

phonological variable can be defined as a phonological

unit which is involved in co-variation with sociological

parameters or with other linguistic variables* Note

especially that there is no necessary connection between

the term * phonological variable* and the term n phoneme1* •

A phonological variable can rather be thought of ao em¬

bracing the pronunciation of a particular segment in a

particular class or group of lexical items, and may

 129

Incorporate more thnrt one phoneme, or be employed to

investigate possible dumemlc contrasts. These lexical

olasses will be left undefined, but a few examples, to¬

gether with comparisons with K.P, and some historical

notes, will leave no doubt as to what items are involved*

Before we go on to list and describe the various

phonological variables, it will be useful to describe

briefly the most salient features of »ch, Sfk and RAn

phonology generally, so that the Nch phonological vari¬

ables can be seen as part of a complex but coherent

system, rather than in isolation.

Norwich
PhonoLojv
The pronunciation of EAn English is clearly that of

a variety of the English of the South, rather than the

North, of England* This is to say, for example, that

^aJ or some variation of this vowel occurs in items such

ae cup* hut* rather than amI a distinction is

made between items from the lexical classes of put and

putt* It also means that a long vowel {.at J, or some

variant of this, occurs before {.fj, [PJ and i,® Jt as well

as in items such as ohance* danoe* plant* rather than a

short vowel [aj* KAn speech also clearly belongs to the

South—East rather than to the South—fc?est or South Central

England* cost—vocalic /r/, for example, does not occur

2
in items such as cart, port* beard, fur.

This brief characterisation of Bin phonology leaves

 130

It undifferentiate*!* aa yet* from the accents of the home

Counties* EAn speech, however, end particularly that of

Nfk and hast Suffolk* la quito strikingly different from

tluit of the counties nearer to London. tfJiat, then, aro

the differences? «fe can first of all ;;ive some idea of

the distinctive characteristics toy listing son© of fch©

points at which the phonology of Mch English differs from

that of the Homo Counties and from 11. P. (ST© can sub-

eequently outline some of tho points at which Nch speech

differs from that of rural LA.) The following is a short

list*

1* hefloxes of ME o
u
and MD ou aro kept
x distinct

in Mch r.n&lish* as /u»/ and /ah/* (Under

?ns ou we must include late ML* ou from 6 ♦ 1*

but not, necessarily* late fffi ou from OL

6 ♦ h(t), whore Noh /-enorally follows R.P.)

This oeana that, in contrast to Homo Count¬

ies speech and H,P., Nch distin^ruishoo

pairs of items such as the i oliowiUt" I

moan I mown

sole t soul

nose i knows

to© s tow

Items in the first column aro pronounced

with a l,Uu di —' u» J vowel* those in the

second column with (,i\u E-u —wuj. Final
 131

unatressed open syllables generally have

/u»/ whatever their source# as in billow*

window# follow# The item n<£ has /axk/ as

the negative particle* /m/ as the ad*

verbt e#g# "No* that*s no good* [hau

ntfu gUdJ#

2# Reflexes of HE o have two alternative

pronunciations* Soice items, such as do,

loose, goon, who, have {,«» —am —-em —3wj

m /wt/ in the epeech of all non~R• P.
speakers• Others, such as boot, moon#

fool# can have either /*/ or /ai/# This

means that in many varieties of Nch speech

pairs such as the following are homophonous t

boot « boat

moon t aoaa

soup i soap

tomb i tome

3# Items such as tune# news# music# <jn©tm,

which have t,Ju» — Jul J in R*P# have no

palatal glide in Nch,*^ and are pronounced

with a vowel identical with the /*«/ de*

scribed above# This means that in many

varieties of Nch English pairs such as

the following are homophonoua*

 nui$i # who

cuto « coot

OUtQ t OOOt

But© I boot

Xtona such as onro are pronounced with

/3»/» which ooaaa tlxat patro such aa tho

following oro honophonous t

puro i purr

euro » our

sorely t Shirley
Slot© that /j/ is conerally rotainod la
jk
owe* valuo (but not ens

In ©OBJ© varioiioo of Nch I -ncliah* iter jo

audi a3 boat > solo* which &ro listed under

1« with /ut/i are ©abject to shortening;*

to /b/* This tioanOf Tor ©aanpl© * that

itoci® each, as road and hood oro perfect

rhyaoo» Tliio "IiAn abort €»* would appear

to bo linked historically to the porhopo

ssoro widely known "Now Cn^land alxort j£a»

with wliioh it 00©jus to kavo may character-

K

lotioa in coonon* Both feature©# for

instance* appear to bo receoaivo* and vary

widely la their incidence fro. 1 word to

word* contest to context * and opoakor to

speaker*** Neither of thcca# raoroower* ocouro

 133

in final opw syllables* aa in #ro. The

New England short jo, on the other hand*

contrasts with the vowel of rot, mil,

whereas the EAa short <> does not* The

EAa shortening* moreover* also occurs* (to

a United extent only in Hch)» with the
/m/ items listed under 2*» such ao boot,
sooon. This is usual in the context

/f/t as in hoof* roof, proof* and is

common in the context _
/o/» as in room.

broom. where /U/ alternates with /••/ and

/««/«

It should be clear from tho above

that a complex system of phonological

contrasts involving /tti/* /«»/* /U/, and

/ss/ operates in Nch English* (especially

when correction towards h»F* and hyper-

oorreotlon are also taken into account) *

which distinguishes It both from R*E* and

from Home Counties speech* It should also

be clear that the latter two varieties

have only a three-way system of phono¬

logical contrasts at this point* compared

to the four-way Wch system outlined above*

5* Items such as hero and there, which are

 13k

distinguished in n.P# Jind Horn© CountAoa apooch

as /id/ ami /ea/ respectively, aro centrally

fj
not distinguished in Kali. This means that

pairs onoli as tho following are hoaophoaoust

fear i fair

hero t hair

boor « bar©

Those itoi.ia aro pronounced with a vowel ©£

the typo with tlio possibility of

a schwa off-clid©#^

At a lower pJionotle level tharo aro

very many other differences that coul '• bo

mentioned* Potable nroi tho hyporlencth

of stressed ion^r vowels* mid tho corres¬

ponding roduotion and disappearance of

unstressed vowolst e.g. thirty-two

il®t3uj| th© tendency to centralisation

of short vowels| tho narked nucleus-slide

differentlation of tho diphthongs* with

second elements approaching {,1J and L«J

rathor than j,! ~ ©j and j.17 ~ $J* ©•£• <ter

td&*ijj the very front ta*J vowel in, after*

carti and iho unrounded nature of the vowel

i** ^0t« tOP*

l£a-st flrt-jLl ctfy PfiprwLojy

Tho speech of ffch differs in aono interesting ways
10
from that of rural IA as a whole# To complete this brief
 135

characterisation of EA phonology* wo shall list son® of

those differences below* Information on the rural speech

It

of EA lias been taken from throe aain souroesi* g&korlts*s

study of the Suffolk dialect* for which field-work was

13
carried out between 1926 and 19301 Low»an*s records

of field-work oarrled out In Hfk and Suffolk in 1936 for

tho Linguistic Atlas of the Eastern United Statesand

the records so far published by tho Surrey of English

Dialects (SEO) for Nfk* Suffolk and Essex*and the

as yet unpublished field-records of work carried out for

the SED by V* Nelson Francis in throe Mffc localities,

SED work was .carried out between 1956-57 for Nfk* 1950*59

for Suffolk* end 1953-62 for Essex* Map 1 shows the

various localities investigated by the various surveys,

Moot Of tho feature# typical of rural speech listed

below are also found to a limited extent in tho speech

of some* mostly' oldorly* people 'in Neh.

1« Of the tfoh characteristics listed above*

the distinction /ot/-/a*^ appears to be

or have been usual throughout Nfk* in most

of Suffolk* and probably also in Essex*

although the phonetic nature of the

distinction is not in all cases the cause*

(i) Sbkerlts has tv»j (HlNiJ) in

alone* coal, clothes, loaves.

soao. toe, otc.* and i«0 — »U — oUj

/

fc.

o Sfi
PPpey
*£)<>cki t. ShonLfiL

BL'ckUn
Horfol K
*
CrCmsfcon •
k- Blink clm
« Ltudhum
& S- WaUhatvi
q f\!e.cb
i\iz<~cori
♦ Wlt/i/
*
Shiydhxin NOdwtCH '-i-:
•
OutFweli ♦
G-ooder&fcone QeeclhcitYT
*
•
(shweU th oefK
U>w*sit>FiS (Z
N /

PlcLhclM oiUaUhkU
9

9
C&fiboldtitcLftl / ChedFto^ IM*Zworth

'••'"Fr'^ljheldX ** Sou tkwtUig

4
% .

• Pit dele nha-M

O
UcnCn(jtoi\ itfXcdhwk-e- X
9

rv *
» «<« e HencllKSkdM ycxfovcL
Suffolk
Mri8/2/j)CE © (iujchull
/
t
4
■

Pp'jtch-
* x
4
\
ke&UiiqtoH IPSlvtCf/ Weodbe-uij e
o
,S ^9% \
* d'-s
ft ' I
MiO tl^dnxK

ESSIES
MAP I

kitru-L iDidUclS Sf,ri'ii/F

Kokerlt-z X

Lowr\\Ciy\ O

£> . t. J). *
 in blow, crow, grow, know,

though, etc* Items stack as

bough11 thought , which gener¬

ally have /at/ in Nch, as in

P»p.» also always have /au/ in

this dialect.15
(li) Lovmn's records also show a

clear distinction for all three

Nfk and four Suffolk localities.

16
He writes * for example!

Neeton clothes >o*d (<or

Stiffkey clothes SO*-5 £dtt'

S» Walsham clothes 0" low >vou<

Ilketahall clothes ?o«3 though (SoU

Martleahaia clothes ao* roU

Honington clothesj «o» though "AttA

Buxhal1 clothes <0* though caU"

Setae localities have a closer vowel*

in some items* of the Kch /u»/ typei

e»ig» 12a l®<«*

(ill) The position, twenty years later*

as portrayed by the SEP material,

is less clear. In the published

material* all the Nfk localities

have an LAUJ type vowel in mow,

colt, or colts-foot, and /ut/' as

Lou — oU — 01U etc,J in load,

unload, ovor or foal, Mono of th©

Suffolk or Isogs localities, how-

over, appears to have preserved

the diotinctlost, except that sou©

Suffolk localities have a high

vowel I. fti — aUj in foal, as opposed

to all other cases. It can

therefore ho asouned, in th© ab¬

sence of further evidence, that th®

CAn /ut/—/a\%/ distinction has been

lost, or almat so, In Essex and in

17
nost of Suffolk, and now survives

mainly in Kfk, including Neh. Th©

cuaount of phonetic differentiation

would appear to b® greater in Nch

than in t-/e rural areas, sine©

Lowr&m writes j.©*3 o* j ami Francis

J^ou — oil J for Nch [Un as J, although

sotte exanplce of hi her vowels do

occur. For instance, Francis writes

road, and toad with j.^sJ in Orlfmaton,

«md coal with J in lut'Xioia* Under

Lu.iham he writes "ME

t
p>^ouJ, not
as hif;U as Cardinal ^uj usually •

This is ft feature of the lower-class

dialect of Norwich, wSiero IE o is

cor.Hsonly Ln»J," Under Asftwellthorp©

 138

he writes» "ME 9 >teuj but some¬

times very high or eron i«t« J*"Xa

2« In rural £4* reflexes of MS 9* generally

speaking, occur with /*«/ or /tJ/» and only

rarely with /»»/• forms with /V/ appear to

be particularly common la Suffolk*

(1) In ShkerltBf reflexes of MB 9 are

most commonly written with the

sign liu, t]t whieh appears* from

his description, to be a close

rounded front or oentral rowel

approaching i«»J* with a certain

amount of diphthongisation. A

number of items of this type also

occur with /*?/» which is much more

frequent than in Nch, and probably

than in l¥fk as a whole* Some

Items'also occur with come type of

H » F *-imitative /ui/. K&kerita

commentsi "The children use l*«J •*•

and alee a kind of hyper-correct

rising diphthong [fieij in moon and

soon m« X hare heard the St* Eng*

rowel tu#J for instance in boot.

19
do. food ..." ' Also noteworthy

are a number of items with /a/»

 4• n \C

01 21

** fi 0- O

1 0C sA 0• 0 IH 4 1

^ OHa i »4 A§ 9 9 9 »t<

J| M{
O 4« »A 9A .P

91 0 0« • 11 *4 A0
I z i S J

« 0« «% O * J K »4 3 * £H ©

i 0 « 5a %m a >« A 9 4« »4 9

»- < »— •

s a & * a i~ «* %■< *

I 0 ni 1 •- IV 1* »*< 0s

3 tf S 0& #1 I S SJ >

000 0« »4 5< 9 «* 2 I a0 0
 140

of boot items, end a closer vowel

than elsewhere in EA In boat items*)

Examples from Khkerita Includet

a) /W » bloom, hoof* root, tooth

b) /«./ * fool« loose* ehoe

e) /B ~ V * broom* roof

d) /*» — V/ 1 nooa* soon, afternoon

e) /mi — ut/ t cool* do* food, moon, school

f) /*» — UI ~ U/ I shoot, boot, goose
d) /** ~ 0 ~ a/ i spoon

(li) Lowisan* a records show a position very

similar to that illustrated by

K&keritat /««/# /U/, /V Rnti /*»/

all occur. /«»/ is written as

(e*j in Mfk, —Inj in

Suffolk, /u*/ is generally written

t«.j# a clear indication of at least

a partial merger with boat items,

which are also written in this way

(see above),20 although /u»/ forms

are not especially common for this

lexical elass, and some items have

i.U • j which is not so frequent with

boat items. The merger appears to

be more frequent in Nfk than in

Suffolk, and more frequent in East

than in West Suffolk! /at/ forms are

most common in West Norfolk, where

 141

they may he the result of Midlands

as well as R.F, Influence * Other*

vise the tendency Is te /at/ la North

and East Nfk, /V/ In Suffolk and
21
South Nfk* Sows items, such as

two, hare /at/ throughout.

(ill) The SEP records show a similar

alternation between /at/, /u«/ and
/tr/ forms, although /at/ forms seen

to bo on the Increase* For example#

stool, which has only /at/ In

Kbkerltas# has only /u«/ In Suffolk

and South Nfk here (see Map 2)*

Map 2 shows the distribution of

forms for brqqi?, root and stoo£.

Generally speaking# /at/ forms oeeur

In North Nfk# /tj/ la South Nfk and

Suffolk# and /ut/ In West Nfk# but

sporadlo /ut/ forms (mostly In stool)

occur In all but two of the Nfk and

Suffolk localities*

Ths position with respect to the

boat < boot merger Is not elear.

Francis vrltss {.ou — oOj for Neh

/ut/ In boat* (occasionally (.utj)#

whereas boot Items# when they do not

have /at/ or /tr/# are written

 M ft P 2 ' fiOOT : 6M0M- STOOL

f,,„ s.g.j).

A;/ •'

/U:/ o

/II/ X
Fo dec^ome
 142

22
l.U*i ~ an* — O ~ tti j* There i«
thus no evidence of a complete

merger In Nfk, (Vo can expect

none from the Suffolk localities,

as hero the /ui/-/au/ distinction

appears to be lost - see above,)2**

5* (1) The absence of the palatal glide

L4j in tune, music, etc* and the

merger of who « Hugh etc* as /hat/
appears to be universal in the

areas studied by KOkerits, who

states* "The Staff, dialect has

levelled US o with HE euf the oobh

Eton sound being {.atJ", He points

out* In addition* that this is also

truo of Nfk and Essex,2**

(li) The same state of affaire emerges
from lowaaa't records for the whole

of Nfk and Suffolk* Be writes, for

example*

No, <1f*>
St,

s.w,
XI.
Ha, >!«>

Ho, "Zw Iu«

Bu, X«>
 Ih3

(The forme fron Hartleeham may

indicate a distinction*

(114) In Francis* rsoord* v« Find the

same piotaro ceoflpswd In respoases

#neh aai kaw i.nw J» haii ilk^n dt j»

4. We have already eeen above that tine ma, abort

£» as It affects refieaes oi >& 1« acre

prevalent in the rora1 dialects of FA, and '

eopooially in Suffolk, than in Kch* This

is also true, and to a larger degree, of

reflexes of ME $•

(i.) Kftkerits Hats, anonget ethers,'

the following items with /U/, and

indicates that the vowel in those

words Is identical with that in

anil. hood, ©to.t

beast,.boat, bene, choke*

cloak. clover. co&oh, const.

coat* d<m*t, folk, goat*

hole, home, hope, load. •

loaf, neat, nest* oak* oath,

oats. over, poa^cfa. nolo.

Post, rood. STOPS, amoks.

JSJfeSSlSt Jfeoad, vitsJLe, s^SMjC*

(ii) Lowtaan,s record.* ale© ©how a largo

number of exauplea ox t :-o short £, and

in fact the short vowel t i? ~ 'O* J
 \
144

ete* Scents to bo mora common than

tho long J, ;©• — <!>•]« Examples

like atone (.stonj make it clear
that the short g does not in all
cases* (usually in Nfk)* represent
a merger with the rowel of pull.

hood (which always hare (jpj) as is

the ease in Suffolk and present-day

Nch. In the records we find the

'following' examplest

/fl/ in all localities - froze.

posts, comb. bone, oats.

m>te* tea*
/u«/ in all localities «

clothes, pole, goal, road.

6321*
/U - u»/ - boat (in one local¬
ity shown as * quick" vs.
•slow"), stone, yolk,
poached, hotel, floats,
33311* S2Sli» 2£2lSS» 2Z9M*
oyer, (/»/ in Suffolk*
toad

/u«/ in Kfk).

(iil) Zn so far as it Is possible to estlm-

ate, the proportion of short g*s seems

to have decreased by the 1950*s. Zn

the published SED material, for in¬

stance* cotrib has short [oj in throe

Sfffc looalities and [Uj in one Nfk and

 1*5

two Suffolk localities* All the

others» however* have a long vowel*

except for one Suffolk locality

which has [aJ*2^ Xn the unpublished

material* yolk* toad, oak* loaf*

don't* all of which had the short

i in the pre-war records* have only

the lend vowel* The loud vowel also

occurs la toes and spoke* while

•tone* goad* &2j&t j&ES&S b°th

variants* Whole has only toj* Xn
his notes» frauds mentions short¬

ened forms at Bllcklinsr* Ludham,

Ashwellthorpe and Griraston* and

writes* under Pulhara (£, Sffk)s

"Evidence of shortened lax forms*

apparently touch acre prevalent in

the dialect 50 - 75 years a&o* was

rather plentiful in the speech of

**« the oldest informant| thus

t-IOd, otUn, kUa* apU'h, tyjUt]"

(* £2fi&* S*22S» SSE&t &BS&&* .&E®£&)«
"The prevalence ef £et»j in the speech
of younger persons seems te be a

result ef St* E« influence*"

5* The tendency to merge /Xd/ and /«a/ appears to

 1*6

Ixi prevalent throughout EA» but not to bo

00 fully- carried out* particularly in Suffolk*

ao in Rob* The merger would appear to be

undergoing a set-back in the South of the

region*

(i) K&keritas otatea* "while the diphthong

i,laj hardly ever oocure ae a sub-

stitttte for St*E« l«a j» the ueo of

tho latter phoneme on the other hand

10 extended to nearly all the words

pronounced with £l<?J in St*£«* whioh

consequently exhibit double pro¬

nunciations in the Suff* dialect.*2^

There ie thus a tendency to merge

these two classes in this area* with

a considerable amount of "phonemic

overlapping"«

(11) The following table ehovs some of

the evidence concerning this point

from bowman*0 recordst

hear here beard queer there chair where care

IS if less

No. 7e*a "0*4 *0*3 0*a Ae*a '1*3 e*a '#*3 0*3

St. SO *3 »•*<? '©*<9 'e*3 "e*a '•a '1*3 0*3 e*a

s*w. "«a *sa *e3 "s ♦a ea •e *a ?ea -®a Ae *a A03

11. "I'd "0*3 0*3 e*a ea Ae «<? Xd" 03' '©•3 >e*a

Ma* Xd id "13 '1*3 e«a A«a Atd •»3 Aea "1*3

Ho. 13 "I'd -e«3 -eJ'a e *a as «a 9*d Ae*a • *3 '1*3

flu. la "i*a "ia 'i*a "9*3 *a 'ea 'e *3 "e *a '1^3
 "S

»4 9 I9 I

»i
I

3
9 A S 5© 9

o <V

i
« &

# V<
£ »♦< 9

9 3T
£

32
>i!
a H% s !t I
\r< o
I

l I I i 1

W >_t 2 9 »• >T« 9 tf
>x< 9 « 5"

*W •
>tr 9 &r 9 tf m 9 *4

1H • 9 • >t< 9 tf >tt

0H 1
$ 1§|f« f#f 1 9* W 99

: fI •« w9J • ® u *** ! M<9 g 41

*

I I9 • I9 V !9 1 # « ♦

£ 9 e<
 the merger as being usual at

least in AahweUthorpe* ^
goto that the schwa eff*glide la

usual la the rural dialects| this

la met the case, except vlth elder

speakers* In Heh#

tb* position concerning tha above

forme la further complicated In the

rural dialect* by the phenomena

diaeuaaed under 6# and fm below#

The elder rural dlaleeta of Mfk wad Suffolk

(but not* It seems* of Essex) retain the

dlatlnetlen between MB a m [et ~ at a*a J

and ME ai as tml ~ el —alj*
(1) KOkerita states that the Suffolk

dialset* "as spoken by elderly

people* clearly distinguishes

bstveen such words as name (pro*

nouneed with i#»j) and nail (pro-
nouneed with i*Xj or teXj)which In
St*B# are pronounced alike"#^2
He else polnte out that this

distinction* under the Influence

of R#P# and "Cockney*, le dying

out* with younger people general*

lalng the !«! — clj or a eoaproalso/

R♦P.»iraitatIve Lei] to both group#
of items#3-5
 4 4 s

s S 1 *4 5 4

11
lli a'.
•

•
•

©©
S

4 i?
M
lli 111 lli
Ij
3
i

*
I
! ®4

£>
sn

«*• » ©H 1
tf

I S
4

3 1^"■#

% »4 I 3 »4 4 © « i E I

4 © ■| 3
1• a 1 —r v i ©i 3 !

si

0< » »4 3§ tf 2 2

s * H |" Y ? I© 3 4 li
 150

pny« tail, bhder Ludlion, Trends

writ00$ "ItC a - several different

variants, perhaps indicative of

change - tc — oj no longer than

half-long with las high off-glide -

forma with j,dJ oay show phonemic

shift with reflex of I IB ai» oi»

wMch is l©lj or l?Ij%

Ue con say, then# tlip.t this distinction

is generally on tho decline in rural

CA*"^1 The distinction can, it is

true# still bo hoard in ?Ich, but this

is generally in tho opcoc i of elderly

pooplo, in casual 1JC speech, or in

humorous or facetious conversation*

In those cases, the vowel of name ©to*

generally does not have the schwa off-

giicl© in Kch, so that fierce, faco

aro both {,f£ isj.

7# Thcr® are traces, quite substantial in tho pro-

war record®, of a distinction between reflexes

of ITD f as ill — ii J and hC © as |_oi

fteflexos of £ appear in sono cases to liav©

fallen in with reflexes of (1T, a, and in others

to have remained distinct,

(1) In tho localities studied by KOIcerits,

iteno of this typo have either tho

 151

if* J of ansae or the tit j of H,P,

ooe. It is also striklniT that

KMceritss records several instances

Of derirwJ froa MS $• TlntSi

4<wp. for example, has {.?» ~ i« J»

(ii) Lawman writes a number of refloat©*

of ME f with this m«l. Thus

Hootea has beans as |.b-o*ansaj.
What is striking hero is tliat this

Indicates a rowel that is Identical

with that of the classes here8 there»

and name* it many cases. Thus,

sfeoton, in addition to beans ^b-e-diiaj,

has bairns {.be* ana j, ear t Ae«a j» and
relations i rat *««a na 1. (This
apparently does not apply to the

areas studied by S&kerlta, since L91J

in deep. aa«e. without the schwa off-

•^lido, contrast with here and there

items with the off-glide ^Xa — ea ~ sa j. )

Lowraan has the L,e*£ j rowel in rrreaeo

ih four localities, inel-tding all three

STfk looalitiesi in beast in one NfSe

locality| in. wheat in both Nfte local¬

ities which gars this item and in two

cut of three Suffolk localities, in

one of which it Is labelled *older form"

 152

and occurs alongside {."Ijj» It

also occurs In brood in Stiffkey

and noat in Kocton.

(ill) Francis states that I2D f and $ have

fallen together in Aahuollthorpe# hut

that informants wore aware of older

forma with {^oiJcflD <| and occasionally

used tJiota# The older forts of the

dialect* he says# appears to have

had a tliroe-way contrast!

J IE a >[ei — £tl —• « t3 J
ME ai >j,ti —eij
MTI o >[©* ~ ©*I J

There is probably no trace of this

feature in Nch, except perhaps in

the tendency of sOae informants to

prorurmco St# Stephens with [oi — o«l J,

Sorae traces of post-vocalio /r/ are found in

all t'li'oe sets of rural dialect rocord3« It is

obviously very much a relic form# and tends to

occur only in items which have R»P, /3»/# On©

B£& informant # moreover# consistently had

after L3*i» It is not clear whether this# too#

is a forct of relic pronunciation# or a personal

idiosyncrasy#
(i) K»>.Grit!! writes i "Th® principal

modern Stiff# equivalents of itL

er# ir and ur , are ^ai a ~3*J#

and the circumstance that the com*

paratively rare t3»J ••• is almost

exclusively recorded as an alter¬

native pronunciation of la»j and

j warrants the conclusion that

t3*i is a late importation frou

St*E»" Jto shows that ta>J occurs

mostly a j a reflex of er, an J j^A j

mostly as a reflex of Ir# ur# al¬

though this is not consistent#

According to Kftherltz, ta,J occurs

witht certain# concern# earn# learn#

service, but also with* bird, burn#

nurse# workt {,aJ occurs wltn til rd#

church, buret# vorsr# turn etc.

Under J.aJ ?5?cerlta includes the

variants {.? ~ 3 aj# Tho Laj,
houev r# may equally well represent

a shortening of ta*i» fc'hieh is also

fairly frequent in items such as

partner.

(ii) Lov/rian* s records show n situation

that is considerably noro co »plex

than Kftkerita's fairly neat distrib¬

ution of K#P# /3*/ items over

 15*

I,a —3i —at J. For example. Low-

man writes It. P. /a/ items such aa

S2» worry, with some form of IaJ.

but also with some form of t3j or

t"2j| R.P. /at/ itemst on the other

hand* always have some fort-i of l»* J»

usually i.a. j itsslf( but occasionally

modified e.g. (, <a« ~ Ja. j. R.P.

/3»/ items» on the other hand* show

a very wide range of variation* The

following is a complete liatt

thirteen wflrtx first Thursday furthest hers

Ne. "P T? A3 ~ <A "B* 'a "£•

St. 3*- P P <A. B *

s.tr. ;a »• 4t ia#
11. >a* V* "p—sA S* a <£•
Ha. ,'Ar~'A* <a* <A <A* \a» 'a.

HO. u. «A <a >a.

Bu. S3 <a 3

vermin learnt church nurse sermon rmrpoa^

Be. .'a* ;a» 'S <a a* "S*
St. a* >a« a« xa* S*
s.w. ;a* ^a a <s »*• <S'
11. a. a* .g p« a*
TSr
Ha. »a« >at 5a "a* >a* ja»"

Ho. 'a. a* *
a **-33 'a.-33 o
33-*3
#

BU. va» ?a» <a» ~3a 'aa,~"a» "3r

This shows that some items * notably

sermon, learnt, vermin, together with

thirteen. Thursday, here in a few

 155

localities* probably have the ease

vowel aa barn, basket. ete« It alec

•hove that some items* notably furthest,

together with nurse, church, thirteen.

thirty in tome localities* probably

have the same vowel ae up. ratjg* eto*

for the aost part* however* there

occurs an intermediate vowel of the

type £.1? — ??• ~a» ~3 ~a ~>»J etc*

(The variant seems te be acre

prevalent in Hfle# the — aJ variante*

ae KSkeritt's evidence suggests* in

Suffolk*) This* (and other similar

phenomena discussed above)* posse many

problems for a taxonomie phonemic typo

of phonology, which* together with

terms like *phonemic overlapping**

"variance analysis"* "phonemic in¬

determinacy" and "compromise phoneme*«

will be discussed in later chapters•

(ill) Francis writes in his tffk field-notes»

[?• ~ «i —?tj
i/Pi — at J
iv
iy
L? —*• ~5J
(.£*
s *i

this reveals an equally large amount

 156

of variation* There appears to bo

some evidence, although loss than in

Lowaan*® records a of a merger of

these itetas with R*P* /«»/ itemst

thus, Francis also writes |.o» — 8.1 j
in, for instance, darning Xn spite

of this he states, in the phonetic

notes for Ashwellthorpe, that

{,3 —5—1 — 2? J occur, short end

toag| in theao items, and that these
are all apparently allophonee which

are distinct from [at *— ««|<MB nr.

Xn If oh there are fairly frequent traces

of this phenomenon, although nest,

especially younger speakers, have

/£•/, which tends to be rather open

£|iJ* Hot# that WC speaker# tend to

have /3«/ also in clerk. Derby.

10* The £ai] in cart* partner is, in the speech of

n few older informants, distinct frost the

vowel of half* mth. glass, dance, which Is

£«tj# There are traces of this in all throe

sots of rural records, but none or very little
in Efch.

11* Just as tfoh £s»j in here* hair is £««a ~ eiaj

ete* in rural areas and with older speakers,
 157

so older rural dialsott distinguish ruor*

pore / paw as iea ~ o<?J / Lot j« Sola gener¬

ally baa ioi] in all eases*

12* We have already seen that /b/ Is retained

throughout rural m» including Bsee** hut

that in Kch it is usually lost# Kftkeritz

states that* in the areas he investigated*

only the urban areas of Southwold and

Halesworth (lp63 populations 2*180 and 2,kk0

respectively) shewed any tendency to *h-

dropping**

13* There la a notable tendency for front vowels

to he raised la eons wordst e*g* head* net

with tZjl sat* entoh with iej* S&kerita

states that *the substitution of ixj for if]
is eaetreiaeiy comaon in the Suff* dialect**"*"*

It* JUS*# /lee-/ often corresponds to /k«/ or (.{£•]

in Sffk and Suffolk rural dialects anougat

older speakers« e»g. quarter ikfitaj.

13* R.P. ^r-/ and /#r-/ correspond to /sr~/ and

/tr*»/ with older speakers* o*g* Lowtaan has

three i.tU$X,j] in Heeton*

16* /!/ is usually "clear* in all positions»

 158

This corresponds to If oh

l.hli ~ 4* J ©to,

JL7. final <*©<1 io often /* t/« «*gi hundred

th^ndUalj*
PhansoLoyL ca.L Va-iTi a hies
Aftor this brief description of ©onto features of E*a

phonology# v« ©an now proceed to an exposition and dis-

ouasion of the twenty phonological variables to bo nood

la thio inveetigatioa of Eeh English* The majority of

segmental phonological elements in Hob English are in*

volved in variation of some social significance. the

variables involved in this study were initially selected

from among these en the grounds oft

(a) the amount of apparent social signif¬

icance in the pronunciation of the

segment or segments involved 1

(to) the amount of phonetic differentiation

involved!

and (o) the frequency of occurrence.

This selection was made (and different values allotted to

various phonotlo realisations * see below) on the basis eft

(i) native knowledge of the speech of the

area 1

(11) results obtained in a small pilot

survey conducted In December 1967t

and (ill) the record# made by Sbkerita# bowman
and Francis (see above).
 160

England, non—R.P.) typo of pronunciation. It was

therefore assumed that, in Nch, there would be a

correlation of the following type between the social

class of the speaker, the social context, and the

amount of glottalisation!

(a) Style FORMAL, class HIGH i glottalisation

LOW.

(b) Style INFORMAL, class LOW t glottalisation

HIGH.

In investigations concerning the realisations of (t),

four phonetically distinct types were distinguishedt

[t1*) (aspirated)
[t] (unaspirated)
[t2] (glottalised)
[2] (glottal stop)

On the basis of the assumptions concerning class and style

discussed above, these types were then allotted values as

follows t

(t)-l [th tj
(t)-2 [t2j
(t)-3 [2]

Using these values for the phonological variable (t),

each informant can be assigned a (t) index for each con*

textual style which is obtained by means of the questionn¬

aire. The index, which indicates the extent to which

each informant deviates from an idealised R.P. type of

pronunciation, is computed in the following way. Each

 162

logical material quantifiable» and which therefore

makes this hind of study possible* Much taor© reliable

and significant conclusions can be dram from the in«

formation that informant Z has a (t) index of 123 in

Vh style and an index of* say* 230 in <33 than from on

impressionistic statement to the effect that the in*

focoaat uses a higher outer of glottal stops in CS than

in v:L style*

Some {tOQfitio difficulties arise in oonneetlon with

this variable, in that it is not always possible to

determine whether* for example* a final unroloased

voloelesa stop is glottalised or not* Zt is difficult#

too# to know how to classify elective J* Doubtful

oases such as these are' omitted from the index colonis¬

ations*

The Variables (n\ and fie)

The variables (p) and (k) are concerned with the

pronunciation of /p/ and /k/ respectively in items such

as tape, taper> bake* baker. Indices for these two

variables euro calculated In the same way ae for (t)#

(as are# indeed* indices for all the variables)* The
values allotted to (p) and (k) correspond to those
allotted to (t)i
 163

(p)-l [ph pj (k)-l [kh k]

(p)-2 [p2] (k)-2 [k2J
(P)-3 [2] (k)-3 [2]

Note that [p2« t2» k2J Imply a pronunciation with

simultaneous oral and glottal closure, the oral closure

being released inaudibly before the glottal closure*

This pronunciation is not identical to that of pre-

glottalised stops which occur, for instance, in some

types of R*P«, where the glottal closure is released

before the oral closure*

Since the value scales for (p), (t) and (k) are, in

effect, Identical in their treatment of glottalisation,

these variables can be investigated with respect to their

co—variation with each other, as well as with socio¬

logical parameters*

The Variable (h)

This variable is the initial consonant of happy*

home* etc*, and is well—known as a linguistic variable

throughout England* In most parts of England, and partic¬

ularly in urban areas, (h) indices are likely to be in

direct and straightforward relation to the education and

social class of the speaker* In Nch, however, the pos¬

ition is likely to be more complex* This is because

 164

Hell is surrounded by what is, at least amongst older

speakers, an h-pronouncini* area even at the level of

rural dialects, as flap 4 shows. This saccosts that

older people in Hcli as well as ircraif-ranta froa the

surrounding rural areas are likely to have lower

(m tsore "H,P.-llko*) (h)•indices than younger people.

The value scale for (h) 1st

(h)-l [hj
(h)-2

An index score of (h) OOP therefor© indicates a con¬

sistent "h-ful* pronunciation# Note tliat weak fotvis of

iterxs like have, hln, which are normally j^avj, Lanj etc#
in west varlotloo of English are excluded froa the

calculations of index scores for (h)«

The Variable (nr:)

The variable (tag) is tho final consonant in tmlfcin/?#

runntn/r# etc#, which is a well-known variable in many

parts of tho Dnclish speaking world • T ho usual non—

P#P# pronunciation of tho -in- suffix in Nch, and

throughout IA, is (.an^nj rather than IXnj* The value

scale isi

(ns)-l J^j
(nfj)-a {.nj
 165

The variable is therefore concerned with the proportion

of [nj to Itj J found in this suffix in a ©iron body of

speech material*

The Variable (r)

The variable (r) is concerned with the phenomenon

known a a the "intrusive and the index measures the

proportion of intrusive £*s that ooour to the possible

number of such r»s* By "intrusive £* is meant a pro*

nunoiation with /r/ whore this is not historically

"Justified*, i.e. whore the £ is not descended from

Early Modern English £« and dees not ooour in the

spelling* intrusive g*s occur only where /«»» o«» 3t«

, sa § tfa » a / or their Noh equivalents ooour immediate*

ly before another vowel Examples from the If eh

linguistic interviews includei idea of ^ide navj,

law of tiauavj# lot of old |.»ltt*la^»u»jt out to eat

|&«2td>jel2j* The value scale 1st

(r)-l |>J
(r)-a LUJ

in the above specified contexts*

The remaining fourteen phonological variables are

concerned with the pronunciation of vowels* They aret

 166

(&)» (£)# (a«)» (*«)» («)t («r), («r), (I), (4r), (o),
(on), (o), (S)» (yu), The method of hsndllng those

variables will necessarily be somewhat more complex

than that used In dealing; with the variables already

described. In the case of fcho consonantal variables

we are dealing: simply with the presence or absenoe of a

particular consonant, or with variant pronunciations

that are auditorily quite distinct. The index value

scales, that is to say, are well motivated. In the case

cf the vocalic variables, on the other hand,there are no,

or very few, auditorily distinct variants. The range

of pronunciation of any given vocalic variable is likely,

cn the contrary, to take the form of a certain undiffer-

entiatod area within the vowel trapeslum, This means

that each variable has a eerie; of infinitely graded

realisations that can fall at any point along the

appropriate phonetic continuum, which can only be divided

up for the purposes of the index value seale in a manner

that is rather arbitrary. It also means that there will

be, as it were, border-line pronunciations, to which the

transcriber will have difficulty in allotting a partic¬

ular value*

The solution to this problem is to divide up the

continuum using Cardinal Towels and other points of refer¬

ence as a guide, and to base the number of values for each

soale cn the amount of phonetic differentiation involved

and the number of different types that the transcriber

 167

can perceive without difficulty. Each instance of a

variable thoa haa to bo allotted a particular value,

according to which idealised pronunciat±ort-1ype it

approaches most closely* Xt is not important, for the

purposes of this study, that the continuum is divided up

in a relatively arbitrary fashion, since the <11vision

is the same for all informants, which is all wo require

to make the results strictly comparable* (Xt may, how¬

ever, obscure details that might be important for other

purposes*) It is also, for reasons of comparability,

important that the transcriber be consistent in his

allocation of border-line oases to particular values*

Xn the present study it is felt with some confidence

that this consistency has been achieved* Moreover, any

mistakes that do occur are likely to be insignificant

in view of the large amount of material involved. He

can illustrate the above points by taking as an example

the variable (a).

(a)

The variable (a) is the vowel in *>M. oaa* matter

etc* In Lowman*s records wo find th@ following phonetic

types in items of this klndt {_• ~ > &> — >m — -a — ~

Mr
~
© ~ c — »• —- *ed j, Francis* transcriptions are

similar, and include the variant tli* The different Nch

variants are mainly differentiated by length, height *

 163

and dlphthomrlsatlon. of which vowel height la socially

t)M «o«t diitinetivei We can distlagtjiih In Heh English

r *
ths following pronunciation typesi [• -a ~ | ~ a - ^
» ~ ~ ««e ~ «« -«» ~ i ~»i ~ »ia — • i*j ate* The
number of possible symbols here, however, dees not re¬

flect the number of discrete pronunciation types that It

does, for example, in the case of (t)« It merely reflects

the limitations of the transcription system in represent-

kl
ins an Infinitely varied number of pronunciations *

According to the principles described above, there¬

fore, we in fact set up the index value scale consisting

of the following idealised phonetic types or approximate

vowel qualitiesi

<a)-l fJ
<a)-2 |»ij
(a)-3 L*** i
(a)~& [si]
(a)-5 l*«®j 42

under (a)-l |.mj we subsuas all pronunciations of tho typo

J^m ~ m — m ~ & ~««J, but not oxamplso of If] oto*, and

so on* Thers are, of course, some cases where it is

doubtful whether a particular instance of a vowel should

be considered long or short, or whether it more closely

resembles IfJ or t«J. This is inevitable, since the

phonetic symbols can only be approximate, and the

transcriber is fallible. This, however, is by no means

 169

an Insuperable problem* and the results are felt in all

eases to be as acourate and, more important, as consistent

as possible*

In the above value scale, (a)-l represents the

typical abort laej sound of most varieties of R.P.i
(a)*A represents a sound, typical of many varieties of
Nch English, which, although Identical or similar in

quality, is longer, allowing for phonetio context, than

the R.P. vowel, Variants (a)~3 to (a)-5 indicate

kk
closer and more diphthongised varieties of this vowel*

Index scores, which can in this ease range from 000 to

400, are calculated in the same way as the consonantal

index scores*

The Variable (a)

The variable (a) is the vowel sound in items such as

name, nail* day, acre. In his records, bowman writes

items of this kind, where they do not have the raonoph*

thongal ieij-»type pronunciation, with! t ?tl' '»*IA ~

KmM* — >»S' j etc* Francis has similar transcript*

ions* It can be seen that there is variation in the

quality ef both elements of the diphthong, as well as in

the quantity of the first element* The value scale for

Nch reflects these facts* The different types aret

(S)-l iml ~ «lj

(a)-2 iel~*lj
 170

(a)-3 |.«eij
(a)-4 l««t—«?tl ~ w»lj

To permit invesfixation of the extent of the survival

of the name / nail distinction described above, we also

add i

(a)-5 L$» ~ «» — ««3 J

to indicate monophthongs1 reflexes of MB a. This type

does not enter into (a) index calculations unless other*

wise stated, (a)*l is intended to indicate an li,P, typo

of pronunciation, (a)*2 a very common MO Kch pronunciation

with a more open first element, (S)-*t the most "extreme"

Nch pronunciation with a more open first element and a

closer second element, and (a)*3 a pronunciation inter*

mediate between these two.

The Variable (at)

The variable (at) is concerned wit;< the quality of

the vowel in after, cart, path, which varies in l?ch from

(at)-l a R,P.-type [a» j, through (ai)*2 (at at j to

(at )*3, a very front [at —- at J. cf • Lowtaan's j, <a> J.
The variant (at)*4 was need to record those items, such

as Derby, olerte. which have /3l/ rather than /at/. It is

not included in calculations for this variable in the

present work.
 171

The Variable fan)

The variable (au) Is the vovel in out, cloud,

louder. This vovel may vary somewhat according to

phonetic environment. This factor is not taken into

account in index calculations• since the large amount

of material obtained means that the possibility of a

blaa being introduced into the results is very small,

Kbkerlta» Lowtaaa and Francis vary quite considerably in

their transcriptions of this vowel, KBkeritz has [euj

with occasional variants jtm&j and [adj. Lowtaan writes

[ a(u< ~ du — J etc, most frequently except in 8ux-

hall and Honington, where [, 'iu^j etc, is most frequent ,

but both typos occur throughout the area. Other vari¬

ants are [ <a»U —ttJ ~ eIT — rThis vowel would

therefore appear to vary widely both within EA and

within each locality, Francis writes - tu - su

itt at?J etc, in the SEP Nfk records, and Ellis writes
[eO ~ eO« J etc, in the Suffolk records. The many differ¬
ent types of Nch pronunciation would appear to represent

various degrees of compromise between these and n,P,

forms* The value scale for (au) 1st

(au)-l {.au — aoj

(au)—2 i,a¥ ~ "8¥ ~ «eU ~ fetf j
(au)*3 [an ~ -eu ~ s« ~&uJ
(au)-h [km ~ tu ~ 3uj

(au)-2 and (au)-3 are difforentiated by the height of the

second element of the diphthong* and (au)-3 and (au)-4

 172

by the height of the first element*

The Variable («)

The variable (e) la the vowel In tell* bell, well.

healthy* where /e/ occurs in the content «• /1/| ami
better, metal* where /e/ occurs In the contexti

I bilabial Cj [2 J [fj i*e» in stressed penultimate

syllables alter a bilabial consonant end before /t/ where

/t/ is (2j «<► (not before (tj or L*2j)« Th« ralue scalo

is ae followst

(e)-l L© jj
(o )~2 1.5 *]
<s}-3 j& * XJ

The centralisation of /e/ nay also occur in some other

contexts* such as ■—>"»»■ * but the details are not

clear* Kokeritse observed a similar tendency in Suffolk

in the 1920*s and 1930*s» where it appears to have taken

the form of opening to {.e J or L»j# and to have occurred

In a wider range of environments« Re also observed

better* for instance* with (.aj* Lowmn* too* writes

twelve (tw aIvJ in Hocton* S* walaham, llketshall and
ftonington* Francis writes (ej in this item* In If oh*
the extreme centralisation to tAJ etc* would seem* ae a

widespread feature, to be a recent development (see

Chapter Eleven)*
 173

The Variables (&r) and (tr)

The variables (er) and (tr) are concerned with the

nature of the vowel in iteraa of the classes here and

hair respectively. The index value scales aret

(er)-l (er)-l [Xa]

(er)»2a (cr)-2a [ei ~ eij
(er)-2b (tr)-2b [ea
(er)-3a (cr)-3& L« s — «»J
(er)-3b («r)-3b [sa — £<9 j

The common scale means that the amount of differentlation,

if any, between the two lexieal classes can be measured,

and that oases of hypercorrect ion can be noted. By

means of the sub-division 2a»2b, 3a-3b, calculations can

be made concerning? the schwa off-slide. One ©an in¬

vestigate, for example, its frequency of occurrence, and

study whether or not it is used to differentiate the two

aets of items.

The Variable (f\

The variable (I) is the diphthong in ride, rl^ht,

rhyme, riper* Comments under (an) concerning vowel

quality are also applicable to this variable, X&keritz

has this diphthong as [al ~ z>I ~ aI j with [dZ ~ 31 j as

variants, bowman writes [Xi°X — yv$ —'a*1"J etc,, and

Francis ~ a! aI j, The index value scale 1st

 17k

<I)-1 Lai ~ &IJ

<I)-2
(I)-3 t«i
(J)—4 L oi AiJ

(I)«2, 3 and 4 are differentiated in the sauno way as the

corresponding values of (an)* In this way co-variation,

if any, between these two variables can be investigated*

The Variable fir)

This variable is the vowel in Items such as bird,

farther* fern* i*e* reflexes of MB er« ur and lr, as

discussed above* The scale ia as followst

(ir)-l L3iJ
(ir)-2 U»i
(ir)-3 L»t ~ A]

The Variable (o|

The variable (o) deals with the degree of rounding

or unrounding of the vowel in ton* cod* box* The un¬

rounding of thie vowel appears to be a feature only of

Nfk speech within EA. Lowiaan writes Lla ~ <5 ] etc*, and
Francis La* ~ ~ The value scale for Heh 1st

(o)-l L®J
(o)-2 La ~ ~ ®J
 175

In addition to these two types, wo have (o)-3 lot J*

which i« not, in this work, incorporated in results

developed for (o). (o)»3 accounts for those items such

as off, frost. cloth. which can have /»»/ rather than

/»/•

The Variables (on), (o) and (5)

These three variables are the vowels in know, boat

and boot items respectively, as discussed above. The

common index scale for all three variables means that

the amount and nature of the phonetic differentiation

of these three lexical sets can be investigated, and

hypercorrect pronunciations closely studied. Note that

(u)~4 is not equivalent to (ou)»*» or (o)-4, but that

(u)-5 is identical with (ou)-4 and (o)-»4« The index
value scales arei

(ou)*l <o)~l (S)-l tAtt —

ou — »u j
(ott}*2 <S)-2 (u)-2 ~

•uj
(ou)-3 (S)-3 (u)-3 i«« —
ui ~
Vuj
(S)4 (u)-5 LtT -~yj
j.*» ~
Vm ~3*j

(ou)-l, (o)*l, (u)-l represent lAn /au/| (ou)*2, (o)»2,

(u)->2 represent R, P. /on/I (ou)~3, (o)-3, (u)-3 represent
EAn /ui/f (u)-4 represents EAn /*»/§ and (ou)»kt (o)-k,
(u)-5 represent /U/«
 176

The Variable (yn)

The variable (yu) ia the vowel in Hugh, tune, rmale,

ami ia concerned with the presenoe or absence of the

glide tJj» The scale ist

(yu)-l (Jut — J*«j

(yu)~2 ~3*J

Note that for all variables, the lowest score 000

represents a consistent idealised R.P, pronunciation)

highest scores, ranging from 100 to 400, represent

consistent use of the raost extreme form of Nch pro»

nunciatioru
 177

Index Value Scales of the Phonological Variables

Value 1 2 3 4

Variable

t
(a) » let et e t

<S) el el el etl

(at) fit it at 3*

(ao) uW ew 3*

(«) e ft A

(er) Id e»(a) «*{a>

(sr) la et(a) e
»(a)
(h) h «*

(I) al el el dl

(1*) 3t gt at

M k KJi 2

(nff) a

(o) 9 a at

(ou) All eu u» U

(S) Att eu Ul 0

(p) P a

(*■) •»
4

(t) t 11 I

(S) Att eu Ul at

(yu) jut
 X78

Chapter Eight* The Co-Variation of the Phonological

Variables with Sociological Paramotera

¥• have stated that one of the chief aims of this

work is to investigate the co-variation of phonological

and sociological variables* In previous chapters wo

have developed methods for measuring both types of vari¬

able» and this investigation can now* therefore* be

carried outi a study can be made of the exact nature of

the eorrelation between phonological and sociological

phenomena*

In order to make this kind of measurement of

correlation* a record was first taken ef each occurrence

of all the variables in the four contextual styles for

each informant* Index scores for each informant in each

style could then be developed* and* subsequently* the

mean index score for each social group calculated. By

means of theas scores we shall be ablet (i) te invest¬

igate the nature of the correlation between realisations

of phonological variables and social class* social con¬

text* and sext (ll) to discover which variables are

subject to social class differentiation and which to

stylistic variation| and (ill) to find out which vari¬

ables are most Important in signalling the social con¬

text of some linguistic interaction, or the social class

of a speaker*

The methods we are using of calculating and portray-

 179

lag Individual and group phonological Indices were

initially developed by Labov. * In some respect#, how*

ever* the present work represents a development of

Labov'e techniques, in that use is mad# of phonological

Indices for investigating problems of phoneaie contrast*

and for studying aspects of what ia usually termed

Nphonologieal space** (For reasons that will be given

later the torts "phonetic space* will be preferred in

this work.) Sons variables are also investigated in this

work which do not produce evidence of social and stylistic

differentiation that is so clear-cui or conclusive as

that discussed by Labov. This increase in complexity

means that it is possible to gain some additional in¬

sight into the Unguistie behaviour of different social

groups, and to learn mere about the mechanisms of styl¬

istic variation and linguistic change.

We can. begin by taking as an example the phono¬

logical variable (ng), the pronunciation of the suffix

-ing» This is well-known as a variable in many different

2
types of English, and seems likely to provide a good

example of eoelal class and stylistic differentiation.

We can tell, by means of information developed for this

variable, whether the Wch questionnaire ha# or has net

been successful in eliciting different contextual styles,

and whether the social class index has or has net been

successful in distinguishing between groups of informants

who have significantly different phonological character¬

istics.
 130

The Variable fit?)

Table X shove the average (ng) Index scores for

the fire social classes established in Chapter Six* in

each of the four contextual stylosi tford List Stylo

(WLS)t Reading Passage Stylo (HPS), Formal Speech (FS)*

Table X

(ng) Index Scores by Class and style

Style* WL3 RP8 FS CS

Class

I MMC 000 OOO 003 031

XX LMC 000 010 013 042
XXI WG 005 015 074 087
xt mo 023 044 033 095
V Lb'G 029 0 66 093 100

and Casual Speech (CS), Table X demonstrates thati

(!) the Jfeh questionnaire has in fact been

successful in eliciting four hierarchic**

ally ordered and discrete contextual styles*

since* for each class* the scores rise

consistently frora 'LS to CS|

(ii) the social class index has provided a success-

ful basis for the establishment of discrete

 181

social classes as these classes are re¬

flected In their linguistic behaviour,

since, for each style, the scores rise

consistently from MMC to L¥C)

(iii) the method, of calculating index scores

for phonological variables is a success¬

ful one and is likely to be useful in the

study of Nch English;

and (iv) the phonological variable (ng) is involved

in a considerable amount of social class

and contextual variation, with scores rang¬

ing over the whole scale from 000 to 100.

The information given in Table I is more clearly

portrayed in the diagram Fig. 1. Index scores, from 000

representing consistent use of [n], to 100 representing

consistent use of are plotted along the ordinate.

The four contextual styles, from WLS, the most formal,

to CS, the most informal, are shown along the abscissa.

The lines on the graph connect scores obtained by each

of the five social classes in the four contextual styles.

The stylistic variation of this variable is por¬

trayed in the consistent downward slope of the lines from

right to left across the graph, representing an increase

in endings as we move from everyday speech to more

 182

formal styles* The variable (ng), it can be seen# i*

a very good indicator of social context* with scores

ranging;* as vt have already noted, from OOP (HMO and

LMC in VLS| MHO in RPS) to 100 (LWC in CS), Note

that stylistic variation is greatest in the case of the

i

WO* whoa a range is frost 005 to 087* and whose line on

the graph consequently has the steepest gradient* We

have already discussed (see Chapter Six) the greater

awareness the WO and ZJIQ have of the social signlfio-

aneo of linguistic variables* because of the border-

line* nature of their social class position* The

linguistic insecurity revealed here in the large amount

of WO stylistic variation for (ng) is clearly part of

the same tendency*

Tho social class differentiation of (ng) is, of

course, shown on the graph by the clear separation of

the lines connecting the scores for each class, and by

tho hierarchical ordering of these lines* LWC • KMC*

The amount of differentiation can be gauged from the

spatial separation of the lines on the graph* Thus

tho greatest amount of differentiation occurs in FS*

where the two HO groups appear to have the ability to

control (ng) forms to a level nearer that of the more

formal styles* whereas the three vc groups have scores

which more closely approach their CS level* Vote that

in CS, which we can as suite to be reasonably represent¬

ative of normal, everyday speech in familiar social

 183

ea7lronncait« f the three t*G croups shoo only- a smll

amount of differentiation one from the ether« 08? <•

100. This Is also true of the two WO ©roups, 031 -

0h2m There Is, on the other hand, a my significant

difference between the (ng) level of the WO as a whole

and that ef the MO* This underlines onoe again the

importance of this particular social division in the

eoeial structure*

We have shown, then, that the proportion of ^nj

to (/r)j suffixes that occurs in speech is a function of
the social class of the speaker and of the social Con*

text in which he is speaking* Moreover, although (ng)

quite elearly differentiates between all five social

groups, it is most important in distinguishing MO from

WO speakers* WO speakers have the greatest amount of

stylistic variation, and MMC speakers the smallsst,

although it is instructive to note that even this class

uses an average ef 31 per cent forms with |,nj la OS*

Sex Differentiation of (wr)

Fischer, in his **tudy of this variable

1
in an American locality, found that males

used a higher percentage of j,nj forms than

females* Generally speaking, this is also

the case in Noh, as Table XI shows, Xa

 njj. Viin&Kie. ^3) Ci'-Vis euvM,

l4nabte (p) hij Class «U«A. ^^U..

ws m H

St^ie

PW3. \Z$-m (oj^ by CWs& cum^ S'bjlo.

Sec
L.W f

145

S
J4 csoj
Gj&

0 Q QjftXttMA&VZ**!
VOS B cs
 184

Table II

(nc) Indices by Class, Style and Sex

Style i VL8 Hps rs £gi

Class

MMC M 000 000 004 031

r 000 000 000 000

LMC M 000 020 027 017

F 000 000 003 067

IJU'C M 000 018 031 095

F Oil 013 068 077

mc J! 024 043 091 097

F 020 046 031 033

LVC W 066 100 100 100

F 017 054 097 100

seventeen cases out of twenty, male scores

ar© creator then or equal to corresponding

female scores, "Je can therefor© say that a

hi£h (nc) index is typical of ml© speaker®

a® well as of \.*C speakers*

This link between the linguistic char-*

notorieties of KC speakers and nalo speakers

is a common one. Almost all the Nch vari¬

ables have tho sane kind of pattern as that

shown in Tablo II, with t/oacn having lower

 185

index sooroa than acnt This is a fact which is

not, on the «nco of it, particularly surprising,

but one that is at the sane time in need of

some explanation. There would appear to bo

two interconnected explanatory factors*

1» Women in o»»r society are nor® status-

conscious than son, generally speaking,

and or© therefore more aware of the

social si.Ttilficnnce of linguistic

variables, There are probably ttiree

main reasons for this*

(i) Th® social position of woncn

in our society is less secure

than that of men, and, fjenor-

ally speaking, subordinate to

that of men. It is therefore

more necessary for women to

secure and sljjn&l their social

status linguistically ami in

other ways, an.< tucy or© there¬

for© more m/ar< of tlx© import¬

ance of this typo of signal.

(il) It is very often th® mother of

a family who transmits social

characteristic^ of the status-

 186

signalling typo to children,

both consciously and unconscious*

ly* For this reason* mothers arc

more aware of tho social aignif-

icanoo of these characteristics,

(iii) Hon in our society can be rated

socially by thoir occupation*

their earning power* and perhaps

by their other abilitiesi in

other words, by what they do.

For the most part, however* this

is not possible for women, who

have generally to bo rated on

how they apponr. Since they

cannot be rated socially by

their occux»ation* by what other

peox>ie know about y!wt they do

in life* other signals of status*

including speech* are corres¬

pondingly isoro important. This

last point is perhaps the most

important,

2* The second, related, factor is that WC speech,

like many other aspects of 1 C culture* lias*

in our society, connotations of masculinity*

since it la associated with the roughness

 137

and toughness supposedly characteristic

of fcfC life, which are, to a certain

extent* considered to ho desirable

masculine attributes* They are not,

en the other hand, considered to be

desirable feminine characteristics*

On the contrary* refinement ana soph¬

istication aro rmch preferred.

This discussion is of co-.rjs necessarily at a

rather simple level, but it is clear that wo have

reflected in these .-honological indices part of the

value system of our culture as a whole* Froti the

point of view of linguistic theory, this means that,

as far as linguistic cliange "from below* is con¬

cerned, wo can expect men to be in the vanguard*

Changes "fron above*, on the other hand, are more

L
likely to be led by women* The type of sex

differentiation shown in Table IX is, in any case,

usual. Only a reversal of this pattern, or a

large increase in the normal typo of male / female

differentiation can be ooiwidorod to be signific¬

antly unusual in any way*

Tho Variable (o )

Indices for this variable measure the degree of lip—

 183

rounding In the pronunciation of the vowel In the

lexical eet ton, fog, lorry, etc* Evidence was given

in Chapter Seven to suggest that the vowel typical of

Nfk rural accents Is an unrounded In Kelt

we would therefore expect there to be class and style

variation of (o) ranging frosa Hfk j,«J to ft,P. [®j»

Fig. 2 shows that this is in fact the case# The pattern

illustrated in this graph is very similar to the one

shown in Fig# 1. As in the case of (ng)* there is a

large amount of class differentiation* with the biggest

division again being that between the MO and the ¥0.

The UtfC and MS»'C have very similar scores in all styles*

This indicates that unrounded vowels are a predominantly

¥C feature* but that the IMO is distinguished from the

ffltG and WC by a particularly high rate of unrounding#

It is a characteristic of the LWC, as we shall have

occasion to note again* that it is relatively isolated

from normative and innovating tendencies# Zt is there**

fore not surprising that this variable appears to be

currently involved in linguistic change* with rounded

vowels increasing. This topic will be investigated

further in Chapter Eleven#

Stylistic variation is also very evident in the

case ef this variable. This can be seen once again in

the fall in scores from right to left aoross the graph.

This fall is quite consistent* except in the case of the

*•

LMC* which has a small rise from 006 to 008 between FS

 139

and xtPS* The LiiC also hao alias-post fall la scores

between C3 and FS, fmt% 03^ to 007* (Fg liavo already

noted that iho HC has tlio ability, or is sufficiently

strongly motivated, to control FS towards UPS level,

whereas tho FC has the ^raatooi fall in ©cares between

FS end HPS. ) nsic op©o&ers bnv© only a very amll

number of unrounded vowels, and their speech must bo

consldorc-d to bo ohoractorisud by noro or leos consistent

use of Foot TTcIi arjoohero, how over, can be said to

Increase the amount and/or frequency of ltp~roundlag of

this vowel as the foraallty of style Increases.

It la Interesting to compare the sex differ-*

©otiatlon of (o) with that of (na)* Ficuro© for

(o) are etven in Tabl® HI* The most eirlfelng

fact ©Iiown horo la that, although ?!C non oaco

acaln have higher scores thr-si women, for tho

tsC this pattern la moro or loss completely

reversed. In ton coses out of tvolvo, wonen have

higher scores than men. This is tho only case

of such a ooriploto reversal of tho pattern of

sex differentiation amon* all tho Flch variables.

This reversal of tho pattern cannot b©

explained as the result of a current "linguistic

 190

©hang® from above", with wnen In the vanguard,

since changes of this type generally- involve the

socially downward diffusion of prestige features.

Table III

{© ) Indie oa bv Classs. Stylo and Sex

Styloi m,s
•mhmm*
PT>S FS CS

Jlass

MMC M 000 000 001 003

F Oi}ij ooo 000 000

LMC H 00k 01k Oil 055

F OOO 002 001 003

uwc M Oil 019 044 060

F 023 027 068 077

mo H 029 026 064 073

F 023 043 071 066

IMC M 01k 050 080 069

F 037 062 O83 090

There are clearly no overt pressures which would

load to an Increase in the uoo of this particular

non-prostig® form, and the HC women have in arty-

case lower index scores than men. The reversal

must therei ore bo re ;ardo? as the teault of a

"change from below", with i.'C men, typically, in

 191

the vanguard* This ©Siange must he ascribed

either to sous kind of internal ayttmtie

pressure* for which there is no evidence, or

to the diffusion of linguistic influences from

i,»J-pronounoing areas such as Suffolk, and

more especially Ipswich* (Lovma records
for bo/? in the three Fffk localities,

and LbogJ in the fotur Suffolk localities*

The symbol (d j tpproximt«« to the IPA alphas-
bet t«J#) If tiiis is the case, we have here
an interesting exactpl© of a change *froo»

below* due to external Influence led by WO

men, and a possible change "fro® above* due

to R*F* influence led by t*C women, working

in the same direction* In any case it is

clear that a high proportion of unrounded

vowels in top. J&g, etc., is moat typical of

kC women in Heh« (See further discussion in
Chapter Eleven*)

lOlSj&S&M&Jl&il

Indices for this variable measure the vowel quality

in items of the lexical set cart* path* etc* Scores

range from 000 for consistent use of (.«»j to 200 for

consistent use of {.atJ* Fig* 3 shows the class and style
differentiation of (at) in the flfeh sample*
 192

The ¥C and (at)

Xt la immediately apparent from Fig, 3

that the two lowest «C groups, the LWC and

!S./C# us© a very high number of very front

vowels in everyday speech - almost 100 per

cent* Tho Tjtjo also has a CH speech index

which Is only slightly lower* For the as

a whole* moreover, thora Is only a small

amount of stylistic variation of (ai),

©specially in the case of t >e L«'C* We can

interpret this as indicat J. tg that, although

vc speech is characterisod by tx pronunciation

of this variable that la slgniflonntly differ-

ent from that of tho HC, little attention is

directed towards this difference within the

Nch speech community* This interpretation is

supported by the a >t that (at) was not the

subject of overt comment or criticism in any

of tho discussions of Tick English recorded

during the interviews.

It is interesting to opeculato as to why

90Q0 variables, although subject to marked

class differentiation, are subject to little

or no stylistic variation, while others are in¬

volved in a great deal of this type of variation#

(It is of course true that the o is a email amount

 123

of regular stylistic variation of (a*)§ but

the relative Insignificance of this can bo

gauged from a comparison of Tic* 3 with Figs*

1 and 2#) We Initially considered (see

Chapter Five) that stylistic variation was a

function of class differentiation! members

of lower class groups attempt to reproduce

«

forms more characteristic of higher social

groups in more formal contexts* How# then*

is it that for sorao variables speakers are

relatively unaware of the social implications

of the forma they use?

There would appear to bo two explanationsI

1» Speakers are more aware of the social

significance of forms which are

overtly stigmatised* (i) forms which

are the object of humour or ridicule»

such as (h)-2t and (li) forma which

are actively discouraged by the educ¬

ational institutions* such as (t)-3*

2• Speakers ere more aware of tTie social

significance of forme which are

currently involved in linguistic change*

or, perhaps, for which an already com¬

pleted linguistic change is still re-

 X9k

fleeted la differential age-group

behaviour. The fact that conflicting

forme eon be heard vl ;hin the same

social grop mean® that attention la

directed to these differences*

Forme which do not come into one of these

two categories will be much less likely to be in¬

volved in stylistic variation. Tho variable (ng)

would appear to come into the first category*

the variable (o) into the second. Tho variable

(at)* on the ofchor hand, la subject neither to

overt ©tignatlsation nor to linguistic change in

Noh. We could speculate further and at some

length as to why some variables are more liable

to overt criticism than others. On© of the princ¬

ipal reasons, however, would see-•< to be ortho¬

graphical! (t)-3* <h)-2 and (ng)—2 can be

characterised as *dropping your jfe*s, h*s and «'»*

respectively.

The HO and (at)

fIC scores for (a») arc considerably lower

than VC scores, an J, ones again, the UG/'AC

division BfeOiis fro 3 tho graph to bo the most ira-
 193

portaat• There la* however* a significant

difference between LMC and MMC scores in ail

styles except CS. It is also Interesting to

note that there is stylistic variation of some

significance in the MHC scores* Theso speakers

are aware of the social significance of (ai)*

This ©an ho ascribed to their greater famil¬

iarity with U.P. forms*

On the basis of those scores we can

state that the typical «C pronunciation of

(at) is the UiG pronunciation {.&« j and

the SMC pronunciation [^*1*

The Variable fir)

This variable is the vowel of tolr U hurt* fern* etc*

¥# saw in Chapter Seven that the typical rural EAn pro*

nunoiatlon of (ir) was markedly different from the K*P.

pronunelationt forms with ~ ai ^ a] were usual*

Pig* h shows that this type of pronunciation is now very

much of a rello form in Nch* slnoe the highest score of

all is 083 out of a possible high of 300* Non-U*P.

forms* moreover* are more or less entirely confined to

the two lowest social class groups* The two KG groups

have 00Q throughout, and the WO has a high of only 001*

Note too that even the IMC uses only 10 per cent non-ft*P«

forms in WL8,
 196

Wo can conclude, then, that (lr)»2 and (ir)*3

aro heavily stigmatised In Meh, and are In the process

of dying out in favour of f?»P* (ir)-»l# Only the LWC

uses more than 23 per cent non-JUP. forms in any stylo,

an indication of the social class barriers to tho

diffusion of normative and innovating influences as

they affect this relatively isolated croup, (The lack

of stylistic variation between C9 and FS is quite

typical of tho LVC, of. Figs* 1, 2 and 3») Ve shall

see in Chapter Eleven that only speakers aged forty or

over have any trace of noiv*R«P. forms, but that these

predominate among the over*slxties.

The Variable (aul

Index scores for this variable measure the vowel

quality of the two elements of the diphthong in out.

down, etc# Scores for the different social classes in

tho different contextual styles are shown in Fig* 5*

Class differentiation is quite marked, with the largest

distinction again being between the vfC as a whole and

tho fiC. Of the three wc groups, the t/fefC and MMG exhibit

similar characteristics, and are both quite widely

differentiated from the IMG» The norm for the LWC is

the apparently ©ld«*fashloned oloso or centralised

i.cu ^ 3*J also found in the rural areas of EA. The

other VC groups have a norm intermediate between this

 £%4, (ir) bij a*A#l SrUj'l\
20&

lbs

,2
H _

0 %ti
L*&>

640

jtii&vtx--*-?*'
COO
Wl£ ^ LM&'MKC. Pi

Rg.C. Vhn&btfe (au) fe4, Cto& tMAriL

F$
SfojU>

Ptij.fe. bKt (&.u^ bvj $(!->'• CW$b ^fACs

240

§ 2eo LMO MMJE

am—wga*

2
H /&& L.M.C.

l£a

otas-
V&& r-fs
 197

(®.u)»k and (au)-3* Tho LMC is clinracterised in CS by

th© torus of (au)-»3, while the fillG has (au)«2 (.ae^J*

It is significant that no group consistently uses the

forms (.airaUJ typical of d* P*

Stylistic tTartntion. and (an)

A striking feature of this variable is

that there Is little or no evidence of styl«*

istic variation# except in the case of the

LMC, which his a sharp drop in scores between

CS and FS. This variable in fact confirms

the discovery made in connection with (at)

that there is no nocosiary connection between

class differentiation and style differentiation.

Some variables are characterised by class

differentiation only* We also hypothesised

in connection with (at) that variables which

are subject to stylistic variation are either

overtly stigmatised or involved in linguistic

change* It is certainly tie cose th.- t no

overt stigmatisation is directed towards the

pronunciation of (au)* There does# on the

other hand, appear to foe hoio form of linguist*-

i© chancre associated with this variable* This

Change# however, most probably comes into th©

same category as two char., o at i octing the

 19B

third-person singular verb marker (see

Chapter Six)# It is# in other words# a change

that has already been carried t'-rouph# but

whose effects con still bo observed within

tho speech community* tT© aust therefore ad¬

just our supposition coricominu variables

subject to stylistic variation to include in

the second category only those variables

currently involved in linguistic chanje*

It is al:JO noteworthy that even the MMC

makes no attempt to approach an r,.P.-typ©

pronunciation* ©von in *7LS» This fact could

b© linked to tho fairly wide variation in tho

X*ronunciation of (au) that occurs in W.P."*

further Problems wft*1. fan)

Two thing's remain to bo explained with

respect to (au).

1. There are unusual slight but uniform

fluctuations in tho ocores of the

three i'C croups# with the pattern,

from C3 to PLS# being down — no — down.

The unifomtty of the behaviour of

the three grou s moans that purely

 199

random variation can bo ruled out as

an explanation for this pattern* The

fluctuations must bo explained in the

following way* The falls in scores

between CS and FS and between HPS and

¥LS are the result of a email amount

of normal stylistic variation* The

rise in scores between FS and BPS can

be explained by reference to ©oraiaenta

made concerning (au) in Chapter Seven*

It was stated there that the pronunci-

ation of thle variable is dependent to

a small extent on the phonetic en¬

vironment . There seems, for example,

to be a greater tendency to use central¬

ised variants before voiceless conson¬

ants than before voiced consonants* It

could be, therefore, that the proportions

of different types of phonetic environ¬

ment in the VL and the BP of the questionn¬

aire do not correspond to the proportions

produced in speech, with the result that

a bias occurs in these two styles* This

suggests that it is only the WC which

has different vowels In voiced and voice¬

less contexts*

2* We must also attempt to explain the sharp

 200

fall in tho scores of the LTIC between

CS ervl FSf for a variable which otherwise

exhibits very little stylistic variation#

3one help on this point can be grained

A

from the infori-at ion portrayed in Fig:# 6#

This shows the scores for LMG male and

female informants separately, together

with the combined LJIC scores and the

fine scores# Quite clearly, the largo

fall in LTfC scores between CS and FS is

almost entirely duo to the fenalo LiiC

informants# (There is of course nothing

unusual in the fact that tho tsalo in-*

fomants have higher scores than the

female inforraants#) The explanation for

the fall in scores Is therefore that LMC

women# because of their consciousness of

tho social significance of this variable,

which is duo both to thoir sex and their

social class position# are alone in the

speech community in attempting: to re¬

produce (ou) forms more typical of a

higher class group, in formal styles#

While the rest of tho community does not

hava its attention directed towards (au),

the border-lino social class position of

the LMC and tho rolativo social insecurity

of women forces this particular sub-group
 201

to bo nor© avaro of the importance of

this variable*

Some further conclusions and problems arise

from Fig, 6# For the phonological variable (au)

there appear to be four normst

(1) a LWC norm of 260 «• 290

<") a ifWC and WC norm of 220 - 252

(lii) a LMC norm of 160 «• 200

and (iv) a MIC norm of 100 - 110
Female members of tho LMC* however, us© the LMC

norm only in CS, In more formal styles they seem

to switch over completely to the :nc norm* This

is evident from the very close approxima tion of

tho feraalo LMC line to the ?LIC line in Fig, 6,

This raises the possibility that some vari¬

ables may not bo characterised, as wo assuiaod in

Chapter Seven, by continuous variation over a

certain phonetic range acoor ling to class and con¬

text, with "norms1* being more or leas fictitious

statistical approximations. It suggests rather

that they nay, on the other hand, bo characterised

by relatively discrete class-group behaviour, and

by norms that are, again relatively speaking,

genuinely phonetically discrete. This would mean

/

that speakers would be nul© to switch fron one norm

 202

to another, rather as the IMC vomen appear to

have done in the ease of (an), instead of shift*

iag gradually through phonetic space with styl*

istie context* If this is the ease* this would

in turn suggest that it may be legitimate to

talk about different class 'varieties as discrete

entities or constructs* and that discrete social

groups may use discrete linguistic varieties

which signal their social status in the community* ^

If there are discrete class variants of

the variable (&u)« we would expect the intra*

style variation of this variable for each In*

format to be relatively small compared to that

of other variables* Tills would be because

variation would tales place over a smaller*

more restricted, discrete phonetic area* We

have* at the moment* no means of Judging

whether or not this is the ease* since the

index scores are merely measures of central,

tendency for an individual or group* We re¬

quire* on the other hand* some kind of measure

of dispersion* so that w© can decide whether

the range of the phonetic area involved in each

style for each class is smaller in the case of

(au) than for other variables*

In obtaining our measure of dispersion*

! I 0© ! Jt *4 s a9 9 a 9U 2< © s«

It © 30 0 I >4 1 I UI

4*© k 18 3 0 3Z • ik0 k e £0 ° »•«

g© 1 © 4< »+ A 0 H9 M 35 0 » »4 0H s m 34 6

• » »♦ « i© i1£ *© 3 $*« • 0u •

11 © ©© k0 © £ 2 1K* 0 : >4 *4 a •

S I *4 3 H % SK 4© o k *V © £ ©0 © «0 3

2 I 4* S * * • 3 0 © » 4♦ 0 24 »4 © «i *. »4 ©

I• k 03 »4 0 »% T 3 »+ 24 k 01 ©
§ © »I
4 k

4« ©2 9 « *4 !J ©< ! »4 k 44 © *4 © lf©

3I 3 3 S wik ( 4' 0* 3 t »4 ©

w £3 $
* : I v 34 © k ® k< © £ © k© 0 3 .

H<o k ♦»* ■ 3 t »4 24 3 *4 0 M<© k H 11 «4« w » 13

 204

tendency* the mean* "The moan deviation is

defined as the arithmetic mean of the absolute

7
differences of each score from the mean."

Thus* if we wish to calculate the mean

deviation for on informant for the v triable

(a) in FS» we look first at the number of in-

stances he has of each variant of this variable*

If he has the following number of instancea t

iihl (»)-g lilz2 lih±

0 2 4 4

we first calculate his average score as follows I

0x1 w 0

2x2 at 4

4*3 « 12

4x4 a 16

Totali 10 32

Moan m £2 * 3*2
10

Ve then calculate the sura of the absolute

differences of each instance of the variable

from the meant

 205

2 x (a)«2 « 2 x 1*2 « 2*4

4 x (S)-3 « 4 X 0*2 * 0«3

4 x (S)-4 « 4 X 0*3 m 5*2

Totali 6*4

Wo finally oaleulate the neaa deviation by

dividing thio total by the number of eaaoot

6.4 « 0*64

10

which to the average amount by which each

©aso differa from the mean*

Rote that an informant oho uses only

one variant of a variable will have a mean

deviation of 0*00. On the other hand, an in¬

formant who has an extremely wide range of

dispersion, sayt

(,5.h* SEtel jC£k£ illdt

5 3 5 3

(a most unlikely event, and one for which there

is no parallel in the scores for the Nch vari¬

ables) will have a mean deviation of only 1.00.

 206

We caw therefor* expect our mean deviation

scores in this case to lie somewhere between

the two extremes of 0*00 and 1*00.

The range of the individual mean devi¬

ations of the 22 informants for each of the

three variables ia FS is as follows t

(au) 0.000 • 0*375

(a) 0*143 - 0.554

(I) 0.136 • 0.636

This, however, tells us very little# It is

more informative to compare the mean mean

deviations for the group as a whole for tho

three variables# The figures arei

(au) 0.3543
(a) 0*4243
(I) 0.3980

The evidence is not conclusive, but it does

suggest that, particularly when we bear in

mind the very email range over which these

scores are likely to fall, the Neh informants

use a signlfleant1y anallor; range of variation

in the case of (au) than in the case of (a)
 207

and (I). Realisations of (au) are aore lively

to l>o confined to a smaller area of the phonetie

continuum than realisations of (a) and (I). for

each Informant* Tiros v© can state that (au)
is ch&raeter 1 scd by a snail ©mount of Intro-style

variation as well as by a small amount of Inter*

St£l£ variation# This# however# does not imply-

that the norm for tho MWC in FS is anything

other than a statistical construct* Haither

does it imply that the area of tho phonetic

continuum used by this class in this style is

a relatively discrete one, particularly since

the mean deviations were calculated separately

for eaoh informant* All wo have demonstrated

is that individual informants vary less from a

personal norm in a given style in the case of

(au) than for other variables* This indicates

that discrete norms for (au) arc perhaps a

possibility# but it is more probably the cose

that the small amount of variation is simply

a function of the lack of stylistic variation

for this variable as a wholet the less inter*

style variation, tho less intra-style variation*

We can pursue this discussion further by

raising it above the level of the individual

informant* To do this we will calculate the

average amount by which the (au) FS indices for

 203

each informant deviate from the overall class

moan, rathor than from his own average, This

will tell us more exactly if the different

social classes have less internal variation

of (an) than of other variables, and if there

are therefore more likely or not to be discrete

class norms and varieties.

The average MifC scores for IS are (a) 194,

(1) 194, (au) 229, The range of absolute
differences from these mean scores for the 22

informants are as followsi

<au) : 2-69

t
<S> 1 - 155

<d: 0 -
130

This again is not particularly informative,

although it does suggest that there is less

variation in the case of (au), Ve therefore

calculate the mean deviation scores for each

of the three variables for MifC PS, The results

ere i

(au) 28.2

(a) 36.3

(J) 34,5
 zoo

Again th© evidence Is not conclusive, but it does

show that individuals * average scores differ

less from the class-group mean, for (au) than

for the two other comparable variables* {au)
is therefore also characterised by a relatively

smaller amount of intra-class variation. Each

class, we can state, is relatively homogeneous

in it© linguistic behaviour with respect to

this variable.

The main conclusion to be drawn from tho

above investigations of tntra-sfcyie and intra-

class variation is that the differentiation of

(au) mean deviation scores from those of (a)

and (I) is not gr e.• t enough to suggest that

there are discrete class norms for this variable.

This conclusion is supported and indeed confirmed

by an examination of the range of (au) scores

obtained by individual informants, in speech,

from the different classest

cs
tmrnm

HMO 000—214 000-243

LMC 033-300 200—200

WC 157-260 150-300
mc 160-293 181-300
WO 213-288»
hi 111 m mm wii m
250-300
 2X0

This shows that only the 33MC has any slgnlfic*

ant non-overlapping areas of the phonetic

continuum, as it were, to itselfi 000 — 022

in FS, and 000 — 150 in CS» All other areas

Of the phonetic eontinums are shared by more

t

than one class* There are therefore, at least

for (au) in Nch English, no discrete class

norms* This also suggests that talk of discrete

class or other varieties, oven as artificial

constructs, is without validity.

The above discussion, however, is by no

means without value, since ?/© are able to draw

the following conclusions!

(i) Some linguistic variables do not

participate in. stylistic (inter-

style) variation*
(ii) Those variables which ore not in¬

volved in inter-style variation

will also exhibit a relatively

small amount of intra-atyle vari¬

ation*

(ill) Those variables which arc not in¬

volved in Inter-style variation

will also exhibit a relatively

small amount of intra-clngs vari¬

ation,
 2X1

(iv) Inter-class variation is independent

of inter-stylo variation*

(v) Soma variables * particularly those

not subject to inter-style variation,

may bo charactorisod by social class

norms which, although not discrete,

overlap to a smaller extent than

others, particularly those subject to

a large amount of inter-style vari¬

ation* The smaller amount of over¬

lapping is due to the fact that

individual informants deviate less

from personal and social class norms

for variables of the typo (au) than

for the typo { a) , «>•

Finally, it still remains to be explained why,

if there ere no discrete class norms (or varieties),

IMQ women appear to switch from one such class norm

to another between CS and FS in the ease of (au)*

The fact is that the close approximation of LMC

female scores to overall mmc scores is largely

fortuitous• It is frequently the case with the

if oh phonological variables that the men of a part¬

icular class have scores close to those of the

women of the next highest class, while women have

scores close to those of the man of the next lowest

class* Bex differentiation is less marked in the

 212

HMC than in other classes, and for that reason

1*KG female scores often approach those of the

mtG as a whole* although not usually 00 closely

as in the ease of (eu).

The Variable (o)

This variable deals with the vowel of road, home. 0

etc*

Stylistic Differentiation of (o>

Class and style differentiation of this

variable are shown in Ft#* ?« There is very

little evidence of style differentiation* ex¬

cept * as in the ease of (au}» in the speech of

the IMC* which has a marked fall in scores from

CS to WLS* Thus* while other classes make

little attempt to correct their pronunciation

towards ft* P.-type forme in moro formal contexts*

LMC speakers* because ef their greater contact

with f* »P* apeakere in the IC1C and their rather

tenuous social class position* are more aware

of the social significance of different variants*

and do make a correction*

 213

Class Differentiation of (o)

As far as class differentiation of (o) is

concerned, note that an index of 100 would be

given, for consistent use of IR.P.-type /eU/ and

200 for consistent use of Nch /ut/. Scores over

200 indicate some (fairly considerable) use of

EAn short o, as in home (.hlfeaj etc* We can see,
therefore, that it is only the VC for whoa the

"short o" is to any great extent a characteristic

feature* This is particularly true of the LVC

and MWC* Note also that the UWC has a slight

tendency to uso R.P.-type forms in WLS, but that

neither the LHC nor i-IMC consistently achieves

the K*P, none*

Monophthongs vs, Diphthongs

We noted in Chapter Seven that rural Nfk

speakers had an (.puj type ef vowel in lexical

items of this set, whereas Noh speakers tended

to have ^ui ' u* J* The diphthongised form is,

however, also used in Nch and appears to be

available to most speakers* Fig* 8 illustrates

how the alternative diphthongised form is in*

volved in the approximation to .H, P, that takes

place in moro formal styles, ami illustrates

that there is in fact a certain amount of

 UuSaite
(&) ^ Class SfcyU.

3JL (p)""^. % f'WopktKcj^ by CJtass <xv& ^bjU.

I6&

&

tC

5"
a 4s

2S

WLS fcPS PS cs
§£^(kL

UibtK^^ OMCL W.^^,cfe:

HelU&dUi*, >

py kwcgyj. % M«vv>pUtkw^s b<j SI^Ica

ICsO

&>
/©
4©

z»

m* fefS B cs
stylistic variation of this variable which la

hidden In Fig* 7* Fig* 8 shove what percent*

age el* (u)»3 forms in each style for each class

is monophthongal [at •—utj as opposed to the more

typically rural alternative (The

scores for the MMC must be regarded with caution*

particularly in CS» since only a relatively

small number of (u)*3 were recorded for this

elaes*) Xi can be seen that there is quite clear*

ly a tendency to increase the number of diph*

thongised vowels in more formal styles* and that*

generally spealcing* the higher classes have the

highest number of diphthongs* The LHC* on the

other hand* uses only 10 per coat diphthongised

forms in CS* Xt is therefore demonstrated that

R*J»* influence is effected through the transfer*

in more formal styles* to the rural* and possibly

older form Lfuj* vhlch is phonetically somewhat

closer to the R*P, form*

Xt was also observed during the course of the

interviews that there seemed to be more signs of

some internal geographical variation of this

feature within the city than of any other* Xn

order to investigate this point we must of course

compare the index scores of informants from

different parts of the city* To avoid the

 H

19 VII

nil I 0 9 1
ft

Sr
ft 0 1 ft »

<»
e

•ft
ft 9 ft 9
M
ft
ft ft ft
&

•4

0 m o »

t-
e 5 O
H- 0 ft

*
•»

ft M S ft
ft ft-
&

*4
« ft ft
? I

«#•
ft
*I
! S 9 —g§
•ft
. V
ft

<4
ft

i
i ft 9 a- » ft
o
«►
cr
•ft

a;&
• 3

i*
S « o
*

sr
« 9
i: 8 9

•ft
si t ?

i O ? ?
5

! I E £
I ft ft M ft ft 0 1
9 »
er

•ft <* <>

«*

? I
ft ft

I & J

1
B5
efr

5 *- M
?
ft
w

ft s-
*3
& ?

ft «•

er
a
5 0 1 S oi*I!
3 9
I t
 216

The East Anglian short o and (o)

Fig* 10 shows in more detail the social

distribution of* the EAn short o» It gives the

percentage of* (o) wliicli were (5)-4 in each

style for each class*

There is obviously evidence of' class

differentiation in the use of this variant*

together with a certain amount of stylistic

variation. Mote* however, that even the LWC

uses no store than 42 per cent forms with short

o. This suggests that tho short o is some¬

thing of a relic form in Nch. (This is con¬

firmed by material presented in Chapter

Eleven*) This point becomes still clearer

if it is realised that the 42 per cent is

composed of a relatively small number of

common lexical items* Details are given be¬

low in Table IV* which shows the number of

instances of (o)-4 recorded in the Nch survey

per word*

In addition to these items* aerodrome*

comb, alone and combed were recorded with (o)«*4

in the ML and the RP, It can be seen from

Table IV that don't* only* 3»p.-?ose and home

account for the vast majority of short o forms

In the material recorded here* 80 par cent

in the FS material and 73 per cent in the

CS material» (including anrroosed. homework) •

Note that aerodrome. Hlonodroma and roller*

drome are almost universally pronounced with

(o)-4 in Neh* hut are not oF especially fra*

quant occurrence*

Table XV shows Further that phonetic en*

vironment lo not the only factor which

determines the presence or absence ot £uj

in a given word* although the word*Final

consonants are all alveolare or nasals* with

the exception of {.vj, and ^kj. Other

Factors are sentence stress* speed oF

utterance, and social connotation* short <»

is more likely to occur in lightly stressed

words* in rapid speech, and in words meet

current in popular speech or within the

Family. (These are all Factors mentioned by

Avis in connection with the New England

short e>. )
 Table IV

Number of Occurrences of (o)»k

FS cs

don't 156 44

O.VAX 51 8
suppose 23 9
bom® 22 11

Bead 8 0

F.f!n£& a 6
both 7 6
broke 4 2

supposed 4 3
whole 4 1

Holme 4 0

over 4 0

Holmes 2 0

bloke 2 2

poet 2 0

stone 2 1

a«Mr* J° 1 4
homework 1 1
moat 1 1

a poke 1 0

photo 1 0

npJMs, 1 0

coats 1 0

road 1 1

roads 1 0

blokes 1 0

rollerdrorae 1 0

Close 1 0

drove 0 1
Hippodrome 0 1

woke 0 3
 i.i&. V/arikbU (€)\ %b«fj Cbss

Si^le

P^.u, KbjieJsSt [p*\) by cu^st ^iy{e.

S4vj?e
 219

The Warlable (on)

This variable le the vovel la know, old, throw,

eto. As this variable and the preceding one# (oj# are

not distinguished In R,P. or In many other accents or

English# It Is Interesting to compare the two In Nch

English. The Index scale for (ou) Is the same as for

(o), so that a score of 100 would be given for a consistent

R.P.-type pronunciation, and OOO for a consistent pro¬

nunciation of EAn IauJ. fig* 11 shows that for this vari¬

able there is clear differentiation between the MMC and

the LUC# and between the LMC and the VC* Within the ¥C#

however# differentiation is minimi. This class as a

whole is more or loss consistent In Its use of Nch /au/.

KMC speakers use, on average# a vowel with a first

element rather more open than that of R.P, # while the

LMG us# a vowel closer to that of the Nch WC than to It. P.

There is also a very small amount of styllstlo vari¬

ation of (ou)# at least amongst the WC» Zn this case

the class lines slope from left to right across the

graph# since lower scores indicate vowels further re¬

moved from the R«P. norm# because we are here using the

same index scale as for (o). The wide fluctuation In the

scores of the f#!C suggests random variation over a relat¬

ively largo phonetic area [eU ^ at?J# and reflects the

large amount ofvariation there is In R.P. In the pro-

9
nunclation of this vowel.7
 220

Differentiation of (on) and (a)

We can now use the combined index scale

to measure the amount of differentiation made by

different classes In different styles between

(a) and (ou) Items* Figs* 12» 13 and 1% portray

graphically the average amount of phonetic space

differentiating the two sets of Items for the

KMC, IMC and WC respectively. (Graphs for the

MtfC and LlfC would be very similar to that for the

W.) Fig* 12 shows that the KMC Is close to

actually merging the two sets of Items* as In

Ft,P., but does not In fact* on average* do so*

The graph of Fig* 12 poses an interesting

question for phonologleal theory* The distinction

between (o) and (ou) is quite clearly preserved*

Is, however* the amount of distinction shown here*

(at Its smallest 31 index points as compared to

around 200 for tho WC)» consistent with the typo

of distinction normally considered to bo phonemic?

The amount of phonetic distinction is In fact so

small as to suggest that a linguist unfamiliar

with ISAa phonology might not notloo the differ¬

entiation* There is also* of course* a very large

amount of what is usually termed "phonemlo over¬

lapping** * much larger than the amount mentioned

by Labov* for instance, in his discussion of

% '2, (cu)j tu^d. (Vj i
 5 H

I © I ©
P © W 3

ct- t*

•0

ft
ft

a 0 3 a
n

o 3 ©
ft
3 4 9

I
«t
t* tr

E ft
©
0 ©
© 9 €

et
»* **

M
I

9 3
1
a

c*
2 ©

<fr
ss & I
a a
I *
* 2
ft

2 0 a 1 O
9
Sf «► ct *■»
«t

g
b

«♦ tr
« i ©
&
©

**
f t

ct-
S a § ft ° g i ©

H 5ft
►% «t
$

!3 £ © ©o © «♦ I*s «* ° . t* = «* t* 2 X %J
%*:©►» •t
I
► H

© 3 a p
$
^5 © a
£ M

«
». S * r

* •
«*

3 P
t * 3

*>»
?
I f

«*
Br
4

•0

4© §i 3 ©
»
©
I
°
I

«*
» g 3 0
9
1 S

«*
t*
*<
t*
«
a
3

i t
0 9 © © § 1 » * I
s

t*
►3 it
or

«*• Cf

i ia
I
5
c

E
g

5
4

3i
**
9 3 §

«♦
p

«*•
I
I I «*•
H- © * O 9
r

tnie
E H H
P « « P O o © 3 I
9 © a & S a 9

tr «fr «t V{
♦3 i*

H S i t 3
3
© O 9 0
o» 9
Si
t* St

•a

i £
« © i * P g- ©
3 a © 1 a
cr

* 9 C « 4 © 0 © H> 3 i o
© © 4

c* «* tr
>1 0^ I*
•a
»4 & o ** a 0 I g e *4 SJ 01

*r

I i s 1 SH ia * Q t*•

5 HH

dif 5 4• *W I * a s« i 1!0 • H s|

o © s tr< « »4 g #4 J5■ lri 0 *4 s © »4 s i i »*• 5 0 a5

i!g »* i *4 3 0 0 *4 « 1 % 1£ 0 44

i < \ S ••

iii
a? ! >9 0 4« I► 8 S0 © .V" O
10 g0 *4 1 »4 s 0 VA 0 a 4* H »rw im I % ai 0 »4 ia ©§

>4 «* 0 0 NS 0 i 01 0< 01 »4 VK«j 1 w4 |

1* wA tif §C 1iTJ g w0( »4 1
 RpjjC \4naA4t. (j^ hxj Class Style.

wts ftps p-%

%(e

F^.lk 1/cin^y.K, (j^ buj Clas^ &*J~

ft&. F*
Style

Rq, r? WnetUe (£) b<j Oa%$ &v\iA Style.

200

(to

I ^
rf Dfi&

0©d
 223

Co-variation of (p) and (k)

▲ quick glance will show that all three

variables have a very similar pattern of class

differentiation and, more particularly, of

style differentiation* This suggests that it

may be worthwhile Investigating the degree and

nature of the co-variation of certain phono-

logical variables with each other* It is, of

course, not surprising that (k), (p) and (t)

should exhibit similar characteristics, because

of their phonetic similarity. At the same time,

however, there is no a priori reason for

assuming that this should be the oase, and the

fact that they do behave similarly is of some

Interest, ¥e can examine the co-variation of

14
the two marked phonological variables of

the three with each other in the following

way* Fig* 13 is a scattergram which shows

(p) and (k) soorss for all informants in WL3,

with each dot representing one informant. It

can be eesn immediately that there is a very

good correlation of (p) and (k) scores for a

large majority of the informants. Those In¬

formants who have high (p) scores tend to have

a similar (k) score, and vice versa. This

suggests that it would be possible to set up

instead of (k)» (p) and (t) one variable of

 22k

glottalisation. which could be treated on ita

own as an indicator of social class and social

contort* (vls was selected for the purposes

of this demonstration since many of the in¬

formants have (k) and (p) index scores of 100

in FS and CS, which Bakes results for these

two styles less Interesting* and there is mere

material far (k) and (p) available in VLS than

in HPS.)

Differentiation of (k)i (r>)t and (t)

For (k) (Fig. 15) there is little class

differentiation in CS* except that the SfMC

has significantly lower scores than the other

classes* The two lowest WC groups, moreover *

appear to be indistinguishable in their be¬

haviour with respeot to this variable in ail

contextual styles*

The picture far (p) is very similar* with

all four lowest classes being virtually in¬

distinguishable by this feature in everyday

speech* and only tha 4M0 notably differ¬

ent charaeteristics* In other contextual

styles* the scores for the three WO groups are

also very similar.

 otwoL (k) SC6imSe$ ; WL$.

P^,i\ \4.mUe (Ja) ^3 SivjUt,

 225

The pietore for (t) Is also similar to

that for (k)» except that the IMC remains
separated from the C even in CS. It is

interesting: to note that the v*C has almost

100 per cent "extreme" forms, (t )-3, in CS,

just as it does for (k) and (p)» where
glottalisation is not so complete*

Word-final and vord-tntemal glottallsatlon

Index soores show that glott&lisatlon is

more frequent in the case of word-final (p)»

(t) and (k) than it is word-internal!y. A

comparison of (t) indices for a random sample

of ten of the sixty informants givas the

average scores shown in Table V.

Table V

Sample of ton Informants! Average Scores

Word-internal and word-final (t) by style

hi WLS WPS FS
mm *
CS

Word-lnternal 029 052 113 13^

Word-final 028 089 151 161

We can therefore state that glottalisatlon

of voiceless stops la inversely proportional

 226

to social class and social context, and Is

core frequent word•finally than word-Internally*

The Variable (h|

Fig. shews that this variable has a pattern of

Class and style variation similar to that of (t), but

with a relatively smaller amount of stylistic variation*

It also shows that the IXC and M¥C arc Indistinguishable

with respect to their use of this variable*

we saw la Chapter Seven that rural £A is an "h-

pronouncing* region, audi commented that Nch appeared to

be an Island of "h-lesaness* within the region* It Is

therefore interesting to see that (h) In !?ch exhibits all

the characteristics of a normal phonological variable,

since this Indicates that "h-lcssnass" is well established

in Mch and has been for many years* At the same time,

however, it should be noted that the LWC and M¥0 use

approximately kO per cent "h»ful* forms even In CS. This

is probably a much higher percentage than would be found

In many other urban areas in England, and indicates that

Neh*a "h-le«snc>ss', is only relative.

Diffusion Processes

¥e speculated in Chapter Two that h-dropping

 227

in Ncli is the result of diffusion processes

which Involve the spread of linguistic innov¬

ations from one member of the urban hierarchy

to another* rather than in waves across the

countryside* If this is so, it would appear

to be the case that Unguistie changes of

this type spread from the WG of one urban area

to the lie of another* since the Nch MC has

only a small percentage of h-lees forms* This

fact* however, could be due simply to the

counter-influences of Tl.P. and education* This

point will be taken up again later* in Chapter

Eleven.

Rural—Urban Dichotomy

It is informative at this point to compare

the scores for (h) of informants born and

brought up in rural districts with those of

informants who were born and have remained in

the city* Table VI shows the average (h)

scores for rural-born informants compared to

the average (h) scores for their sooIa! class

as a whole, by contextual style* No MMC

scores are included, since none of this group

was rural—bom*

Tabls VI shows that informants born and

 223

Table VI

(h) Indicest Rural Horn vs. Total Informants

Stylet VLS RPS rs cs

Class

LMC Toti OOO 005 oo4 014

RB 000 000 002 012

UtfC To*i 001 007 024 040

RB 000 000 007 012

Mfe'C TOtl 004 012 043 059

RB 000 004 023 050

UTC Tot l 005 013 041 061

000 003 000 000

brought up In rural areas hare conelatently

lower (it) scores than city-born informants.

This is particularly true of the WC and the

LITC. This means that the rural-urban dichot¬

omy is still a linguistically valid one in the

area around Nch. As far as (h) is concerned,

it is not a distinction which Is valid only at

the level of the conservative rural dialects

of the elderly.
 229

The Variable (e)

This variable is t'e vowel of tell, well, better,

•to, (see Chapter Seven), Fig, 20 shows that the

pattern of class differentiation is an unusual one.

Here we have a case where the uwc has an index soore

which is significantly higher than that of the MWC in

CS, and where both the UWC and HWC have scores that are

significantly higher than those of the LWC in all con¬

textual styles. Thus, for the wc as a whole, the whole

pattern of class differentiation is completely reversed.

Note, too, that in CS the LUC score approaches those of

the LMC and KMC very closely. How can this be explained?

We have already noted that the LWC, as a relatively

underprivileged group, is Isolated from innovating

tendencies. Since we find that in this case the LWC is

differentiated from the UWC and MVC in an unusual way,

we can hypothesize that high scores for this variable

(that is, a large amount of centralisation) represent

an innovation in Nch, The phonological variable (e) is

involved in linguistic change, in that centralisation of

this vowel is increasing. In the vanguard of this

change, we can hypothesize further, are the upper members

of the WC, The LWC and LMC are also participating in

the change, but at a lower level, an i the MMC are not

participating at all, or very little.

This hypothesis is confirmed by the pattern of

 230

differentiation shown In Fig* 21, This gives scores for

(e) for each contextual style by age-group. We can see*

first of all* the stylistic variation of (e) in the

clear separation of the lines connecting the scores for

each contextual style* The main point* however* is

-that this graph shows that there is a very marked

amount of a^e differentiation of (e). Younger people

have much higher (e) scores than older people* This is

particularly clear in the case of informants aged ten

to twenty-nine* but there is a clear general increase

in scores for (e) across the graph from right to left*

Centralisation of (e) is more prevalent among younger

speakers* and is becoming increasingly so.

The full implications of the study of linguistic

change in "apparent time", the relatively higher scores

of the over-sixties, and the role of the UWC and MWC in

linguistic change will be discussed in Chapter Eleven.

Tho Variable fa)

This variable is the vowel in bat, bad, carry, eto.

Fig* 22 shows that there is an overall tendency to a

regular kind of class and style differentiation* On

the other hand* however* there is a considerable amount

of overlap in the scores of the WC* and stylistic

differentiation between HPS and WLS is. for four of the

 231

fir# social classes* la the * wrong* direction* One

explanation lor this can perhaps be adduced from the

fact that the overlapping of different social class

scores is largely due to the LWC. We have already noted

that this class is to a certain extent Isolated from

Innovating tendencies* It Is therefore possible that the

overlapping of scores is due to some kind of linguistic

change In Nch English which Index scores for this vari¬

able stay have obscured*

In fact information from the rural dialect surveys

discussed In Chapter Seven leads us to suppose that two

somewhat conflicting dlachronlc tendencies are at work

here* The Nch (a) index has the following variantst

(a)-3 « I***J etc*

(a)-4 * l**"" «*j etc*
(a)-3 m £et®J etc#

Now K&kerltas*-^ states that his informants used vowels of

the quality for this variable* He

considers that |,sj forms may be partly due to "Cockney*

Influence* However* the fact that the Suffolk children

had only — at J suggests rather that £« '— * * J 1» the

older EAn form* and that it is in the process, at the

time when K&keritz is writing* of being replaced by

R.P*-type L®jU This hypothesis is partly confirmed by

Forby*s statement (and he is writing around 1830) that

 232
\

bad, nan, etc,, are often pronounced » bed", 'men*, etc.

In Nfk, especially in the region of King*s Lynn, We

can therefore hypothesize fur thai- that older Jfch speakers

will tend to have a higher number of (a)-4 forms,

and that middle-aged speakers will tend to have the

newer R.P.-like forms of (a)-l or (&)-2, This In turn

will mean that the diphthongised (a)—3 and (a)-5 forms

must oocur in predominantly the speech of younger inform¬

ants* If this is the ease, then, within the WC, age—

group differences will be more important than class

differentiation, and the overlapping shown in Fig, 22

will occur.

This hypothesis is confirmed by the facta shown In

Table Til, which demonstrate that the older EAn form of

—«t J Is more commonly retained In the speech of older

Nch people, but that the more typical modern Nch pro¬

nunciation is which in the speech of many,

particularly the young, is diphthongised to j,«i®— mi1

«*ej. This suggests further that for some, particularly

older or LWC speakers, (&)-*» represents a more formal

pronunciation than (a)-3» For them, (.ej functions Just

as the more modern [mj does for younger speakers, Diph-

thonglsation and vowel height must therefore be regarded

as Independent variables which cannot successfully be

incorporated into a single index, as we have tried to do

here. Table VII gives the percentage of each particular

variant of (a) used by each age—group in FS and CS combined.

VWUs. (\) huCb%. Sttftg,..

%)bj Class Siy^S.

 233

Table VII

Percentage of (a) Variants

Per cent

M-* (*,)-£ iaJbl l*l-k .Mr,5 laisi

10-19 5 40 44 6 4 99
20*29 26 31 26 10 6 99
30*39 22 45 25 5 5 102
40*49 21 37 14 12 14 93
50*39 32 35 25 8 0 100
60-69 26 34 27 a 4 99
70* 13 £3 26 25 3 100

Xt shove that both the 10*19 had 70* age-groups use a

email percentage of (a)*lt but for different reasons* the

10*19 group uses instead a high percentage of diphthongised

(h)"-3» whereas the 70* group uses a high percentage of

(a)*4 * t j. We can see that (a)*4 is employed to a

large extent only by the 70* group,, and is therefore

clearly dying out. It has largely been replaced by the

more R.P.-like (a)-l, but this too is clearly ceding

ground to (a)-2 L*»J» which is the most common Nch variant
at the moment. (The demise of (a)*l is most clearly in¬
dicated toy the very low score, 5 per cent, among the

10-19 group#) Even (a)*2, however# would appear to be

ceding ground to the diphthongised forms, (a)-3 and

(a)-5, which between them are the most common form of

(a) for speakers aged 29 and under. We can theroforo

postulate a diachronic progression* j_e > » >mt >mie ( >«*•)].

 234

The increase in diphthongisatlon is more clearly

revealed la Table VIXI, which shows the percentage of

total raonophthongal and total diphthongal pronunciations

of each age-group* The scores that stand out as except**

lonal are the 51 per cent only total monophthongs and

the 48 per cent total diphthongs of the 10-19 group.

Table VIII

(a)I Monophthongs and Diphthongs

Age-group Per pent

Total Monophthongs Total Ptphthongs

10-19 31 48
20-29 67 32
30-39 72 30
40-49 70 23

50—51? 75 25
60-69 63 31
70e 71 29

All groups thus show some usage of all (a) variants,

but no group except the 70s group has above 12 per eent

(a)-4. All groups use at least 68 per eent total mono*

phthongs, except the under—thirties. No group uses more

than 31 per cent total diphthongs, except, again, the

under-thlrtles. Further implications of this linguistic

change will he discussed In Chapter Eleven, as will a

 235

further (a) variant* centralised L®J» which appears to

occur in contexts similar to (e)»

The Variable (a)

This variable is the vowel in name, nail, plate* etc*

Fig, 23 shows that (a) is subject to both Glass and style

differentiation, although the styllstio variation is not

especially groat and the WO and WO scores are more or

less undifferentiated except in WLS. The WO has scores

substantially higher than the other two vc groups, but

this may be due to the fact that Fig, 23 shows scores

including (a)-*5 {»/•*/)# WG j—type diphthongs, in

other words, may have a first element very little more

open than the WC and (JSC. Scores for (a) without

(a)-5 will be discussed in Chapter Eleven.

For the moment we can say that there is a norm for

1WC speakers in CS between [elJ and t»»ij1 for the KMC

and CMC of [eij| for the LMC between Lei] and [eXjt
and for the MMG of Surprisingly enough the KMC

approaches [elj or [«lj only in WJLS,

The Variable ill

This variable is the pronunciation of the vowel in

right, ride, rye, etc. Fig, 2k shows that (I) differ-

 23 6

eatUtea between the social classes in the usual way#

except that the iilfC has higher scores than the LWC in

three of the four contextual styles# We have suggested

before that this kind of overlapping pattern 1® a sign

of a linguistic change in progress« and this therefore

could be the case here.

The index scale for (I) ie as followsi

(I)-i . L«ij
(l)-2 - fcBXj
(I)-3 a l*ij
(IM = L=»iJ

Many of the rural dialect records indicate that the older

KAn form for this variable is {."Pi J • (a)-3. As (a)~l and

•2 represent R,P.-influenced forms* it may therefore be

the case that j, oij is a fairly new development In Nch,

with the MWC again being ins trunoutal in introducing the

change# We can in any case see from Fig# 2b that the

ySHC is tli© only class which has an index score of over

200, which indicates a fairly consistent use of some

(J)—k forms# We can investigate this possibility further

by studying a more detailed breakdown of figures for this

variable#

Sex distribution of (?)-4

Table XX shows numbers of male and female

 237

informants who use any (J)-4, together with

a breakdown of those who use more than certain

percentages of (i)-4 as opposed to other (I)

variants* It shows that over half the in¬

formants in the sample tise some (I)-4, bat that

only eight of these arc women* No woman* more¬

over* asas more than 20 per cent {£)—4* The

aex differentiation of this feature is there¬

fore very marked indeed* and indicates that

is a characteristically male form*

Table IX

Informants asIng (X)-4

50 per 20 per 10 per 5 par

cent ♦ cent * cent * cent ♦

Male 27 6 12 15 24
Female S 0 0 3 5
Total 35 6 12 19 30

Class Distribution of (%)-4

Table X shows that use of (I)—4 is also very

much a MV0 characteristic* since 95 per cent of

this class use some (i)-4, and 50 per cent of

the MWC use 10 per cent or more (sT)-4 compared*

say, to only 37 per cent amongst the LWC* (Note

 Table X.

Percentage info man ts of

each class-using (iT^iT

Per¬ Any 50per- 20per- lOper- 5 per-

cent .
cent+. cent+. cent+. cent+.

MMC 17 17 17 17 17
LMC 25 12 12 12 12
UWC
'

37 0 12 19 37
MWC 95 18 36 50 82
LWC 62 0 0 37 50
 233

that the HMO 17 per cent and IMC 12 per cent

scores represent on© laforront each only. These

two Informants are both men who have been

socially mobile In an upward direction*)

Age ^Distribution of Ohh,

Table XX shows that there Is in fact some

evidence of linguistic change for this variable,

since a higher percentage of younger people,

aged thirty-nine or less, have (£)~4 forms

than older people aged over forty. This is

particularly noticeable in the case of those

informants using 10 per cent (I)-4 or more.

Table XX

Percentage of Informants
in Two Ago—groups ualng (I )-4

Age Any (l)-4 50 per 20 per 10 per 5 per

cent » cent ♦ cent ♦ cent ♦

10-39 61 11 18 43 61
40* 56 9 12 22 41

We can therefore state that (I)-4 is character¬

istic of male speakers, of tho MWC, and, possibly,

of younger speakers. If a linguistic change is in

 239

progress, it is being led by younger, male

members of the ®G# One© again, this point

will be discussed in greater detail in Chapter

Eleven,

The Variables (er) and (er)

These variables aro the rowels in here. Tear, idea,

etc,, and there, chair, care, etc,, respectively. The

patterns of stylistic variation Tor (er) and («r) are

given in Figs, 25 and 26.

Class and Style Differentiation of (or) and (cr)

Fig. 25 shows that there is a significant

amount of stylistic variation in the realisation

of (er), but that, especially in CS, there is

little significant differentiation between the

four lowest social classes, and ©specially the

WC groups • The fact that in CS the tR?C and LMC

have scores higher than the 3RfC and LtfC suggests

that a linguistic change may be in progress

here, although the differences in scores are

very small (see Chapters Seven and Eleven),

The scores indicate that in CS even the MM0

uses a vowel more open than [ea e»j, while

the norm for the other classes in this style is

around Lf«~ ©tj.

 2s*. Vars&biig Cjc-ss

UJUi

fiq.Zfe ^5,r'io.ble (£*") bvj Cfes O.KfA

4§ ^

Oft

<£t

C£&
Vl« ftf$ FS c&
St«jb>
 240

Fig* 26 shows that there is# on the other

hand, no stylistic variation ot (*r). Social

consciousness# and probably therefore linguistic

change# Is concentrated on (er)• Mote once again

the pmimity of the three ¥0 scores#

Differentiation of w.n*'
(er) and
wne. m ■■ ww—awnnMie'ii
(tr)

Because the same Index is used for both

variables# (It is for this reason that MMC soores

appear at the top of Fig* 26# around the R.P,

norm of 200)# we can use scores for these variables

to investigate the nature and extent of the phon-»

etle distinction between items of the fear and

fair sets# as it is made by different groups of

Nch speakers• This is exemplified in Fig. 27*

Fig. 27 shows the scores for HMO speakers for

both (er) and (er), by contextual style. The

index scores shown along the ordinate have been

reversed so that 009 is given at the top of the

graph and 200 at the bottom. This has been done

in order to symbolise "phonetic space*t so that

scores for more open vowels occur at the bottom

of the graph# as on the conventional vowel trap**

eziura, and close vowels at the top. Fig. 27

shows quite clearly that the MJiC makes a definite

distinction between the two lexical sets# but that

 zbi

ill© amount of phonetic ©pace employed t© effect

this distinction decreases In the more Informal

styles* Thus even the MMC is involved in the

Noh tendency to merge (er) and («r)» the dis¬

tinction in CS being on average only that be¬

tween {.e J and ifj-type vowels*

Pig* 23 shows that the LMC make# a certain

amount of distinction* bat much smaller than one

would normally associate with a ^honewic dis¬

tinction* between (or) and (er) in formal styles*
In CS* however* the merger is more or less com¬

plete* That is to say that in normal* everyday

speech LMC speakers do not distinguish between

(«r) and (er) items* but are able to do so in

more formal styles, including even FS, This of

course argues very persuasively for a theory of

phonology which distinguishes between a surface

phonetic level and a deep phonological level,

and in which it is possible for two distinct

underlying phonological units to have one ident¬

ic
leal surface realisation. '

Fig* 29 shows a similar picture for the

UWC, but with an even smaller amount of dis¬

tinction of (er) and (er)* ami a complete merger

in CS* The more open vowel for (tr) Items in FS

can be interpreted as the result of pressures

A
(er) ft(fx) Wj Sfcjlfe : KHC
 2kZ

operating; in phonetic space, with more open

(or) trowels resulting in more open vowels for

(er) as well, in order, presumably, to preserve

the distinction* Where the distinction is not

made, in CS, («r) vowels can revert to their

"natural" level again*

Fig* 30 shows the same sort of picture

onee again for the Mfe'C, but with the distinction

only minimally preserved* There is again a com*

plete merger in C3# Mote that (er) vowels are

again more open in IS, but revert to the natural

level In CS.

Finally, Fig* 31 shows that LWC speakers

distinguish the two sets of items only in VLS,

and perhaps UPS, and then minimally*

Forms with Schwa

It must now bo remembered that the (er) and

(er) indices we have been discussing are concerned

only with vowel height* They aro not capable,

in other words, of distinguishing between t®*i

and i,ea J, or |.e t j and j. In order to confirm
that a merger of (er) and (er) items has taken
plane in everyday speech in Neh, we must supple*

meat these indices by figures measuring the amount

(prj <X*4 (Srj by Stvjle > HWC.
 Zk3

of schwa used In the two seta of Items• If it

can toe shown that there is a significant differ*

once in the percentage of schwa for the two sets*

then it may be the case that an («r) / (ear)

distinction is toeing preserved as* for example,

j.e»j / J^ea j. However, Figs# 32 and 33# which

give the percentage of schwa forms for (er)

and (tr)i show that in CS no social class uses
more than 13 per cent schwa for either variable*

This means that schwa is not need to preserve

the distinction, since it is mainly in CS that

the merger in any case occurs. This is more

plainly revealed in the figures given in Table XII.

Table XII

Per Cent Schwa in (er) and (er) Items toy Class In CS

fer) (*r)

mc 0 0

LMC 12 11

U¥C 2 0

bWQ 3 7
LtfC 10 13

It can be seen from Figs. 32 and 33, however, that

schwa is much more commonly used with (er) items

%£. % (ef) w,tk £3] bij C'ass awA. Sfejle.

6V8. % (ir) loll-k Ci] U &«ss

 2kk

than with (tr) Items in contextual styles

other than CS» and this fact must he taken

into consideration when interpreting the

amount of distinction between (er) and (tr)

shown in Figs* 23 to 29*

Stylistic Variation of Schwa Forms ■

It is clear from Figs# 32 and 33 that

there is some considerable variation in the

percentage of schwa forms used* We must there¬

fore revise our previous statement, and say that

there is some styiistio variation associated

with (er), since Le * J tends to be replaced by

j.*a J in more formal styles#

Linguistic Change and Schwa Forma

It should bo noted that in almost all

styles, for both variables, the IfiiC and MWC

use fewer schwa forms than the other classes,

Including the LWC# In C3 this is also true of

MMC speakers, as Table XII shows* This can be

Interpreted as indicating that the MHO, UWC and

M¥C are leading the L&C and IMC - in both cases

the two groups using most schwa forms in CS -

 245

in what It most probably a linguistic change

(eee Chapter Eleven)* If this la the ease* it

la a linguistic change that la being Intro*

duced into Nch English in two different wayat

1* from P.P. into M?!C speech, with

the IMC as yet not so fully

affected|

and 2* from the WC aoeents of the areas

earrounding Nch, of M as « whole#

with the WC end MWC being at the

moment the two groups the moat

influenced* There are thus certain

parallels with the introduction of

rounded (o) vowola described above*

The MWO phenomenon * note ©specially the big

UNO etyliati© variation of (er) from ¥LS 97 per

cent schwa forma to CS Q per cent * cannot be

entirely due to R*P* influence, since the ajshwa-

lose type does not ocea to occur in R»P. in the

. , « is
case of (er) items*

Wo can conclude, therefor©, that (er) and (er)

are not distinguished in Nch, except by the MMC,

in everyday speech* Moat Nch speakers, however,

can make the distinction In more formal contexts*

He must therefore postulate that there are still

two separate underlying phonemes in the Noh in¬

ventory.
The Variable (u)

This variable is th© vowel in boot, spoon, roof,

et o.

Stylistic Differentiation of (5)

Fig-, 3% shows that there is a significant

amount of stylistic variation of this variable,

with a vowel on average closer to j.a» J being

more common in WL3, and one nearer to i«tt J in
C5* This in fact means that in CS there is a

higher percentage of item pronounced with

/Hi/ as opposod to /ui/. It does not so much

Indicate a gradual shift of vowel quality in

phonetic space, (although this does also happen,

end several intermediate forms were recorded)*

Speakere generally switch from an i«ij-»type

vowel to an ^uiJ-type vowel with increasing
frequency as the formality of stylistic con¬

text increases*

Class Differentiation of (u)

There is on the othor hand little or no

social class differentiation of (u), except

that the MMC is quite clearly differentiated
 2ky

from tho other classes in CS, One explanation

lor the apparently Irregular behaviour el the

HMC in this etyle ia that speakers

do not have the Neb phonemic distinction

/ui/ i /*/1 but instead have a whole range of

vowels i«SU » ei j as Iree variants* or as con-

textually determined allophones# (u) le the

only Neb variable which la not Involved in

class differentiation,

Tho Bast Anglian abort 0 and (u)

Fig* 35 deals with the other alternative

pronunciation of (u), namely /IT/, It ehows

the percentage o! Bin abort «►*» in (u) items

by class and stylo, (Note that (u)-5 Is

not included in the index scores of Fig# 3^»)

Xt can be seen that there ia a certain amount

ol class differentiatlen of this feature, al¬

though this is very small, (Age differentiation

is more significant - see Chapter Eleven*)
What stylistic variation there is, moreover,

appears to be, with the exoeption of tho IMC

shift between CS and FS (which should probably

be interpreted as a genuine stylistic shift of

the normal type), in tho wrong direction. This,

however, is quite easily explainedi short o in

VftnKbte (u.) hij Class GAccSjtini,
(u) items occurs in Nch most often In the

coco of a relatively small number of lexical

items which are well represented in the WL,

less well in the RP, and which occur even less

frequently in speech* The best illustration

of class differentiation, therefore, oan be

obtained from the MLS figures, where all

Informants pronounced the same (relatively

large number) of (u) items* This shows that

the HO is significantly distinct from the

WC in its use of short o forms*

Items subject to Shift> CS and FS Material

Tho extent of the stylistic shifting of

/u»/ to /«»/ in Hch, and vice versa, can be

gauged from the following list, which cites

all the (u) items recorded in the interviews

in CS and FS*

Item with /ut/« /at/ and /IT/t

room

Items with /ut/ and /ut/«

school* move, to. fool* boot* improve*

afternoon* mood* smooth

 2k9

Item witn /a*/ onlyi

pool, wound* rvox.tr*» approve* troop*

food

Items with /W onlyi

too* two* do* lose* you. t?vr o«^h*

who* ahoe, soon* Waterloo* root

It eras with /XS/ only*

proof, Tombland

lit addition* the following items were reeorded

in the interviews only with /at/* hut can fre¬

quently be heard in Nch with /»»/ •

saloon* scooter* hoop* IMonican*

goose* scoop* croupier* honeymoon*

Toiabland is also pronounced with

/ui/ and /**/•

WLS and HPS Material

Of the above items * the f ollowing also occur

in the WL and the RPt

room, proof* boot (WL only)

 250

and /rood, too, do, you, who, soon, school,

move, fool (WL and RP).

The much larger amount of material obtained in

these two styles shows that proof also occurs

with /«»/ and /u»/, that food also lias /*»/, and
that too, who and goon can also occur with /«»/,
X "

The full list of items ooenrfinff in either the

¥L» or the ?/L and the RP, is as follows« fool,

cool, school, food, move, spoon, roof, soup,

goon, too (in the SL and th© RP), and hoof,

proof, room, broom, boot (WL only). (You, who

and do also occur, but are overwhelmingly al¬

though not entirely pronounced with /wi/, and

will not be discussed further.)

The vory large amount of variation that

occurs in this variable in Kch can be gauged

from the fact that, although only ten of these

items occur in both the WL and the RP, as many

as 48 informants had at least one case of styl¬

istic shiftt there was at least one iters that

had a pronunciation in RPS different from that

in WLS, This differentiation cannot in most

cases properly bo termed stylistic variation,

%

since there is no clear tendency to shift in a

particular direction. In the case of cool, for

 251

example* twelve (» 20 per cent) of the Inform*

ants had a pronunciation .in EPS different from

that in VLS* In tlx of these cases* the in*

fermant switched frost VIS /u»/ to EPS /at/, and

in the other mix. eases the opposite switch

occurred* Furthermore* although altogether

♦

enlf t*enty*five instances of (u)*ttems occur

in ths RP end ifl combined* only two croups of

informants* ono consisting of three* the other

of two* were identical to each other in their

usage of /ui/« /*»/ and /W with respect to

(a) items in those two styles* Further die*
euselon of this distribution will be made in

Chapters Tan and eleven*

Monophthongs vs« Plphthoarei

For the variable (©)* we obtained some

information on the proportions of monophthong jai

to diphthongal typos of pronunotation * which

was portrayed in Fig* S« This has also been

done for (u)* and Fig* 36 shows tho percentage

of (u)*3 which were monophthongs! {.si ^ ut J as

opposed to diphthongal There is

little evidence of stylistic variation hero*

but there is> a certain amount of class differ*

entiatlon* Zt is* however* significant that tho

 252

highest number of monophthongs occurs in CS

(100 per cent for all but one class group)

where one would expect the fewest R«P.•type

pronunciations• The explanation for this la,

of course, that the merger of (o) and (a)

items moans that the least R.P.-like pro-

nunelation of (o), i*e» the most monoph-

thongal, also affoots (u) Items*

Differentiation of (o) and (u)

The nature and extent of this (©) / (u)

merger is illustrated In Table till, which

shows the amount of distinction made between

the two lexical sets by members of the differ¬

ent social classes in different contextual

styles* The amount of this distinction is

calculated by subtracting (o) index ooores

from (u) Index scores* This procedure, however,

cannot be followed when both the (o) and (u)

scores are over 200, since this implies the

usage of some (o)-% and some (u)-4. It will

be remembered that although the rest of the

index scales for (e) and (u) are identical,

they diverge at this point, and (o)-4 m

and (u)-4 m [aij, so that the use of (u)-4

and (o)-4 implies a di f f extent la ti on* In these
casesi therefore, we must subtract 200 from

each score, a-sd sura the results* Thus for

examplei

(u) 185 I (o) 143 « 040

but (u) 283 • (o) 240 - 130

The figures shown in Table XXIX can,

of course, represent only a rough putdo as

to the extent of the raergor of (o) and (u),

since some lexical items, such as do* who*

you* only very rarely hare /at/ as opposed

to /*»*/, so that {©) and (u) scores are never

likely to bo Identical* The table shows,

however, that in FS and CS, where the finjures

are most reliable since the proportions of

(o) and (u) items most closely reflect those

of normal speech, the biggest differentiation

occurs in the two most extreme class groups,

the MF4C and the LVC (and also, in CS,the

LMC)» In the ease of the MC# the largo

differentiation is due to an H,P««type

/•U/ * /at/ distinction, while in the ease

of the LKC it is duo to a maintenance of tho

distinction by usage of more (u)-4 and (o)-»4

forms* The LMC and fWC also have a fairly

high differentiation rate* It is the WC

which has the greatest amount of merged

forms, more particularly in WLS, and which

 25 4

Table XIII

Difference In Irtdcat Points between (o) and (u) Scores

£s 03
££3
Class

WIG 127 124 131 142

LMC 115 113 030 126
mo 037 043 070 083
wc 043 063 101 103
L¥C 040 O63 117 120

eaa be said to be in the vanguard of this merger

as a linguistic change* because of Its re¬

luctance to use (u)-4 and (o)-4 forms and its

inability or unwillingness to use (o)-l or

{©)-2 forms* The general increase of differ¬

entiation as the fermllty of style decreases is

attributable to an increase in the number of

(u)-4 forms.

The Variable (vtt>

This variable is the presence er absence of the glide

iiJ In the lexical set tune, music* nneue, etc, fig* 37

shows that there is definite class differentiation of this

variable* although the IMG and MMQ are undifferentiated*

and the WC has an index considerably lower than the LMC

 255

in CS* The stylistic variation is slight# and must

probably bo disregarded, because the pronunciation of

LJJ ©lid®, where it does occur, depends for its

frequency on the phonetic context and on the actual

lexical item itself* This means tn t while 1?LS and

HPS scores are comparable, since they Include the same

items, FS and 03 scores are not* This perhaps explains

the otherwise strange overlapping of VWO and LMC scores

in CS* Note that there is reliable evidence of stylistic

differentiation between NFS and WLS, and that the LUC,

LMC and tWC groups use 100 per cent {.J J-lesa forme in CS*

The Variable (r)

This variable is the use or non-use of intrusive

/p/« During the course of this work it was discovered

that full indices could not usefully be developed for this

variable, since almost all informants used 100 per cent

intrusive ft/ in speech. Figures were obtained, however,

for WLS and HPS, and are given below in Table XIV* This

shews that the UWC* MNC, and LMC us* most intrusive ft/* e»
with the L¥C at almost the same level* Nothing can be

deduced with respect to stylistic differentiation, since

XLS examples involve only word-internal intrusive /r/,

as in drawing* sawing* Figures for HPS also need some

further discussion! what is clear from a close study of the

details is that all members of all classes,with a very few

 W

©\

VA

& 8
«

•3

© 9

VA
o-a
CO

nagiooo o©©G3
«r

oooN^MVOW»
o
o o V3 M
©
O
H oU
o U

<r
10

§
!

*
[£,
 )

257

Table XV

No* of Intrusive /r/*a r?or .Xnfomarvt by Class

£2L £2.

KMC 0.333 0.000

mc 0.125 0.000
wc 0.562 0.187
mc 1.091 0.313
tMC 1.530 0.750

ConelugAon

This concludes our tuuJLysls of the Nch phonological

variables as far as social class and stylistic differ-*

ontiation arc concerned» The followin.; infomation con¬

cerning the variables lias e^ierjedi

A# 1) The ¥0 is undifferentiated for (er)«

(on) and (t).

2) The LVC and M¥C are undifferentiated

for (at), (h)» (I), (o) and (yu).

5) Th© ITJG and WC are undifferentiated

for (a). (&) and (o)»

 258

4) The EWO, LMC and nm are undiffer¬

entiated far (ir)«

8» The relieving variables are Involved in

linguistic changeI (a), (a), (e), (er),

(«*)t (I)# (if)# («)• (o), and (u)«

C* All the variables are involved in style

and elaae dlffer&ntlation except!

1) ( u)t which is Involved only in

stylistic variation! and

2) (ai)» (au), (I) and (er), which

are involved only in claae vari¬

ation*

S* The following variables are the best

indicators of ety-listie contexti (a)#

(er), (lc), (n%), (o)» (p) and (t).

£« The following variables are the beet

indicators of social elaesI (a), (at),

(an), (ng) and (o).
L". .CTTO r D t T'V, • :i
 259

Chapter Nlnei The Theory of the Diaaystom

The heterogeneity of the Nch speech community has

been illustrated in some detail in Chapter Bight. We

hare also described and discussed some of the rays in

which different groups of Noh speakers are distinguished

from each other linguistically. We have shown that

different social class, sex and age groups have pro¬

nunciations of the phonological variables that differ-*

entlate them from other groups, and that most speakers

have variant realisations of many of the variables in

different social contexts.

The fact remains,, however, that in spite of these

significant differences, all the informants investigated

in the course of this study were distinctively speakers

of the same variety of English! Nch English. This

chapter and the next will in part be taken up with a

discussion of Nch English that will be ae concerned with

the similarities to be found within the Nch speech

community as with the differences. As in Chapters

Seven and Eight, this discussion will be limited to

phonetic and phonological features of Nch English.

The Nch Speech Community

It must of course be recognised that Nch English

is not an isolated, distinct or discrete linguistic

 26a

variety. Geographically, it merges into the other

varieties of Nik, IJAn ami Southern British English.

Socially, it merges with the non-localieed varieties

of R,P. and Standard English, Xt is not, however,

unreasonable to regard Nch. as if it were a distinct

variety. We have already shown, for example, that

there is some evidence of the existeneo of a linguistic

rural-»urban dichotomy in the areas surrounding Nch*

There are, moreover, many linguistic characteristics

which are peculiar to the urban area of Nch., Speakers

of Nch English, that is to say, belong to a particular

speech community which has very many common linguistic

features. Many of these features are not shared by any

other speech community.

In this and the following chapter we shall be

attempting to establish a common framework that will, in

some sense, incorporate all types of Nch English, We

hope to show that these different types are derived from

a single underlying framework that is the same for the

whole community. We shall call this underlying framework,

(together with the rules which produce the different types

and relate them to each other), the Nch diasvstem, We

believe that this diasystem can be most meaningfully

constructed within the theoretical framework of generative

phonology. By this means we ©an draw on the great in*

sights into the nature of the linguistic competence of

 261

the native speaker which are provided by the theory of

transformational generative grammar, The diasystem,

in other words» should provide a model of the linguistic

competence of the individual aa a native speaker of Neh

English and as a member of the Nch speech community*

Ve shall also attempt to show that a taxonomio phonemic

diasystem is an unilluminating construct, and that it is

impossible to devise such a diasystora for an urban area

liks Nch,

The Reality of the Dlagyaten

There are two main justifications for attempting to

establish this kind of diasystem. The first Is that a

model of this nature provides a common basis for the

description, comparison and typology of mutually compre¬

hensible dialects* The second is that there is some

sense in which members of, in this case, the Nch speech

community can be said to have internalised this dlasystera,

in that they are able to draw upon it in their variable

speech production, and to use it in the comprehension and

social interpretation of utterances produced by other

speakers.'"' We therefore believe that there is a noed to

establish and investigate some kind of model of this

2
internalised diasystem* The diasystem, it appeal's, can

be said in some way to exist, and must therefore be

described in order correctly and fully to describe the

 262

linguistic competence ef the native Nch speaker*

There seeria to be in principle no reason why a

diasystcni of this type should not be expanded to in¬

clude all mutually intelligible varieties of English*

We shall here* however* confine ourselves to the Nch

material*

A New York City Plaaystea

If we aooopt that* in order to describe correctly

all the possibilities Inherent in a particular urban

speech variety* it is necessary to set up this type of

dlasystem* then it is clear that this Is one point where

criticism can be levelled at LaboVs major work in urban

*5
dialectology* In his work on New York City English,

Labov falls to discuss explicitly the common underlying

structure of New York City speech. In the main he

illustrates and discusses only the differences within

the community* (although he does hint at a common under¬

lying framework in his discussion of the fact that the

New York City speech community lias a set of shared

linguistic noma)* This failure is particularly evident

la his treatment of New York City vowels* Labov analyses

these as comprising a series of discrete social class and

ethnic group vowel systems* He explains, for example,

that the merger or non-merger of the lexical sots of

 263

here and hair la evident only fro® variance analysis*

There is, in other wards, only & statistical tendency

in the speech of some individuals to pronounce items

from these lexical sets with identical vowelst the

distinction between the two seta, and between speakers

who have the merger and those who do not, is a more ~ or -

less distinction, rather than an either - or distinction*

Labov nevertheless considers it legitimate to set up

two separate and distinct vowel systemst the one com-

pletely withont the merger, and the other completely

with* This gross distortion of the facts Is of course

the result of an attempt to force speech material from a

heterogeneous speech community into the taxonomic phon¬

emic straitjacket*

The Taxonoialc Phonemic Piasystera

The term diasystem was first used by Uriel Weinrelch

h
in his article "Is a Structural Dialectology Possible?*

This article is in the main an attempt to bring about a

rapprochement between the methods of structural linguistics

and dialectology* In the course of this article, Wein-

reich develops two important points* The first is the

introduction of the concept of the diasystem itself* The

second is his Insistence that dialectoXogists should

compare forms from different dialects not Just as "the

8at»®" or "different", but as constituent parts of thoir

 26k

ova systems* In other words, at the phonological level,

phonetic similarity should take second place to phonemic

similarity and the overall comparison of phones!o

systems* ffeiurelch*s diasystera is therefore a combined

system based on phonemic correspondences* It illustrates

the importance for dialect comparison of the similarities

to be found in the systems of different dialectsI "a

diasystem can be constructed by the linguistic analyst

out of any two systems which have partial similarities1**

The two aystens of the varieties under study can be

compared with respect to (a) their phoneme inventories*

and (b) their phoneme "distribution"• It should b©
noted at this point that there is a certain amount of

confusion in the literature concerning the term distrib¬

ution* Veinroich* for example, uses it to signify the

distribution of phonemes over lexical items or "lexical

■§
correspondences"* Other linguists, such as De Camp,'

nee the terra to refer-to permitted phoneme sequences in

a dialect* The two are of course inter-related* For

example, in some dialects of English, /©/ is not per¬

mitted before /r/# with the result that certain lexical

Items have /e/ in these dialects corresponding to /o/ In
others* To avoid confusion, we shall restrict the tens

distribution to the latter sense, and discuss the former

under the heading of incidence*

To exemplify, we can say that a comparison of tho

 265

vowel systems of two di&locts, A and B, with respect

to their phoneme inventories, might yield a (partial)

dlasystem of the following typej

differences of incidence, however, "cannot be directly

inferred from a comparison of the differences in

inventory, although the two ordinarily stand in a definite

6
historical relationship"« Xt is the stress on * partial

similarities" underlined by the notation // ^ a » © «***//

which quite clearly differentiates the dlasystem from the

The overall system is merely a sun of all the phonemic

distinctions made in certain English dialects, with each

dialect having In its own phonemic system only a subset

of the maximal set of phonemes* Xt has little value

except as an aid to "broad transcription"•

Xt is interesting to note, with reference to the

claim made above for the "reality* of the diasystem,

that Welnrelch himself dees not regard the diasystera as

a mere artificial construct* "A diasystexa is experienced

in a very real way by bilingual (including *bldialectal*)

speakers, and corresponds to what students of language

contact have called 'merged system**•

 266

Criticisms of the Taxonoroic Phonemic Dlasyatem

There are several very Important drawbacks to

Kelnreich*s theory of the diaaystem. These stem

very largely frota the theory of taxonomic phonemics

itself* and mean that the establishment of a taxonoraic

phonemic diasystem (particularly for the different

types of one urban variety) is at best unrevealing and

very difficult* and at worst impossible*

1# The first* relatively minor drawback is that the

type of diasystem proposed by Weinreieh is totally

uninteresting for the comparison of dialects which

have identical phonemic systems. These dialects may

be phonetically very different and be quite dissimilar

with respect to incidence* but these facts will not

be evident from the diasystem itself* A solution to

the problem of phonetic differentiation was proposed

by Cochrane,® who suggested that a diasystem should be

established at two levels* the dlaphonemic and the

diaphonic* The dlaphonic level of analysis would

compare the physical realisations of otherwise Identical

phonemes. (The problem of incidence is dealt with

under 2. below.)

2. Zt is clear from the hypothetical diasystem

illustrated above that the two ooisnon dlaphoneeies

 267

//a«=?o// have couch In common with the unit

described by Daniel Jones as the "diaphone". Jones

sayst "It is convenient to have a name for a family

of sounds consisting of the sound used by one speaker

in a particular set of words (said in isolation)

together with corresponding though different sounds

used in them by other speakers of the same language*

o
Such a family may be termed a fdiaphone**"^

The phrase that we need to discuss here is "in

a particular set of words0* Weinroxcii states* as

we have already seen* that the incidence of phonemes

in a particular dialect cannot be inferred from the

diasystem* But it is obvious that* before a diaphone

can be established* the particular set of words in

which it occurs must be taken into consideration* What

then must we do about lexical correspondences* the

incidence of phonemes in cognate lexical iteros?

Obviously* two dialects may have identical phoneme

inventories, and yet be very different because of

incidence* And even if Inventories are not identical*

some indication of lexical correspondences must be

given in order adequately to represent the relationship

between the two dialects in question*

One attempt at a solution to thie problem ie

10
provided by Cochrane* He proposes that the degree
 268

of relatednoss of two varieties can be indicated by

taking lexical correspondences into consideration in the

following way. If every instance of a phoneme A /P/

corresponds in the cognate morpheme to an instance of the

phoneme b /p/» this is termed a "reliable simple correa-

pondence" and indicates a close relationship, at this

level, between the two varieties. If, on the other hand,

instances of a phoneme A /p/ correspond to instances of

phonemes B /P» P* , P"_ / so that the particular B

phoneme la eontextually predictable, this is a fully-

reliable complex correspondence. Tfhere this type of

correspondence is not contextually predictable, tho degree

of reliability of the correspondence must be indicated,

Cochrane says, by some kind of count* If A /?*/ corres¬

ponds to 8 /P/ and 8 /P/ in the ratio of four to one,

this can be represented ast

A /P/ « kB /P/

IB /P'/

and the correspondence is "89 per cent reliable". Ex¬

actly how or why this is to be done Cochrane does not

say* It does of course indicate to some extent the de¬

gree of reletedness of two dialects, but it tells us no¬

thing at all about the nature of the relationship. Neither

does It Indicate which particular phonemes occur where,

or why. The most important drawback, however, is that

Cochrane has first of all to establish the diasysteraie

 269

inventory, and only subsequently can he begin to in¬

vestigate lexical correspondences. Clearly, however,

lexical correspondences must be taken into account

before tho diasysteia can bo satisfactorily established.

If this were not done, then it would be possible to set

up a diasystem containing two totally unrelated languages

which Just happened to have identical phonemic systems.

The same also applies, however, to comparisons of

olosely related dialects of the same language. For

example, the fact that London English and some types of

North Country English have vowels of the type LaJ does

net mean that we can establish a valid diasyateia for the

two varieties containing the diaphonemio element

L, NC //£=* a//, since this vowel occurs la London in

Items of tho lexical set but, hut, and in the lexical

set bat, hat in the North. To treat //a// as a valid

common element would give a totally false impression of

the similarity of the two dialects. Similarly, a

comparison of two particular Noh Idiolects within the

Weinrelch framework would lead to a partial dlasystem

of the following typei

1 / ii

This, however, merely states that where idiolect 1 has

three long front vowels, idiolect 2 has only two. This

is of no particular interest, txrnl, as vioinreich says,

 270

aothlag ooaeerains incidence can t»* iniemu from

thil diasysteaio Inventory. Moreover, it Is not oven

the ease hero, as Volarolch suggests, that inoidenoe

and inventory can bo aeon to stand in a definite histor¬

ical relationship. It onto with 1 /si/ do not correspond

to cognate items with either 2 /ii/ or 2 /e »/, hat

rather to 1, 2 //tei//» A diaeyetoa of the foil owing-

type would therefore he much more significant and

realistic*

1 /•* mi/
li^ci VAlai-si

1, 2 // 2 /mi/

ife must, therefore, take lexical correspondences

into account when establishing the diasyetom, It can

also be shown, however, that absurdity can equally well

result when lexical correspondences are taken into

11
account* For example, Moulton, In hie study of Swiss

Gorman dialects, develops a method for taking lexical

correspondences into account and for stating thorn more

explicitly than, for instance, Cochrane does. This

method revolves around the fact that a study of the

historical development of modern Swiss German vowel

systems permits the diagramming of these systems together

with the historical sources ef each vowel. Moulton sayst

"This at the some time gives an Indication of the lexical

correspondences from one [modemj system to another*•

 271

If we portray the Middle High Oeriian s:hort vowel system

thus t

migl

u,

n
e 0,
A

M
a, a,

we can show the historical sources for tho short vowels

of a particular modern Swiss German dialect in tho follow¬

ing way i

h h "x

®1,2 Sl,2 ®1,2

*3,4 *2 °2
a4

Tliis type of diagram permits a diachronlo comparison of

the evolution of modern system, as well as a synchronic

comparison of different modern systems and the lexical

correspondences that obtain between them#

In the case of two particular Swiss German dialects,

Moulton shows that the Uiasyaten which results from this

type of approach is as followsj

'<■»
/io~9l~'2~''3y„Bu /»2~o2~S2---/
LU ,AT'll AP /i.~x «J~ V '*2a\~Z h--/
 272

This indicates that the two sjratens have only* one

completely shared diaphoneme in common* The absurdity

lies in the fact that these two dlaleots are spoken

only fifty miles apart and are completely mutually

intelligible*

Even by using this system* moreover* only a rough

indication of lexioal correspondences can be given* as

Moulton admitst "Since

/*2/ ^2/ ar® here for
both {.dialectsj* these would seem at first glance to
constitute the fully shared diaphoneraes //Og Sg//* This
is not the ease* however* Whereas all the lexical iteras

Identified by the subscript /2/ belong to hV $

only part of them belong to AP /Og Sg/, the rest

belonging to AP /. S4 2/."

Moulton concludes that, "if in constructing a dia—

system of two or more varieties ef n language we disregard

lexical correspondences, then we are treating these

varieties as if they were totally unrelated to one another*•

(And in the case of these two Swiss German dialects*

Welnreich*» method produces a system totally devoid of

interest which over-emphasises the similarity of the two

dialects t LU, AP //ite )

Alternatively* "if •*• wo treat tuem as related and hence

take lexical correspondences into consideration* then

the usefulness of the diasystem in dialectology becomes

12
questlon&ble*» in the words of Pulgrnm, wa diasystem ««•
 273

that takes into account certain conditions which historical

linguists* dialectologista* and the speakers will regard

as indispensable > and that then shows as little agree*

stent between closely related dialects as to make thera

seats foreign to one another* distorts the facts no less

than does LaJ dlasystent •.. which makes the dialects

appear identical0*

It should also be pointed out that Maulton's system

shares with ¥etnreich*s the defect that it fails to show

exactly which lexical items have a particular phoneme*

It also fails to show the degree to which a phoneme is

descended from one historical source as compared to

another*

It seems clear then that* while the problem of

phonetic differentiation within the taxonomic phonemic

diasystem can be overcome* the problem of incidence

cannot* Both taking lexical correspondences into con¬

sideration and omitting them altogether result in gross

distortions of the facts* there therefore seems to be

little merit in Pulgrara* s suggestion that a diasystem

should consist of a combination of both types of approach*

3* the third major drawback is that the taxonoraic

(autonomous) phoneme is not a suitable unit for dialect

study or dialect comparison, this is already clear from

the points discussed under 1. ami 2* above> but there are

 »4 o *t « © © »4 4 iff »► 3© « iT 4« m< 2 • a

a I »4 1► 0 6■ $ M « ,i « a « 4©C 2 m s> a *4 2 a • 34 !X e «©e

s « a *« 0 k e 3" »*• 0 0< © •« 0 « 20 J »> 0§ O 31

« • «v I

0 »4 a 1 © 44 34 « i i ►«

1© €o w i »4 © ® & f H ^> 4* e- 4 © »4 4 5 a «• m © t0 2
.

«
cH I U%# © * a I »tV »4 ^© ef 0 * • ee «so
 273

"Structural dialectology*, and, in particular,

the use of the taxonomic phoneme within dialectology,

would therefore appear to be impossible* This is, of

course, the strict structuralist view which Veinreich

was presumably trying to overcome* However, in view of

the failures that we are able to point to in Veirireich's

theory, this view would seem to be the correct one*

4* As an extension and elaboration of the points mads

under 3*, we can show that the most striking failure of

Velnreich*s inventory (and distribution) taxonomle

phonemlo diasysteraio model is provided by urban dialect¬

ology* This is because linguistic variation and the

heterogeneity of speech communities cannot be ignored in

the study of the speech forms of urban areas, (although

* 1&
some linguists have made valiant attempts to do so)*

(i) Because he is working within the taxonomio

phonemio theoretical framework, Weinrelch#s

diasystem inevitably involves the comparison

of two dialects as discrete systems* This

has no unfortunate results if the two dialects

are quite widely different, as in the case of

•
(T

Cochranc*s J comparison of Yorkshire and

Australian dialects (although the usefulness

of such a comparison is open to question)*
It is, on the other hand, extremely unhelpful

in the treatment of closely related dialeots,

 27$

and in particular for dealing with the type

of situation tliat obtains in Nfch and other

large urban areas* Here* the different types

of speech variation must be regarded as

representing cllnoa. The setting up of dis¬

crete varieties is therefore a gross distortion

of the facts* For this reason* it is not

possible to divide Nch English into several

discrete types and allot each of these types a

separate line in the dlasystem* At the same time*

there is too much variation and differentiation!

as we have shown in Chapter Eight, to permit

us to dispense with some kind of dlasystem

altogether* and to regard Nch English as one

single type*

(11) The taxonomie phonemic approach is also

unable to handle the wide range of variation

that occurs in urban areas within groups and,

more particularly* within the speech of in¬

dividuals* Attempts to fit this kind of vari¬

ation into a taxonoraie framework can lead to

some absurd results* Consider for example

16
Slvertsen*s work on London English* Slvertsen

manages to avoid most of the pitfalls in the

way of the taxonomie urban dialeetologlst by

eliminating a major factor contributing to the

speech community's heterogeneity! she uses only

 277

a very small unrepresentative snraple of

speakers* She is still, however, faced with

problems like the following* There is a

series of stylistic alternations in London

English of the type* better [^betaj l*be2.a],

where [tj and £2] are both considered to be
allophonea of the taxonotaic phoneme /t/«

There is also, however, another series of

alternations involving the replacement by

1,2 j of the taxonomlo phonemes and phoneme

clusters* /k/» /kt/, /p/t /pt/, /g/ and /d/*
Because of her unwillingness to recognise

that London English is not a single discrete

system and her reluctance to abandon the strict

principles of taxonomio phonemics» Sivertsen

is forced to adopt absurd and counterintuitive

phonemic representations like the following*

big /bit/ directly /dretlij/

garden /gahtan/ couple /kital/
reckon /retan/ kept /ket/

According to this interpretation, stylistio

variation la a question of incidence, rather

than of variant (and overlapping) realisations

of common underlying units.

(Hi) Further examples of similar difficulties

 278

with the taxoriOBilc approach can b® found In

I a
be Camp*® work on >an Francisco English*

be Gamp is at pains to stress* more than onoe,

tliat "rigid struoturaliat procedure" mast be

abandoned in order satisfactorily to handle

the urban speech material* The large amount

of variation he finds in San Franeiaoan speech

also leads him to postulate the existence of

two separate taxonoraic phonemic systemst not

Just within the community* but within the

speech of individuals* He is* in addition*

led to discuss features which do not fit into

these systems as compromise pronunciations*

be Camp states that low front vowel

phones in the area of |,a] were recorded

fro iuently for some informants and more rarely

for others in the lexical set of calve, rather*

pass* aunt* but that no single informant used

low vowels in all the items in this set*

"Rigid structuralist procedure would require

recognition of 3 separate phonemes* /«/* /a/

and /«/•" But* be Camp says* i.antj and ^kafJ

for aunt and calf are ** inter~dialectal loan

words and, like many foreign loan words* are

at variance with the phonology of tine borrow-

19
ing dialect # Fries Pike liave shown that

two or more differing phonemic systems can

 279

eoezist in the speech of one individual«

Therefore* "for some San Franciscans one must

recognise the coexistence of San Francisco

and other phonemio systems* in this case

Eastern Nov England *«•*

In our view, aunt and calf, however

pronounced* can hardly be said to be loan

words* (Loan pronnnoiatlons might be a more

appropriate term*) This type of phenomenon*

moreover* can be handled quite easily within

a generative phonological framework* as we

shall attempt to show in the next chapter*

This is also true of Bo Caiap*s " compromise*

phones* particularly when* as In this case,

the compromise phone is identical with one

of the phonemes in the secondary (Mew England)

coexistent phonemio system* (X>e Camp ex¬

plains j.aj in aunt as an " inter-dialectal"

borrowing, but ^aj in Nevada and ra these* as

a * compromise articulation"1 between f'mf and

/o/m) It is clearly, on the other hand* a

difficult matter to fit this kind of material

into any taxonomio framework* however much

we relax "rigid structuralist procedure*.

A further argument in favour of using

the generative phonological approach is

 230

provided by Do Camp's difficulty la handling

phoneme distribution and Incidence la San

Francisco* "Some Informants are also in*

consistent In the distribution of phonemes*

Again rigid structuralist procedure demands

a clearcut decision! Does /o/» for example,

ever occur before /r/ or does it not?* The

answer* of course, is that it does sometimes*

but be Camp has no satisfactory way of

stating this*

He also points cut that the hoarsei

horse distinction has a "psychological

reality* for some speakers, in that they

insist that such pairs are not comparable to

pairs such as salet sail, but do not them¬

selves make any distinction* be Camp has

no adequate way of explaining this, but it

would seem that there are two distinct

underlying representations for these two

items, and that the two systematic vowel

phonemes in question have optionally ident¬

ical realisations before /r/» Thia fact can

be adequately expressed only within the

theoretical framework of generative phonology.

(iv) The impossibility of using the taxonomie

phoneme in urban dialect study is further
 281

underlined by a brief* examination of the type

of approach that would have to be adopted in

setting up this type of diaaystem for Nch.

The firet problem we are faced with ie

which of the contextual etylee to take as

basic* Ve have eeen( for example* that some

troupe of Nch speakers do not distinguish

feari fair in CS» but do make the distinction

in VLB. Now taxonoraic phonemes are usually

established on the basis* amongst other

things* of minimal pairs* These can only be

elicited in a style of the MLS type* If we

were to do this for Nch English we would be

able to set up two distinct taxonomio phonemes

/Zd/ and /ca/* but w® would fail completely

to characterise the normal everyday speech

of many Nch people* Me could partly resolve

this problem by setting up a partial stylistic

dl&systera like the followingt

This* however, entirely obscures the fact that

the distinction feari fair in Nch la a more -

or * less distinction, rather than an all - or -

none distinction. It implies that some speakers

 232

simply have the distinction, and that others

simply do not# Or, if we regard the diasystera

as purely stylistic, that a speaker switches

over from one discrete system which has the

distinction to another that does not# It is

entirely incapable of indicating the nature of

the stylistic variation that was illustrated

in Chapter Eight*

A second problem occurs in the ease of

phonological variables like (ir)« Xt Is clear

from the material presented in Chapters Seven

and Eight that a linguist working in a tuxo-

nomio phonemic framework would have to regard

realisations of (Ir) variously as instances of

/3t/# /»»/# /a/ and /«/* It is also clear

that there would be many doubtful cases and

"compromise* pronunciations. Variation in the

pronunciation of this vowel would have to be

regarded as a case of multivariate phoneme

Incidence in "cognates**

A third major problem is that if ws

wished to set up a taxonomio phonemic diasystem

for Nch English, we would have to account for

the fact that only some Nch speakers make the

following distinctions, an > no speaker makes

all ef themt
 283

fear » fair boot t Bute hat i hurt

name t nail boat i foot heart t hat

face $ fierce boot » foot oar j are

prare » purr pore i poor fire t far

dew t do pore i paw box i bartea

so t sew heart t hart player i plan

boot i boat hat i hart (gwer i mob

(Bach lexical Item stands for a lexical set*)

There are also some speakers who do not

distinguish the vowels of the following pairsi

fired 1 c€>d player t half

back t bake

There is* moreover* a certain tendency to

merge the vowels of hid and head* and of set

and sat* In addition to this* many younger

speakers have a series of long pure vowels *

which older speakers have only in part* of the

following type»

Is 1 i ®S£t air* skier» see it

L*«J Player* play it

Ia*J are

13*i bird* pure, brewer* knew it

La* J Poor* pore* paw, going* employer, go it,

enjoy It
 234

L®» J lower, know it

L«ij fire, higher, our, allowance, try It,

allow it

Ve therefore have a maximal set of Nch vowel

phonemest /X ~ *— «—U a— p ~ it ci—

Ai — oi —*i ui — «m -~ou — tt — ast •— at —

3t ~ »t ~ bi ~ et ^ ^ Is ~U9 ~ oa/«
No single speaker has all these "phonemes"» so

they clearly constitute a Trager-Smith type of

"overall system", which can tell us nothing

about the various inter-relationships between

the different phonemes*

Even if we attempt to set up a dlasystem

which does not take lexical correspondences

into account, the number of different discrete

systems we would have to establish in order to

incorporate all these contrasts would be very

large* Of the above twenty-six phonemes, only

20
the following ten constitute ths "common core"

Of Nch English* /X^-e — m — U — a i» ~ Ai-~-'

oi -~«u ^aj/. It is certain, however, that we

must make some attempt to indicate lexical

correspondences and to illustrate the relation¬

ships that obtain between phonemes. It is an

important fact, for example, that /la/ and /e*/

can only occur where /t*/ does not occur. The
 235

dlaeystem* In addition to the above * common

core"| will therefore have to contain the

following diaeystemic blockst

A. 1 /» «* — ri — ©» — la — «a /
2 /a* a»i —' ti/
■
'—'
r i —

2 /m —
HP! R?1 —''X»~ e5/
fm ~
n — rl — « i/

/
'
J /*
-
e*l '—' ei ^ la —

6 /* —
«?1 -— o» € %/

1 /® —
ri ■—' la ~ **/
3 ~
ri —'

B. (*) ~ ~ ni ~ t»n/m
(il) /u — at — »a/
(iii> /h — n» — w/
(iv) /ti- on/

c# (p\ / ot- oa -
~
txa/
(a) /o» ~
na/
fr) /o»- oa/
f») />*/

0< I /A — 3» ~ at <~ at
"
0 *" t»%/
ir /a ~ at — ai — a —
'1>«/
XII /a ~ 3* ~ at — 0 — >»/
IV /A ~ 3i ~ at — at ~ 0/
-

V /a — a* — u —
>•/
VI /a ~ Rl ~ at — V

vn /a — 31 — at — a/
VIII /A — at —
 286

(This, of course, con only bo done If we are

prepared to pretend that there are discrete

varieties of Nch English*) A diasystem of

this type Is absurd for several reasonst

1) Even the very large* complex and

unwieldy diasystem that would result

frost the conflation of theso blocks

and ths common core would not be

correct* This is beeauss we have

not indicated above the alternation

of /m/ with /a/* /at/ and /3»/ in

items suoh as church* first* This

means that to obtain the "correct"

diasystem we will have to merge

Block A with Block D. This will

lead to an even larger* store un¬

wieldy and more unrealistic diasystem.

2) There is no way at all of indicating

lexical correspondences exactly* If

we were to attempt to do this by

constructing the type of diasystem

proposed by Moulton, even more com¬

plications would arise* In Block B*

for example* wo would have to dis¬

tinguish between /IT«— at -—' ut ou/

where /U/ includes FAn short o itsms
 287

and the same system where It does not*

3) The different "systems* Indicated in

the different lines of the above dia-

eyeten are not* (as we have already seen)*

distinct systems or varieties employed

by different speakers or groups of

speakers* They are merely an indication

of the contractive possibilities open

to Noh speakers* The above system

would appoar to indicate that speakers

could switch from one system or partial

system to another at will* However*

this is not at all consonant with the

facts of linguistic variation demonstrated

in Chapter Eight* The separate systems

are in fact totally artificial constructs

of no validity* They cannot be "dis¬

covered* in the speech of any individual*

There is no correspondence between any

particular line in one of the ?<bove

blocks and any other line in any of the

other blocks*

When we add to the above drawbacks the fact

that a dlasystem of this type is totally incapable

of indicating or expressing the class and style

differentiation Illustrated in Chapter Eight* it

is clear that the taxonomlc phonemic diasystem

 \

233

must be rejected for the purposes of describing

Nch English and the linguistio competence of

the native Nch speaker*

The generative Phonological Plaayatem

We have already indicated that a theory of phonology

which permits a distinction between a deep systematic

phonemic level and a surface systematic phonetic level

is much more suitable for the description of many of the

Nch phonological variables* The facts cf linguistic

variation suggest, for example, that all Nch speakers

have different systematic phonemic representations of the

lexical sets of fear i fair* but that theee are frequently

realised at the systematic phonetic level with identical

articulations, particularly in C3*

Di&lectologlc&l work that has already been carried

out within the generative phonological framework confirms

that the generative phonological model is particularly

21
suitable and valid for this type of work. Keys or, for

example, has shown that some of the American dialect

22
data recorded by Kurath A McDavid involving dialectal
dlfferentiation is the result of differential rule

ordering* Both the dialects in question have the same

set of relevant phonological rules, but the fact that

the rules are ordered differently results in surface

outputs that are significantly distinct* Saporta^^ has

 239

also shown that different dialects may have the same

underlying systematic phonemes but be differentiated by

rule ordering, rule addition, or by rule simplification*

In a major study that is more relevant to tho

24
present material, Vasiliu has demonstrated that Daco-

Rumanian dialects can be derived from a common set of

phonological entries in the lexicon by rules which are

differently ordered for different dialects or which

apply only to a subset of the dialects. The generative

model, Vasiliu shows, also allows for an accurate

typology of dialects which adequately describes the

relationships between them and which cannot be made on

the basis of the taxonomie inventory and distribution

model.2-* Thus, for example, Muntenian Daco-Ruraanian

dialects are distinguished from Moldavian dialects in

that, although both types have tho following morpheme

structure rulest

1) After /!*, a*/ only /e, 1/ oecur and

not /s% i/,

2) After /s, a/, /e, 1/ become /d» I/.
3) /a*, »*/ become /», a/»

Muntenian dialects have the rule order 1 - 2 - 3 - Q

(where represents a further unspecified block of rules)

and Moldavian dialects have the rule order 1 - Q - 2 - 3.

The Moldavian dialects can be further sub-divided into

 290

two types, X and T, The type X dialects have the

morpheme structure rule h) which ether Moldavian

dialects do not havet

*> /*/ /♦•/

/d/ /d»/ /. front

vowel

/«/ /»«/

Type X dialects themselves can now he split up Into two

further sub—types• Sub-type 1 has the two phonological

rules 3) and 6)t

5) ♦ Obatr

♦ Stop ♦ Acute
>
— Acute ♦ Strid

+ Sharp

6) ~e Obatr~

♦ stop

-
strid ^ j~- ActtteJ
- Nasal

♦ Sharp

In dialect A of sub-type X the two rules appear in

erectly this form* In dialeot B of sub-type X they occur

in the conflated form of 3a)»

 291

5a) ♦ obstr
♦ Acute
4»
Nasal »
♦ Strid
— -J
♦ Sharp

se that instead of operating consecutively they "apply

simultaneously*« Sub-type IX* on the other hand# which

coneiota only of dialect C» has neither 3) nor 6)« hut

a different rule 7)*

7) /**/ /a/
/d*/ /s/

The presence or absence of the above roles therefore

gives us ths following typological classification of

Daeo-Huiaanian dialects i

Daeo*RunMiian

Moldavian Huntenian

II

A taxonomic phonemic approach* on the other hand* is

onabis to provide this type of classification and is

 292

incapable of expressing the true relationships that

obtain between the different dialects* Dialects A and

C, for example* hare the same taxonomic phoneme in¬

ventory and distribution* They are* in other words*

identical. Dialect B lacks the taxonomlo phonemes /k*/

and /g*/ which are present in both A and C* and must

therefore be eoneidered to be different. The resulting

elassifioation isi

It is only if we compare A and C with another dialect*

for example the literary standard* that we realise that

they are net identical - that their phoneme incidence Is

different* A comparison of inventories and distributions

falls to reveal that A ami s are more closely related

than A and C* and la incapable ef expressing the hier-

archieal nature ef the inter-relationships which the

generative typology produces. This can only be done by

means of a "correspondence table*1 established with

reference to another dialect* and this is in fact "no¬

thing but a non-formal statement of (.the above J rules".

Further examples of the way in which a generative

 293

phonological approach to dialectology can ho much more

interesting* revealing ami Intuitively correct than

24$
the taxonoraic approach are given by King. Drawing on
27
an example iron Lamb, * King states that in one dialect

of Monachl, there ie an alternation between (.»] end ^vj

such that the former never occurs after vowels and the

latter only after vowels, *»g»t

{.miyawaij "will go*

^ tawiyawainaJ "onr future etotms*

In another dialect of the earn© language the equivalent

alternation is between [s»J and fcwJ* In addition there

is in both dialects a phone Iwj which does not altemato
with Loj but rather contrasts with it. Using the gener¬
ative phonological approach we can simply state that

there are two underlying systematic phonemes that are the

same for both dialectal /&/ and /w/« For the first dia¬

lect there is a rulei

1) /*/ ► l»J /*

For the second dialect there is a rulet

2a) /a/ ■ ■ ■ ■) |wj /V

or, alternatively, as King has it, a rule which states

that glides are ncn nasalt

 Z9h

2b) * TOO
* l-
* eons

This mult* In the same output t

i) M > t«J /*
2b) 15} » l«]

According to Kind's Interpretation, tho difference

between tho tvo dialeota lies solely In the feet that

dialect IX has an additional rule* The result la dialect

XX la that in certain contexts the two separate under*

lying phonemes hare tho identical realisation iwj»

Tho roleyant part of a taxonoraie phonemic dla*

system for theoo too dialects would be //a w// "which

reveals oxaotly nothing of tho way in which tho dialects

differ*,2** In dialect I* tho [sji i.wj alternation must

ho considered to ho allophonio* whereas in dialect XI

the alternation is morphophonemics Tho two dialects

therefor# differ, according to this approach, at two

different levels, "This example demonstrates that

meaningful dialect comparison does not necessarily emerge

from comparing the phonemic inventories of two dialsets,

for In this case tho phonemic inventories are ident*

leal, and whatever differences there are arise via rules

that state realisations ef phonemes. ... To gain any in*

sight into dialect differences we must concern ourselves

with the /arammaro of languages, not their vowel or

 295

consonant systems .** The study of dialect differences

29
is the study of how the grammars of dialects differ*"

0*Keil#*^ in his work on the dialects of Faroese*

provides further exemplification of this type of

approach in dialectology* 0*Heil*s method is character*

ised by the fact that he takes one particular dialect as

basic* and derives ail others from this basic dialect by

various seta of rules* (This type of approach can also

be found in Kliiaa*a-'1 study of syntactic dlfferontiation.)

0*Neil states that! "We begin by proseating the matrix

description of Southern Streyraoy Faroeee because the

other dialects can be most simply presented in the form

of rules that show in what ways they depart from the

Torshavn system"* This method has been criticised by

Thomas* who has shown that it cannot work in the case

of certain Welsh dialects* since one dialect has to be

taken as basic for one part of the system* and another

for other parts. Thomas shows that North Welsh is in

many ways more conservative than South Welsh and is for

this reason well suited to provide a basic form from

which South Welsh can be derived by additional rules*

However* the following set of correspondences (together

with others) has to be taken into consideration!

North Welsh South Welsh

/so in/ "mention" /sotn/

/ant/ "zoo" /sut/

 296

Here there is no way at oil of predicting which North

Welsh /s/ forma will correspond to South Welsh /s/ and

which to /is/* The only possible way of relating the two

dialects is in this case to take the South Welsh forms as

basic* The different dialect forms can then be generated

in the following wayt

1) NW, SW /soin/
SW /cut/

2) m /ant/

Am it is therefore impossible to take either North Welsh

or South Welsh as basic* Thomas, like Vaslliu, generates

both dialects from one underlying source that is the same

in both eases* Since this common underlying source is

not in its entirety characteristic of any actually

occurring dialect, it must be considered to be abstract*

"The generative device is clearly most powerful in

comparative statement when it has an arsenal of abstract

underlying forms, from which all dialect forms are do*

rived through a set of ordered rules,"

We have seen that one of the most important failures

of the taxonomio phonemic diasystem is its inability to

handle problems of incidence* Thomas points out that the

generative statement Is able "to formalise the type of

dialect difference which is not covered by an inventory-*

 297

distribution statement « the incidence of a phoneme in

parallel lexical oetsi where two dialects share the same

phoneme inventory, but select different phonemes in

specific sets of words*• The following dialect differ*

enoes in Velsh, for Instancet

£JL m D2

koiid kuied koid "trees*

ma i in main main "stone*

so in soin

*
puter puier power*

can be generated by the following set of rules acting cm

common abstract underlying phonemes t

1) 01
Dx / D2, Dj koiid

AS /(.»«*] Dj / D2, D3 rial in

2) Oli Luisj kuied D,

3) vxii > [^ij I>2» »3 kotd

02
mam
£>2, Dj

The generative phonological d la system la therefore la

many ways much more powerful and real 1stlo than the

taxonomlo phonemic dlasystem#

 298

In the follovlns chapter ve shall attempt to

establish a generative phonological diasystem for Nch«

The work already performed in this field within the

framework of generative grammar shows conclusively that

an approach of this kind is to a vary great extent more

suitable for this sort of work* In contrast* the

taxonomie phonemic diasystem is an unwieldy and even

impossible device* particularly for the treatment of the

speech varieties ef urban areas* A generative phono¬

logical diasystem for Nch English will more accurately

model the linguistic competence of the Nch spsalter* and

will provide a more realistic description of English

as it is spoken in the oity of Neh.

 299

Chapter Ten* The Norwich Dlasystcra

This chapter is an attempt to present the abstract

phonological nysiem and the phonological roles vhlch

underlie all forms of Nch English. This underlying

system, together with the rules which generate the

different types of Rch English and relate them to each

other, constitute the Heh diasystem.

Our attempt to develop the Nch diasystera is sot

within a generative phonological framework, for reasons

given in the previous chapter, and involves the recognition

of four phonological and phonetic levels«

1) The physical-phonetic level

2) The articulatory systematic phonetic

level

3) The auditory systematic phonetic level

k} The systematic phonemic level

In this chapter we shall be dealing mainly with 2) and %),

and shall make some referenoe to 3). We shall also be

introducing an additional articulatory phonetic level,

which we ehall call the performance Y>honetic level,

Distinctive Features

For several reasons, we shall not be working within

 300

the distinctive feature framework moat frequently

associated with generative phonology. It has been

shown* notably by Fudge,* that this type of approach

has many drawbacks and inadequacies* There are* more*

over* many arguments for Introducing feature# at the

systematic phonemic level which are not quasi-phonetic

or phonetically orientated, as ars the more usual

distinctive features associated with, for example,

Jakobson, Chomsky and Halle, but are instead completely

abstract* Some of these argument# are presented, very

briefly, below*

Arguments Against Distinctive Features

2
1) Fudge has demonstrated that the facts of Tswana

phonology can be handled by means of tne distinctive

feature approach only in a way that is quite clearly

arbitrary* There re several alternative ways of allo¬

cating distinctive features, and each of the alternatives

is equally unmotivated* Xn addition, all the alternatives

fail to reveal the underlying structure of the language,

and may even obscure it* Difficulties such as this,

which have parallels in other languages, do not arise,

Fudge claims, in the abstract feature framework proposed

by him*

Fudge's findings concerning Tswana coincide quite

 301

veil with the type of pietore that emerges from the

discussion of claasificatory distinctive features made

n
by Harms* In particular* the two conflicting analyses

of Finnish vowels presented there stress the arbitrary

nature of the distinctive feature approach. Other

factors* sttoh as the lack of agreement as to the number*

typo and identity of the distinctive features that a

linguistic theory requires* underline the difficulties

that occur in much of the work in this approach.

2) Fudge claims that* if a distinctive feature

approach is adopted, there is certain to be a conflict

between auditory and articulatory requirements! "any

criticism of auditory features on the grounds of under-

specification in the articulatory field is matched by an

exactly corresponding criticism of artieulatory features

on the grounds of over-specification for recognltlonal

a
purposes". For example* the dental allophone j.j|j in
tenth has to be distinguished from the j^nj in tent in

articulatory terms* because there is no auditory feature

which can distinguish the one from the other, On the

other hand* the (articulatory) distinction between

dental and alveolar is for recognition purposes totally

redundant in English, (except for fricatives). We there¬

fore prefer features which are neither articulatory nor

auditory, but aro neutral with respect to the two. These

features are therefore abstract.

3) It is an important fact that identical artic-

 302

ulatlons must often bo regarded as realisations of

different under-lying abstract elements* We have

already seen from the Nch material that even where»

for Instance* the lexical sets of fear and fair are not

distinguished* they are best regarded as having different

under—lying representations. In the distinctive feature

framework* this can have the undesirable effect - since

phonemic and phonetic elements are treated as formally

indistinguishable * of implying that one systematic

phonemio element has become another* rather than that

the two have identical realisations*

Fudge gives an example of this from Mandarin

Chinese* In this language, the effects of tono-sandhi

moan that Tone 2 ♦ Tone 3 has a realisation that is

identical with that of Tone 3 t Tone 3* When distinctive

feature notation like

I- High J - ■> 1+ high J

is employed to handle this phenomenon* the implication

is that Tone 2 1ms changed into Tone 3«

%) It is also the ease that "morphophonemie con¬

siderations often lead us to make groupings which arc

definitely counter-phonetic, and which cannot be handled

in distinctive feature terras"* There are, for example*

very good reasons for regarding Hungarian short {.»j as a

 303

vowel that, although phonetically rounded, belongs to

the phonological subsystem that consists of vowels that

are phonetically unrounded. This la because it functions

in vowel harmony alternations as an "unrounded" vowol*

There are many other examples from other languages.

5) The advantage of employing an abstract system¬

atic phonemic level of this type in the development of

a dlaeystem is plain. "Xt enables us to generalise the

phonological description beyond the limits of one speaker

without having to postulate a whole series of vague ab-

a tract ions 4 la Daniel Jones**. •• Just as a whole set of

feature—bundles can be made to correspond with one string

of morphological elements (e.g. a set of varying pronunci¬

ations of the word came may all be assigned to

jcoraej +
|past| on the morphological level), so that whole set
can be made to correspond with one string of systematic

phonemic element bundles (e.g. the varying pronunciations

of came may all bo assigned to the sequence /kam/),"^ We

thereby avoid situations such as that described by

8
Chomsky A Halle, where, for example, the ♦ back
♦ round

systematic phonemic vowel /u/ has to be converted to a

phonetic vowel that is neither "back" nor "round* in most

dialects of English, but must be kept "back" and "round*

in some dialects.

The usefulness of this type of approach la illustrated

o
in the work of Tay. Toy*a diasysteraio study of Hokkien

Chinese is greatly facilitated by the employment of

 304

abstract features.

The Ahstractnesa of Features

According to the approach outlined above* the

systematic phonemic elements are as abstract as the

syntactic and morphological elements* They are* of

course* closely connected with the stream of speech* but

actual articulations are of Secondary importance at the

systematic phonemic level* (Actual articulations also

vary considerably from speaker to speaker* of course.)

The Phonological Component

Fudge has stated that in this type of approach the

phonological component should be independent of the

10
syntactic component. The grammatical model* that la to

say* is not of the type that has been characterised as

"monolithic*• This is clearly a debatable topic* but as

it is not one that is central to the present study we will

adopt this formulation without further comment.

The first section of the phonological component

consists of a syllable generator.** (Fudge*s syllable

generator is an extension of the syllable grammar devel-
12
oped by Cheng in his study of Mandarin Chinese phonology.)
The function of the phonological syllable* once generated*
 305

is to provide a basis for the application of distinctive

prosodic features and to account for constraints on

13
possible phoneme sequences*

The English syllable generator has the following

rulesi

Syllable >■ ■) Onset ♦ Rhyme (♦ Termination)

Onset ■■■■) 1 ♦ 2

Rhyme ) Peak ♦ Coda

feak ■■) 3

Coda » k ♦ 3

Termination > 6

1* 2, 3, %, 5 and 6 are places in English syllable

structure* Place 6 is used only in word-final syllables*

The syllable struetore produced by these rules Is illus¬

trated in Fig* 1*

A different systematic phonemic system can be shown

to operate in each of the syllable structure places* To

demonstrate that the members of these systems are abstract

and non-phonic* they are tabulated as in Tables 1 to 3,

with the phonetic symbols acting as purely mnemonic de¬

vices standing for the complexes of abstract features

represented by the letters and numbers* A* a* 1* etc*

A full justification* baaed on etymological and

orphologioal considerations, for these arrangements,

 306

Figaro It English Syllable Structure

Including the counter-phonetic position of /s/, /a/ and

/r/# has been given by Fudge.2"11 (For example, /k/ 1 /a/

alternations of the type opaque 1 opacity are partly

responsible for the position of /a/ at places 1 and 5«)

Table It Phonemic System at Place 1

a P t c k

b b d J g

c •P st ek

f $ e

▼ 9 a

3 v 1 r

k » n

<r~ h
 307

Table 2i Phonemic System at Place 2

A B C D

3 V 1 r

k a n

Table 3* Phonemic System at Place k

A B 0 D

3 1 r

4 <V ■
N " '
/
\

Table 4i Phonemic System at Place 5

A D C D

a P t c k

1 b b 4 J e
e ep et ek
V

a £ e s a
2 V
b V £ e

3 1 r

% m n

Table 5< Phonemic System at Place 6

A B C D

1 a/b T

o St

2 • S

s
 308

Place 6 "may be occupied by one of the members of the

system operating there, or by a stria# of two (exception*

ally three) of these members* Thus boxes is represented

as /b*o,kSS/ (one phonological syllable although two

phonetic syllables)* and sixths as /s*l*kS$S/t where the
symbol • indicates the selection of wro".1'

Cooccurrence restriction rules which govern con*

etralnts on phoneme sequences are of the form*

2(®3) ■ ',i^X 1(B) t

This Indicates that tl*» dl*» stl-, «1* etc** are not

permitted initial clusters in English.

The above systems* and the co-occurrence restriction

rules* would appear to be very similar for all types of

English* and wo shall not discuss then further*

The Phonemic System at Place 3

It is prcbablo that the whole of the above schema Is

Intended to be diasysteraic* Fudge states* for example*

that "the inclusion of post-vocalio r (places 4 and 3)

must not bo taken as implying that the scheme does not

apply to *r-less* dialects> D3 is an abstract element

which in some dialects (notably Jft.f*. ) may often have no

realisation of its own* but which will* so to speak*

*
 W9

ȣ
contribute to the realisation of the preceding vow®l"»

It is quite clear, however, that the system proposed by

Fudge for syllable structure Place 3 - the vowel system <•

is not adequate for Nch English# Fudge's Place 3 system

is shown in Table 6#

Table 6» Phonemic Syst® m at Place 3

X T Z
*=*<•
I3 P f
mt

I i A I A an ol
em 4t*

II e U e u
«a>

III a o a o at 91

This system does not permit us to generate the

distinction made in many types of Nch English between the

lexical sets of nail and name# For this purpose we

clearly need an extra systematic phoneme in addition to

//a//# (Systematic phonemic elements in the ffch diasyatem

will be symbolised by enclosure in double slants#} Xt is

also clearly impossible to produce the Nch distinction of

the lexical sets of nose and knows from the above system#

We therefore need a further systematic phoneme in addition

to //o//# Note that it is also impossible to generate the

wheat t week distinction made by a few older V-An speakers,

if we should wish to do this#

 310

The solution to this problem is to amend Fudge's

vowel system to that shown in Table 7* This is the sys¬

tem that will be produced at Place 3 in syllable structure

by the syllable generator of the Nch diasystem* These

vowels are the abstract elements that underlie phonetic

vowels in all types of Nch English*

Table 7' The Nch Phonemic System at Place 3

Y z
^ X r—*

P P
/

I i u I u
mm

a e a

II e a mm
ol
b e au
mm

a a o

III a o
b al ou
>

The addition of the elements mneraonically represented

as //«//» //ai// and //ou// permits the production of the

Nch distinctions Just mentioned* The element //z// will

be realised as /et/ in wheat\ //a// as , /ti/ in name;

//ai// as /©i/ in nail| //o// as /Uu/ in nosei and

//ou// as /uu/ in knows* The transfer of //u// and //u//

in Table 7 to the places occupied by //a// and //a// in

Table 6, and vice versa, and the inclusion of //au// under
T vowels instead of under Z vowels, makes the system more

symmetrical. More importantly, it means that morpho¬

logical alternations of the type profound / profundity

 311

can bo bandied as well as those of the typo produce /

production* It also lias the effect of indicating the

peripheral nature of the phonological element //el// in

the English vowel system since, as wo have now omitted

at and at, //oi// is the only remaining Z vowel. The

elements at and at have been omitted since they do not

occur in many types of English, and can usually be pro¬

duced at the phonetic level from various combinations of

phonemic elements, for example!

//o// ♦ //r// > /at/

Morphological Alternations

Morphological alternations can be handled most

satisfactorily within a framework of this type by rules

of the formt

T «■'■■■-•')' % / . (la certain con¬

texts).
This rule will produce alternations of the typet divine /
divinityi ecrone / serenity» opaque / opacity* This

rule exemplifies one of the advantages of the use of

abstract features. The aboy© arrangement of the system¬

atic phonemic vowel system means that the vowels which

alternate most frequently differ by only one feature - %

as opposed to Y. Comparable rules that have been devel¬

oped within the distinctive feature framework are

 312

17
relatively complex and cumbersome*

Note that the two Y phonemes //»u// and //a//

correspond to the single X vowel //a//* This means

that the distinction of //an// and //a// is neutralised

under this kind of alternation* In the same way, //o//

and //ou// must be made to correspond to the single X
vowel //o// in order to handle alternations of the type

//ou// ■■■ ) //o// i know / knowledge as well as thoss

of the type //©// ■"""""" > //o// t provoke / provocative*

This is also true of //al// ») //a// and //a//
//a// alternations* where neutralisation also takes

placei explain / explanatory t profane / profanity*

(This would presumably also apply to the neutralisation

of //«// and //©//» as in clean / cleanliness t serene /
serenity*)

The vowel system shown in Table 7 will be called

the Neh Diasystetaic Phonological Vowel System*

Diaayateraie Inventory Rules

The Nch phonological vowel system shown in Table 7

is characterised as diasystemlc because, as It underlies

all forms of Nch English* some of it© units are redundant

for many Nch speakers as far as speech production is con*

cerned*

We shall now* therefore, develop a set of rules

 313

will eh will relate this dlasystemtc rowel system to the

particular phonological vowel systems used by speakers

of various Hch accent types in their speech production*

For example* as we have already seen* many Nch speakers

do not distinguish the vowels in the lexical seta of

name and nail* We therefore require* for these speakers*

the diasystosiic rule* ruls li

1) /A HI
(J) // > /A HI//

That is //a// )
> /AV/
//*!// )

(Note that the choice of symbol to the right of the

arrow in the second formulation is totally arbitrary,

since all the units involved are abstract*) Puis 1 is

a dlaaysteriie inventory rule*

One Justification for the development of this type

of rule is that it permits all typos of Neh English to

be derived from the sane underlying system without any

redundancy for individual speakers* A further Justiflo-

atlon is that one must assume that this type of rule is

used by speakers who do not themselves make a particular

distinction in the interpretation of utterances in which

the distinction is made*

We can Illustrate this point in the following way

 Fig, 2i The Act of Communication

Fudge has proposed the model of the act of linguistic

communication illustrated In Fig* 2, which *shows the

steps involved in Inducing the hearer to select tho same

id
phonological element E as the speaker has selected**,

Xn the case of Tone 1 In Mandarin Chinese, for example*

the following takes placet

E i Mandarin Tone 1

A t Vocal cords tense, vibrating at frequency

which is high, and remains constant through¬

out the syllable

F * Fundamental frequency of sound wave constant

and high
 315

P « High level pitch

E » Mandarin Tone I

In the case of this particular Nch distinction,

however* it would seem that selection of //&// on the

part of the speaker can induce selection of //ai// on the

part of a hearer who does not have the //a// i //ai//

distinction* This is made possible because the hearer

has internalised the liasystemie inventory rule, rule 1*

This assumption is partially Justified by the fact that

many speakers appear to be able* in some way, to reverse

the direction of application of the rule* Xn other words,

speakers who normally never stake this distinction are

able to do so, without error, if they wish to for

humorous or other similar purposes*

As far as other diasysteiaic Inventory rules are con¬

cerned, it is a fact that all speakers who have the

wheat | week distinction also have the name t nail

distinction, although the reverse does not apply. Sim¬

ilarly, speakers Who do not have the nnmo t nail distinction)

do not have the wheat t week distinction either. Wo can

therefore postulate additional diasystemic Inventory

rules as follows,

2) //r ^ II
[£j // » //Y« XX//
That is //©// )
//;//1 > //://
 315

This rule applies to oldor speakers, except for the

very oldest who preserve this distinction. For younger

speakers , who do not have either of the two above dis¬

tinctions, the store general rulo, rule 3, applies.

XX
3) //Y= 1X3 is! " 4 //TM //

That is //S//
//•//
//«//

and //«//
//a±//
//at//

As, however, a and b occur only under XX and XXX,

rule 3 I* better expressed in the store general, simpler

form of rule 3a*

3a)//Y ft!/' + // t®*//

This rule then applies vacuously to //Y =>< X// •» //I//

as wsll.

Xt has been suggested that rule simplification plays

an important part in the process of linguistic change*

The increase in generality of application of this dla-

systemio inventory rule, as expressed above in the ewitoh

from rule 2, which older speakers have, to rule 3a for

younger speakers, would appear to provide a very good

 317

example of precisely this phenomenon. This point can

be more clearly illustrated if we restructure the above

rules to read as followst

2) > a '
T II

3a)
aj > 0 / T°<

(This also shows that at least one theoretical advance

wade in the field of generative phonology is not

surrendered because we have rejected phonetically based

features in favour of abstract features•}

Rules 2 and 3a are the only two diasystemle in¬

ventory rules in the Nch dlasystem. This means that in

addition to a phonological vowel system which is identical

to the diasyetemio phonological vowel sjrsten Illustrated

in Table 7, there are the two additional phonological

vowel systems, produced by application of the diasyoteraio

inventory rules, shown in Tables 8 and 9*

It should b© noted at this point that the failure

to distinguish the lexical sets of nans and noil is not

of the same order as the failure to distinguish the

lexical sets of nose and knows. Some n,P„-influenced

Nch speakers wight appear to have merged the two latter

 318

seta* since they make little phonetic distinction

botween Items oi the two types (see Chapter Eight)*

The apparent merger la, however, by no means total*

Table 8i Kch Phonological Vowel System

X
^
ft A
li u

X
IX
au ol
m
a O
XXX
b al ou

Table 9» ^oh Phonological Vowel Syetem

X r
<=K
a
a u

xx
an ol
«e»

o
xii al
on

Thle can be demonstrated most clearly by results ob~

talned In the Pairs Test, All those speakers who, during

the course el the Interviews, made little or no distinction

between items of the type nose t knows could always be

 319

induced to mke this distinction by pros on tin;? thorn with

a pair ot this typo to bo read side by side* The same

did not apply to pairs or the type days t daze (except

in the case of some older speakers), even in those oases

where the distinction could he made for humorous purposes,

Where an apparent merger of nose and knows items does

occur# therefore, this is the result of a lower level

rule dealing with the realisation of two distinct phonemes#

and not of a diasystemio inventory rule*

lexical Entries .

It would appear to be at this point in the phono*

logical component that the link-up with the lexicon must

occur, so that items can be listed in the lexicon in

terms of the phonological systems shown in Tables 7# 9

and 9.

For speakers who have the nail i name distinction#

name will be listed asi

//B4,Y~< IIIb,AW/
or* //n*a#m,//

together, of course, with the necessary syntactic and

semantic information* For speakers who do riot make the

distinction, the lexical entry will bet

 320

//B*.Y ex

on //mai.au//

The fact that lexlo&l entries are made In terms of a

particular phonological eyeten rather than the diasys testis

phonological system means that the tii&syetemlc inventory

rules must be allowed to operate before the lexicon etage

is reached for recognition purposes as well as for pro*

duction. ¥e must therefore postulate a model of the type

illustrated in Fig. 3.

Fig* 3* The Place of the Pi&eystootle Inventory Rules

Syllable Generator

Biasystemic Phono*
logical System

I
Diaeyetemi c Phono¬
Inventory logical LEXICON
Rules System

Input Output
 321

tt this is tho correct interpretation of the

situation, then the seeond stage of the phonologieeil

component must come into operation after lexical items

have been inserted in the correct slots and all the

syntactic rules have been rim through* The rules in

this seeond section of the phonological component that

are of dlasystemie interest are rules which we shall

call diasystemic incidence rules*

Plaayateralc Incidence Rules

As ws have seen in previous chapters* there is a

large amount of variation in ffch English of the type

that is usually described as resulting from differences

in phoneme incidence* The two main sources of this vari~

ation are the following!

1) Vowels which are reflexes of HE o are

subject to shortening in Nch English

from lutj to LtfJ In closed syllables -

the EAn short o# The item road* for

example* can be pronounced either

[_ru»dJ or [rtfdj.

2) Items such as boot* soon* hoof* in

which the vowels result from HE o*

*

can be pronounced in Nch English either

with an t,w« J-type vowel * with the i.u» J

 322

vowel also found in road# or with the

short j^Uj of road# hood.

Now It eould bo claimed that these alternations in

pronunciation are simply the result of a series of

alternative forms in the Lexicon* The item road, for

instance) might be listed both as //r.o.d,// and as

//r,u*d*//. This, however, would suggest that the

various alternative pronunciations were entirely un¬

motivated. and would fail to reveal the underlying

regularity that these alternations affect whole lexical

sets in exactly the same way. It would suggest, in other

words, that the fact that alternations like (.mid J /

(,rOdj» |,stuinj / tstWnJ etc* are identical is entirely
accidental* We propose* on the other hand« that there

is a significant generalisation to be made here. That

is to say that the relationships that obtain betwoen the

systematic phonerale elements //©//* //u// and //u//» as

exemplified in the above alternative pronunciations. are

not simply aooldental relationships of alternation among

phonemes* They are. rather, subject to rule, and ought

to be accounted for in a phonological gracssar. There is

a definite and constant relationship between the pro¬

nunciations of these sets of lexical items, and the

amount of variation is so largo (and determined) that it

cannot be handled satisfactorily simply be listing altern¬

ative forms in the lexicon*

 323

20
The t/p« of rulo proposed by Thomas to handle

features of this type in Welsh, as illustrated in

Chapter Wine, is relatively complex compared to the form

of the rules required in Not* English* for example, to

haadlo the alternations associated with the lexical set

of road* we must assume the following* The item road

is listed in the lexicon as //r*o#d»//. We take the

//»// form rather than the alternative //«// form as

basic, sine# rules of morphological alternation must

have T (*long*) vowels te operate on if they are to pro-*

dace the correct output* If, for example, we listed

provoke with f/vj/ rather than //©//, tho Item provoc¬

ative would not have the correct phonotlo form* It is

aloe true that tho variant [raidj is the more common,

particularly in store formal styles* Historically, too,

tho long vowel is the original one*

V« then require the diaayatemio Incidence rule, rulo

%) //T(?IIIa// > //Xfil//

That le, //©// //*$//• This eventually produces

the output ^rtktj*

To handle the alternation of items such as boot*

we assume that ve have lexical entrloo of the typo

//b*u*t*//. This is partly because [ btt 11 j appears to bo

tho older KAn form, and partly because we have already

 \ fI ;5 *I V s N tf • 91

N tu \U • '9 V * \ .N \ >f • 01

I »*< I 2

\u \ Fe» \ \6» ot • V *s \\ tf 0* *
•

>« x^ \\ *t * 9 • »« • \ \ \f ♦ 9•

>4 i i tfP >4

 &

M *4 1o■ gs 3 g ©1 $© £ S( u l ►

g£ O *« 5 O lf * © O BJ ©

*I * It © i W 5 O fT ©g 9 •« ♦ §o e 9 © oj e 1 ©« 9

©9 « M# 4« 9 k >*• g & « * 3 M« H 4•■ © K © M4 :J \ ot\N I1 se

i © k S H 14 © © 4* s a© 1 0e © »4 M• © 4•• s tr<k I S t—
i 0 » 4i « 11 ! M< 31 © 9 *t SH

3© >% ©■ c © •» ©A © 4< a »4 o SI © © © £©

11J © »k

© gk

1
34 © e5 © BJ

«k «

it£ ©
!
3 kpi© © «

© « © *•«

i:
ft ©
*4 9 »* 9 © *

3s
©k © »♦ i

S •
«i

!i s* | © ii H «& §WN« Is8^ IS ©

©g ©
I «»

4i ©©
©©

© 4* © © © ©
«© ©£«

8• © © © ©
§ s i©k

I 211
 rs « i
g «£
* 8 &H .* *I
4 »4 J« S H H «u

i © H( *« u» a. »4 SJ § 0* t*< < *I »4 1 18 5< Jt 0g

1w

1«* 0A * i »4 »4 1 iv »* 01 0 »+ A 0 »4 xa »♦ © 2 8IH tf

& »4 tw< O1 H K 3 O tf 5 »4 A a© o£ e w

91 »♦ * a2 1•
" Ii

I s »4 1 0 ® tf £ 5 ,c tf£ 2a * sI• :i

©a *4 1 0 H ! >* © "I 3 *4 a ^ ai^ - -v,S- :5fi"

93 tf © »4 I *4 4* i S « h S( d £ a 9§ H< O 8f*T 0 tf tt x «

© tf »4 1 • §|© £ tf « H© »4 X © © © « »4 © tf >t >* tf oC SS 3

 3*7

as presented above, But the number of speakers re¬

maining who nee this distinction in their speech pro¬

duction is so small that the rule sill soon apply-

Obligatorily to all Heh accents, with the result that

tho phoneme //«// (and th© rule itself) will drop out
of the diasystem altogether.

Tho diasystemie inventory rule, rule 3» described

above, provides a similar example• That section of tho

rule presented ast

//£//
//*i//
//Si//

was discussed as an inventory rule, It is clear, how¬

ever, that in tho speech of some older informants, this

ruls is a diasystestis incidence rulet

//a// » //si// •

These speakers, in other words, can choose, in normal

everyday speech, to proncunco words liko name either

with 1* i j or with (.aiJ, (These speakers will Include

among their ranks those who produced the //a// t //ai//
distinction only, or mainly, in the Pairs Tost in the

Interviews)*
 328

Further Problems with Plaaystemlc Incidence Rules

There are some complications that arise concerning

the two diasysteraie incidence rules* rules % and 3* Ve

hare seen in Chapter Eighty for instance* that they are

applied to some lexical items much more Frequently than

to othera• Some items* such ae do and lose* moreover*

suit he marked {.-rule 5 j in the Lexicon* a a they are

pronounced only with {.mi* (They cannot be lleted in

the Lexicon with the //a// vowel of tune* music* since
they never receive the j. J j glide which does occur in the

lexical set of time,) Rules 4 and 5 are also restricted

to certain phonological contexts for some speakers.

And* in addition to this* the application of rules 4 and

5 is to a certain extent dependent on the social class of

the speaker and the social context. This point was

illustrated in Chapter Bight in our discussion of the

variables (o) ami (u)•

Ve will discuss the last-mentioned complication

first* A useful method for handling rules containing

variables of social class* social context* and other

sociological parameters* has been developed by Weinreich*

21 , .

Labov & Horace* For example* post-vocalic /r/ in New

York City is a variable1 tho omission or inclusion of

/r/ in post-vocalic position varies with social class*

social context and a^e-group. The relevant phonological

rule is therefore formulated in this wayi

 929

/*•/ ) el**J /. { M
gj.rj m t (Style* Class* Ago)

Tho «zpr«>»loii g|,rj "As the linguistic variable defined

by tho rule", usually donated (r)« "Tho category /r/ is

rewritten as the variable (r) in final and preconsonantal

position* equivalent to the frequency of the eonstrieted

consonant j.ri» m function of style* class* and age level

of the speaker."22

We can therefore indicate that the Not diasystetaie

rules are not only optional but dependent on sociological

factors by the following type of notations

•>)
e//o// ► //«•//
e (rule k) m t (Style, Class* Age, Sex)*

Pule 5 can be rewritten in a similar vayt

5) &//Vf > //*//

g (rule 5) m t (Style, Class, Age)*

The value ef g In each case can be deduced from the

material presented in Chapter Eight. In a slightly

different formulation of the same New York City rule*

Labev gives an example of the sort of values g can have.

The rule here appears in the fortai

 330

(r) - X LrJ
X m t (Stylo, Class, Ago)

•Tho function £ la a linear function, and it Is estab¬

lished by shoeing that small changes In any of the In¬

dependent variables ^Style, Class or Agej produce

corresponding changes In the value of X»* And them

"The proposal Is also seen as one means of shoving the

structure which relates competence to performance. Thus

the competence of a speaker Is given by the general

rules

(r) y x LrJ x ) f (S, C, A)

This may mean that In actual performance, the speaker

never uses Lri In the relevant positions, so that hie

performance may be

f (S1, c3, Ak) « 0

or f (S1, CJ, Ak) - SO"23

A second complication in the treatment of dia-

systewic incidence rules is that for many Speakers applic¬

ation of rule 5 plus rule k for //«// Items only occurs

ia certain contexts* Some speakers, for exampls, have

tuj < //u// only In roof* hoof, prooft others have

only in tills lexical set and the lexical set of

room* bloom* broom* Tho fact that there is no set of

 331

comparable phonological restrictions on the application

of rule 4 to //o// items means that w® need a third

diaeystemio incidence rule* rule 6•

6) * //T(31// » //Xgl//
8 (rale 6) • f (Style, Class, Age)

That la, //n// ""*'" """"■) //a//. The addition of this rule

means that the three incidence rules can form an un¬

ordered set within the diasysiem, since it is no longer

necessary to run //u// items through rule 5 and then

rule 4 to produce outputs with (9j* This can he regarded

as a simplification.

For speakers who have jitrj only in hoof, roof, proof*

rule 6 has the fens

//YflT// > //X$X//L //A2a//

For those who also have j.0j in room, broom the rule must

read

//rfz// > //xf,i///. /A //

A further complication that arises concerning incidence

rules Is that, as wo saw in Chapter, Bight, some items

are much more likely to he subject to rules 4, 5 and 6

than others of a similar phonological composition. For

 33a

instance* spoon is smelt store frequently realised with

^.0j titan is moon. ¥e hare seen that items like do and

loss must be marked rule 3j« It ie therefore also

possible that we may hare to mark some it erne (,♦ rulo 3 J

and others ru^® 5j|* This is a rery speculative

proposal* but it would appear to bo difficult to build

in indications of which items are most likely to bo

affected unless we do something of this kind* Another

possibility would be to allow the ralue of g In theso

rules to depend on some semantic features* Items eon*

nested with everyday or family life* for example* appear

to be catch more likely to bo run through rules k and 6

than other items* There Is no way* howerer* to indieato

which //u// items are more likely to bo susceptible to

rulo 3 and which to rulo 6* (except that items / ■ «■» f

are more likely to hare rule 6 than rule 3)*

An Additional Incidence Rule

Rules Ut 5 and 6 are the most Important of the Nch

diasystemie incidence rules# Pule 7* the final Incidence

rulo that we shall discuss that is concerned with rowel

phonemes* applies to the unstressed word-final rowel In

items such as city# moneyed# lively# Items of this type

are listed In the Lexicon with final //i//» which giree

as output the /Z/ or /d/ characteristic of older speakers#
For younger speakers* the olasyatetaie incidence rulo*
 33*

rule d, api»U«a«

8) //x// » //«//

This gives as output the /li/ vowel in these iteras more

characteristic of younger speakers* Items of this type

will be listed in the Lexicon with final //!// rather

than //«// because of alternations of the type likely /
likelihood* where many younger speakers alternate

/IA/~YI/.

Phonetic va. Phoneate Rules

One apparent weakness of the above diasystemie in*

oidenco rules is that they appear to be better motivated

or more explicable at the phonetic than at the systematic

phonemic level* Rules of the type

l**J ■"" ') t««J

L«» J ■"> L^i
ixj > Li»J

appear te be much more reasonable, ami much easier to

explain (in phonetic terms) than rules of the type

//U// ► //o//

//o// > //V/

//i// » //•//
 335

Would It not, therefore, b« better to consider that

these rules should apply at the systematic phonetic

rather than the systematic phonemic levelY We have

already given the counter-argument to this above in

ths ease of //©// |""1"1""" "■"> //«//• In the ease of //u//

//©// there is one very clear reason why this

treatment is net possible! phonetic i«tj is a realis¬

ation net only ©f //n// but also of //a//. Items in-

%

eluding realisations of //a//, however* do net have

alternative forma with {.UJ or latj. hiasystemic in¬

cidence rules at this level* therefore* would have to

take the systematic phonemic source of l«»J into con¬

sideration*

It is net surprising, however* that phonetio

factors should have repercussions at higher phonological

levels* Schane has shown, for example* that it is very

common for changes at the phonetic level to lead to

24
Changes at ths phonological lsvsl* V* can* moreover*

provide an additional example from Keh English* In the

speech of a large percentage of the !?ch informants who

were aged under twenty, items like this, there* then

were pronounced with initial tvj, and items like thistle.

thing with (,f J* In the caee of informants who never

produce [tj or i'j we can assume a diasyatemlc inventory

rules

9) //132//
» //A2//
//**//
 33$

since they clearly do not have //#// or //$// In their

phonological systems. For other speakers who sometimes

produce {,#] or j re can assume the diasysteoilo

Incidence rulc» rule iO.

10) //B2// > //A2//.

That is, //«//

^ //r//
Mv //»//

This aliows them to produce (vl5J, itZrjJ optionally* The

causes leading to these rules • the failure of young-

opeakers to produee distinctions mads by older speakers -

are clearly* at least to a certain extent, artloulatory

phonetic in origin* They have not acquired the artieulat-

ory feature distinction« dentalt labiomental•

Diasystemic incidence rules and the phonological

rules of morphological alternation are mutation rules,

which deal only with elements at the one (systematic

phonemic) level. Note, incidentally, that the dla-

systemie Incidence rules must he ordered after the rules

dealing with morphological alternation. If this is not

done, stems like //pro..-v.o.k.// will ho liable to

"shortening* of //o// to //«//# and produce the item

provocative so m /p*V,?vUk* at* Iv/, Other rules of this

typo include many of the rules discussed by Chomsky A

Hallo.2-* including; those relating to stress. The other

 337

rules In this staffs of the phonological component are

of little interest to no here, since they are the same

for all types of ffeh English and Indeed very many other

types of English*

Phonological Realisation Rul t>a

One characteristic of the approach adopted by

fudge ie that there is a strict separation of the

systematic phonemic and systematic phonetic levels*

This avoids the ambiguities and uncertainties that result

26
from an approach like that described by Chomsky* where

there are many intermediate levels of no systematic status

at all* Hiit«tion rules apply at or within a particular

level* with the product of the rule or process being "an

27
element of the same stratum as the input to the process** '

Rules which relate one level to a lower level are realls*

ation rulee*

In our development of the If eh diasystem we can now

moire on to a discussion of the rules that realise element9

from the systematic phonemic level* (after they have been

run through the phonological mutation rules)* at the
systematle phonetic level* These rules convert the ab*

street underlying phonemes into potentially concrete

systematic phonetic elements*

The problem we now have to face is twofold1 what

 338

fern are thee* phonological realisation rules to takej

and what enact status is the systematic phonetic level

to have? According to Fudge*s early formulation, phono*

logical realisation rules are of the following forat

//A// ) Front of tongue highest

//2// > Tongue low

for the realisation of vowel phonemes * and

//8// '"*"'""" ■) Volar plaoe of articulation

23
for tho realisation of consonant phonemes* A later

formulation has realisation rules of the form*

//A// > Labial ity 2 / //2//

for the realisation of consonant phonemes*2^ (Fudge

gives no examples of realisation rules as they apply to

vowels according to this later type of approach*)

In these rules, th® symbols //A//, //B//, //2//t

etc*, refer to the abstract features of tho systematic

phonemio level* Features of the type* Labiallty 2, are

derived from Ladefoged'e attempt to establish a sot of

30
universal phenetio categories*

Tho disadvantages associated with rules of this

 339

typo are that they (l) tie the abstract phonemic

system to the concrete in a way that can be very in¬

hibiting • and nay lead the linguist to take phonetic

footor® into consideration to too great an extent in

the eettlng up of a phonemic eyatemi and (ii) that they

often appear to attempt to produce generalisations where

there are none to bo made. For example* Fudge sets up

the "Latin0 phonological subsystem for English conson¬

ants as portrayed in Table 10,-*1 For very good reasons

associated with patterning and morphological alternation,

the elements mnemonically represented as it, g, s, J and

r are presented in the same column in this table - they

all share the abstraot features //J>°//» This arrangement

has many advantages at the systematic phonemic level,

and enables us to handle rules at this levsi in a very

simple and convenient way. Fudge, however, attempts to

Tablo 10i The "Latin* Subsystem

A B D
«><

/?.
a P t k qu
X
b b d * e»
a f S s
2
b V X J
3 1 r

4 m n
 3*0

extendi the advantages gained at this level to the

realisation rules themselves* It is* of course* a

desirable objective to discover generalisations and

establish natural classes of the type illustrated in

rules ouch ast

//8// ) Alveolar place of articulation*

but rules of the following type show that* onco wo have

abandoned phonetically based features in favour of ab¬

stract features* the fact that natural classes are

phonetic natural classes means that this Is a vain hope*

//»//
Alveolar p*a« 2a
Volar p*&*

j.Glottal constriction 2 ..r -m

Glottal constriction :in

Glottal timing

syllable
//«// ►
Glottal constriction —i
Slottal timing
llottal constriction 9
^
jalottal timing *

Since we do not have a natural phonotio class under D*

 341

there la little point in, setting up a realisation rale

that pretends that we do* Natural classes at the

phonemic level can only be i ml lea ted by meaita of phonet¬

ically baaed feature* such as those used by linguists

working within the distinctive feature approach*

Similar disadvantages 00004 with this approach, in

the realisation of vowel phoneme#* For the realisation

*12
of Hungarian vowels, for instance, Fudge* has rules like

the following!

! Front of tongue highest / la

Middle of tongue highest

Back of tongue highest

2a(ll)

Ve can obtain no generalisations because there are none

to bo sade at precisely those points where the phonoatie

system Is counter-phonetie and where the abstract nature

of the system Is most clearly emphasised* It would there¬

for# appear to be more suitable to permit phonological

realisation rules to apply individually to systems!la

phonemes* This is the approach vs shall adept here*

3»hc Status of the Systematic Phonetic Level

The function of the systematic phonetic leve^ has

\
been defined as to "indicate the way the physical system
1*1 \
of articulation is to perform*• "tJ Typically, the system¬

atic phonetic level forms the output stag® of the phono- \

0 *4 I »4 a 4* 4r 8 © $

IU '

s 4r 4* tA

t03 its *4 94 o • 3i a £& © lV 0 94 4•

II I £' 4 11 I I i S 3 I I9 H

W l 4*• « 04 U S 94 §1 * 0H «© 5 s0© o- £ 0 9

ft 1

«© .y 45 £ K8 0U 0o »♦ 94 2 §® » «r © 4* 0 ® *4 0 0© e 4* 0 5i i

© ©£ 0• »4 £J i« m 0 SS | * © ► 90 4* 94 £ H« 0^ CS © « 4« «4 e S

£ KJ > u 04 « © « 0 « 4* ©* ££ Ig
H w «1 3H • tr 0 « 0 »4 © o ©3

©H0 © iS »* « 0 » © ©0 c » ' * o£ iO § 4» © 3A .*4 © tr

© £< 94 94 © o © I t 94 © *4 a ©©0 4* 33 © »4 a
 3*3

following perfectly normal Nch utterances traaaerlbed

from the tapa-reoorded interviews,

r~—

A, U?« 33 s -ei °ew&nj

That's lust how It go on

B* iwo ft fa °g$w »a a sfali nAiij

Veil, .if .f m going out on a Saturday night

C, Iua a §2 0b /la It lililoij

No, X haven't been down there lately

Those uttcranees are quite typical of casual

speech. They are also quite distinctively Hch utterances*

even as they stand on paper with no indication of proaodie

featurea,and for that reason the Nch diasystem should be

able to produce thea as output* They do not, of courso,

much resemble the outputs that are usually associated

with phonological grammars*

It is therefore clear that if the systematic

phonetic level is to indicate how the articulatory system

le to perform, it can only do so very Imprecisely, There

is therefore little virtue in having narrowly articulatory

outputs of the type developed by ladefeged and Pudge at

this atage, since the outputs do not correspond to actual

articulations #

The systematic phonetic level in the Nch diasystem,

 3 kk

therefore, will not deal with actual articulations or

pronunciations» but with potential or ideal pronunciations*

Units at this level will represent unambiguously idealised

pronunciation types* The level will therefore be an

extrinelc allophonlc level, since intrinsic allophonee

result from actual overlapping articulations*

Phonological realisation rules in the Heh dlasyetem

will be of the forai

//rpina// > /u./,

whore /u«/ is an extrinsic allophon© indicating the

approximate region of the phonctie continuum where title

Idealised but phonetically determined unit ie to bo

realised* ft represents, in other words, the phonetic

"area" to bo "aimed at"* (Note that systematic phonetic

elements are enclosed by single slants, phonemlo elements

by double slants*)

L&defoged* s articulatory features will be reserved

for the realisation of the extrinsic allophonea at the

lowest phonetic level, which we shall call the performance

phonetic level* It will be at this lowest level that

intrinsic allophones will be produced ami other "perform*

anoo phenomena" will be handled* At this level we will

be able to Indicate much more precisely how the articulp¬

atory system is to perform* There will, however, be two

 3%5

major alteration* in the way in which these features

are treated by Ladefaced and Fudge*

1) The Ladefoged features will be totally

specified for sash segment* Xf a parameter is

relevant for i?ch English, its value must always

be given, where possible, if the physical

system of articulation is to receive exact

instructions. Much of the Information would

of course be redundant at higher levels,

2) Only consonants will be specified in terms

of Ladefoged*® features. These are by no

moans sufficiently specific for vowels, and

L«defoged*s approach is in any case phonological*

contractive rather than strictly phonetic.

Xn other words, if there are thres distinctive

vowel heights in a particular language,

Ladefoged's parameter auditory height t 1, 2,

3, will be perfectly adequate if we wish

simply to indicate which vowel it is, Xf,

however, we wish to indicate how the articul-

atory system is to perform in order to obtain

precisely the right pronunciation, then it is

certainly not adequate, . This is particularly

evident in the construction of a diaaystoM

where it is necessary to indicate slight

differences in pronunciation. The outputs of

 3h6

realisation roles applying to vovrels at the

performance phonetic level will therefore he

given In terns of XPA alphabet symbols which

will indicate as exactly as possible the

vowel quality te be produced* This level

will form the output of the dlasyatemle

phonological grammar* and will be a linguist*a

attempt to produce and reproduce as closely

as possible what spoakers actually say*

As an exemplification* we can give the

following series of four rules which will

convert an abstract systematic phonemic

element into one of its possible realisations*

Phonological
Realisation ftulot //A2b// > M
Phonetic Mutation
Rule* M » /€/ /— /t/
Phonetic Bealis*
ation Pulet /f/ » Labiallty 2
glottal con*
striction 9

Nasality 0
Stricture 2
Affrlcation 0

Glottal timing 3
Laterality 0
Lip rounding 0
Secondary
articulation 0

That is* /f/ "") j.fj

 3*7

Performance Mutation H\ ile i

Labiality 2
Glottal con¬
strict Ion 9 5
Nasality 0 0.3
in
Stricture 2 3
Affrication O 0 / certain

contexts
Glottal timing 3 2

Laterality 0 0

Lip rounding 0 0

Secondary
articulation 0

That is# [f J L^J* Performance mutation rules

produce intrinsic allophones and alter pronunciations in

rapid casual speech to produce the type of uttoranco

shown above (a, B and C)»

Phonological nealianticn Pubs in the Nch fllasyatera

we can now return to a discussion of the phonological

realisation rules* (which convert systematic phonemes

into systematic phonetic elements)* in the Nch diasystera.

We will begin with an examination of the realisation

rules for the consonantal systematic phonemes - members

of systems at places 1, 2* 4* 5 and 6 in syllable structure*

Wo ore first of all faced with the following problem!

should the phonological realisation rules take form A or

form B?
 348

(/ph/ !m _L
11) //Ala// > !/p/ / m
1/pV

B*
</ph/ !*— ,

//Ala// » >e(/p/-• /«*/ ~/l/V—M

U(/p*/ ■— /J®?/ ~ /i/^ /y)

g 0 f (Style* Class)

Clearly* features such as lateral er nasal release of

J,pj before laterals and nasals mist be produced at the

lower performance phonetic level* since they are due to

overlapping articulations* It is also clear that /p11/*

/g»*/ and /p/ are extrinsic allophones which belong at
the systematic phonetlo level* But what is the status of

pj., b and It

The solution we prefer to adopt here is that those

are not oxtrineie allophones* bat are variant realisations

of the extrinsio allophones /p*/ and /p/ at the lowest

phonetic level* This means that we are able to confine

our study of the "phonological" variables discussed in

Chapters Seven and Eight to the purely phonetic level*

This is clearly where they belong* since it is actual

articulations which ve were comparing in the figures

presented in Chapter Eight* It is* moreover* the ease

that |bJ and resulting from /p*/ and /p/ oecur only
in casual or rapid speech* anil it is the performance
 3*9

phonetic level which le designed to handle features

of this type. The glottal stop alee ocoure autoiaatle*

ally before heavily stressed Initial vowels* in vhleh

ease It can hardly be said to be an extrinsic allophone*

The low«*level nature of this feature is therefore beet

emphasised by confining the glottal stop to the perform-

anee level* Perhaps the most important reason* however*

ie that this arrangement permits ue to handle mutation

rules at the systematic phonetic level in a much more

simple way. ¥« can* for example* have an aeeimllation

rule ef the type

instead of

-We will therefore have phonological realisation rules of

type A* and produce L$j» i<2j etc., by means of

phonetic realisation rules or performance mutation rules*

The ether consonant phonological realisation rules

include the followingi

 350

13) //Ale// (/»/ ♦ M //4

/•/ ♦ /p*/

Ik) //A2o// /*/

15) /0V 1/4

//A2t>// /T/ J—H
KM

Note that it Is also possible to specify the context# for

realisation# of this kind in terras of the places in

syllable structure at which the systematic phoneme occurs*

16) //A3// ■» M

17) //A*// 4 /m/

Systematic phonetic elementa resulting from other phono¬

logical realisation rules are the following!

/*"/, /*'/, /*/, /&/, /<i/. /a/, /.%/, /of/,

/<}/. /l/. /*/, /n/. /f/. /.«/. /$/, /<%/. ///.
/./, /„«/, /«/, /V. /}/. A*/. A1/, A/. /°c/.

/!/# /e/» /r/t /d/t M*

There are* in addition* other elements that result from

phonetic mutation rules* which we shall discuss shortly*

Special rules are also needed to realise the archlphonemes

at plaees k and 6* In the esse of //H//, these rules

 35*

rosttit in systematic phonetic element

a new
/7)/*

//»// ♦ //&// 1111 1 > A)/» for example*

Note that these rules apply to all Jfch speakers* Dia*

systemic differences at this level result only from

differential application of the diasystesiie inventory

and incidence rules and from the phonetic mutation rules

(see below). They can also result from some of the vowel

phonological realisation rulesI

Phonological yteallgatlon tm» fo for Vowel Phonemes

Phonological realisation rules for the vowels are

as followst

18) //X-I// -4 /!/ as in Mg

19) //**XI// -4 /«/ as in bed

20) //X^III// — /»/ as in

21) //xw/ ~4 /»/ as in

22) //X/3IX// —-4 /»/ as in bud

23) //X£XXI// —-4 /*>/ as in £2*

2k) //t^I// ——-4 /»i/ as In buy

25) //T-IXa// —► /Xi/ as in bee

26) //TKtXb// —-4 /o./ as in K&Wfc,t

27) //rKXIXa// —-4 /«i/~-/*•/ as in name

23) //Y<*IXtb// —-4 /*i/ as in nail

29) //Y/3I// -4 /«*/ as in £2221

30) //T^XIa// -4 /tfa/ - as in tune

31) //T0IIb// —-4 /W as in now

 35a

32) //TPIlia// » /Uu/ aa in jsa

33) /A^nib// 111) /»u/ aa in know
3*») //Z// ')' /*&/ a® in boy

Note that rules 29 ami 30 mean that //a// and //3// eon

have identleal realisations at the systematic phenotio

level* The two alternative outputs to rule 30 aooount Tor

variations like* tune /tJWm/^ /t¥un/# which occur in the

speech of some, particularly more educated people* It is

possible, however, that this feature may instead • or in

addition * be due to a diaeystertle incidence rulei

//"// » //y"//.
%

There is also a lower level rulet

/«H/ » U«*j

or /¥a/ "■) /J/ ♦ /«u/

which produces hypercorrect forms such as

so on /i4«w/*

Systematic iihonetio vowel elements in addition to

those already discussed result in part from combinations

33
of systematic phonemes* Fudge has pointed out that one

of the advantages of having both mutation and realisation

rules JUt a phonological grammar is that the necessity for

 353

rule ordering Is dittinl9hed« Together with Fudge and

Uob»^ we regard this as a desirable simplification*

For this reason we will generate these additional

systematic piionetle elements by rules el the type

//•// ♦ //!// > /*•// <//*//

j/A//

rather that by ordered pairs ef rules of the type

//a// » /at// (/A//

\/Mf
//»// ► 0

The full set ef rules for the production ef /a»/» (apart

from the combination //a// ♦ //r//» which will fee dis¬
cussed shortly)) Is as follows*

35) //a// ♦ //X// ► e/»«/~ A-.// (//f//

}//<v/

36) //V/ » s/a./~ /«./ (/ {//f//

[//*// In
//s//
certain
//nt//
positions*
//no//
//ns//

g • f (Age) ^
 ne m

s© £© m X tft *4 I *4 44 «

X© 3 >© © tr k © t« »♦ 2 ! 44 I w■ ee >\ ♦

$ »4 I ©I S « >N © © ©© k «m

©©I *< © s© tf S fr4 © I In ©

a I 4* 1 44 * 1Jit a
i 4« 4< « X O 0< 44 £

11 i »k « 44 . »4 I ." i i> -s »4 *4 X * £ tI £•

© »k S £ k © « 4* «*• « © *4 44 g o * © »4 I X *4 I »4 44 ©

11 2 »4 5 £H •

o X 4r 1 © H« ES 0 A I © tr ©a

Ok O ©© «k s© »4 4• m23 »4 h t

! •
! © k* *© g 8 »4 I3 3 ©k I ©k O X 4■ 5 0 %4

I: s i© © 4« 4» © > | A ! SI « e ©© e t©

»4 £ 44 2 %© I © »4 44 O 3 44 © » 1 i© 2 »4 44 © 1 w

44 A 4* N # ne

S *<ne 2 S •• N

U \ ♦%a 2 S »% ifC m \

g« I 2 ©2 X 44 « *4 © k 44 2 4* 1% 9© »4 0 ©U A© u H©

as *

tf k© *# \ \ »4 >

91 km 00 \ II C<°« \ .0 >
 355
fair ft* •/ store /atai/
ik

t over ftutf
employer /teiplai/

These are historically products of long vowels plus jr«

Consider now the following list of Sch pronnneiatlono«

C. fthratt%/ booing /Dba«n/

£££&!£ /»«««/ knew it /n3it/

fttifig / °e»*n/

sszlas /win/ knowing /no in/

allowing fdlatnf

employing /stnploin/

tilth the exception of saying (and with the addition of

?<nowlnf») these revels are identical with those produced

historically from form with jr* These forms here* however*

have no historical jr* They are produced from the sane
underlying vowels* but in conjunction with the reduced

vowel schwa* (in the above eases resulting from unstressed

//I//)# Vc can be quit© certain of this because* whereas

younger speakers generally have /'gain/, elder speakers

often have /"goiaa/, We have also shown already, in

Chapter Eight* that "schwa retention* in items of the type

fear ami fair is a function of social class,* styls* and

ago in Well,

Clearly* if we solve this problem by having two

 I 94 © H 0s *4 e II 9 *4 £• Iw k W< ©

9 4*< *4 £ 91 0 eO »4 £ 9 tI tV e tI *I e 1 £ 94 £ B 34 £ 9© 3I
M
'

31 § 9 t ! H eC »H H H A9 HoS *© 9 £O »4 £ £i

9 ©3 1 9 tI H 9 V > 4« 3* 9© 6. 94 »4 a • »4 |> 6 9 1 4* 9

3
IH © !

i2 H 9

« © *4 £
2

: 0 tt • «
»4 9

© tf
I £ « K< O tI

I ia ti

t* S £ a9 i
Ie

M i« 1
& tf ttI

s9

9 N<
sil »t •

19 « tt •*
*e

94 i ti 9 M«e
»4 sj

»
k0
9.

9 2
9 ta « 9

3?

3 *• 94 a *4 44 tt *4 0 e »cf S 9 h >4 1 »4 K O H tf 4# § 94 S | *

J3 *4 8 4* 1 I 9 *4 9 * 3 e 9& ©

I > t-» g 4« 1 1 <ef ©k & tf 9« »4 «. 94 £ 9 I H

.N" 5 »• < *© I IS \H> ltt !I 9■ i H © I 4»> © 1 ! I

3a
3S IH *4 * >• 3 ©© 7 44 I >! 94 I o. si I

i » »4 s « 1 ^\ > .N" »4 tf g © t% »4 £ e

! tt H > I
s> * - > *

H 44 ; » 9 tI 9 I►9 »4 s 9 »4 tt © w< 1w % I 9 tI

I 9 iH 3 I9 »4 *4 B 9 9 M« X 9 £ JiS i

© £ » i H Q* 1> 9 »4 0 £ 9 ©© H 1 H tt 0 ©a
M

tI© H 9 rO

& 1 8 0 k 9 9 M tf I tt tf ©© © tV ti 9 tt tt g■

1
Ia £ * 9 H tt * £ 9 tI 9 H 9 94 © 31 3g 9 9 tt 0

IH © £ 9 tI 9 h« © w 9 3►9 »4 © ©9 »4 ti 0 i *4 © ©© 9*
 357

Phonetic Mutation Rules

The solution to this problem is us follows, first

of all wo require tho phonological realisation rule,

rule 37,

37) //r// »

(This rule must of course have the form //r//

/r/ for those members of the speech community who hare

poet*vooallo r,) The /a/ element which forms tho eutput

of this rule is of course identical with the schwa

produced by other realisation rules aetiiw? on unstressed

vowels. Subsequent rules to generate the above vowels

must for this reason operate at the systematic phonetic

level* They will therefore be termed phonetic mutation

rules# and will deal only with elements at this level*

We next have a sot of obligatory phonetic mutation

rulesi

38) /I/
37) A/ ♦ /a/ > «/3»/~/a»/
HO) /»/

gut (Age, Class)

Hi) /»/ ♦ /V """) /a«/
H2) /V ♦ /?/ > /oi/

This produces the required vowels in third, fern, fur, far

 358

and for. These are, incidentally, the only rule* r#«

quired for the oldest Kelt speakers, who retain /&/ and

have no vowel natation in it eras like tl^ro ■

For other speakers, we need the following set of

phonetic nutation rules, which are optional (except for

rale k9)»

43) /»u/ 8 /3«/~ /3*V

44) /Uu/ 4 8 /»«/—/»» /

45) /Ui/

46) A*/ ) # /5/ ■ ■■ ■■» a /at/— /u» /

*7) /«t/

48) /Xi/ *» « /oi/~ /eta/— /ci/~ /••*/

4S» /mi/ "» 0 /t l/ A I 3/

30) /•i/ * g /**/ ~ A |/ — /»«a/

51) /mi/ ■* 0 /»•/■—'/*>%*/~/*%/~/*%d/

The value of £ is a function of Claee, Style and Ag*e«

Kote that an alternative formulation would be to reetriet

ruloo 43 to 31 to oatpute with A/, and than to have an

optional roloi

0 A/ ► 0

Rulo 43 handles forme like pure, brewer, knew iti rule 44

 339

forma such as store, ^ginq, flo itf ruls 45 ofljgylnfi,

M,i rul« ^ fMt* &*£&«*> iOLJkil r**® *7 towe£,

rule 4S pier, seeing, see, it | rule 49 theret
*«i® 50 player, say ltj and rule 51 knowing, nw^,

Note that in the ease of tour, poor, pronounced

with /at/, it is necessary for the dtasystemic inventory

mile //u//""'" //»// to have applied in order to pro-

duee the required /oi/ from /th^ ♦ A/# Failure to do

thie produces /3»/» which can in fact be heard in tour,

Xt is much more common, however, in items like booIra?,

which can be /0b3»n/ as well as /Jboin/.

Mote also that these optional rules are applied

more frequently raorph-lnternally (as in fire) than word-

internally (as in higher), and mors frequently word-

internally than across word boundaries (as in try it).

The Problem of Intrusive ,a*

Mow that we have written out post-vocalic r at the

systematic phonetic level, by realising //r// in this

position as /a/, we are faced with the problem of how to

generate intrusive and linking r in forms like law and

order and lore aqd lan/^uagc.

One solution would be to amend the above phonetic

mutation rules in the following kind of wayi

 360

This i* to some extent nil unfortunate arrangement • in

that eye teats tie phonenjle //r// is written out by phono*

iogioal realisation rules * only to be written in again

by this nutation rule* This, however# is neoeesary#

sinoe the above rule writee in /r/ even where /a/ is
not derived from //r// * intrusive r# It is alee

possible# moreover# to deduce the relationship between

this particular rule and the phonological realisation

rule# rule 37# from the fform of the two ruloei

//r// M
/**/

Thero le« in other words, an indication# if rather an

obscure one# that only accents without post*voealie /r/

have intrueive /r/.

An alternative and in some ways more attractive

solution ia te have rules of the type*

thus prdaoing the /3»3/-typo forms of older speakers#

Vs would then have the optional rulet
 • *J

s tS © 1 « ©
a© tt S tr« 5J © c * »* ! \H \
i: \ >n \ V s iV »*• Rj « *I e 4# s s i? s ! 1* ©u

I iU • tf m 1 # © iC© H • tf »4 ! tf 4 *i
a,*«

1 t ti if 04 Ua M> «H 9 © ©©© §

»t s I • e *<
I i 1 *« 1 1 i 04 u m

©U »t a • I*

11 ♦ ;♦ * s

H► I »4 © »4 R © & 4• © O 9 s
 3 X H ♦ a *■ X

sa 3 1 Ii «£ 1 13 § 6U »4 £ I

> ♦ a x

g :5a s J

H3 ?j s ai i 3 »4 § 1 * H t« ?% Jt

U3 « ! -v !! £ £s * I £ a 4* * £

S§ »▼ 0 £ c 3 M> 1

we 2 « s* o 4% i aH !
iS
 M

t*

I I f E ©& § »* • s ?
I

© •
v

*♦
0
» 8
J- S » * • © M S !
«1 © 0 *
£r
© 2
1 ? ? 11 s ?

N
u

9 N»

2 ° t• * r >» © \ ' > \ ^ • •» » V» >s

-
V I
! & *
• © © e © 2 • © © i
C I
* cs tr *
*%
*|

S
5 » ©
1 0 « « 1
sr
<* ♦
s.
I

■s.
♦

0 1 © 3 ©

f-
*<
«<

*»
&
©

Vj«
U) < o
sr
1

I
!
& .& ? I i* »
I
1
£

«♦
t
§
«.

*
•

VP!
w
I o
&
n
&
©

•%
0
©•

8?
©

sr
©

♦I •i
©

a% •
i
&
•

I
u

V*
•»
 365

Speaker* of this type* 1% seems* rely heavily en

spelling in order to achieve tiie eorreet /r/*less result

in items of this typo*

01 the above eyetematle pbenetie elemente» /«•/«

/»)/* /»»/ and /at/ result only free phonotio nutation

rules* This is in aany eases also truo of the element

/*•/« which* however* can also ho a realisation of //«//

and of //a//. Put in most eases /at/ is an alternative
Cora to /si/ In ite<as of the set fear* For example* the

lexical entries for fear and fair aroi

//f#e#r#// //f*al*r*//

4t the systematic phonetic level this results ln»

/flia/ /fsia/

The relevant phonetic nutation rule for fear (43) is

optional* that for fair (4?) obligatory* Thie means that
we can have the following contrasts!

/m*/ /fe ia/

/«**/ /ft i/

If rule 4S selects /et/» we have the contrastsI

 /ft %d/
or /fet/ /**«/

Zfi rule 48 selects A «/» ths result 1st

/t%%&/ /fsta/
er /f* 1/ /ft %/

The consequence of ibis is th« variation portrajrsl in tho

dioouoolon of {or) and (»r) in Chapter Eight (from which

information concerning the values of g in rules 48 and 4?
can also ho obtained)*

At least one other phonetic nutation rule has tho

offoet of resulting in a now consonantal eystematlc

phonetic element*

5») /«/ ♦
j%| ► /y / *

(This rule is paralleled by rule 33*

53) /»/ ♦
{ /$ ) » /X/ /- *•)

for example, pleas tiro will bo derived from the lexical

entry*

//pis*a*// plus the suffix //y.A.r*//.

 367

This will giws //pls»* • y*A*r*// by the rule of

norphologioai alternation* This will result in the

eystetaatlo phonetic representation! /phli«j0i /, PhOnetie

mutation rules will delete one of the final aehwaa. and

rule 54 will sire uai

/phl«3a/

Rules 5^ end 55 are obligatory*

Other Phonetic Mutation Rules

A* The following phonetic nutation rules of

assimilation are optional1

5$) /a/ » /V/— /«/

as in ten minutes /t^eraolnlt®/

57) /*/ ) /p/Z—— /ph/

as la outrwat /«mpp,4Uust/

58) /»/ > ,y/m ///

59) /■/ 1 ) /j//L_ /«/.

There are of oourso many others*
 363

8# Also st this lovel w* mat have pales of gilds

Insertion, which to a certain extent parallel the in*

trueire /r/ rale#

60)

From the Input go on /gUu *m/t Cor example* this gives

the output /eon *»«/•

61)
A/

This will give the output /ni Jem/ from the input /oft ma/
1 am# Vo will see those particular phonetic mutation

rules in operation later in this chapter*

C* The final phonetic mutation rule that vo shall

discuss is the rule of resyllabifieation* This optionally

takes syllabi ©"-final consonants and transposes them to

syllable*initial position in a following word beginning

with a vowel* This rule can be formulatedt

6a) /e/ e /*/ > /-/ ♦ /c/ j¥ /v/

id
where /-/ represents phonetic syllable boundary*
 369

The Syat»aatio Phonetic Vow 1. System

is * result of tho jtboaologlsal miisatisn and

phonetic taut at ion rales, wo now have a phone tio eye teas

of voveie and consonants consisting of extrinsio

allophones. Tho vowel eyetorn consists of a series of

svtlMystsesi as portrayed in Table 11* A majority of

epeahere have sab*s]rstses i, 8 and 8 exactly ae shown in

Tablo 11* Thero are, however, several alternatives to

aub-systen C. Two of theso are illustrated in Tables

12 and 13,

Table lis The Hah Systematic Phonetic Vowel System

A ■ 6
a b
i

X ui
V tt | Wtt Uu
i
« a
i
st 3* Ol

m » ml »i J tws on mi ot
i
at as
i
i

«£*•
a

The eymbolisation used for these systematic phonetic

elements shows approximately how the artleulatery system

is ts perform, and stresses the phonetic symmetry of the

 370

Table 12i Sub-System C(t)

«• 3> si

Table 13i 6ub~Sy*toai C(ll)

CI Of fees paw

eta Old HSI22 J22SS

a# ( £5ES&
( nsai

system* This symmetry is by no means entirely a linguist*#

artificial construct♦ Indices for the variables discussed

in Chapter Eight* for example* show that there is a close

correlation in scores betwesn (£) and (au). There is

therefore Justification for tabulating /»i/ and /mm/ as

parallel elements* There ie aleo considerable symmetry

la the development of A and B vowels in conjunction with

the I> vowel to produce C vowels* Thus the four highest

A vowelst /!/»/«/» /«/ and /»/ all become /3»/ in eembiiv*

ation with /a/. The two parallel vowels /ui/ and /®«/

both become /»*/$ while /li/ and /»!/ ©an both become
 371

/e »/ Mid /Ou/ and /»u/ can both become /»»/* Note, tee,

that Be vowels trigger off the /,j/-ineerti©n rale, Bb

vowels the /w/-insertten rule, and C and i> vowels the

/reinsertion rale* This suggests that it might be

possible to express phonetic mutation rules 43 to 51 la

a simpler, sore general form (see Chapter Eleven),

Phonetic space

One ef the drawbacks of Fudge** scheme, (and ene of

which he le aware), ie that hie phonological system does

net allow for the possibility of indicating cr explaining

how innovating structural pressures eon lead to sound

change, There le no system in which forces of this kind

can be played out.

Fudge sets up, as an alternative to the vowel system

shown in Table 6, the system given here in Table 14,

Fudge etatee "although Table 6 is the one we adopt, it

must be admitted that the relationships ef Table i.1^1 «*°

exert an influence en the phonological system • sound

changes set up phonetic structural pressure in an innov¬

ating direction, while morphological relationships which

persist through a sound-change tend to pull in the direct¬

ion of conserving the eld phonemic system* There is there¬

fore a struggle, and the implication of our approach le

that this struggle ie one between present-day phonetics

 **t »»r 10

I *m TJ ►« I3 1 e > s * k« ot R *4

*
*2<
~S< ..

f* 9 tr« •« £ 9m

• © • 4« 0 k< »< m

0 < J« S e0

1 1 « 0 »4 1 9 K 8 k1 €0 4• 0 ii 9© 0• 0 3 *4 c »* e i a

0 10 0 0 3 k< Q• k0 0 »4 4* O »4 £ 0 k 0 »4 k« 3 k« >©

S 0 0 »4 }X I 9* 4* 0 « 1 4• 3 «0 »4 « 1 0SH« v

k 0 1 4< « 0 >* 4*< O 0 0k .S* 0* 0 % & 4* i :

!? fI
k©

I *0

»8 © 1
i

i-

!s 0 »4
I« I k. 4*

s
0

1 0 k H t*
•

© 4*
0 2
-

57S *J st
% *4

| t£
>

2i
a

cs 3 H # m m
31.
« »0 I s k« 44 s « 1 0 >4 »4 r (« tt *4 ej »4 3 1 »4 £ 0 k t* iH #* 8

0 »4 g I :i jX0 • k« 0 c1 A 0 *4 » 3

I• I »4 S H1§9 »4 k 0 m 0 w< . a1 • 0 ot 0

o »4 3 k4 *4 *4 3 k« 0 I 31 © © O k k tif »4 k4 » 6 i ?3 © K 8
©

eif 1 ot © k k« »4 k< © k< 9 i e § 0U

!Hk > « H a a »4 k4 « H a )X X k *• 9 9 »4 k• I

H 1 k • ki *4 .cf 1 H©99 5 « k >X k■ k k< »4 kJ O k< ©XcH 0 k4

% I
0 4« £0 S« £» ,c A§ 1 0 4-t 1 if 3 if >* £ 0 # • 001

6 2 0 if sI 0 ► • «f af 30 9 .
af af i a s!

I ®4 3 £ ; *4 09* £1 ®4 2
.I
aI 5 i

9 ?I IJ ?

I if t 8 M* ®4

2 *4 2 kI 1 s »4 SJ ! ®4 0 if » *4 ®4 A *4 a !3

ot 8 a ! ®4 A 8« 0 4* ®4 e tf e 4* 0 I s *4 A8
B

»■( !t if i SJ af I4® 4m 5 | .

9 if 1w afI
4* ®4 4i § | £

af 0 0 !* ri - af £ « if ®4 »4

U .*2 $£
O

®4 if & 0 af 9• e ® ! £ ®4 ®4 A• ®4 0 4* 0

t t|
f *4 4* © I ®4 I 0 *4 af 9 O »4 0 4♦ 0 »8 : « »4
 37%

this level generally# nay veil be found to be one

c*uao of phonologloal changes ouch a« rule eiwplif-

icatlon and rule reordering* What 1« even more likely

le that purely phonetie change* and dialect difference*

which raay be ef little phonological interest but of

considerable dlasysteaie importance are the revolt of

*

thie type of preeeure*

Justification for this belief ie provided by the

41
vork of Martinet and ef other linguist# such ae
42
Moultoa* Martinet ha* shown that the concept of

"phonological** space ie a very useful on© in the study

of Unguistlo changet and that terua such as "holes in

the system** and "drag chain" are certainly not without

valuo in thie field*

Martinet himself docs net overemphasise the import¬

ance ef thie particular cause of sound changet "On ne

ee laseera pa* de rep©ter que personne n*a Jflrnii

pretendu que lee facteure phonologiques internee sclent

le* seals* nl memo neceasairecient le* plus deeieif*",^

but sosse of hie claims can be substantiated by reference

to tho STch material*

1) "tine opposition phonologlqtxe qui sort a

maintenlr distinct* de* centalnee d« mot# parmi le* plus

frequents «t 1©« plus utiles n*opposera-t-elle pas una

resistance plus ©fflease a 1*elimination que cello qui

44
ne rend service qua dans un tree petit norabre de eae?"
 375

One of the Bch interfere we have dealt with in thle

wrk is that between items of the sets fear and fair*

Fry**5 baa «hown that A3 / and /!<?/ are respectively

seventeenth and eighteenth out of a total of twenty

HiPi vowele in their degree of frequencyt and that they

between then account for only 0*55 percent of nil P.P.

••gnents. . i>y way of comparison. /%/ and /a/

between tuem account for 19*07 .per cent.

2) "Som avene ... indlque ••• comment lee

aeeeesitee phyaiologiques peuvent eentrecarror 1*integ-

ration phonolo-ique • 11 y aura, ontre «t iuj una

plua petite difference d'ouvertare qu'entre t#J et iij.

bien que 1*angle emxillalre eoit le memo dane lea deux

la other words« *tha physiological asymmetry

of the vocal tract Interfere® with the achievement of

permanent symmetry (for example* in the matter of

distinctive degree# ef height in vowel phoneme#* there

ie more latitude for keeping the front vowele apart than

for the back vowels),*^ Martinet himself given as an

example the fronting of At/ in Assoree Portuguese, which

releases more space for the maintenance of the contrast

between the remaining back vowels. /o/» /o/ and /a/.

Ae far aa the Beit material is concerned, in the case

of the short (A) vowele the foot that there is less space

for the realisation of the back vowola /V/, /#/ and /a/
has lad to the fronting and lengthening of /»/ in the
 377

Table 16t Sab«Sf»ita A (11)

I ©
e »

The deficiencies of Fudge** scheme with respect to

phonetic space are made good in the Iffoh dlasystsm by

the phonetic vowel system or inventory shown in Table 11,

Xt is this systematic phonetic vowel system which de¬

monstrates how and why structural pressures of this kind

©an occur. This system is, of course* in conflict with

the systematic phonemic vowel system shown in Tablo 6*

But events at the phonetic level are important from tho

point of view of phonemic change, since, through time,

they can affect lexical entries and tho systematic

phonemic system itself (of, the remarks made concerning

//»// and //<?// above), Xt is possible, for example',

that the less of the dlasystemle clement //?// la Itch is

the result of pressures at the phonetic level, (a con¬

sequence of the proliferation of long front vowels, /li/»

/•i/, /«»/» /««/» /•*/)# leading to a merger of /o«/ and

/XI/, - together with the fact that there are no important

morphological alternations which require //*// to be

distinguished from //#//, This proliferation of front

vowels may also have led to the merging of fear and fair

items, of course, particularly since all C vowels have

 n +t ot

%S tf & i A « 2* 9« af »* I* tf 1i &C ®2 1 0 af SX

!w 2as ©c Ie 0< tf « s af £ £ «* a M BS ©

a ** I©2© ; ot © i e af af © s af 1 > I

S 5- ^ef 11 €£ ef ef tf tf tf £ ► 34 af

af a

# :a A© K ©£ * - O efM « af A af m af af t * !

s i af « IH * fI• a 9* af 5O af af > 9 aft I

© af A © ef * sI a *T af a © af I © © *v if t& 34 4•< §

11 © sa af i IH » a© © 9 S I »4 C• « I © § af ? afO ef 4v< © *4 a a« S

3 9 © af J af a af a » a tf © ©£

! ©* I a1I I x 1 af © a af 1 H tf af af 4

31 af s» SJ af § © af •2 © I8 « I

« »4 «i
i af © f ef af tf af af © A 5 O af £ © ef > ©S af tf af A © i af a

I af ef © tf af I © tf © « ef 33 5g Wtf af £ tf tf af tf tf © ef 9 tf ©« 3

8 £ © af ©©£ © tf £ tf af « tf IS ©af$ 9 af 1 af af af tf « 9 Ief£

 379

acute variation, this ia an in«xpUeabl« foot in tamo

of diotinoivo f«tur« theory* Any theory vbioh dissolves

the geometry of articulatory apace will have difficulty

with this phenomenon, la New York City# a complex oat

of ©©variants has provided empirical verification

for the existence of relatione in phonological apace#

Covariation of /oh/ and /ah/, of /oh/ and /*#/, or of

/oh/ and /ay/ ,,, are not easily described ia the present
form ef Igenerative phonologicalJ theory#**®

Phonetic Realisation Kuleo

The funotien of phonetic realisation rules has al»

ready been discussed above* They convert extrinsic

allophonoa into intrinsic allophonee, ideal pronunciations

into actual pronunciations* Obvious rules of this type

are«

a"/
/*/ 111

/•/
/3»/ > Iff]

We have already stated above that rules of this typo

should in faet be formulated in terms of the framework

developed by Ladefoged* as far as consonants are con*

earned* except that all relevant parameters for a

 380

particular language should be totally specified for oaoh

segment. The phonetic realisation rule for /kh/ above

should therefore readt

Articulatory place 4.4

Glottal nonetriotion p
Nasality 0
Stricture 1

/kh/ Affrication 0 /.\ /Tl/

Olottal timing 3
laterality 0
lip rounding 0
Secondary artieulatiea 0

Otter realisations of /k / will dtpead on the following

vowol and other item* constituting the phonetic environ*

merit* For example. the full phonetic realisation rulo

for A11/ would also have to include i

Articulator/ piaeo 5.7

Olottal constriction 9
Nasality 0
Stricture 1

/kh/ Affrieatlcn 0 /. ii ii mi
Hi/e •/
Glottal timing 5
laterality 9
lip rounding 0* 5
Secondary articulation 0

Rules of this type oan also ho made more general oo no to.

in this ease, apply to /g/ as well, for example.

 381

It is at this point that w* art? able to produae

the different variants of the phonological variables,

ll&jt]# t*J# arwl that we discussed above* For

examplet we have the diesestestis phonetic realisation

rule, role 63*

63)
A,p. 3 A,p. 3 A*p* O

a.e. 9 1 o*o* 1

Has* 0 Has* 0 Has* 0

striot* I Strict* 1 Strict, 1

/t/ & Affrio* 0 Affric. 0 /"•>w

Affrio* 0
o.t. 3 o*t# 3 o*t. 3
Lat* 0 lat* 0 Lat* 0

L*r# 0 L»r. 0 L,r* 0

S.a. 0 S#a# 0 S»a» 0

That is, /t/ * « L«i ~ Lt»i ~ Ui

f • f (Style, Class, Ago).

This rule specifies that the variable (t) is the realis*

ation of, in this ease, the systematic phonetic element

/%/ at the performance phonetio level, and that the real**

ifl'dioa of /%/ will he it], it£j or (.3) according to tho
factors spscifisd. This ruls is a diasysteuic ruls which

applies to the whole spesoh community* Wo can assume

that all speakers have this rule as part of their linguist*

ie competence* Performance, however, will vary widely

from speaker to speaker, Rules of this typo are pro*

•umably also applied in a reverse direetion in the social

 38S

interpretation of utterances md« by other speakers.

There vlll be similar diaaystemie rulea for (p) and (k)«

The rules will also be somewhat more complex than pre*

seated above because intrlnsle allophonic featured euoh

aa lip-rounding oust alee be indicated* to must aloe

build in a restriction on the selection of the variant

UJ. Th« following fidh pronunciations* for example*

nuot be taken into considerations

MS&J& i but not a i'eitllj

not at all [nailst»jt*J but not m ina«£a2o«*]
isn't it tint'aaj but not « tlalal]
it (.phot*3lj but not « tph§l3l]
about it
[bjWt'aa] but not
fi°* ,*v *° , h. , , .
come igea st'ak |®J but not * [geaaiak

The following* however, do occurt

won't have to go to i.v§a«*aag|?3j

✓

don't ought to td§2»%za j

something to eat [a^fnd2fitj
Lwsafaj

The restriction is* therefore* that a glottal atop cannot

oeour both before and after /<?/ or unstressed /!/♦ (The

went into example ehows that this possible in the case

of stressed /%/«) Note too that in the phrase get her to

 383

come, with /3i/ wther than /$/ in her* two glottal

•topo are possible* The reotrletion also appear* to

ferae the first of the /%/ or /t*/ consonants to ho

realised ao ^tj* t**2 or L^tJ* rather than (,2j* except
where the second /%*/ is in initial position* ao in

/ret her to cone and not at all* (The fara at all is

pronounced /a«*t*e«l/ rather than /st-o»*/)»

Va nasi therefore nodlfy rale 63 to read«

W ~ t*jtj /
/%? "Iai" ' * 0

i*i — L*ai ~ UJ

Symbols like i,tlj are of coarse short-hand devices

representing complexes of Ladefogred-type features as

shown above*

This# then* is the forts that phone tie realisation

rules will tako for eonsonants« We have already in-

dioated above that the realisation rules for vowels will

have to ho of a different ferst* This is not without

Justification in the literature. It has been shown, for

example* that kinesthetic and tactile feedback are catch

 •

8 H• 3 8 m »* £ « wkM • 8 Votj 5 1 8 • 3• 0H» « • I 4* *t ec * s< « e 4*

0" tr« «< *« • • a# 1 M4* e »4 *« © 1 tS »4 »

1 1 fH «0 »4 Vk4 4* ►s0 4• 4 1 »4 H■ « £3 « 0 *4 4« g
rf 1• •

& tI »♦ £ • sa 8 0• fI 3 & cf»- i • 3 SH 4* 8 * #

s<
 386

63) S i** ""V / /1/

A/ >

S m t (Stylet Classi Age, Sex).

66) is -e !•*•«iwtee«*!.«* ij/M.(/g/

/•/ » [«-« e<Hs«s» t«*&«»/
. t /
.11/6/
^
t e •« e<Hs«SM«HRC «*eeS"ME i j

g ■ f (Style, Class, Age).

67) ( l*-?i //Ph/ )_

/D/ »5 ' /M )
< iHi

68) /»/ *

g m t (Age).

69) M

8 m € (StyIs, Class, Age, Sex).

70) /IV ♦ c Lli-ft-ji-?ii

8 » f (Class)

71) /«!/ ■> ff J

8 m t (Style, Class).
 387

72) /uiy ► e

get (Class, Age),

73) A»l/ -—•» g

g » t (styla* Class, A&s, Sex, /««/)*

7^) /<£*/ • e

8 * t (Class}•

73) /mm/ ) g Lao-a6W «-;>*]

get (Class* Age, /»!/)•

76) /Ou/ """■ ""■"") g j.uf-u-' j-ga-so j

get (Style, Class)•

77) /»«/ g t fcu«aa-»u«aoi-0o j

get (Class* Age),

78) /#«/ ■» ■■ ■«"-) if*** *** *"*7* i

79) /* »/ '•") i
m si**
l C
i«4? i«J« J
*■ i-

60) /»»/ —■« "-> t

.-a.-e
t t-
.-t
«•
i j*

81) /a«/ ) #

get (Class)
 383

82) /»t/ ) i®- i-aft ]

83) /»•/ » ' '0 L OH | «•»*< *|J

84) At/ ■ )

85) /),/ >

These ralM are in tom« eases toaevhat simplified (al*

tlten^i they give a feed idea of the fera phenetle real¬

isation relies in th« l»ch diasysten msi take)» ¥• will

aXao need, for exaapie, rales to lengthen vovels before

veleed consonants*

Not* that those rules allow for the possibility of

tho following pairs of systwatie phonetic vovels not to

be distinguished at the performance phonetic levelt

A/ At/

A/ /»»/

At/ At/

/m/ At/

fof At/

/3«/ /a»/ oto

Other cases of overlapping realisations include /ot/ and

/#i/* /»»/ and /»»/» /ss/ said A */, and /X/ and /*/.
 389

Kot* finally that all realisation* of A, C and D

vowels have voiceless counterparts which are realis¬

ations of /h/*

Performance Mutation Pules

V3 have reached a stage in our Soh diasyatemic

phonological grammar that is probably much mors d*tailed

than that normally achieved by the phonological compon¬

ents of generative grammars* He have the stated aim,

however* of generating an output which approaches much

more closely what different Itch speakers actually say*

The level we have so far reached is adequate for fairly

slew* careful Kch speech* Xt Is not adequate for rapid*

casual speech* To produce forms of the type illustrated

in the three Not* utterances quoted earlier on, and to

handle casual rapid spssch in general, we will new

develop a set of rules which will be relevant only for

this type of speech* These rules* which will incidentally

demonstrate that at least some performance phenomena are

subject to rule* will of course bo optional* Sineo they

will apply to and produce elements at the performance

phonetic level* we shall call these rules performance

mutation rules*

Xt is preferable te have this type ef phenomenon

handled at this lowest level* because it is then a

 390

relativity simple tatlw to derive casual speech frou

careful speech* It Is net d«sirttbl« to handle par*

fornance phemjtsena of this type by phonetic realisation

rules such csi

Ant/ ■"■"") iisj

because this complicates the realisation rules ami asone

that, in addition to the rules produeitv: intrinsic

al1ophone», there will have to be sever®1 alternative

rules* with the one wo have lust postulated as only one

of the alternatives# In this particular cose# for example*

the omission of Inj in item® Ilk® plenty occurs only

where ft/ is 12j» so the above rule would have to bo

prosenfeci in just that form* if a complication of levels

is not to arise, We will therefore have Instead rules of

this type«

(nj » fi t|j l2J

The trua forts of this phonetic mutation rule is the

following.
 n \C 4*

OS 3V

• .tf * • •

* 0• •

094(«AS )»3i}(<6j04[ o^rnw

i^#4*
04 « « #

tf 4( tf « »

«%»4W *• e i »*« r v* r• •

• tf •

it >4 • 4* •

. tI * *

®
« »* *
<0 «

»rf t 9 kI *4 1 » t 4- s ► I tf I 9 tf •% tf #4 * & tf

1 o£

H « tf tfi

tf 9 % s * i• < » tf tf 1 *4 I I*J tf I 1 *4 0i

s «i I satm « 9 tf S I 9>

I P X tf tf tf m4 SJ tf tf tftf I tf■ *4 1* § 1 1 1 **I * BJ

£ tf I *4 £ #4 A tf tf I tf 4* I I I a 4• tf » tf t m tf tf

tf a a

:i r* T

I I tf «i tf tf i« #• # « et 1

B £ »•- H tf tf tf • 8K » tf

tf tf tf I I g

tf t tf tf tf Itf9 tf Ib tf tf 0 tf *I tf tf *

»» I *4 tf QJ ef

tf A tf 5 tf tf tf I ii »g I tf tf £ «tf »I15 jtfi- *-tfa

tf tf v> g4 tf c© «►
: £ tf tf tf£ tf I 3 tf tf tf tf 1tf 1
 I 3o• 9 & e 9 »* jt i o i i• *4 9 £ 9 Is

II i 1 s *4 ij' i 9 4% »4 4 • U9 i e »4 O M« 5S

1w 4* 4 % 9 4 9■ 4* s K 44 S 40 5 « 9 9 41 « *% « 4 9 e 9- 9• •► 2 !

5 9 5 »4 4 9 *I »4 i »4 4 »4 4<« ► 9 SK *S 4§ « 9 9 cf . & © a 9 9 « 9^ 49 s.

I 4s i »4 s1

i9 2 »4 3 3< O 9 « SI 9 « *4 4 9 »4 E 9 I *4 5 9 9 ef »4 4 9 1 * »

4 • A• 4, • tf 4<, * tf te

O# 9 * « O• 9 %*
*8

0©MHI«%* 4lt»N<cf« 0d4N\>l<i

55 C t

t4«»*
49 es4 9 >g

#f h 9 »4 « *4 9 *4 m »4 U 4< 9 # * \r
4 NI W 4< 9 4 NI 4* 9 < *% *V tI 4« 9 • o

O • )t
© • e• o ♦ »4 • *«

5* 4O • 44 3* 4* < n

4 « tI • )• 9 a • il * o

ift • lf • 0 %• © « 4 \r
 »*> »

92)

93) i£4i > i<Kj

9k) if ■■■■ > La-j

95) l«-aj L&i

These rol*0 centralis® and open short vowels in.

rapid speech* Pots* toe* that rolo 92. applies not to

/«/ hut only to |»j*

96) L3'~**i '> L»-£«aj

This rule shortens the vovol in bird, first, etc.*

ami produces the typical can short forms with the wide

range of phonetic distribution discussed and described

by Kokorits and howma* and in Chapter Seven (the vari¬

able (ir))« Pete* however* that as it applies to a

phonetic area rather than ts a systematic pnsnetie

element it h»«« the effect of optionally shortening all

i««J-type forms* not only those derived from //!//♦

//«// »»d //a// e //r//* This is a desirable effect*

sinee pronunciations such as partner Lp^lnsJ were ro-

eorded during the survey* V« have* moreover, soon that

 393

/»/» too, eon have a realisation in the region of /a*/»

Xt is therefore satisfying for oar thesis concerning the

input to these rules to note that feme such as hob

.lob j (rhyming with huh) were frequently

recorded during the surrey* The input is therefore,

once again, a phonetic area rather than a systematic

phonetic element* [.ttj j.'PJ whatever its origin*

97) Ibn-an-tfu-dJuj »"•>

This rule deletes the glide from all diphthongs

that result from systematic phonetic B(b) vowels*

98) ^Stricture l) 1- ^Stricture 2J

99) i,Stricture 2J •> {.Stricture- 3j

These rules convert stops into fricatives,and stops

and fricatives into approxiia&nte. Thoy account for ffch

pronunciations such act

100) {.Glottal timing 3j ) {.Glottal timing 2—1J

This converts voiceless segments into segments that

are partially or completely voiced* Xt accounts for IVeh

 396

pronunciations such a«f

he celled

101) ta-aj ► p

102) t*i —-—1»

100) |.nj il ■) $ Zwww*i^g6

io*) tlj p/ t«i

103) Ivoiceless vowel J "»" "■) p

These ruloe all deists sounds as indicated* Koto

that rule 105 east also have a corresponding phoaetle

realisation rule*

/!</ > *

to account for speakers who "drop their &*»* even In

careful speech*

106) tVowelj ) {Nasalisation Ij

This rule nasalises vowels which precede nnsai

coneononis*
 397

107) 0 » UJ / 00*

This rule introduces a glottal stop befaro initial

rowels in «tre«s«d syllables, as in«

awrt& Ur if*J

Many other performance mutation rules are of course

required, in |>artieu2ar to delete vowel* and consonants

in certain contexts and to shorten long vowelo •

%*e are now in a position to demonstrate how our

three Rch utterances, 1# I* and c, can be generated by

the Kch diasystea,

A# M ,.Se,.aa

*fe can assume that the systematic phonemic under*

lying form for this would bo something approaching the

following*

Systematic Phonemic towel

//$»a,tS ♦ ♦ h«au,•* ♦ »,i«t, ♦ f*o*»* ♦ •#o»n»//

There are no dlasystemie incidence rulee that apply

in this case. and. subsequent to the phonological rules

of stress, boundary delation etc,, the phonological

 398

realisation rules eenvort this into the systematic

phonetic representation!

Systematic Phonetic jLevel,

/pmto d399t bom *t °gttu »n/

Phonetic imiiatioit rules now apply, in this ease

the rules of glide insertion and recyliahifieation, 60

and 62, to produce the follouingt

/pmtc /*j3#t hew st 4gl7u van/.

The next cot of rules te be run through are the

phonotie realisation rules, which convert this to tho

level of actual articulations* The result is something

like tho followingi

Performance Phonetic Level

§«*» dX 'fi-m wenj

This represents a possible slow, careful typo of

Nch pronunciation* Now the final cot of performance

mutation rules come into play*

 399

1) Rule 88 delete* the Initial In that**.

2) Rale 104 delete* the glottal atop before

1*J An that**.

3) Role 98 convert* i&jj into t3i-

4) Rule 97 convert* i£*j and iauj into i£ j

and 1,3] reepeotlvelf.

5) Rule 92 centralise* and levers the vowel*

of that** and how te i|j and ivj respectively.

8) Rule 101 delete* id j In jy| and In i?o.

7) Rule 108 naealleee the vowel ef on.

8) Rule 105 delete* the i»J

9) Finally* rule 10? ineerte the Initial glottal

stop.

The reeult 1* the required outputt

i*u 3ao -pZ "gvlnj.

 hoo

0« ¥mtt, if X*b>'.twins? out on a Saturday nirrht

Systematic Rhoaonto Level

//w*e.l# ♦ ##i»f. ♦ ♦ £*«e. •*»».iftfg* ♦ «#8Uit»

♦ ••o«n* ♦ ♦ S.W.*»t. e. r«»d#e««• ♦ n«I»*.//

Systematic Phonetic Level

/wet If <*« gtJuan sent pa? eset'a dli at*it/

The phonetic mutation rules convert this tot

/we* I fdtt "gOuwan «wt ana «et<adli m»it/

Perfornancs Phonetic Level

tv$* & faw "gfuwaa aUg ana s|-2adoi tiAilj

1} Rule 102 deletes tile in well.

2) Rule 101 deletes ^3 J in * *m. /;oinff. a,

and Saturday.

3) Rule 97 deletes the ^uj in noknrt and the

luj in out.
 koi

k) Rule 106 na«aU*«» the yovela lit ^ and

RQtnn*

5) Rule 103 deletes the Lnj's la oij sad notwr.

6) Rule 83 nasalises the In Saturday*

This produces the required output#

I,we 8 fai 6e§v «£ a sflei nail]

except that we need an additional rule to delete the Idj

In Saturday, The exaet environment for the operation of

thie rule is not yet clear.

Cm Ko X haven*t been down there lately

Systematic Phonemic Level

//n,ou,,« ♦ »«I«m ♦ , *lfft, ♦ b«e,n# ♦ d*au,n, ♦ $ ,a,r, ♦

l,a, t"*l» i, • i//

The diasystemle Incidence rule* rule 8» changes the

final //I// *o //©//.

 402

Syatewatic Phonetic Level

/mm d Int Jban ^dama pmts Icitlli/

The phonetic mutation ruloe then produce*

/n&u a Int 0t»d» ,d*M»A • ImitlXl/

Performance Phonetic Level

Liiau a r«a 4ba n 9de«n t- * l^Ulk j

1) Rule 97 delete* the t**i in go and the i*j

in down.

2) Rule 86 delete* the ^nj In haven* 4»

3) Rule 87 produces fro** InJ and y^J.

4) Rule 106 aaeallaee the vowel in down.

5) Rule 101 deletes idj in been.

6) Rule 103 deletes the tni** in been and down.

V* also require a rule to shorten the vowel in

there. We now have the correct output*

1*4 3 t2 *b *d® *r ifiii«fi

 403

We have finally, th*refor«, reached the »laf«

where we are, in theory at least, able to generate all

types of Heh English, including casual, rapid, we

speech, iron the same underlying system* One ef the

obstacles in the way of achieving this in practice is

the complexity of the phonetic realisation rulee and

the performance mutation rules# In the next section we

ehall be making a few suggestions en hew te attempt to

simplify this problem*

The large amount of variety te be found In Meh

English is the result ef the numbers ef optional rulee§

at all levels, and the socially determined rules with

variable outputs, which conetitute the ISeh dlasystem.

ft have been able te present only a part of this dia«*

system above* Wote that all ffch speakers have, in

principle, access to tho entire diasyatem* in their

actual speech production, however, optional rules are

either employed or rejected, with the result that any

given a peak* r only reflects part of the diasystcrs in hie

performance*

Artlculatorr Settings

Finally la thie chapter we ehall briefly discuss a

tcplc which should be of some assistance in formulating

and simplifying the phonetic realisation rules and per*

fonsanee mutation rules*

 405

Titer* are alee fear sab-groups of type A

B* (a) longitudinal modifications

(to) latitudinal modifications

(e) tension modifications

(d) nasalisation

This classification of setting components enables

as to analyse and describe reasonably accurately the

distinctive sotting employed toy Heh we speakers* This

typs of setting distinguishes Rch WC speakers not only

from HO speakers tout also from fAn rural speakers and*

of eoaree* from epeakors of most other types of English*

Xt also characterises some otherwise MC speakers as hawing

coma from a WO background*

The sotting typical of heh «c speech can toe described

in the following way*

Settings ..of, ,the Larynx

Phonetion Typesi There is a tendency for heh WO spoakers

to employ the phonatlon type which has been characterised

56
as "creaky voice"* There is no corresponding tendency

in the speech of MC informants*

Pitch Pangeat The pitch range used toy both male and fe¬

male Rch WC speakers tends to toe high rather than medium

or deep*
 1I i e. H I HI * m 4r S »4 % tif KI tt €

*5 oh i 1 s *© I 2 i 1 4* i s s*

8 4• C
M

3 *4 fr*«•1»*4 % »4 M »4 a
1 tf 0s
II
9 *4 *4 * J 2© s3 im !i »f 9u

*4 »4 2 £1 5 5 2 9 t9 ! i1* J
.

« 9 j

!I 9 | 5 $9 2t «4 V »
S a 3
 407

■nasality i An editor/ which can bt loosely

iabtUcd as aasftllty is a frequent component of the Neh

VC setting* end is particularly noticeable in Association

with the pronunciation of low front vowels (see per*

fermaase mutation rule* rule 87)* We shall see below

that nasality may well be an inadequate term for thia

feature*

The combined effect of these components of the

setting is to produce a type of speech that is somewhat

harsh and motallio* Jlany of ths HG informants in the

survey considered this type of voice quality to be

particularly unpleasant*

Host of the various components of thia sotting would

appear to have little or no effect on the articulation

of individual segments* This is evident from the fact

that *MC segments* can bs superimposed on ths WC raised

larynx, voice (and WC segments on the HQ neutral larynx

voice)* The features ef nasalisation and vowel fronting

and lowering* on the other hand* could be considered to

have a more significant effect en individual segments*

Thle* however* is entirely due to the fact that they

annear to affect only certain vowel segmentst front and

low vowels* It is not in fact possible to divorce eon*

elderatlon of any of the component# from a consideration

of the setting as a whole* sine© there are good reasons

 408

far believing that all th« compoaentf are anatomically

and physiologically linked in some way* The nasalisation

and vovel fronting and lowering, in ether words* cannot

be considered to be properties of these particular ««g*

taents where their effects are auditorily the nest marked#

They must be considered as integral feature# ef the

overall setting*

The general connectedness ef the different components

of the setting can be described* somewhat tentatively* in

the following way* ft would seem to be the case that

the component of tension modification is the most import*

ant and fundamental component in the production of this

particular type of setting* It is net simply tension

modifications of the supralaryageal vocal traet that wo

must consider* however* but modifications of the vocal

tract as a whole* Me must also take into consideration

muscular tension and effort at the sub-glottal level*

The facts suggest, for instance* that the If oh ¥€ setting

is characterised by a relatively high degree of effort

and muscular tension throughout the spoeeh production

apparatus* Thus* the three laryngeal components cf the

ffeh «C setting - with the possible exception of Phonatlon

type * appear to b© controlled by this single overall

tendency to high muscular tension* The high degree of

tension* for example* produces the high pitch range

typical ef this kind of setting, since increased vocal

cord tension leads to a higher frequency of vibration*

I « '•H H ! tf *4 1 t% «9 0 4% tf O 4< 4• *4 *4 g »* tf I >4 «

i1 I « 1* J£ H N 4# »4 :3£ « i tI I 9 m* >+ C 9 I9 O

.*
»4 H tf t% I 9 5£ I s< *4 04 a Pa 3 1 !
.

I I 5" •

2stI 1 tf t% ►I % tf S tf £ J * g • *Stftf »4 4■ a tf t%
"

5♦ tf- I 4< »4 tf 4• 1 »4 0 04 WhA 1 tf 3 W »♦ C n ! tf

#4 Atf I* 0• I 9 § tf » tf tf *< I& tW 3 tf tI tf 4# tf

tf *tf »4 3£ tf ► tf 4* > tf 2a 5 S 4* tit 4+ tf 5

, 5 tf 9i
\e*
* *

H2 tf ftt tf 1 tf »4 &> 1 3 *4 s ttf tf tf ttf H tf tf c 0 tftf f tf tf H »

s tf 3 *4 tf 3 tf tf tf 4# »4 4 tf tf tf tf tf 9 tI 4f 3£ 1 * 11
£

i tf 1 2 1 4* § »* tf tI tf 6 tf 4* tf tf & tf tf tf tf tf »4 s I

I 4* t 3 M i tf u »4 2 t t 0

% tf H S tf ?tf A 0< tf i ttf tf tf 04 tf tI tf § *4 4* £tf *4 9 A9

i »4 1 3 »4 3 tf tf tI »4 2 tf .I tf • H *4 4< « a tf tf

tf « 4*< 0< 4* *4 * tf ttf tf

*4 a « > tf tf tf iff tf »4 K 0* 9 tf

4< 3 4tf t »4 9 !i tf tV tf 3S »4 «e tf « « e tf 4*< tf »4 » 4* 04 A

 klQ

the high Bttssttlw tension of the wails of the pharynx*

The ten» "nasalisation" employed above is for this

reason probably not adequate*

In view of the fact that all the component * of the

•ettln® appear to be eo cloaely intsr-conneeted, it

would seem legitimate to rolntorpret rule 89 ae a setting

rule* ef a rather different type fvem the performance

amtation rules which delete segments ami handle certain

assimilations* The nasalisation associated with low

front vovele will then be seen not simply as sua «n*

motivated phenomenon to be found in casual ¥0 speech*

but as part of an overall artioulatory process* Motet

too* .that not only rule 89 Is involved here*' .for

example* rules 9® •» 99* which centralise and open short

vowels* ean also be accounted for* at least lit part* by

a setting where the centre cf gravity ef the tongue is

optionally lowered in casual speech as a correlation of

the raising of the larynx* It would clearly bo an ad*

vantage if we could replace these six performance

mutation rules by one setting rule* There is therefore

some Justification for proposing a set of setting rules

of the following typsi

103) focal Organs *) g (Sotting 1~ Setting 2)

g m t (Class)

109) Setting 2 * High Muscular Tension

 411

110} ( High Pitch Range

High
Muscular
ToneIon i tiigh Loudness Range
Tense Yeeal Tract

| Raised Larynx

111) Tongue Fronted and Lowered / Raised

Larynx

112) Valtam 4 Lowered / Tongue Fronted

(Rulee 111 and 111, at least• are optional,sine®* For

example* the fronting of the tongue that ie facilitated

bp the raising of the larynx is not physiologically in*

writable! it is possible for muscular adjustments to bo

made eo that the eoutre of gravity of the tongue is not

fronted*} The rule* prodoeet

(a) The high pitch* loudness* metallic

quality and reload larynx typical of

Neh ue speakers|

(b) 11»e fronting and lowering of vowele

in phonetic spaee§

{©} Blade rather than tongue*tip artie-

miction of alveolar oonaonantei

(d) The nasalisation of lew front vowels.

 kl2

Ttass© rules indicate that the components of the

*»'ch wo setting are initially dependent en the relatively

high muscular tension in the vocal tract* They alee

suggest that some int ©reconnect »d sound changes oeearring

ostensibly as the result of pressures in phonetic spaos

cosy in fact be due to changes in the setting* Koto* for

Instance, that the above rules account not only for the

fronting, lowering and nasalisation of certain vowels -

which do not occur to a very narked extent in the speech

of older informants - but also for the fronting of /»/

from id"** J to ia1-* j» which we have discussed above as a
sound change, Vs can hypothesise that the fact that

many young and middle-aged speakers have a front vowel

in the lexical set ef box is due in part te the fronting

of the centre of gravity of the tongue* It is also

possible that other Neh sound changes, such as the is(J

vowsl in items of the lexical set of fear* are due to the

lowering of the centre of gravity of the tongue, although

this is mors doubtful. It is instructive, on this point,

to coispare the pronunciation of elder #eh speakers with

that of younger speakers, (The categories "older* and

•younger* represent only certain statistical tendencies

associated with different age-groups, of course,)

Consider the following!

Toungoq

hymn ihiwj U«J

better 1*42* J
&22SSE ihf»*pj i£»<? j
Changes of this tyv« ean be deal ft with buc nor® easily

if they ere re/carded ns i?r-- result or a sin;:!® chatv?® in

setting than if each change la rr.-rarded as en individual

event*

If ve incorporate synchronic rule® el this type

into the -oh feiasyotco* ooreover# the statement, of both

the phonetic realisation and pcrfor«.diuce smt ;tion rules

ean he much ciftpiifiod* U eon# far cfeonplct replace &

phonetic reallection m$e <&i t-.-.co

/t/ ■■) G ti--©J

C ® f (Claa®, Af»c)

by the rulet

/*/ > 11-J

and allow the setting rules to produce t.ho nor® open

vowel where appropriate*

Another advnnta -® of sottin.-r rule® is that tney can

relate different types of "leh rnjlieh to each other in

the diasyotent in a tauch no re /generalise--, and significant

way tiiAn a eerie# of individual rule-:# different social

types ol Nch "".owlish my bo charactc ris<vl by the presence

or absence of# say# rule lli| rather than by a whole

series of rule** This 1-i cm important mint# since it is

 klk

elcar that perhaps the alalia most socially ai^ifieant

feature of linguistic differentiation in Itch is the typo

of voice quality produced by the particular type of

setting employed by o speaker. It is in any. ease this

feature which most clearly distinguishes VC from MO

speakers. This point* of course* did net emerge at all

from our atomistic analysis of the co-variation of

linguistic and sociological phenomena in Chapter Eight.

Setting rules can be regarded as ancillary to the

phonetic realisation rules« and must come into play

either before or simultaneously with these rules. If vt

in fact place the sotting rules before the phonetic

realisation rules* this will permit us te realise

systematic phonetic elements according to an already

established type of setting* . The outputs of the phonotie

realisation rules will then have been filtered through

the setting rules to produce the particular output

required in each ease. Exactly how this is to be done

60
in our model is not clear# but we might propose a

mechanism each as the following*

if Tongue - Fronted and Lowered

then /«// ■■■■ ■■" !«-•♦ j

M — ► • J

Hots# however* that if we are to allow setting rules to

play any part in sount changes of the "phoneme merger*

 *13

typo, we arust allow for the filtering offset of the

setting to he skewed in a©»© way, so that, say, book

vowels can he nore affected by fronting of the centre

of gravity of the tongue than front vowels*

Surinary of Pule Tvpea

Syllable generator Hole ■■

Rhyme 111 ") Peak ♦ Coda

Co*»occurrewco Restriction Pule

2 (153) —1 <B)

Piasystonic Inventory Rulo

//Y-<XII // ■■> // r^nt//

Phonological Mutation Rule

//Y// > //X//

Piasyatemlc Incidence Pule

//Y^IIIa// > //Xpl//

Phonological Realisation Pule

//A2b// » /V/

Phonetic Mutation Polo

/W ♦ /3/ » /3i/
Phonetic neallaatlon Pule

/«/ ) 1 e*»e —c-ej

Performncc Mutation nnlo

L#J * ft/**

Artlculntorr ^ettln'r Rule

Velum ■■"■ •■■•■) Lowered /

/ Tongue Fronted
 aKY.v,t.j

j-'doxxo-K a • n^'.iiixsno
 418

lit our sample* we ©an make inferences about linguistic

©hang©*, and describes this technique as the study ©f

linguistic chance la "apparent tlne".^

hare already seen* in the oaee of the variable

(e) la Chapter Eight, that it ie possible to plot the

covariation of phonological variables against the ago

croup as veil as the social class of the • poofter. Fig.

21 in Chapter Eight shows very strikingly that (e) has

undergone and is undergoing linguistic change. Xn spite

of this possibility* it would of course bo preferable to

make our study of linguistic change in real time * rather

than in apparent tiwo* either by comparing the results of

a survey with those obtained in earlier surveys* or by

conducting another survey Of the sarae typo after a period

of a number of years# Xn this particular case the first

alternative is not open to us* because no previous sur*

vey of Neh English has been carried out* while the second

alternative* although possible* is of no assistance In

our present analysis of linguistic change*

The disadvantages of a study of change in apparent

tine are that one must ta'..e Into consideration, first* tho

fact that older speakers nay well have changed certain

aspects of their linguistic behaviour during their life*

time# and* secondly* the fact that systematic changes of

the typo known as age-,tradtnc? may occur in tho cock unityi

differences between older and younger speakers may be

 kl9

repeated la each generation. These disadvantage* can,

however* he overcome# First# chancrea made by speaker*

during the course of their lifetimes will generally he

minimal in CS» and ao the study of linguistic change ean

he coaoantrated on this particular style# Secondly, the

exact nature of theee changes* (if cutty)* together with the

possibility of age-grading* ean ho investigated and con*
trolled for hy reference to historical records* Indeed*

if the study is to he accurate* this must be done* In

this particular case v« are able to cheek the results of

our analysis of change in apparent time hy referring to

the rural dialect records made hy ttekerlts, bowman and

kelson Francis* The two types of material are not strict*

ly comparable* of course, as they were obtained by entire*

ly different methods* The earlier records* moreover* are

rural CAn rather than urban if eh* They can* however* help

to confirm that apparent changes in rich English are not

due to age-grading, and can act as a general control*

Phonetic and Phonological Change in Hch

ire have already discussed certain aspects of phonetic

and phonological change in Hob and A, especially la Chap¬

ters Seven* Eight and Ten* - Particular changes that we have

mentioned include the following!

1# The apparent decrease in frequency of the

FAn short £ in items of the type boat*

 420

2. The toss of til® /Vtt/ » /au/ distinction

between the lexical sets of nose and

knows in Essex and Suffolk* although

apparently not in Nfk*

3< the increase in the use of /Uta/ rather

than /9m/ or /U/ in item of the lexical
set of hoot.

4# The aerger of the vowels of the lexical

•«* thyre.

5* The loss of final j in iteas of the

lexical sets of here, there, poor, Pore.

6. The merger of the vowels of the lexical

sets of name and nail.

7* The loss of older EAn realisations of

(lr)» fP j» l**J» lAJ# l®i* which are

being replaced by L3f] in items of the

type church, first*

8* The spread of h«»dropping to Nch bat not

to the surrounding rural areas*

9* The phonetic change of the variable («)*

as portrayed in fig* 21 in Chapter Eight*

fren it j to ^aJ*
 421

10) The phonetic chan;e of the variable

(a) fro® through j.wj and t^*J

jaiej.

11) The phonetic change of the variable

(Z)t with an increase in th« frequency

Of tho variant (I)~4 * tsij*

12) Tho phonetic change of the variable

(o) froa i«j to t»i* *r from t«j to i,aj«

£**9 ftAn Shor^ 0

In this chapter we shall discuss those changes in

rather more detail* ¥o shall examine the canoes which

led to the changes # and the mechanisms by means of which

they have become* or are becoming, spread through and

established in the community* We can begin by analysing

mora exactly the form that the loss ef the fin short 2

IS taking in Heh*

material presented in Chapter Seven suggested that

the short o was losing ground in rural BA. is far as wo

were able, to Judge# there appeared to be a lower per¬

centage of short o forms in the sfd records made by

Francis and Ellis than in the pre-war records made by

bowman and K&kerits* W© also saw* in Chapter Eight*

 kZ2

that even the Mob LWC uses no mere than H2 per cent of

the possible total of forma with short £ in CS» and that

this k2t per cent consists to a large extent of a relative*

if small number of lexical items of high frequency. We

can expect that the loss of the short £» If such it is*

will be more marked in the city of Nch than in surround*

ins rural districts* so that information concerning the

current position in Mob will be suggestive for future

trends in the whole of v. A*

The extent te which the fa short £ is or has been

involved in linguistic change in Mch can be Investigated,

in apparent time, by calculating the average index scores

for ths variable (o) obtained by each age-group in the

Neh sample. This information is portrayed in fig* 1*

which shows the distribution of (o) in apparent time*

as in Fig. 21 in Chapter Fight* average index scores

(ranging from OOP to £00) are shown along the ordinate*

and the different age-groups (ranging from the tea to
nineteen year-old group to the seventy year-old-and-over

group) along the abscissa# The lines on the graph connect

average scores for each of the age-groups in each of the

four contextual styles* The slope of the lines across

the graph from right to left will indicate the direction

in which the linguistic change* if any, is progressing#

It is clear from Fig. 1. first of all, that we do not

obtain the olear overall drop in (©) scores from right to

 423

left fteross the graph that we might have expected from

ear supposition that ehort o it dtomting In frequency.

CeHainly, speakers la the two oldest age-group# have

higher scores# generally speaking# than younger speakers#

particularly In CS (on which we are eoneentrating)• There

Is also a general fall In seorea* indicating fewer (o)*4»

from the oldest group down to the forty to forty-nine

group* There Is# however# a marked rise in eeores from

this levol to the scores of tho younger age-groups# IIow

does this fit in with the supposition that (e)-4 forms

are decreasing in frequency - and does it indicate that

short © forms are again on the Increase?

In order to solve, this problem wo must compare the

pattern shown in fig* 1 with Fig. 2* fig* Z shows tho

age differentiation of the variable (ng>* There is no

reason to suppose that there is any Unguistie change

currently associated with (ng) in Rch. General observ¬

ation and comparisons with the rural dialect records

suggest that (fig) Is a stable marker of social class and

stylistic context in Nch# and that it is fairly constant

throughout all ago-groups* CS and FS (ng) seorea ehow a

symmetrical pattern of age differentiation# with a steady

fall from seventy-plus scores to forty to forty-nine

scores# and then a mirror-image rise from forty to forty-

nine scores to ten to nineteen scores# Comparison with

other variables which are not involved in linguistic

change in Nch suggests that this is the normal or neutral

 ) tji/j »?V>| k,VL £UA»A, I- fi

A^e

ftfj. *». Vft.i*iOk •;!2/'v^ J OtV*£\_ hfijG,

¥•

%3> % (p)"4 OM ShjU- OA'^i 'iyO _

 k2k

form of ago differentiation pat torn* All variables not

undergoing linguistle change have a very similar syw*

metrical pattern of differentiation*

This means that the rise in scores for younger in*

formants shown for (o) in fig* 1 is typical of the Nch

phonological variables as a whole* and is therefore not

indicative of any change* The significant deviation from

the neutral (ng)*type pattern lies in the fact that

speakers over slaty have higher scores than the typical

pattern wonld lead one to expect* is can therefore say

that there is a linguistic change associated with (o) in

Nch* in that elder informants have higher (e) index scores

than younger informants* This suggests that younger

people use fewer (5)«4 (short o) forma than older people*

Before we discuss (e) scores further* it is interest*

lag to speculate as to why what we have exiled the "neat*

ral* age differentiation pattern,which characterises

variables like (ng) that are net involved in change* has

the fens that it does* Xt would appear to be partly a

result of the fact that age and social class are not
independent variables in British society* fiteauat of a

general increase in educational opportunities and the

standard of living during the course of this oentury*

younger people in Britain tend* on average, to be of

higher social class than older people (see Fig* $ in

Chapter Four)• Thie partially explains why the seventy*

 h26

That there are in fact two conflicting patterns hero

i» confirmed by the pattern of age differentiation of («g)

for VL3 and PFS, shown in Fig* 2* VLS and RPS are formal

reading stylos where the influenoe of the educational

institutions in stigmatising (ng)*2 forms will he moat

marked* For this reason* in those two styles* pattorn

(a) completely obliterates pattern (b)» there is a general

decline in ecorea from right to left across the graph*

from the oldest group to the youngest* In this sense

there has boon a change associated with (ng)t younger

speakers hdve a much wider range of styllstlo variation

than older speakers*

To return to our discussion of (e) scores* Fig* 1

gives us some reason to suppose that there has boon a

decline in the percentage of (o)-4 forms used in hch* and

that the abort £ is a recessive feature in Neb English*
The evidence is not conclusive, however* since the vari¬

able (o) measures all typos of pronunciation of items of

the lexical set of boat * The decrease in scores could

therefore equally well be due to an increase in* eay*

(o)—2 forme* We can therefore confirm our hypothesis

concerning the loss of (o)4 forma by the material pre¬

sented in Fig* 3« rig* 3 shows the percentage of possible

(o)-% pronunciations that were (o)«4» on average* for each

ego-group in the four eonteatual styles* The pattern of

age differentiation is si liar to that shown la Fig* 1*

but the fall in scores from right to left across tho graph
 427

is rathe*- mora pronounced, There is a fall from a high

maximum of 1»3 per cent for the sixty to sixty-nine group

in CS| to a low maximum of 12 per cent for the thirty to

thirty-nine group in PS* Once again# the twenty to twenty*

nine &m! ten to nineteen groups have higher scores than

the thirty to thirty-nine group.

We oan confidently assert* therefore# that the EAn

short £ has dsereased in frequency in Nch* and that (o)-*»

is mors common in ths speech of older informants than in

the speech of younger informants* The deolins# however#

is not extreme * and the evidence suggests that after an

initial drop in frequency around the time of the First

World War# (Illustrated on the graph by the sharp drop in

scorns between the sixty to sixty-nine group and the

fifty te fifty-nine group)# (5)4 has maintained ft fairly

constant Isvsl of between 10 and 20 per cent in Kch

English# There is no real svidoncs# that is to say#

that (o)-4 is dying out# susrely that it has declined*

The diesystemic incidence rules

//«// > //»//

is therefore still prsssnt in the grammars of even the

youngest speakers in the ffch sample*

The variant (5)4, then* has decreased in ite fre¬

quency of occurrenco in ffch English# Why has this

 k2B

happened? What was the cause of this linguistic change?

What was the mechanism by means of which it became

accepted in the speech community, and which ensured that

this change, rather than some ether (or none at all)

took place * and at this time, rather than at some other?

Zt seems clear that the main reason for the change

lies in a prior-existing linguistic difference (whose

origin it Is not cur task to discuss)« neither the pres¬

tige accent, R.P,, nor the accents of any of the areas

surrounding EA have the short o. This is the initial

cause. The mechanisms which subsequently effected the

spread of long ©, forme into Weh are the processes of in¬

novation diffusion, by means of which now forms arc trans¬

mitted from one area to another, and the soeial pressures,

within the community itself, for the adoption of more R,P,«

like forms, particularly in tho educational institutions,

Tho reason for the change occurring at this particular

time, rather than at any other, must bo ass timed to be the

increase in geographical mobility that took place during

and eince the First World War, and the increase in educ¬

ational opportunity that occurred around the eame time.

It ie also possible that there is a further linguistic

cause, Forby indicates that in the 1830*s, when he was

writing, the EAn short o was common* He states that the

letter g "has also in some words the common short sound

of the diphthong oo (in "foot")"»"* and cites as examples
 kZ9

hone* stone. whole. He also indicates that the letter

j§ la pronounced oj lit items of tho typo full. mill.

bash. (Lovesn and Kokerltz alao record items such as

roof and soot with taj.) This moans that pairs such as
coke > cook, foal t full* lake • look* must has# boon

distinguished by vowels of the typo l&j c |>j rather than

by louj t illjt It is therefore possible that# as Hfk

lAJ ^ fall was replaeed by l»J# there was sane pressure

to replaee ivi in items of the type foal, in order to

preserve the lexical distinction. This may# at least,

have encouraged the change from /%// to /tfu/.

of J00/

Wo stated in Chapter Seven that the older EAn

dietinetlon between items of the types nose and knows ie

apparently being lost.in EA generally# but not # as yet #

in Nfk. Kbkerits give# evidence which suggests that it

tad died out in Essex by the 193<>*e# although it wee at

that time still typical of Suffolk speech. The published

SEP material suggests that it had died out in most of

Suffolk by the I950*e* In Chapter Eight# however, we

stated categorically that ther? was no evidence of any

linguistic change concerning this distinction in Itch*

This statement new requires seme modification. Fig. %

is an age differentiation diagram for the variable (ou),

which is the pronunciation ol the vowel of the lexical

set of know, old, low. Comparison with Fig* 1 shows that

 *30

there is no general tendency for (e) and (on) to ho

merged in Heh English* even in the speech of young in¬

formants* fig* * does shew* however* that (ou) is

subject te stylistic variation in the speech of inform*

ants of all age*gr©upo orecot thooo aged seventy or over*

This means that (ou) is not a varieblo for the oldest

age*group in the Koh sample* The introduction of (ou)

as a variable, (with vowels mors closely approaching

those of (o) in formal styles)* therefore represents a

linguistic change in Reh* The situation described in

Chapter Eight, where M*iC speakers make a consistent bat

email distinction between (ou) and (o)« must therefore

be of fairly recent origin*

Shevelov^ has suggested that the prediction of

future linguistic changes is a useful and Insightful topic

for study* Can we predict that this change* the merger

of (©) and (ou)* will take place in Rch at some future

dateT There is no linguistic change currently associated

with this distinction in Neb, but the fact that (on) has
become a variable during the eourse ef thle century

suggest* that this is certainly possible* There are

three ether factors which also help to suggest that a

change is likely*

1) Koh ifC speakers taak# less distinction

between (on) and (o) than WO speakers*

their realisations of /ou/ and /Xh%/ are
 & 4-
jWj bi] S'b.f.e «u-A Ac
C&> j

75^ iy&i §E-g-i vo-i'i ■%•*•■

'J'"

fcj,£ *V&nAK<L (ft) bt* a^i Aq

»4«t
xfcV

ttkuv .J
L
jo-^i a»-2? 5o-3k, CJH$ to-CA "fc-V
4qs

f^,(g ?o (p)S b>, OaAVL *Vj&

if* a
»** I

50
 *31

phonetically closer. This weana that

prestige way well become associated with

a phonetically email amount of distinction*

and that the distinction any therefore

become increasingly email.

2) Neither the prestige accent* nor the

accents of those areas adjoining and

surrounding Nfk« make any distinction

between the lexical eete of nose and

Jjgeews. It is therefore likely that the

diffusion processes which hare already

been responsible for the loss of the

distinction in Essex and Suffolk will

continue the spread of the merger into

Rfk. It is also likely that ±n~aigration

and London orerapill projects will hasten

this process.

3) The phonological rowel systems presented

in Chapter Ten suggest further that this

merger any take place in ft eh. We saw

there that the diasyetemio inventory rule*

rule 3t

Y^ll

has become generalised in the speech of

younger people to rule 3&»

 <»3 2

a! ♦ 0 /-
T ^

V© have also seen that the systematic

phoneme //T^IIb// « //I// ia likely

to disappear very soon from the Reft

diasystem. When this occura the relevant

diasystemio inventory rule will then bet

•j
b' > „ / T-<III

It would therefor© not bo (surprising if

this rule» in turn, became generalised toi

*)
b'
* ^ /
Till

That 1st

//S//
\ > //ai//

and.

//S//
//ou//'
} » //ou//
 *33
The PeDlacemcat of /Mm/ ^nd M *>y /*'*/

We saw In Chapter Seven that the alternation of

/¥«/« /hu/ and /C/ in items of the lexical set of bgot

Is probably the result, at least in part, of a 1incutstie

ehangc. Zft older rural speech, /Mm/ was most coneson in

Nfk, and /M/ in Suffolk, Forms with /Vti/ represent an

Introduction frets SUP* or frost Midland accents. The

result is the diasystemlc incidence rules discussed in

Chapter Ten#

Vo can now analyse (u) scores obtained by the Sob

informants to see if there is any trace of this change in

Itch itself. If /tfu/ is, generally speaking, supplanting

/Mm/ and /X3/ in items of this typo in Nch inglish, wo

would expect scores for younger speakers to be lower than

those for older speakers# Fig* 5» which Is an ago

differentiation diagram for (a), shows that, this is in

fact not entirely the ease. There Is a slight but clear

tendency for scores to increase with decreasing age, in

VLS and MPS, That is to say that only the pattern

associated with increasing contact with prestige forms

occurs, and not the pattern associated with increased

educational opportunity (see above). This is because (a)

is ths only Kch variable which is subject to stylistic

variation and pot to elass differentiation. The higher

social class of younger people Is therefore of no eon*

sequence with respeot to this variable, (Note that this is

further confirmation for our "two overlaid patterns* hypo—

thesis,)
 *3*

The scores for C3 and FS aro difficult to interpret*

but on® possibility is that they represent something

approaching the neutral pattern* but with the drop in

scores from £00 at thirty to thirtjuntas in CS, to 283

at ton to nineteen* indicating a recent doorcase in the

frequency of (3)*4 and (3)*5* The evidence is in*

conclusive* but perhaps the most reasonable conclusion

is that a major change involving this variable* if there

was one* mist have taken plaee before the torn of the

century* and therefore does not appear in the age differ*

satiation diagram of Fig. 5* Fig. 6 shows, on the other

hand* that there is some kind of change currently assoc¬

iated with (u) in Ncht the ten to nineteen group has a

significantly lower percentage of (u)*5 short £ forms in

tfJUS (where all informants read the same lexical items)

than the other age-groups. This* together with the drop

in scores for the two youngest age*groups in Fig. 3,

suggests that /Utt/ forms, perhaps after an initial in*

crease some @0 to 100 years ago, duo to the replacement

of /9m/ forms, are now beginning to replace /V/ forms* as

well* in the speech of the young*

we can ascribe any increase in /Uu/ forms to the

influence of non*FAn accents. It is not clear, however*

if we can include R,P, amongst these accents, since many

11. *». speakers, including those in the Mch sample, use a

vowel that is phonetically intermediate between Nch /Uu/

and /9m/$ and indeed generally more nearly approaches

/9m/t (except before /I/). It is more probable that this

 *33

change Is the result ef the influence both of Midland

aoeents ami oi older types of »<«P. (aimson writesi

"Considerable centralisation, i.uj, appears increasingly
to be a feature of certain types of K.P*, especially in

the London region"•"*) In any case, /*?«/ appears to have

bseomo sua unaeeeptable pronunciation in many words for

sevsral MC Heb speakers, who produced it only in the

Pairs Test* to differentiate pairs such as boot i boat*

which are normally himophonoua in their speech*

The Merger of (er) and (tr)

It was suggested in Chapter Seven that the phonetic

merger of (er) and (sr) was a completed linguistic change

in Nfk# Xt was not clear, however, if the merger had not

yet been fully carried cut in Suffolk, or if it was under¬

going a setback there, nnder the influence of Home Counties

speech# Evidence presented in Table 1 gives the currant

situation, with respect to the merger, in Noh. Table 1

shows the difference in index points between average

Index scores for (or) and (er) by contextual style and

age-group# The difference is calculated by subtracting

•ach average (er) score from the corresponding (er) score.

The symbol * indicates no difference# or an (or) soars

slightly In excess of the (er) score* a complete merger#

The figures in Table 1 therefore relate to the amount of

phonetic space used to differentiate the two sets of items*

 kl6

Table 1* Difference In Xitdtx Point®

between (er) ami (tr) Scares.

qroupt 20r,22 22-22 Jtfete 32r-52 2°i

as *» «* * •
17 35
rs • 26 56 20 40
9 %3
HPS 4 3 49 51 53 33 33
SfLS 23 43 61 76 63 36 5%

There is no complete progression of scores from younger

to older# but the figures do suggest that the merger is

still on the increase in Neh, and is particularly marked

in the speech of younger people*

We have already suggested two possible causes of this

linguistic chango* They »ret

1* The small functional load carried by the

distinction between (or) and (tr) in English*

2* The proliferation of long front vowels in

phonetio space in £An English (which may

have exerted some pressure towards a re¬

duction in the number of those vowels).

There is* however* a possible additional cause*

Many commentators have discussed a linguistic change

that has taken place in ft.P, and in other accents of

 *37

English* iAClttdiag those of the Home Counties* Gimeon

writes« */aa/ now rarely exists in n.P., having been

levelled with /»«/, «*g» in aftre, saw" and "the level*

ling of earlier /ifd/t /e.9/ and /»»/ is now common. e*g«

la noor* oore. paw*/ we have also already observed

that there is a tendency in P.P. for /ea/ to bo realised

as I* t J« We eon therefore postulate that P.P. is in the

process of changing from System 1 la Table 2 to System 2.

Table 2i P.P. Vowel Systems

System 1, Syste^ 2

/la/ /vV /la/

Aa/ /ea/
/*•/ /««/ A*/

The series sf parallel changes that led to this

shift are the following*

Aa/ > /a*/ > /a*/

/«<?/ > A */

or* in the sytabolisation we have been using for the Nch

systematic phonetic vowel system

/Utta/ > /a*V > /a»/

A •*/ > A «/
 *38

Mow* the eldest Tfeh Informant, who woe eighty-nine

at the time of the survey#(and thirteen years elder than

the next eldest informant), had the following types of

pronunciation in lexical items of this kind*

here [hii^lJ /hits/

J&£££ /$1d/
poor lph9U*9 J /p*W/
r>oro [phou*9 J /pho» V
S£Z iPh»«j /pb»«/

The points to note here are!

(a) thero is no evidence of th» merger of

(or) and (cr)» which helps to confirm

our analysis that this merger is in-

Creasing in Nchf

(b) the phonetic mutation rules which lower

certain vowels before schwa in the ttch

diasystem do not apply in this informant*#

speech (including the rule /«!/ J /a/

) /• tsf% which we described as

obligatory in Chapter Ten)*

(c) the rule deleting final schwa in these

context# doe# not apply to the speech

of this informant*
 *39

Because these forms occur la the speech of our

oldest informant, wo can assume that the speech forms

typical of younger Belt speakers must bo the result of

the following types of changer

1* /aia/ > /eta/

2* /Bus/ >/»ia/
3« /Xia/ >A«a/
4* /eta/ > /»*/
5* A fa/ >/««/

(This also means that the relevant phonotie imitation rules

in the Reh diasystom are synchronic reflexes of dla*

chronic changes*) Thus, parallel to 'the' series of R«P»

changes «

/Baa/ > /a»3/ > /o*/

/««?/ >/m/

there is a series of Keh English changest

/Bua/ >/si3/ > /oi/

/Ii3/ > /sis/ > /«t/

That is to say that the same change occurs la Boh English,

but in a more generalised form than in R.P. This point

is further underlined if one compares other ftch eliangos

of this typot
 kko

/mU/ > /®i/

/Via/ > /*»/
/wia/ > Ai/
/«*a/ > /3»/
/«wa/ > /«*/
/<ma/ > /»«/

with change* la R«P« and London English!

/»i5/ > /«*/

/»*p/ > /at/

King has written that rule simplification in the

transmission of rales from one generation to another *

in diachrontc change, in other words, - also occurs, at

least to a certain extent, in spatial change, and "In

particular in the spread of a rule throughout a geo¬

graphic (or socially defined) area9* He states further!

"Borrowed rules are common in th« billn^ial tihrntion and

in the vicinity of a prestige dialect* Urn hypothesise

that in Borrowing, in general, rules are simplified

rather than complicated* That is, a rule is borrowed

with the same or greater generality *.**^ The material

we have Just presented supports King's hypothesis* A

ruls that* in London English and R*P,» involves the lower¬

ing of back vowels before oehwa and the loss of schwa*

also occurs in Nch. But in Nch English it involves both

back vowels and front vowels* lore generally,the rule

that, in Ti*P« and in London English, involves the

monophthongisetion of certain diphthongs before schwa

 Hill

In Neh English involves ail diphthongs* It would there*

for* oeem to be the ease that a rule that is present both

in London English and in u # P* • has been transmitted, as

the result of spatial and social diffusion processes, into

Jfeh English in a more general, simplified form*

this general development also accounts for the loss

sf schwa in items of the lexical sets of here, there* f>oro

etc. It is a change which has occurred in many typos of

Pritish English, and which has spread to tfch English,

probably from the Homo Counties, in the form of a phonetic

mutation rule. Historically, it represents the final

stage in the weakening and loss of post-vocalio /r/ in

certain types of English*

The loss of schwa in here and there items in Neh is

illustrated in Figs* 8 and 9 respectively. They show

the percentage of (cr) and (tr) forms that were pro*

nouneed with final schwa in each of the contextual styles

by the different age-groups* Fig* 8 shows that there is

a big drop generally in schwa-usage in the speech of the

under-thirties, and that speakers under fifty use hardly

any schwa at all in CS* The pattern for (tr) shows a

more general ami gradual decline in schwa usage which

suggests that the loss of schwa with this variable began

earlier than with (er), but that the change has taken
place at a slightly slower rate* In the case of (er),
it is speakers aged under forty who use no schwa in every¬

day speech.
 r*„
m
VariV'ti (er\ &•»
'
j J
ct«*&
-j
*

I5&I

its
X
£' * A

i-5

JAA?
IVvt

0^
li*H 2?'A "J&i*
Acs

Pq.2' % (Sr)f S-a^LCL &?;£.

er?
<!>

101^ Z&'i^ I'iO'-rl Sb'^ T'&f

% (rr) 'i'ibj tj S'j Li C-W& Ky.

fv-

ct
ft-
/o

Li.*! <1 <>4'i &-<J\ fc-tf "/:••!

 442

The Merger of the Vowels of nnne and ii|all

ft saw in Chapter Sem that older rural speakers

in EA consistently distinguish between it oris of the

lexical set of iwie, which are pronounced with /#«/#

and items of tho lexical oot of nail, which are pro¬

nounced with /mi/* The post-war SBi> records* however,

an&gest that this distinction is being losti somo Kfk

speakers substitute /«!/ for /e»/, or pronounce /e»/ as

1*1 j, in some words, (x&kerits also remarks that tho

distinction appears to be dying out in the speech of

younger people,} We have stated that the distinction is

now only infrequently preserved in Mch itself.

The extent of the loss of this distinction in Net*

can be gauged from the following material. Fig, 10

shows the style and age differentiation of (a) in foh.

There appears to be a very slight decrease in scores from

right to loft across the graph, but this is minimal, and

does not indicate convincingly a decline in tho use of

(a)-5, the older EAn variant /et/« Index scores C»r (a),
however, measure both the systematic phonetic different¬

iation between /si/ and /®i/ and the performance phonetio

differentiation of /«?!/ itself. Fig, 11 shows style and

age differentiation of (a) without (a)—3 forms, A com¬

parison of Figs, 10 and 11 shows that the two patterns

ars distinct only for those speakers aged over forty.

From this we can deduce that only these speakers make

any distinction between the two sets of items, (Fig* 11

flj.lo
—'jo
1/amUs (fx) fib fcvvsl Am
J

% U Variable (s}b/jSv7ji£ cl'-ACL Aoe

 443

also shows that scores for (a) are Increasing through

time* if we omit /•«/ from our calculations* In other

words, the more extreme forms of the diphthong,

are en the increase In the speech of younger Nch people*)

Tho lower scores for the older informants in Fig* 11 also

suggest that the merger of /ei/ and /ml/ has not resulted
simply from a replacement of /©t/ by /ml/$ but also from

a tendency for /e«/ to bo diphthongised to

A closer examination of the details of /et/ t /ml/

differentiation in tho Nch ©ample shows that only eleven
out of the sixty Neh informants used any (a)-5« All

others eons latently used some variant of /ml/ * All the

eleven informants were members of the VfO (6) or wc (3)»

and all were aged over forty-five, fable 3 shows the

percentage of possible (a)*3 that were (a)*3 in the

epeeeh of these eleven informants, by contextual style

and age-group. (Koto that speakers aged between forty-

five and fifty-nine are treated as one group.) This

table Illustrates quite clearly the loss of /et/ in Nch

English, and demonstrates the extant to which this

feature has become recessivet even the very oldest

speakers distinguish between the vowels of nana and nail

in only one out of every five instances, we can there¬

fore state that /e»/ < //a// occurs only very rarely in

Nch today* In the speech of older informants this is due

to the operation of the dlaaystemie incidence rulei

//%J/ > //»!//

 w
rs
I f a.
«*

& I H ?
I
a S § » O *
9

«► ** ef
«* *»

«0

a
<*
H
M
& £

a>
H

fli «*
i 0
H £
* a

«*
?

*1

T
S a
i i
0 « 0 0 91 S M
U» v»

5? «*
to*. cr

*9

91 a

***. u. .J*
1 U*

(4
**
C3
/*•
•»
u»

©«
I* ©V*
»
o
yt
#
* MOW
H
H
 %

f m99 (D
9
9 m

Q)

N 9 6 0 % 9 9 9 9
5 6* 3 ft 9 0 ft n ft ft
4 H f ft
sr
*! ♦ft
•5

9 1 I
9 9 &
3 ft § ft 0 g 0 O ft 3 ft P
Sr
sr
•ft

9 9 9
9 9
ft ft 0 0 9 1 9 n s 0 o 9 0 w I

sr
JL
{? •ft

•e

9 e- f I
Hi

ss*
r
3
sr
ft

ii
9
I
9
?
E

a 9 I f 9 1 9 * 9 f 9
ft 0 ft H % s 0 * ft ft
ft <

sr
•4
•o

I
&

< ft ft
9

sr
s
i
*4
0

•ft
i
« i
ft*
9 9
O

♦ft

T
9 % u f 1 % 9 9
9
» 9 1 ft ft PI 0 0 0 ft 0
ft ft
St
*% Sf
5f
»*

9 9 I
9 ?
I H $ I o a 9
9 H o o »
at

♦ft »i

9 f 9 9 ?
I & I %
m 0 9 9 n i e
2

•ft
S3.

•
•
 kk6

the Introduction of (lr)-l

we saw la Chapter Seven that the record# compiled

fey lovma and Kokerltz indicate that older rural EAn

speakers did not, la the pre-war years, use the variant

(ir)-l /3«/ in itetae of the type bird, bum, fern, but

had instead vowels of the type iat], L?J» L®J*

I*J* The post-war SEP records have a series of similar

vowels, but with the addition of the R«P«*llke 43;]* This

suggests that. In FA {jmrally, the older pronunciations

formerly typical of the area are being: replaced by (.31J

(and we have already stated, in previous chapters, that

this is in fact the case).

Fig. 12 shows exactly to what extent the- older EAn

variants of (ir) are being? lost in Roh# Fig, 12 is a

style and age diagram of (ir), where consistent use of

4.3s j will give an index score of OOP, and eoneistent use

of the older FAn forms a score between 100 and 200. be

can see straight away that no informant aged under forty

usss anything but 43*J i° any contextual style. Speakers

aged over seventy, on the other hand, use little or no

13«] at all in CS« the average (ir) index score for this

group is l$fr. Thus, in the spacs of thirty years or so,

the R,P.-type form has completely replaced the older

forms* Moreover, the fall from the scores of the

seventy-plus group down to those of the sixty to sixty-

nine and fifty to fifty-nine groups is very steopi the

change was a very rapid one. Speakers born after 1910

 kk7

have virtually tie (1 r)-2 or (ir)-3* This eoggeots that

the introduction of i.3tj nay have coincided with the

First k'arid war*

The causes of this change and the mechanisms of its

spread and establishment are very probably the same as

those associated with the loss of (a)*5» There is

evidence that the spatial diffusion of /3«/ has been

effected by gradual spread across EA as well as by trans*

mission down the urban hierarchy* sines although no 13fj

forms occur in Lowtsan's records* forms with |.3aJ do oeeur

in Honington and Buxhall, the two most south-westerly

localities investigated*

The boss of /h/

The fact that "h-droppiag* occurs in Well but not in

the surrounding areas of rural Nfk has already been

mentioned several times during the course of this work*

Fig* It is a style and age differentiation diagram for

tho variable (h)« It showe a pattern of variation in many

ways similar to that for (ng)» in Fig* 2* This suggests

that* since (h) has the characteristics of a "neutral-

variable not involved in linguistic Changs* the intro¬

duction of (h)-2 into Nch English cannot be of recent

origin* (ll) is subject to Just as much etylistic vari¬

ation In the speech of the elderly as it is in the speech

of tho young* What linguistic change there is associated

 ft p ft *
© 1 » © n c © © © © 9

«t S? sr
•© «♦ ft

H
P w

*5
m
k
P
:
9 9
| I

91
&

3 P E
W s 9 © © © ©
A a
*% «t
t* ft Sf
1*

i p S
4 9 1 X 9 n © 9 © O
*
ft ft **
I*

IP

ȣH & p g H" 9

*
S <
I 5 H
f S 9 9 © » ©

sr

&
p

sr
© 9 9 *
© o §

ft
9 ©

«*■
© 9 ^
e
I
i

«t sr
© n ©

p e & P

© ! 0 1 © * © § » 9 © 9
H < © 3

ft ft
»l
«4

.ft

f 3
3 t © © 9
» « 9 © H»
sr sr
ft t* *% ft **

•0

S* «t © *«
5 •
1 P

t*
r
i
•
P

I
Is

f K * &
%
« 3 n © i ©
SS sr
«t «♦ «t

« & I
© © © 9 9
er
«r
ft «t 3*
«t
 w * 3 &
1 % 9 9
a 0 9 9 9 H» Hi "i H S 9
Hj
si
sr 90
<* 9® IT
*S fm I*
iW
*9

0
9

«*
«
9 8
A as

rffe
w
0
g
&>

£*
«£«
3
1 t
0 * 9 09 IS1 |9
«
—*
9 0

*•*
H | 9 w
i
s

«♦
r

& 0
3 3 9 A 9 0 S 1 H H- g &
w 9 i 9 9 « 9 9

*►
V# ®s

5w w «i 3 H
§

*1
a 3 5 % § 9
n

«*-
0

i*
a
a H
9 H 9 3 9 9
1

9 0
9 B I I §
g 9 0 9 9 a S S H> 0 9
** t* sr
«*>

at
9
0
1

<o
¥*
9 * S
1
£

ft
a 9 9 t 9 9 3
sr
«►
«+ I* I*
i*
•3

H 5
g A
9 9 § 9 3 £ 9

**

0 » & H
a 3
9 9 9 9 9 o 9 3 9 a
*S 9 3 9 * 3 as at a 9 o

**
i* •1 I*
'S

it-
3
H
I s

<*- sr
§ < 9 n ®
i&
s s
&

>*
9
1 s 9 9
f

C 9 3
0
O
4

0 H !
r
o o a i o 1 9
as

«♦ **■
•< *»

§
11 o
0

(9
n

i*
as
§- a 1 s

j,
9 m m
0
i% g
f

E 0 3 H I
1 g » 9 » » o ! « S g
a 9

et
»-

"9 •o
&U2.
(irj bu S&jte CUAOL'A^G.

%i3 Various. (l<\) b^ Sbjlsi c^A. fcjCt.

■ l-r Vari&bb. ^0*J bij Sti*{£ Cu*-'l. .

XS
 *30

The centralisation of this vowel la not now In FA.

Ferby has no mention of it# but Kokeritz and Lawman both

giv* examples* Lovmn, for instance# wrlteo j.twAlv] for

twelve, in the two southern Iffk and two northern Suffolk

localities* In recent years, however* there has been a

very marked increase in the frequency of centralised

variants of this variable in Wch# This is illustrated

very clearly in the etyle and age differentiation diagram

for (e)» Fig* 21 in Chapter Eight* People aged under

thirty* this shows* have very much higher index scores

than older speakers* and the upward elope of the lines

across the graph from right to left is very marked#

What are the causes and mechanisms associated with

this Changs? In this case there is no evidence to suggest

that the cause lies in any influencs from R,P« or other

accents* The change must therefor* be in some way inter*

ml to Wch* Mm have* in fact* already suggested, in

Chapter Ten, that this is the ease* and proposed that

this change may in some way be due to a change in setting*

Older Kffe and Meh speakers tend to have clear /I/ in all

contexts* they do not have dark /!/ in those positions

where it occurs In «,P, Thus, Lowman writes i.lj in April

for all three Hfk localities and for Xlketshall in Suf*

folk# The three remaining Suffolk localities have L&J*

Ws can postulate that# as a result of some of the diffusion

processes and standardising pressures already discussed*

dark /A/ has replaced clear /!/ in Hch English in those

positions where it occurs in n,P» (and indeed, in the

speech of a few younger male wc speakers# in all positions)*
 *51

Olaton states that clear /£/ in R*P. has a resonance

that "is often of the {.fcj typo*, while dark /§>/ has a

resonance vhioh "is often of the type l,oJ or l»" J#*

Furthermore, "variations in the quality of the back vowel

resonance associated with are • •• to be found among

f*«P« speakors, with a range extending from loj, l^J# or

i,3r J to i.oj or j,!]**10 Now, a change in the pronunciation

of 1| with an [Jj-typt resonance being replaced by an

11j-type resonance, is exactly parallel to thechan&e in

vowel quality that has occurred in the realisation of (o)
in Noh English! j,ej has become [aj. We can therefore

hypothesise that the change from clear /l/ to dark /*/

has led to a concomitant change in the realisation of /«/

before /!/• This hypothesis is strengthened by the fact

that /I/ tends tc be somewhat lowered and centralised

before /I/ in the speech of the young, and there io

similarly a lowered and centralised variant of /«/ in the

same position* In fact, hell, Hal* hull are often all

pronounced thllj by younger Nch speakers*

The centralised variant cf /«/ was recorded fre¬

quently in the Nch survey in the items manner* mtter*

that* as well as before /*/. The environments for the

centralisation of /«/ therefore seem to be the sane as

the environments in which (©) occurs, except that that

is of course monosyllabic, and that there is the addition

of the environment t /— " na/, The centralisation of /«/

was particularly clear in the Pairs Test, where the pair

ma?} I manner frequently produced the distinction!

 43a

laesienj • Imimdi

This atigsoais that the context for (•} ought perhaps to

be extended to include /— ■ - ■ na/, although only one item

with a centralised (•) was recorded during the surrey in

this context*

pcnn*orth [p^inadj

The above explanation for the centralisation of

rowels before /I/ dose not, of course, apply to the other

contexts. We shall see below that there may be a socio*

psychological explanation for the increase in centralis'*

ation in these contexts and also before /*/,

The Variable fa)

We discussed the linguistic changes associated with

the variable (a) at some length in Chapter Eight. There

wo saw that the overlapping pattern of class differentia**

tion was due to a change in the phonetie realisation of

(a) from i#j to l»J and from (,ej to let J. for younger

speakers we noted the additional change from let j to i,*tej

to Ictsj*

Fig. 14 is an ags differentiation diagram for (a).

It shows that the index scale for (a), although as ire saw

in Chapter Eight not altogether successful in handling

 *53

class differentiation, does produce a clear enough

picture of age differentiation, w® can see that older

speakers have higher scores than younger speakers, at

least in CS. The particularly sharp fall in scores

between those of the seventy*plus group and the slaty

to tlxtf»niae group indicates the replacement cf (.sj by

L»J*

An examination of the causes and mechanisms asaoe*

lated with phonetic change of (a) must be carried out in

tvo parts* tfe oust first of all consider the change

from to and i«s«j* '.*e must assume that the

initial cause of this change is the pronunciation

typical of tho P*P* prestige accent, which has become

established in Itch as the result of social and education*

al pressures* The form (,st j| moreover, la now also

typical of R*P*t "th© lengthened [«j is equivalent in

quantity to the longest varieties of /it, at, oi, ui,
11
3t/*« Tliis, then, is most probably a "change from

above", and certainly a change that has been diffused

down the hierarchy of social classes*

The second change, however, • that from L«t j to

l^cue j * must be considered to be a change "from below",

since forms of this kind do not occur in R*P* (although
pronunciations such as [m j are reported)* They do, on

the other hand, occur in London English* Slvertsen

writes, of fm/% "Vocoid glides, or L«*j# are mors

common before /C+/ and /City/ than the long monophthong

 k$k

l**J% Ve ean therefore ascribe the diphthongal

term* in Nch EngUsh to the influence of London and

Homo Counties forms* Mote* however* that the diphthong**

isatlon of Nch /#/ is not so closely tied to particular

contexts as is the London variant* except that the

diphthongisation in Seh is more pronounced before velar

consonants. (Sivertsen* on tho other hand* specifically

states that in London speech dlphthongisation is less

common before voiceless stops* and cites as an example

the lexical item bach with a monophthong rather than a

diphthong*) Wo can therefore interpret this change in

Nch English as the result of the transmission of a

phonotie realisation rule in a more general form than

that in which it occurs in ths source dialectt the Nch

rule is relatively context-free*

Finally* it is also worth noting that the diphthong-

isation of /«/ permits a clearer phonetie distinction to

bo made between /«/ and /»«/# Some speakere* for example*

make the following kind of opposition!

Plan tphl»»Snj
playing j.phl»»«j

The Variable ft}

We saw in Chapter Eight that there appear# to be a

linguistic change in progress in connection with ths

 *55

variable {!)* An apparently new variant» (1)4 a [aiJt

la mora prevalent in the speech of the young than of the

old, and la particularly common in the speech of male

MHC informants. However, in Fig* 15, which la an age

and style differentiation diagram for (I), there is

very little evidence of an actual change in Index scores*

the differentiation pattern is very similar to the

neutral (ag) pattern* The stain deviations from the

neutral pattern are the sharp drop in scores between

those of the seventy-plus group and those of the slaty

to sixty-nine group, and the very high score of the

thirty to thirty-nine group in CS* What change there is

appears to be slight. The occurrence of the i.»ij vari¬

ant is obscured by the general index scores* This is

duo to the fact that it is confined to a relatively small

number of speakers, and that those groups which have most

*

(I)-* forms are the same groups which have mors (I)-l and

-2 and less (t)-3 than other groups• The fall in scores

between seventy-plus and sixty to sixty-nine reflects

the replacement of t£ij forms by some ialj forms, be¬

cause of the Influence of R*F* The high and rather

erratic CS score of the thirty to thirty-nine group lo

due to a high concentration of male MWO speakers in this

group*

The Introduction of the L®iJ variant is obviously

not due to the influence of ft*P* It seems likely that

it Is in fact due to the influence of the speech of the

Home Counties* Gimaon states that the first element of

fcnJS Vi\na4te (i) h, Paz

Rot. & VWtcvUc (z) btj Ac-2.

 436

the /at/ diphthong In London speech is generally [aj

or SFD records also show that forms with [cij
occur commonly in Suffolk, Hertfordshire, Bases,
l4
Bedfordshire and Buckinghamshire* &e can therefore

assume that (I)«4 occurs In Nch English as the result

of the diffusion of this type of pronunciation from the

Home Counties,

What now remains to be explained Is the exact role

of the WtfG In the Introduction of thle variant* The

material presented in Chapter Eight concerning the in¬

cidence of (I)-4, together with scores for some of the

other variables, suggests that the ?WC» and, on occasions,

the WC, are the first groups of Nch speakers to adopt

non-standard forms upon their introduction into Nch.

Mm can put forward two reasons for this* First, the $SMC,

LMC, and, often, the WC, adopt only those features trans¬

mitted from geographically external dialects that also

occur In R*F«, or alternatively adopt non-standard

features only after the £!WC, This means that although

they are not usually genuino Jl*P. speakers, P.P. is, at

isast overtly, their only model* Secondly, the LWG is,

as we have already Indicated, a socially isolated group

which is late in adopting all types of innovations* This

is becauss many of this class are elderly, they are all

relatively poor, and have had relatively little education*

Thus, the ?B»*C, alone of the social classes, is both

accessible to the diffusion of non-standard innovations

 *57

and not resistant to them* For tills reason« this Glass

will always be Instrumental In the Introduction of changes

of this typo Into Nch English* ami probably In the dif*

fusion of non-standard forms throughout Britain generally*

The Variable (e)

In their dialect reeorde of rural Nfk speech* Lovoan

and Nelson Francis write pronunciations of this variable

with variants of the symbol i«j* Vs suggested la Chapter

Fight* however# that the variable (o) Is currently under-*

going change in Nch* with n*P»-*type rounded vowels of tho

typo |.a»J Increasing In frequency as compared to t,«j*type

vowels* This is confirmed by Fig* 16* the style and age

diagram for (o)* which shows a clear decrease In seoroo#

indicative of an Inorease in rounding, from right to left

across the graph*

No have also* however* mentioned a further type of

change that Is associated with (©) in Ncht the fronting

from taJ to l*j* and the lengthening to or j.a«j*
which In come cases leads to a merger ef /»/ and /ai/#
or /»/ and /at/* Complete mergers were suspected in the

speech of some* particularly younger speakers who did not

have the rounded variant* Even by tape*outting, though*

it was not possible to decide unequivocally that there was

no distinction in the realisation ef these vowels* We do

 *53

have »ome evidence from the Pair a Test* however* thsrlag

the ©ouroe of thie test* speakers vers asked to read''

aloud two pairs*

arm t on

SS*M * £25

Oat os the slaty Itch informants, as many as seven made no

distinction at all between h-arbs and bo^ and/or the vovele

of arm and on» It should ho remembered * moreover* that

the Pairs Tost represents what is in many ways the most

formal of all the contextual styles. The mere fact that

two itoae (which are distinguished in n#P, and many

types of Ifoh English) ars plaosd side by side foeusees

the informant•s attention on the potential contrast* and

encourages him to differentiate the two items* It is

therefore legitimate to assume that several other speakers

have this phonetic merger in CS*

We must* further* consider the fact that the

seven speakers who were shown to have the merger fall

into the following age-groupsi

10-19 1 50-59 *
20-29 0 60-69 1
30-39 0 70s o
*0**9 1

There is* it is clear, a preponderance of elder speakers*

 k59

This does not* however» counteract our assumption that

this merger is on the increase. This is because there

ere two different phonetic chancres of (o) currently in

progress in Noh English!

£e] >

La] > i.nj

Both and j^aj are more common in the speech of younger

informants* There is possibly* therefore* a higher per*

coutage of unrounded-vovel-uaers who use j.*J among

younger speakers» (although raore older than younger

speakers use unrounded vowels)*

The cause of the change from La] to L»i clearly lies

in the presence of L°i no* enly in B«P« but also in
Suffolk English* This suggests that speakers who have

the highest proportion of L»] forms will be younger

members of the BE1C and of the MWC*the former because of

the influence of B«P.* the latter because of its role in

the acceptance of linguistic innovations transmitted

from other geographical areas* We have no direct evid*

once of this* because our sample is not large enough to

permit a study of class and age differentiation simul¬

taneously. We have some indirect evidence* however* for

part of this hypothesis* We have already seen that it is

male ffiC speakers who are overwhelmingly responsible for

the introduction of into Nch English* We also saw*

in Chapter Eight, that male wc speakers, unusually, have

lower (o) scores than female 'C speakers*

o CS « I■ S • »4 34 P< >♦ tf « tf m I 53 I * tf © A

Att .© S ££ 1s s »* © 24 • ta o :• 5 £I i »4 £ I 8 tt

I 2 if © *>•"« tt © I « « 1© 1 34N *tf 34 e i£ *4 © |

iS 98

I% a • ! t&f © 04 © • K »© tf 3 © 3" 1 *4 I tf s )8 i 04 © *tf »4 I

6 3I • *tf »► © ?s 4* : .tf 1| © 2 »4 © 31 St © ,tf - 3 *tf © >tf 3I #

2. 0 i 2 © *4 & u tf • *4 tf i nI *H © * ©I tf e »4 A S 2 *4 9
©'
© .tf g *4 34 © I 51 5 tf ?2 9 tf ©© »4 tf ©£ *tf 34

£I tu tf S ti £ *I
tf a tf © »tf # A« © *4 34 9 tf g tf 2 © *4 « 4« t

tf : 1 ?2 >; 0 g *I tf g *I A © * 3 *tf 34 © O © $ I Lif

tr I 1 £ © tt I 4© 3 2« *« § 1 > tt 4♦ tt § tt © I *tf *

«cf 34 © 19 *tf4(0 sI »4 © 3 § >4 »*< ©

»4 34 tt tf c *tf © tt O *tf S *tf • H *4 9 tf £ tt O 9 i M »4 i»► a tt « »* X H *tf

be due to a change in tongue setting (see Chapter Ten)#
Support for this theory is provided by the parallel change

ei /ou/i^ruj > The ©hangs is novo likely# however#

to be connected with the Introduction of the new C vowel

/at/1 which results from earlier /s&mV and /ria/# A

front /d/ m (,a»j makes it possible for a distinction to

bo maintained in spits of the lengthening of /*?/» al¬

though of course it does Isad to the loss of tho /'&/ t

/at/ distinction in some cases# The twe possible causes

my therefore1 both have been operative*

Change, in the *fch Vowel System

We are now in a position to view socio of tho

developments we have described above collectively rather

than atomlstlcally# We can aleo analyse some of tho

ways in which# as a result of these developments# tho

Hch '-phonetic vowel system has changed and is changing#

and the implications this has for the phonological vowel

system# The changes that we described above that are

not purely realisatienal and that have seme effect on the

systematic phonetic vowel system are the followingt

1. The merger of the vowels of beer and bear

2* The lowering and monophthongisation of

eortain vowels bsfore schwa and the lose

of schwa

3. Tho merger of tho vowels of name and nail

 462

The introduction of (ir)-l

5* The merger or the voxels of boat end barks

A very useful starting point in our discussion of

change in the systematic phonetic vowel system is the

speech of our oldest (eighty-nine yearmold) informant§

Mrs* B* , which differs in some significant ways from

the speech of all the other informants* tfe will take as

a reference point the systematic phonetio vowel system

shown in Chapter Ten, and illustrate various differences

in terms of the pronunciation of the following items*

which stand for lexical sets*

/I/ bid /ti/ bee /««/ here* there

/«/ bed /®i/ bag, name /W plage*:
/»/ bad /Vi/ boy /at/ fay* both
/V/ nut /»i/ buy /at/ paw, pore* poor, employe:
/•/ Putt /¥*/ lewd /«•/ lower
A/ bot /»V loud /«»/ fire* tower
/Vu/ load /3i/ bird, pure

AV lw

Mrs* D**s short (A) vowelo are the same a® thoso of the

other informants, although they arc realised rather

differently* Table 3 illustrate® these slight differ-

enoes, with the "Wch* column representing an idealised

type of average Tfch pronunciation, the main function of

which is to indicate the distinct nature of Mrs* £>•**

system* The sane is also true of the diphthongs (B

vowels)t the inventory i® identical, the realisation

somewhat different* This is illustrated in Table k.

 *63

Tablo 3t Comparison of Realisation of (a) Vowel*

ttch

b£d l»J iii

Lii Lfl
bad L«»J
J2S* Ifi i»J
patt 1*1 L^i
not Ld^j L«J

Table 4i Comparison of Realisation of (B) Vowels

NCh

bee
tM-i LfiJ
'i»ij
hsx 1**1 1**1
buy
«Mm«W 1**1 L®*1
lewd L3*i L**J
land L?*J l««J
load Lttuj touj
low 1&«J t»»J

The main differences between more recent vowels and

Mrs* S>**0 are associated with the lexical sots of rnne,

l$£E£r there, player, JTar, pore, j>oo£, employer, |ow^,

fire, tower, bird and mire* Table 5 compares the

systematie phonetic realisations of those items, and

also Inolndes tho retaining C vowel*•

 k6k

Table 5• ComparUon of the Realisation of (mainly) G

Vowel*

Rel| q»

name AV /•I/
HSLS. /«•/ /Ii3/
ihSKS. A«/ Ala/
£*&£££ At/ Ala/
£&L At/ /ma/
path At/ /at/
poor /»»/ /Una/
pore At/ /eta/
paw At/ /*»/
*££££ At/ Ana/
fire /*•/ Aia/
bird /3«/ A* a/
pure /3t/ /«*«/
igX2E /«!/ Am a/
-"nAyff At/ /Ola/

Thla means that corresponding to the system of C vowel*

shown in Chapter Eight•

/«»/ /3t/ /*•/

At/ A*/
/at/ /««/

Mr*. !>• has only»

/•*/

At/
 Apart lro.-s the long monophthong In name, the

differences between the speech of Mrs. i>* and that of

younger Neb speakers are accounted for by the feet that

where she ha# S and Q vowels plus achw*. they have C

vowels* The phonetic mutation rules which lower certain

vowel# before schwa and then delete the schwa. (which wo

assume to have been transmitted from another dialect In

a more general form), do not therefor# apply to lire* l># *•

speech* They are, In ether words, a relatively recent

introduction Into ffch Ihglish, and represent both a

diaehronic sound okange which Is still (just) reflected

in the linguistic behaviour of the speech community, and

a synchronic process* Because of the evidence provided

by the speech of lire* h*, we can be certain that the

phonetic mutation rules we set up on purely synchronic

evidence in Chapter Ton do also represent sound changes*

It la of course dangerous to generalise from the

speech of one informant to the speech of a whole gener¬

ation* We do have some evidence, however, to suggest that

Hre* h* is in fact typical cf speakers of her generation*

Thi«» is provided by the rural dialect records* Kokerlts,

for instance, writest

hKFJH with the vowel Lis j

poor .
with the vowel L»sj
JB2ES with the vowel L»sJ
19K*V. with the vowel Laos J
fire with the vowel l««J
pure with the vowel L*sJ
tower With the vowel t«U9j
 4

$ I I

HK *5 t »4 3, © tI « tV © © « ?S3S »♦ 5 .C H © © © 8 § ** o •*

1*3 \ o< H I K€ ©I H \e o c * * H *4 a-
_8

% ► tf- i »4 !X 1s 0 u " S © iH i0 v 9 © tI © 0 &K

1 ©© s c I £ « tI ©
g:I I © > © tI © SJ H©©© 0 t%

3 1sr 9 *4 H 0 >I 9H H H 9 ? 91 »4 i © ©• >* © © © 4%

I ,a ©3 i w ► e »© H N1(1 tr«tI Q
1 *4 94 © sa 0 A 0 9 st 1 H 9« ©■ £H

I H © 3 a tI 9© » O ©< 34 © tS © o tI © © © s? 31 94 3 »* H© 3

I *} © tI Hl © tI

3» a% 3 O H O &t H o «
 The third major change would appear to he the

introduction of the unit /Jt/» jFadging from the figures

presented in fiff. 12» thio appear# to hate started

around 1910, raaybe In the form of /3»a/» (as it appear®

in Uwan's rocorda) ♦ Tlie system of C vowels ♦ schwa

was thus augmented by one*

The Introduction of the development# /ola/ > /«•?/

and /eas/ > /ata/ also appears to have occurred fairly

early# since /ata/ occurs in the speech of most of the

Ifch informants* This was then followed by the develop*

meatst

/Utta/ > /©#<?/

/ltd/ > /e »3/
/»aa/ > /3I5/

probably via stages such aa /Uua/ > /uta/ > /o»a/ > /»«<?/,
The second stair© of application of this process* Judging

by the age levels in the sample where they begin to

occur* involved the changes*

/mta/ > /mtd/

/»ua/ > /»ia/
/bis/ > /eia/

ami the loss of schwa. Figs, 9 and 10 indicate that tho

loss of schwa began with /ei-?/ some time around tho turn

of the century, and has gradually been spreading to other

 468

vowels since* In this ease, then, Klt*g*s hypothesis

concerning the spread of a rule in a more censral fors

is only valid if the change is viewed macroscopieally*

The change began In Nch as it began in London* with

Itoae of the typos fire and tower, and was only sob*

sequently generalised to other vowels*

ye oan summarise these changes by presenting a

series of artificially discrete systematic phonetic

vowel systems which are valid only as an indication of

the overall nature of some of the recent changes in Neh

English. (A and B vowels reaain unchanged throughout*)

The different systems are shown in Table A*

Stage VII is the result of the recent tendency to

merge low A and C vowels* It also results from the

merger of /at/ and /o»/ which occurs in the speech of

many younger people* /»»/ la a particularly unstable

element, and is also merged with /at/ by some speakers*

The legitimacy of regarding these changes as only valid

when viewed esaeroscoplcaJLly is underlined by the pessibll-

lty Cf presenting several alternatives to Stage VII*

Many WC speakers at the moment, for instance, do not

distinguish between fire and tower on the one hand and

far on the other* Those speakers have /«»/ in both

eases. VC speakers normally do make the distinction*

Only in several years* time will it be possible (artific¬

ially) to describe either system as typical of this

particular period*
 U69

Table 61 Stages la the Development of the Neh System*

atic Phonetic Vow* J System

C vowels 0 ♦ schwa fi + schwa

1 lis U13 Uma DU3

ei

Sire. D# 01 aid

ala old said wild

at ata

II lis Ula ¥«d Uu3

e« oi eta 3ta »t5

1910 ala via aaa mgd

at aid

III lid Uia Vas D?»a

Intro# et • 13 3«3 aid

of PM rla "<£>Ud

rules at &I3 a*d

IV U13

Spread 31 «»3 3l3 Old

of PM ala inaa

rules SI aid ata

Cample* at eta 3»->

tion of aid v%3

Spread at atd aid

VI

loss of «. t 3» at

at ®t

at at

VII
Recent «t 3« ot

merger#
at
 470
Metarules

Vt tisggoated in Chapter Ten that the phonetic

emtntion rtU.ee that operate on trowels in combination

with schwa eould be expressed more generail/, The

systems portrayed in Table 6 suggest a way in which

this could be done* ffe claimed in Chapter Ten that one

of the advantages of the type of phonological approach

that we employed in setting up the Hch diasystem was

that it produced systems at both the phonemic and phon-

otic levels* * and it was the setting up of phonetic

level systems that allowed for the portrayal of forces

operating in phonetic space of the type discussed by

Martinet* Another advantage of having systems or in¬

ventories of this type at the phonetic level is that wo

can allow for the possibility of expressing the fact that

both synchronic rules and diachronlo changes apply to the

Inventory as a whole* It has been suggested by Anderson

It
A bass that rules of this type should be considered to

be metarules which apply to the whole granraar* The meta¬

rule in this ease 1st

Diphthong ♦ schwa ■« Nearest long vowel

or»

8 vowel ♦ 0 vowel ■» Nearest C vowel

(There is an exception to this rule* namely that /ui/ and

/am/ do not become /»»/ and /©»/ but /at/*) Thus /II/* /Uu/,
/Ui/ and /¥u/ are lowered to the voxels closest to them

in phonetic apacet /« §/» /»*/» /«»/ and /3«/*

The Theory of *,tfty?ntstle flhanrre

la oar discussion of the various phonetic and

phonological efaagM (which we shall loosely refer to as

* sound chaise*) that have taken plane in Weh, we suggested

the following footers that nay have brought about those

change*!

1. Tike influence ef S»P,

2. The influence of Hone Counties speech

3* The influence of Midlands epeeeh

km Soeio-psychological factors

5« The need to preserve lexical distinctions

6. Functional load

°7* Pressure in phonetlo aaaoe

a* Change in setting

Changes due te these factors o n ha divided into two

typess changes due to prior*exia ting linguistic differ-

enoeo| and purely internal lin,Tilatie changes. The former

can be subdivided into« (1) changes due to differences

between Hch English and other types | and (il) changes due
to differences already existing ia feh English (which saay

set up various types of social and psychological pressure)*

The latter can be subdivided into t (a) changes resulting
from ether changest and (b) "spontaneous" changes* All

types of change cast be assumed to involve certain

sociological factorst since only then® can "decide* which

changes should occur and which not, and when*

 kJ2

Before we move on to si more detailed discussion of

this classification of sound changes* we shall briefly

discuss some different approaches to the problem of

sound change* Wo shall be particularly concerned with

tho distinction between causes, mechanises and effects

of sound change*

Theories of Sound Change

We shall now consider four of the different approach¬

es to sound change which can be found in the literature*

We shall label these four approachest

(i) the physical approach

(ii) the functional approach

(ill) the sociolingulstlc approach

(iv) the generative phonological approach*

The Physical Approach

The physical approach can be characterised by re¬

ference to the work of Hockett* Ilockett states that "when

a person speaks* ho aims his articulatory motions more or

less accurately at one after another of a set of bull's-

eyes* the allophones of the language".Host hearers*

Hockett says* are able to be quite charitable in their

interpretation of utterances (which consist of shots at

 *73

these bull*s-eyes) because of the amount of redundancy

there is in language. The shots, therefore, need not bo

too accurate. "The shots intended for the initial-/*/

allophoae will be aimed in the general direction of that

bull*s-eye, but will fall all about it - may quite olose

»oae in the immediate vicinity, a few quite far away"

The single point around which the shots are .clustered is

the frequency maximum, for the hearer, the frequency

maxima of the speakers he hears build up expectation

distributions. Sound change is the "slow drifting about

of expectation distributions , shared by people who are

in constant communication"♦in controlling expect¬

ation distributions, utterances more recently heard by

the hearer are more powerful than utterances heard son®

time ago. Thus, as frequency maxima change, so do

expectation distributions, arid the hearer now aims his

own shots at a new buli*s-eyo determined by his expect¬

ation distributions. This "slow drifting about of

expectation distributions* is caused by "the vast

multitude of factors which can contribute to the deter¬

mination of the physical properties of any bit of heard

speech. The physical properties depend only in part on

tho *speaker*e intention*. Before the sound reaches the

inner ear of the hearer, other variables come Into play"*

These variables Include, according to Beckett, muscular

tics, alcohol, drugs, and wax an i dirt in the outer ear.

Hots that although wo have been quo tin,'? from llookstt,

similar, although perhaps less extreme views have been

 k7k

hold by other linguists* Bloomfield, for example,

describes sound change as "change in the habits of

20
performing sound producing movements".

The physical approach has been criticised by many

linguists, including ^elnreleh, Labov A Hersog* who sayi

"Linguistic change is not to be identified with random

drift proceeding from inherent variation in speech*

Linguistic change begins when the generalisation of a

particular alternation in a given subgroup of the speech

comaunlty assumes direction and takes on the character

of orderly differentiation* *a*

This type of approach has also been strongly critic*

iaed by Postal as *an extreme* physlcalist account of the

22
nature and causes of sound change** Postal sayst *fho

basic idea seems to be that variation in the performance

of discrete artieulatory •intentions* statistically leads

to drifting of the targets** This view* Postal continues,

*is really quite incredible* For this claim that per*

formanee deviations in the behaviour of adults can lead

to systematic change totally ignores the fact that thero

ie linguistic knowledge underlying speech perfornance* * •*

However, the crltiolsw made by Wclnreieh et al ig*

neres the causes which brought about the "particular

alternation in a given subgroup of the epeeeh community*

in the first place* Postal*c criticism, moreover, ig*

mores the fact that overlapping articulations, at the

 *73

performance level* can lead to the siearner of systematic

phonetic elements and, eventually* to the leas of

systematic phonemes, as we have «««n in the &ch diasystea*

Thle final looey it le true# oay well be effected in the

transmission of a gramar from one generation to another

(by rule loos)y bat tho Initial Impetus for the change
may equally well have occurred at tho level of arlie*

ulation* Another generative phenologist* Ring# writest

"There is unquestionably something to be said for tho

2*
phonetic basic of phonological change** ft is alee

clear that Postal and tlockett have different ideas about

what sound change actually is* Some of tho Jfch sound

changes (as we have called them)» for instance# have

occurred only at the performance phonetic level • changes

in the reallostion ef A vowels* for instance* This hind

of change we have postulated* can eventually have roper*

suasions at the phonological level* Postal* however•

would regard them aa of no interest*

The most doubtful aepect of floekett** approach Is

the view that change ie random* Xt cannot« at present#

be proved or disproved whether or not this view has any

validity* Xt is much more likely* however* that change*

in articulation are due to artlculatory causes* Causes

of this physical typo have often been described as being

duo to the factor known as "ease of articulation"* This

is a vague and for some linguists rather controversial

25
concept* but is nevertheless fundamental to an explan¬

ation of certain changes* and in particular to an explan*

atlon of phonetic changes of the as a initiatory type*

(tfe amet of course look to sociological factors to

explain wiiieh of these possible changes occur*and when*

it is not* for example* accidental that changes of the

"ease of articulation" type are seat common in situations

where redundancy is lees important as an aid to common*

icationi in colloquial speech* for instance* and at times

of political instability* where breakdowns in commtmie*

ation moan that intra-group contacts become relatively

mere important, or where the language in question does

26
not have to be used in formal contexts*) We shall

therefore add to the above list of causes of sound

change the "ease of articulation* factor*

The Functional Approach

thio type of approach can be characterised by

reference to the work of Martinet, which wo have already

discussed at some length In Chapter Ten and elsewhere*

The particular aspect of his work which wo accept here

is ikathpr secures in phonetic space can bring about

certain typos of sound change* both as the result of

other previous sound changes and the "asymmetry of the

vocal organs"* We also accept that low functional load

is a factor leading to merger*

Martinet*# approach has been oritlclsed by King*

Kipareky and other generative phonologists* King*

 sp

ISw © ©8 1 • 94 2 >« s! © «% a »*<

3H 5
3 si oi3 £®
©

.• «I M© §© tt • 4%
© 3 tif *M © « »4 © :s 4* tf 5

I 8 E © »*< © tS I tf *4 sI © *%

«2 1 93 a i 3 !; ♦«< K tr■ c sI3 2 3tt

4C NI
3 © 4t 0< ©C1 4« cf *«< « i s 4*• tt 4* 94 tt • H *4 © 94 *4 © 5 1 C «

s i iS » I >« • tt tt e $ 9 U 9 A 2tt 3 tt © I ©© « G ttt

1
» » fI2k« « »* ? S « £ tt 2 M 9 & u « ■« a 94 *< § if « b »S »tt 4* KK

>* a 1 0 4% iGiS I 3 >• 2 C0 iS m 1 94 5 3 a 3 M4 3 »4 4* O if •

S tt g ttX tt 8 «! © • 4# N ; .S 4• 9 Jt tt 9* t• 2 i j• 3* m4 i

:•
-

I* © *1*1
\ tt c uI tt tr« tt • i © tt • *4 0 »— 4 *'— .
! tt G 4* © 2
 $78

fe'e saw in Chapter fen* in oar discussion of setting

rules* that two rules can he conditionally- linked* in the

following ways

if

X Y

then

:c 0

Andersen A Laee^® hare proposed that the English Great

Vowel Shift* and other similar changes in other languages*

can be explained in the following way* A restraint is

first of ail imposed on changes such that* other things

being equal* lexical distinctions are maintained* Then*

we can propose the rulei

long V — * high

This is a metarule which operates on the whole grammar •

it applies to the entire systematic phonetic Inventory -

and has the offset of raising all long vowels* as ooeurred

in tho Groat Vowel Shift* High vowels* however* cannot

of course bo raised* If lexical distinctions are to bo

maintained* then these vowels must undexgo some other

development* Tho above metarule* therefore, entails a

subsequent rule*

long V ) ♦ high C long high V dia«*

placed
 h79

In the case of English, the displacement takes the fom

of diphthonglsatien* la Swedish and E. Somglaa it

takes the form el* fronting* We therefore obtains

within the fro»otfor& of generative phonology, formal

recognition of the foot that pueh chains do occur*

The SociollafTOietie Approach

This appro&oh ean be ©iiaracterised by reference to

the work of Laboy, whose seeiolingulstie study of New

York City English has provided ouch material for ueo in

a study of sound change* Labov, it.is worth noting,

makes reference to eoae aspecte of the functional ap¬

proach in his work, and indeed puts forward specific

empirical confirmation for certain points in Martinet1s

theory*

to have stated above that we must look to sociological

factors for an explanation of why certain changes occur

and, more importantly, why others do not* We must, in

addition, look to factors* of this kind in order to ox-

plain why a particular change happens when it happens*

Soclolia&ttistio research can also facilitate a descrip¬

tion of how changes happen* Labov, for example, proposes

hiroorcorrection as one of the sociologically determined

mechanisms which produce linguistic change* v?e have also

put forward sonic fairly specific sociological explan¬

ations for the introduction of E*P.-like forms into Nch

English at a particular time, for example*

 480

Work of thin typo is undoubtedly of great value

to linguistic** but sOiA« criticism of this approach is

Rmrth«l«9* implicit in Postal** eonenti *Tho claim

that change is stylistic is not incompatible vlth the

kinds of result* reached by such investigators as Labov*

Those latter ratter* concern more properly the social

explanation for the spread of the ohange, a natter which

aeons more properly sociological than linguistic*#2^ Ve

have ourselves already criticised the view put forward

la velnreleh et al for ignoring the initial causes of

change#

The Qencratlve Phonological Approach

The theory of generative phonology regards linguist*

ie change ae chaste in gramaars# Sound change* accord*

lag to this theory* is the addition* deletion* eimplif*

ioatlon or reordering of rules* We have already given

examples of rule addition* deletion and simplification

in the Wch diasystem. The phonetic mutation rule*

/»!/ ♦ /<?/ » /»•/

has been added to the Nch dlasystem In the course of

this century* The diasyateraio inventory ruloi

//s// )
♦ //»//
//?// '
 481

has# on the other hand, been lost from the diasystem#

or almost so, Purine* the course of this loss It was tn»

vclved in simplification from

*
/V^f
to

*) ,.,
) t /
b' ¥<*

It is reasonably true to say# however, that gene*-

stive phonelogtsts have not as' a rule been particularly

interested in the causes of eourui change, Postal writesi

"There is no more reason for languages to change than

there is for automobiles to add fine one year and re-

move them the neat, for Jackets to have three but tone one

year and two the neat, oto,*^ This is a very short-

sifted view that Ignores both the fact that changes in

fashion of this sort are generally fairly highly motiv¬

ated in rather interesting ways, and, worse, the fact

that investigations into the causes of smart change can

tell us much about the structure of both language and

society, King, on the other hand, has approached the

problem in a somewhat more satisfactory vayi "To use one

of the better-known putative causes of phonologies! change

as an Illustration, it is all very well to attribute a

number of changes to *eaee of articulation*, ©*g, ooto>

fttto, but why do «© many languages so successfully and

so persistently resist ease of articulation? Why have

not all languages assimilated to the utmost# parallel

 kB2

t« »kt* > »tt- ?* The aocioliiiguistie pei*0^«oti?«t

combined with osrlaia aspects of the functional approach*

gives us some of the answers to this problem

1* Compotinj against the tendency to "sate of

articulation" is tho need to preserve lexical die*

51
tinotions and the "desire to be understood",

2* There are certain psychological and socio-

logical factors* which we shall discuss shortly* which

act as conservative forces and mice for resistance to

change*

3* If it were not for these factors* ease of

articulation changes would be talcing place all the time*

As it is* changes of this type will only take place if

the sociological (and linguistic) conditions are right*

There are saany potential changes of the assimilation

type present in the performance saltation rules in the ffoh

diaeystem which operate in colloquial speech*. These

changes are inhibited frost spreading to all types of

speech (and are prevented from becoming phonetic mutation

rules) by (a) the need for speakers to be understood out¬

side the immediate environment of family and friends* and

(b) pressures leading to the desire to use prestige or

otherwise "correct" forms• Thus* at any given stage in

the history of a language* there is a layer of colloquial

speech where* as it were, sound changes have already

taken place* When and if the conditions are right, certain

 %£3

aspects of this layor will become established la the

language* £iaohronio ohaage is, in ether word#* merely

synchronic variation given scope by time#^2

King alto point® out, as we have done* the "largo

number of phonological change* in which a phonetic basis

such &i assimilation la clearly iiifieiniWe"*^ lie

suggest# that other changes which cannot be explained in

thie way may bo "secondarily induced by other changes

themselves assimilatory in origin* •

Causes end Mechanises of Change

We are now in a position to note a major difference

between the four approach®# we have discussed above* The

physical approach is concerned solely with the cause of

"spontaneous* sound change# <• with th« cause of "prior*

existing differences** The functional approach is con*

eerned both with this phenomenon aty* with those changes

due to other previous changes* The sociollnguistio

approach* on the other hand, is concerned with the

mechanisms that lie behind the spread and acceptance of

a particular change at a particular time* Bat how can

wo best characterise the generative phonological approach!

Zn so far as it is concerned with the phonetic motivation

of phonological change, it is clearly concerned with the

cause cf change* £t is, moreover, clearly not ooneerued

 s " £'I 1

d■ o * ! 3I t 1 1
* S> 0u » O »5 .oi i © I

© H »* * I- 5 2 I XkO e ©S

i0 3£ O H■ 0 st s a i9 scf© » » • © O1 . 5 2 1 '"

i 1 Ui st »* c • dI c & 4« 4»" t 5c !I

1 i »4 1 1* st ** 3■ • 0I 9 0 1 1O■ 1* *4 di I *

&« >6 tf iH I i*« « 0 ©a0 0 I W ff »4 1 w< w< © 0 »* ©

 435

*
aporitaneous* changes.

1. raae of articulation* Tendencies to

apply the "principle of least effort** lead to

the simplification of articulator? movements

and especially to aaeimiiatory changes* Given

the'possibility of the effort involved in

articulation being reduced, this will be lone*

all other thing® being equal - which they only

rarely are. Changes of this type are inhibited

by the need to be understood * the need to pre¬

serve lexical distinctions* particularly those

with a high functional load* and the social

need to b© correct or to be acceptable In

other ways* An example of this type of etiang©

in Heh Ragfish is the introduction of the

phonetic mutation rules of tho type

/si/ * fdf /ssi/, which represent a

simplification in artlculatory movements

(although* as we hare seen, this rule was

introduced from another dialect* so the

initial change took place elsewhere).

2* ■ Asymmetry of the Vocal Or/mna. Vm have

already seen that Martinet claims that tendon-

clss* perhaps psychological in origin* to

symmetry in phonological (and particularly in

vowel) systems are In conflict with the basic

asymmetry of the human vocal apparatus* This

 I c H 8 M u
© © 3 ©

9* «9 9*
49

s
9

49 49
I

O a « ©
©
© P
S

fl ft 49 g-
I
ft * 9 ft
© S
ft © tt ¥• ©
•
sr
CP 9* •©
49

I
ft

*1
©

<9
II
1 ft
>•
3 g

49 I*
g
ft

HI
?
I
ft

f ft H £
g g
©4 tt © e 2 » • 9
«
*s 9*
«9 •I
49

8 f
3 \ 49 tt .
I
©

«9
H» 8 o 3

«9

JM9
m*
© 3 tt O
H

O H g

© 9 §

«9
»-

a§
i*
tt s

'49
•I
«
& 2. M
N

*<
©
i l
§

tt *•
<9
?
©
©
H
© ff
ft ft

*1

B
ft ff g

49.
tt © §

CP
«*

I1
iIft
5 H 8 5
II % a

♦
©
9
H
• O

I
© § ft
8P
«9

I*

& I© „ I ©
I
£I I

9»
? -
¥>
tt

9© 8 49 ft
5?
«
t
© S
!
I

»*
§

<9
t*
©

.3
£
9

a n ft &
I g 9 9 g © 9 O 3 O
©

9* 9* ST 9* 9*
»* «*
49 49 •#
<9

tt S«
if
3* 9 H tt

49
?

© O V tt 9 ©
9

•%

Oft

*
1 I
1
*
1

49 9»
CP
*
E IW» *
I 1

w 2 H

. !*&;
!
©9
s
9

4% If ^
49
SP
i

i*

g 1
l » © tt 8 1 © «
9

tJ »» ©fc 9*
!• I*
49

ft
!
g 9 ©

II &9» «
I
& I
9 8 © 9 9 9 *
9*

•9
!*• |rt
CP
«* CP

§
H
H 8

I-
9*
9*
s tt
ff
1
© |
I
S © 9

«9

.9

M t F
*
9 9 g 9 9 M ©
9
sr CP
9* «♦ *%
<9

9 & 5
3 M © tt 9

01 «9 *»
>»

•«

«*•
9
• 3 ft 9* g
ft 1
§ ©
s
©
E
! § 9Hs o*. 9 1 © 9

5?
i o

♦t
*♦
9

» I ?
9 © 9 9 3
a ©>
3 9 3* 1

«*-
or
9* »* «♦
49 I* i*

? 9 ft
9 tt © S 1S s © M W © 9
3 W 3 1

9*
f" 49
49 tt©
jm
Ct >*» •I
49

I g W &
s g W 9 o • A9 0 3* 9
© $ w> 3
fa© >
gk
• «S
<9
•© *1

9 % &
9 I
® n 9 9 i 9 tt
© i 1 9 © a
9*
«9 5* !▼ 9*

lr
 487

was mad# possible by the lov functional load

the distinetion between the two vowels carried*

Not© that th© phrase *oad© possible* suggests

that th© drive to • traplific a tion la la fact

the cause of this particular type of eitonge,

and that the low functional load of this

distinetion merely falls to Inhibit th© merger*

4* Grammar Simplification* The abov© eaases

of ©hang© are all operative at the systematic

phonotio level* Me now move on to causes that

operate at the systematic phonemic level*

King claims that rule loss* rule reordering

and rule simplification "can all be understood

as belonging to simplification In its broadest

34
sense* * They are all due to the ability of

children to constract optimal' - most simple *

grammar# from & body of heard speech material*

Simplification* once again* is therefore a

fundamental factor in this cause of change*

Simplification of this type* however* most often

takes place in the development of the child*©

grammar as the result of a change in the adult*©

grammar of the rule addition type* This must be

the case* since each adult* as a child* must

have constructed an opt 1ml grammar* Change

therefore only takes place if the adult*© grammar

/
 438

is not optimal. Changes of this typo are not*

in other words* truly nspontaneous* changes*

The only "spontaneous" changes are those that

take plaee in the grammar of the adult*'"

These, King hypothesises, "are typically

limited to minor alterationsi addition of

items to the lexicon, minor modifications in

the formulation of a rule, addition of at

most a few rules to a component of the grant*

mar"* Since these minor alterations are

themselves most likely to he the result of

sociological factors, this type of cause of

change is relatively unimportant* We will

show below that all cases of rule simplifies

atlon in Wch English are duo to borrowing or

to other sociological factors, and are not

therefore * spontaneous* *

3* parked and tflrtmrfeed. The theory of

markedness in phonology has been discussed

by members of the Prague school,^ and mors

recently by Postal,**® and Chomsky & Halle,^

amongst others* King has described the part

that markodnees can play in phonological

changeI "In the technical sense of thle word

phonological systems of a natural type* e.g.

/i e a o u/ and /p t k b d g e/* are said to be

less highly marked than systems such as /i e at & vt/

 *89

and /p g e f d a a/* Markedness seesis to play

o part* not completely understood at present*

in the evolution of sound systems* For ex¬

ample, the original stop system of Indo-

Europeani

p t k

b d g

bh dh gh

was highly worked in that it required both

volet and tenseness for its classification

(most languages require at most one or the

other)* The vast majority of Indo-European

languages have developed less highly marked

*0
obstruent systems** In ether words* phono-

logical systems move from more marked states

to less marked* This movement towards less

marked systems is a further eauss of •spon¬

taneous* change* Note that this too can be

regarded as simplification (again perhaps of

a psyehoXogic&i typo) in that loss marked systems

can be regarded as more simple than more marked

systems*

If there is this movement towards less

marked systems, the question arises as to how

marked systems arise in the first place* In

what way can systems become more marked? The

 4^0

answer lies in the interplay of entases of

sound change at the phonetic and at the

phonological levels* we can illustrate this

point from the history of German* It is

generally considered that high front un¬

rounded and high hack rounded vowels are un¬

marked * and that high front rounded and high

hack unrounded vowels are marked# ¥hy then

did German* during the course of the 0110

period* tsovo from a relatively unmarked system*

where it had only /i/# to a more marked state*

with the addition of /u/t The answer lies in

one of the causes of change that operate at the

phonetic levels iuj became |.uj befcre /!/ or

/J/ in a following syllable# The change is a

phonetio one* of assimilation* with LaJ assim¬

ilating the frontness of /!/ and /j/» it first

this was simply a phonetio change* but ultimate¬

ly //I!// aas added to the phonemic system (which

thereby became more marked) when the /!/ or /J/

which had brought about the change was lost#

Thus a simplification at the phonetic level

leads to a complication at the phonological

level (see the reference to Schane in Chapter

Ton)# goto* too* that it Is because of changes

of this type that systems are not continually

being reduced in else because of changes due to

functional load! units are being added to as

well as lost from the system*

 *91

These, then* ar«- socio at least el' the ©auses of

sound changes factors vhich lead to changes *in the

first place** St is notable that they are all* in some

sense* reducible to the principle of simplification*

which* intuitively* seems to be a sufficiently realistic

explanation for why changes occur* This exposition has

been very brief* but was necessary in order to put other

types of changes* on which sociological urban dialect¬

ology has mors to offer* into a coherent framework* Note

especially that none of theso causes is on its own

sufficient to bring about a sound change* Sociological

factors* some of which we shall discuss below* *deeide*

which of all the possible changes will *occur* (i.e.

become generalised te more than one occasion* more tiwn

one speaker* or aor© than one style) and when they will
occur*

P» Secondary Causes of Sound Change

We have listed above the factors which cause sound

change, as we have said, *in the first place*« These

changes can, in turn, lead to other changes, and we will

now move on to a discussion of thoss purely linguistic

causes of sound change which are dependent on and stem

from changes due to initial onuses* Changes of this type

can also be brought about by tertiary changes (see be¬

low)*
1. Pressures in Phonetic,Space* We described

above how the asymmetry of the vocal organs can

be responsible for certain sound changes, by

setting up pressures in phonetic space. By no

aeons all the changes of this type* however,

can be asoribed to this particular cause* Host,

in fact* result from other previous changes*

Wo can say, for example, that the diphthong-

isation of high vowels in the English Great Vowel

Shift was the result of the raising of low and

aid vowels* Similarly* the introduction into

Weh English of the systematic phonetic element

/<*»/ (which itself resulted from an initial

change due to ease of articulation) caused the

fronting of /&/ In phonetic space*

2* Secondary Ease of Articulation* Wo have

already described above how the introduction of

/h/ into Weh English may well have brought about

the centralisation of (o) from [ej to {.*]•

this way an initial or tertiary change (in this

case a tertiary change) which makes articulation

less simple, can lead to a secondary change of

an aesimilatory type which cancels out the now

"difficulty*• Ease of articulation la therefore

a secondary as well as initial cause of change*

3* Grammar Simplification* We stated above

 *93

that moat grammar simplification takes place

during the child's acquisition of a maximally

simple grammar which is based on his exposure

to a certain body of performance data* This

simplification is generally duo to the adult's

grammar on which the child's is based having

become less than optimal# because of initial

or tertiary changes# Simplification of this

type is therefore# strictly speaking# due tc

previous changes. It is# however# of a some-

what different type from 1* and 2* abovet it

occurs only across generations# and has the

effect of consolidating rather than actually

bringing about a change#

As an example# we can cite a case from the

Nch diasystom# As a tertiary change# certain

adults under the influence of R»P» and other

accents began to roalise thoir systematic

phoneme //a// In the same way as //ai//# They

replaced the phonological realisation rule

//S// > /•*/

toy the diasysterale incidence rule

//*// > //ai//.

Since the realisation of //a// is now Identical

with that of //ai//# the child constructing a

maximally simple grammar on the basic of this

data will acquire only one systematic phoneme#

//ai//# and only one realisation rule# Instead

of two.
 *9*

C. Tertiary Causae of Sound Change

Initial and secondary causes of change are purely

linguistic. They result in linguistic differentiation

both between and within speech communities, because the

sociological, psychological and cultural factors* which

control which (and when) changes are to occur* wary from

place to place and time to time# (it is also posslbl#
that noefc*tt*s hypothesis of the random nature of sound

change may play some part in this differentiation*) A

situation therefore arises where linguistic different*

iatlon is correlated with geographical and sociological

differentiation* The different linguistic varieties*

in so far as they are able to interact* give rise to

further linguistic changes. The causes of these changes*

which derive* as we said above* from prlor*enistiag

differences* we shall label tertiary causes of change*

Tertiary causes of sound change can be divided into

two different types* This division is one of convenience

rather than fact* since the two are overlapping rather

than discrete types* It is* however* useful to bo able

to distinguish between changes due to prler«»exietiitg

differences between speech communities and those due to

differences within speech communities* Mote that as

these changes are net purely linguistic* we shall have

occasion to discuss not only those sociological factors

which decide which changes should happen when* hut also

those Involved in the mechanisms which bring about the

development» transmission and acceptance of these changes*

 4<?5

X*Causit of Resulting from Differences

between Communities

l# The Adoption of Prestige Features* The

most important cause of change due to prior*

existing differences between speech commun¬

ities that we have noted in Neh is the influence

of the prestige R»P, aeeent on the Ncti accent.

Obviously# many different change* over long

periods of time have brought about a situation

where R.P, la quite clearly differentiated from

Neh pronunciation# There are therefore various

pressures to modify some aspects of Ifoh pro¬

nunciation in the direction of ft#P# Koto# how¬

ever# that in principle prestige features can

be borrowed from any other speech variety# The

prestige feature can be drawn from a different

eoclal dialect in the same community# or from

a different geographical dialect* The intro¬

duction of the variable (r) into New fork

speech is the result ef the shift of the pres¬

tige model from one different geographical

dialect# that of Eastern New England# to other

geographical dialects whicii have post-voealie

/r/« Zn the English ef England # these two

different aspects of prestige borrowing are

often closely linked# because, since ft#?# is a

non-localised variety of pronunciation (as far

as England is concerned)# the influence of U,P,

 k96

O&n ales both from within and from without

a eootsonitf. fixaaplas of the influence of

R*P. on Hch English have been gives above*

They include the introduction of the systematic

phonetic element /3»/ into the system of C

vowels* This change was effected by re¬

placing the phonetic mutation relet

/X/ ♦ /a/ »'"•) /a#/ or /pt/

by»

/V h» /a/ ■■ ■) /3»/

A rule, we ean say# was borrowed Into MeH

English from It*P.

We also noted the gradual loss of the BAn

short g. This is a result of the Influence

ef ft«P* inhibiting# to a certain extant# the

operation of the optional Incidence rule*

//o// ZMf

That there are socio-paychological

pressures loading to the adoption of prestige

features is obvious ami indisputable* In

practice* however* these pressures aay be of

 *97

a very complex nature* and we ahall not

disease them further here* They are overtly

most apparent In the speech community in the

educational institutions* hut are probably

most important in faee<*to*»faee contact situ*

ations* This means that features of this

type are introduced first of all into tho

speech of the MC, and only later penetrate

to the lover classes* We have been able to

state that some of the changes of this type

in Neh occurred when they did because of the

increase in geographical mobility and educational

opportunity*

2* The Adoption of Features from Non-Prestige

Varieties. We have also noted in oar study of

changes that have taken place in Nch that certain

of these are due to the borrowing of features

from the dialects of Suffolk* the Home Counties*

or the Midlands* Xn some oases* it is not poss¬

ible to distinguish this type of change entirely

from changes due to borrowing frota R*P* For

example* we can associate the rounding of (o)

both with the Influence of P.P. within and with¬

out the Neh speech community* and with the rounded

vowels that occur in the accents of the Home

Counties and Suffolk* Xn cases like this* the

two typos of causality of change reinforce one

another*
 0*

CO

NP

H 3 £

a 9 S 9 9
0 €

«* c?
jy ♦s

*53

S3 «*■
3

93
0

sar t*
0

{?
H

5?
• 9

t+ sr
I 9

«#•

f 0 3
1 0 9 0 3 9 0 3 0 m 3 3 o 0 a
9

• p»
«* ** c* *% «♦
I*

? & ff
3
s 9 a 9 ? § 9 0 1
H, P»

«* «0
c*

0 tt

Ci3 & S
aS
*

•
3 *
I i I

«*
3 9

!f 5II
i . i
&
« 0

««■
a a
1
3 9

«*
tr

z
» p P 4 0 9 rt-
«s
« « 9

sr
t* s*

1 o

& ? E I
p. 0 a o
e s 0

**■

I
«
4! w i

P.
* ar
*■
'or *
#»
I I ?o g 0

«*
§ o

*%
*>

1#
I y
0 p. 3

t»
5
 b99

The iiost common form of spread, as tre

•av» ia from London and tbo Home Counties into

Moh» la what some ia the WC speech of London

more prestigious or influential than that of

the Neh VC? Obviously, London is the political,

cultural and social centre for the whole of

Britain, and for that reason has a very great

deal of influence# But why should this bo

associated with its WC? for the answer to this

question wo must loole to tho theory of the

diffusion of innovations* Wo havo mentioned

this topio before in a rather inexact way. We

must now attempt to make a more rigorous inter*

pretation of the way in which linguistic innov-

ations are diffused in general and into Kch in

particular.

Innovations of all trues - although this

is particularly noticeable in, say, the world of

fashion » are diffused outwards from London to

the rest of Britain, This is because of the

dominant cultural position of London and its

status as a "world city", Xt is also duo to the

more size, in terms of population, of the city.

The level at which innovations are diffused is

generally that of personal interaction* Thus

any larger population will tend to dominate a

smaller population. Linguistic innovations
 500

spread tress London in the eatnc way* an4 «r«

eennrally ««nftned to the we siaply taMtot

of tlio inhibiting ami cooaamtiv# effect that

f?#F, has on tho MC* (Innovations vlthln R*P*

will ganeraUf taks tho oarao course» •advanced*

It*F« forn viU bo Bora common in London thou

in tho pmlaeaft) Tho London tfO i« more

•prestigious- than tho Heh WO simply boenuso of

tho gMterai dominance of London*

Linguistic innovations, than* spread eat

from London to other enjor eittea, and from

thooo to minor cities* ami «o on* (This does

not noon to say of course that all linguistic

innovations start In London* They cioarly do

not*) Innovations* in other words* travel down

tho urban hierarchy. ll&gerstrand into writtont

•The urban hierarchy canaliooo tho course of

diffusion* la addition to tho influence froa a

oeatro on tho neighbouring district* wo find

short cirouits to tho more important places at a

greater dlstaac#**^ This explains why (h)«2 is

found in Boh but not generally in $fk# As well

as spreading sore gradually serosa tho country

into tho hinterland of London* (b)*2 also

Jumps from one member of tho urban hierarchy to

another*

Innovations* however* do not simply novo

 501

down tha urban hiorarahf* If this vara the case®

Glasgow speech vonld more eloaeljr resemble

Uadtm speech than Ncti speech dees* Obviouslyi

distance alto play® a vary lopoHant part in

diffusion* Innovation diffusion Is subject to

distance decay* If we wish to study th« exact

aiaturo of tbo Inflaono® of Isadoa English on

Hot tr® asst take Into consideration (l) tho

transmission of innovations down the urban

hierarchy* and (ll) dlstanoe decay* Wo must

also consider the spread of.Innovations across

country.

Geographers have developed several .

models of tho diffusion of innovations* some

of thsa rather sophisticated and complex* For

our prosont purposes* we oaa adopt one ®f th®

alnpler sodolo in order to demonstrate in a

aero rigorous way the nature and extent of the

linguistic influonoo of London on Neb* The model

we shall adopt is the so-called gravity model,

which has also boon applied in many other sci¬

ences in addition to geography* *!tiis model

noes the Newtonian theory of gravitation, and

suggests that the movement between two centres

is proportional to the produots of their popul¬

ations* and inversely proportional to the square
42
of the distance separating them*B w« can
 502

therefore use this model, as a measure of

the relative amount of Interaction and in¬

fluence between two centres# It can give us a

simple numerical demonstration of the in**

fluence of London speech forms on Neh speech

relative to the influence of other centres*

The formula for the gravity modal that

wo shall uso is*

HtJ . PI F.I
<-ujr

whers Mil 1® the amount of interaction between

two eentres 1 and 1* which have populations

of Pi and PJ respectively* The distance be-

tween the two centres is dij*

Where 1 • London and 1 m Noh. the equation

reads as follows (with populations in thousands

and distance in miles)*

M.London*Nch • 10.000 x 120 m 1*200*000 • 99.2

(110)2 12,100

Where, however, i ■ Birmingham, we have the

following result*

If.Birmingham*Nch » 1.000 a 120 m 120*000 » 5*4

(150)2 22,500
 503

If v« multiply these results by 10 we have

the staple Index scores t

Interaction of London and tfch ■ 992

interaction of Birmingham and Nch • 5^

The influence of London on Xdb would therefore

appear to be at least 20 times that of Blraing-

hoot Even this difference* however* is clearly

a distortion of the facts* The number of changes

that spread from London to !fch is significantly

greater than 20 times the number of changes that

spread from Birmingham* since very few* if any*

changes follow this latter route*

There are two main reasons for this dis¬

tortion* The first is that we have not taken into

account the fact that it is a linguistic situ¬

ation that we are dealing withi prior-existing

linguistic similarities must therefore bo taken

into account* JCch English is more similar to

London English than it is to Birmingham English*

London features are therefore both linguistically

and psychologically more acceptable into Neb

English than are Birmingham features• Since

linguistic similarity is not simply a function of

distance - the English of* say, Canterbury is in

many ways more sltollar to that of Nch than la the

 50k

English of Peterborough • we shall modify the

formla in the following ways

MS « a. J2UU
(<UJ)2

where e is a variable expressing prior-existing

Unguis tie similarity* Linguistic similarity

and difference is very difficult if not imposs¬

ible to measure accurately* fa"© shall therefore

solve this problem by allowing the value of &

to depend en dialect area differences* The dia¬

lect areas will necessarily be somewhat arbit¬

rarily defined* In the ease of Nch, the value

of £ will vary with the dialect area of the

ether centre in the equation as fellows«

• m k for centres in Kfk

s m 3 for other centres In HA

• m 2 for other centres in the South-Bast

a © 1 for other eentres in England

e m O for all other oentreo

The revieod formula produces the indices shown

in Table 7 for the interaction of Nch with

other centres* Note that this does not necess¬

arily mean that the speech of Glasgow can have

no effect on Nch speech* Any effect it does

have (and at least one writer has postulated

 305

Table 7i Indices of Interaction of

Nch and Selected Centres

London 1,98k Cambridge 66

Yarmouth 500 Birmingham 5k
Ipswich 215 Peterborough 17
Klfi&b Lynn 113 Glasgow 0

that the spread of the glottal stop realisation

however* be indirectly transferred through some

other centre such as London* Those figures show

why the speech of London has such a strong in*

fluencs on that of Mcft, particularly In those

oases where London features are also found In

Ipswich and/or Cambridge speech* According to

thie formula* London exerts approximately 40

times as much Influence on Kch as does Einafng-

ham* This figure* however* depends on the very

arbitrary allotment of values to &» which makes

London speech exactly twice as similar to Nch

speech as Birmingham speech is* This assumption

clearly has no validity* All we can claim is

that the influence of London is clearly touch

greater than that of any other centre*

The second reason for the distortion Is

that the above formula provides us with a

 506

measurement of the interaction of one centre

and another, not the influence of one on

another* Wo hare given the index score for

the Interaction of London and Neh ae J$84,

Part of this interaction, however, consists

of the Influence of Nch on London* The in*

flaeaee of ono centre on another, other things

being equal, will depend on the relative size

of their populations* in view of the big dis¬

crepancy in the size of the populations of

those two centres, the influence of Neh on

London will b# extremely small, but it must be

taken into aeoount* The ratio of the popul¬

ation of London to that of Neh is approximately

Bill, No can therefore introduce the concept

of direction into our measurement of interaction

in the following way* If the Interaction of

London and Nch is expressed as 1,934, then the

one-way lnteractioa, the influence of London on

Neh will boi

82 x 1,984 - JU260
34

Tho formula for obtaining an index of the

linguistic influence of mm centre on another

is therefore!

Ii4 « io*9* pi PI . PI
(dij)2 PiePJ
 507

tf« mult therefore revise the list of index

scores given in Table 7* The revised Hot

appears in Table 8*

Table St Indices of the Influence

of Selected Centres on Set

London 1»9£»0 Kings Lynn 39

Ipswich 106 Cambridge 33
Yarmouth 50 Peterborough 6

Birmingham 43 Glasgow 0

Thl«« then* provides an explanation for

why features are borrowed from some dialects and

not from others* for why non-prestige features

are borrowed at all* and for why changes are

diffused down the urban hierarchy* The question

etlll remains as to why features are borrowed

when they are borrowed* The question might* in

fact* be better rephraeedt why are features not

borrowed instantaneously? The answer to this

problem lies in what we have referred to above

as distance decay, which is provided for in the

equation by the expression (dlj) . Distance

between two centres makes for friction* which

can slow down or even halt the progress of an

innovation from one to another. It also lies in

 y
S 9 ft"
©
as

f#
I
ft

•ft
ft ft 9 9
ia
&
3

ft ?9 <* s ©. 5 «♦
«+
S
ft ft
W>
*
ft
t

«*
I sn H ts
ft

&> «;»
«» I*
o p

© » a ft § ft ft ft M
ft
*a
•4

■» .9

ft

i
i

as
ft

sr
M & 5 ft
S ft " i c" ? • S •©II
S I g
ft I
g

ST
ft

a * 9 ft
o. 9
1 % I ft
ft ft
**
sr
«*
«*

« « » s © © ©
©

t*
«♦> «»

?
I
§& *(•
ft>

?
1

& K I
&
« a B S O ft s ©
o ©
Br

f*
S3

%
a ft i
 E f ! a $ H
i M 9 9

<9

3 3 S

*
tt

If •*
9 0

•9
P*
o 9 a

e* J* «♦
8

•1
9

"9

I 3 9

(0 1*

s3 I 8
? C
I
9 i
9 I * 9
*
Ui

JL

Z
s
9
Rf
g .
i s! § ♦»
*»
E

59a sI »
3

<*
**
Z
I e
8
9
9

«
5
s

S
IP

E
& ^
I 9 9 9 9 9 a 9

tsr *!
<*

p
n
E
i

i*
§ 0 s %

«♦
§ 9
I s

«*
9
at a»3
9
8

m P t s
9 S

I
9 I
g 9
S! % s 8
f 8
s

f 91 ? a 9 t K ?
9 » 9 9
St
 510

/tf/1 /Co/ or The origin of those In¬

cidence roles la obscure, but we did see in

Chapter Seven that; there la some geographical

variation in the use of these variants in

rural LA* It is therefore possible that the

variation to be found In ffoh ia dialect mixture*

resulting from ia-migration froat surrounding

areas of Hfk which have different variants in

items of this lexical set*

Map 1 shows the childhood homes of the

rural-born informants who have /Vu/ in soup

and those who have /<h/» (Mote that in order

to obtain genuinely comparable material these
4

maps are based on forms produced in VLS and

ftps,) If the speech of these informants is at

all typical cf the areas in which they grew up,

then it is clear that there is some genuine

geographical differentiation in the use of these

two variants* This is also true of Map 2, which

contrasts the uss of /Uu/ and /X1/ in room*

Ve would of course need a larger sample of

rural-born informants in order to be able to

draw more convincing conclusions, but we are at

least able to suggest that immigration has

played some part in causing the linguistic vari¬

ation that occurs in Nch.

 511

XX, Causes of Chang* resulting from Differences

within Speech Communities

Jiust mm Initial and secondary causes of change, in

combination with various sociological factors, bring

about geographies! differentiation between speech

communities, so they give rise to differentiation within

communities, As ws have demonstrated in the easo of tho

Neh speech community, this differentiation is generally

correlated with social factors such as class, ago and

sex* Differentiation within the community leads to

various types of linguistic sod social interaction and

pressure#, which can in turn result in further, tertiary

changes. Sons of the causes of change due to these

linguistic and social pressures are listed below.

1, ffypercorreotlon# Labov has shown that

one typo of change due to prior-existing

differences within the speech community is

brought about by hyporcorrection. for ex¬

ample, IMC speakers In New fork,and particularly

middle-aged speakers from this class, not only

adopt characteristics of Upper Middle Class

speech la formal contexts but actually surpass

UMG speakers in the use of these characteristics.

In the case of post-vocalic /r/, *lower middle

class speakers go beyond the highest status

group in their tendency to use the forms con-

46
aiderod correct and appropriate for formal styles".
 512

Labor interprets this as the result of the

linguistic insecurity of the LMC (which has

also been pointed out in the case of the 2feh

sample). Crudely put, LMC speakers are anxious

to assume the prestige associated with auoh

forms, and therefore use them as much ae possible.

Change results from the fact that thie

hypereorreotion represente an inerease in the

use ef a prestige form. It also results from

the faet that middle-aged speakers of the LMC

generally adept the formal speech pattern ef the

younger CMC speakers, which "provides a feed*

back mechanises which is potentially capable ef

accelerating the introduction of any prestige

feature"• The acceleration is due te the faet

that the L?IC youth is "in contaet with the new

ky
prestige pronunciation on two fronts*t ' they

are in eontact with them, because of their ago,

in the speech of young CMC speakers, and, be*

cause ef their social class background, in the

speech ef middle—aged LMC speakers,

Labor statesi "The suggested role ef

hypercorreetion in the acceleration of linguistic

change has been put forward with the expectation

that further empirical studies may confirm or re¬

fute thie possibility. Similar investigations may

 313

profitably b« carried out la other eitleo«

perhaps la those which do not show as great a

range of stylistic warlation la the speech

JLA
community** We can present supporting evidence

for the linguistic insecurity of the LMC from the

ifoh survey, there is, on the other hand, no

evidence at all of hyperoorroction of thie type

in Nch. Isolated instances of hyperoorreotion

do of course occur (such as novo /nuvuv/). But

there are no examples of the kind of structured

hypercorrection that occurs in Hew York* Hyper-

correction* therefore* cannot he considered a

universal mechanism of change, even in urban-

areas* It is probable* however* that hyper-

correction stems frost particular tendencies

associated with group identification which may

be universal (see below)*

There are certain differences between Hew

York and Nch speech which can explain this

difference between the two eotaaunltles* The

prestige model for Now York English is a geo¬

graphically external one* Speakers from these

external areas will use* say* 100 per eent post-

vocallo /r/« Because* however* the model one

that is foreign to the city* even CMC speakers

la New York will not achieve this particular

level* This means that there is* as it were*

room for the LMC to overtake them* The prestige

 51*

model for Hen «pe«cht however* as for all

types of English English speech* is R»P, *

which is a non-localised variety* the prestige

model is an internal as well as external model*

There are R*P. speakers in the Wch speech

community. If post-vocalic /r/ were a prestige

marker in England* all R*P* epeakers would nee

something like 100 per cent post-vocallo /r/*

The presence of these speaker* in the Hch MMC

would make it that much more difficult* al¬

though of course not impossible* for L&C speaker#

to *overtake* them* We are therefore much less

likely to find this type of hypercorreetion in

an English city*

Zm Group Identification* It was suggested

above that hypereorrectioa may be due to some

form of universal or widespread tendency assoc¬

iated with group identification* We can now

examine these tendencies in more detail*

(i) Prestige Groups* One type of

change due to tendencies of this kind

is that which results from a prestige

group'# desire to remain distinctively

prestigious* Members of groups of

this type can wish to prevent "over¬

taking" of the type Just described

 515

from taking place* la order to preserve

their dietinet upper-class Identity•

If* therefore, a prestige feature la

iavolt«d la linguistic ehange la a

particular dlrectloa* the prestige group

must continue to develop the foma the?

use in this same direction la order to

bo able to signal their group membership*

Failure to do this may result in the UfO

also acquiring the prestige marker*

x This Is of course an unconscious process *

but it is one way la which changes of

the "drift" type may be produced. Qae

possible change of this "vanity factor"

&0
type Is described by Joos, 7 Another

possible example is the continued fronting

of P.P. /©h/ from S.00J to L©fj Le^i«

The last variant is described by Olason

as an "affectation",but may well be*

come more widely established in K#P*# in

order to preserve a distinct advanced

P.P. form,

(ii) Pon-prestige flrnmns. Changes of

this type are not confined to prestige

groups• The desire to express group

membership and to maintain or exaggerate

 H

4* i 3 *IH H sa

»4 o O 3£ 4* S s t3 C ot II

sa G e• » » I G I S 51 »4 O 2

I 5£ § _
01
'

S• ll * Sst 5

% *► s 9< jt 4« * »4 5 1 32

& a© Q 2 2 I II

»4 I « »4 « ou »4 J 1 2 i

9* s »*• »*« 2 15 i G 4V s« © 4
 517

Because this feature occurs mainly

in ¥0 speech in Nch, (and certainly does

not occur in P.P.)» wo can postulate that

the cause and mechanisms associated with

this particular change are the following*

Centralised variants of (e) have become

recognised as a marker of membership of

the ¥C in Neh, and have therefore acquired

favourable associations with anti-

authoritarian attitudes, masculinity,

or other features. The desire to express

group membership* which can perhaps be

described aa *WC solidarity* has led to

a consistent increase in the centralis¬

ation of this vowel, in order to differ¬

entiate this group mora clearly from (e)-l

users* (This point is underlined by the

fact that the UtfG has higher (e) scores

than the MfsC, and both have higher scores

than the UJC - see Fig* 20 in Chapter

Fight*) This thsrefors represents tho

complete obverse of the hyperoerrection ef

a prestige features exaggeration ef an "anti-

prsstigs feature* •

Ve can provide seme confirmation for

this hypothesis in the following way, Labov

! ? M *
i e

<* »*

#►
? ?
a I
? 0?
s

SR
I §
§

s
I

3 I i a «

M. ft 5§
s
I a
I a a I
3 a g 3 1 a o *

n a
o s
a

i*
a
$ s, ft

n 0 0
I
a 8 3 i a

«♦>

f n
« a a a

*►
t*

I
| gI|Is
1 I ft a
0
s 8 3 3

*<
«e

;
r s
g"" igf
° i • 8 3 a
s
rt- S
2 3

q- J
i ' * O1
H
I i
P

IE 1
• i
• • I*
•
S
I
.? $ I 1!
i*
&
!

»
0
 \K tf

t-< H tm »* M. 5 I »*■ OK I3

O ©U $ 5 A« 1 I g tt £

e <i ©i H 0• K3 >* 2 «f & (* 1 1

5 sa « 2 » *« 4« >§ I

I g

»4 © S5 »4 2 « « if »♦ 3< 1 si « 3

»g « i w&
•
v> 2 fH« iV © SI s 5w i »♦ A fe 29

•§ H© JJ «— ti »* H1 * »4 s 95* »4 3E« M » « wHJ £1 ■

§ a»j 3 e

© 4r ► © lf © 3S » ©U ©H ©1
 520

used over 30 per cent (r)*i (m poet*

vocalio /r/) were, generously, considered

to bo (r)-l users. 70 per eent of those

who, la this sense, were (r)*l users

reported that they normally used (r)-l*

hut 62 per cent of those who were not

(r)*l users also reported that they

normally used {r)-l. is Labor saysi

*Xn the eonselous report of their own

usage ... Hew Torh respondents are very

inaccurate**52 The Inaccuracy, moreover,

Is mainly la the direction of reporting

themselves as using more prestigious forms

than they actually use* Labov claims that

*no conscious deceit plays a part in this

si
process**'*' and that "most of the respond¬

ents seemed to perceive their own speech

In terms of the norms at which they were

aiming rather than the sound aetually

produced**511

The full results for this test ars

shown In Table 9* (Mote that (r)-0

signifies nonius# of poot-vooalie /r/*)

Table 9 shows that 62 per oent of (r)-0
users *over-roported" themselves, and 21

per oent of (r)~l users "under-reported"*

 PS*

Tabid <M $olf»«valuation of (r) - Now Tork

for oenl (r)

Foportcd

03 ed A 2
1 79 21 • 100
O 62 33 m 100

although tho latter were probably simply

being aoourate in view of the 30 per cent

dividing line*

Now* in comparing the resulte ob»

tained in the self•evaluation teat to

forms actually used in Nek* CS was used

rather than FS, einee C3 more closely

approximates to everyday speech * to how

informants normally say words * which is

what they were asked tc report on* In*

formaate* moreover* were allowed no

latitude in their e elf•evaluation* They

were considered to use in everyday speech

the variant indicated by the appropriate

CS index* For example* a (yit) index of

between 051 and 100 indicated a (yu)»2
user* If* therefore* the characteristics
 523

the percentages of speakers under- and

over-reporting shown In Table 11* Male

speakers* it will be seen* are strikingly

more accurate In their self-assessment

than are female informants*

Table lit Percentage of Informants over-

and under-reporting (yu)

22*21 Male

Over-reporting 13 0 29
Under-reporting 7 8 7
Accurate 80 9* 6k

There ls» however* some considerable vari¬

ation in the assessment of different vari¬

ables* Tables 12* 13 and Xk show the per¬

centage of informants over- and under¬

reporting their usage of (er)« (o) and (a)

respectively.

For each of these four variables there

are more male epeakere who state that they

use a less prestigious variant than they

mutually do than there are who over-report*

and for one of the variables* (c)» the

 524

Table 12« Percentage of Informants over-

and under-reporting (or)

Xntal Male Female

over-reporting 43 22 63

Cttdor-roporting 33 50 14
Accurate 23 23 13

Table 13t Percentage of Informants over-

and under-reporting (o)

ixtaA Male tWBM

Over-reporting 13 12 23
t?nder-re porting 36 54 13

Accurate 45 34 3?

Table 14* Percentage of Informants over-

end under-reporting (a)

tsm. Male ESWH&ft

Over-reporting 32 22 43
Under-reporting 15 23 0

Accurate 53 50 57

difference is very striking* 54 per cent to

12 per cent* Comparison also shows that the

 525

Nch informants are much more accurate in

their than are the Hew

York informants, and are also much more

prone to under-raport• Neh fernalo in*

fonaaats are more inclined to over-report

than the stale informants*

Xn this tendency to report themselves

as using less etatusful forms than they

actually use, it is possible to oeo an *

expression of the forces that lead te the

exaggeration of stigmatised forms* Xt

would seem that, particularly for Nch men,

the connotations of tfC speech arc in many

ways very favourable* This is not overtly

so* As in New York, many speakers expressed

"linguistic self-hatred"* But their res¬

ponses to these teste shew that these overt

statements are for public or conscious con¬

sumption, Privately or subconsciously, a

large number of speakers find the need for

group solidarity more important than the

noed for prestige, in the more usual sense

of the word.

Xt is therefore fairly simple to see

how group identification of this typo

functions in linguistic change* Labov has

 5 26

postulated that New York speakers haws

as their norm a variety (which they "hear"

thesis elves using) which is more statusful

than their own. This leads to change in

the direction of the prestige nana* The

Kch material suggests that Neh men have

at least to a certain extent a norm that

is a less statusful variety than the one

they themselves use* Change therefore

takes place in this direction* and (c)

becomes increasingly centralised*

Resistance to Change. Group identifier

ation* as veil as causing change* can

also bring about resistance to change*

The fact that a particular form may be

identified with a particular geographical

area or social class can lead ether groups

to resist the introduction of this form

into their own speech* in order to pro*

serve their own group identity* This may

explain why some London forms are accepted

into Nch English and others not* or why

some may take longer than others to b#

accepted* (we have already mentioned this

typo of resistance in our discussion of

the resistance to (h)«2 in rural fffk*)

 527

V« hav« regarded the causes of phonological and

phonetic change as consisting of three types* Initial

caases are constantly leading to change* but theee

change* are inhibited from re-occurring or spreading

by the need to be understood* and by various socio¬

logical factors* Where they do occur* they can lead

to secondary changes - when the linguistic and socio¬

logical conditions permit* Both types of change result

in the social and geographical differentiation of

language* vhich leads* by means of various pressures

and mechanisms (socio of which we have described above)

to tertiary change#• Tertiary change# can in turn lead

to further secondary changes*

 528

NOTES

Notes to Chapter One

1. See W. Bright9 "Introduction• The Dimensions ot

Soclollngulstlcs" t in V* Jiright (ed.)» Soclolingutat-
les* (The Hague* Houton, 1966)f A. Capell# Studios
in Soclollngniat les. (The Hague 1 Mouton, 1966).
2. See T« Hill# "Institutional Linguistics"* Orblo 7.
(1958) » PP. 444-455*
5. See li. Koijer# "Anthropological Linguistics" # in
c. fiohrmann. A* Sommerfelt A J* Wkatmough (eds.)#
Trends in rinropean and American. Linguistics* 1930-

I960. (Utrecht* Spectrum. 1961).

4. See 11,J.L. Berry A A.Prod, Central Place Studies..
A Bibliography of Theory and Applications. (Phila¬
delphia* Regional Science Research Institute» 1961)*
5* See refereneos in Chapter Two*
6. See P. Raggett. Looatlonftl Analyols in Human Geo¬
graphy* (Londont Arnold( 1965)*
7. See T. lingers trend, innovat i ons forloppet ur Koro»
loglak Synrmntet. (Lund* Gleerup. 1953)*
8. IS. Kurath & R.I. McDavid, The Pronanclation of
English in the Atlantic States. (Am Arbor 1 ©d-
vorslty of Michigan Press# 1961).
9. Soe J. Lyons. Introduction to Theoretical Linguistics.
(London* C.H.P., 1963)# 34-35.
pp.
10. C.y. Hoekott» A Courso In Modern Linguistics * (New
York* Hacraillan, 1953)# p. 486.

11. Sec the discussion in Z.S. Harris» Structural

Linguistics. (Chicago* University of Chicago Press#
1951). PP. 9-11.
12. See W. Labov, ^Tho Social Stratification of English
In New York City. (Washington* Center for Applied
Linguistics# 1966a).
13. Ibid*
14. Ibid., p.v.
 529

15* S«e» for example, the reference in P. Kiparsky,

"Linguistic Universale end Linguistic Change*, in
B, Bach. A R,T, Harms (eds,), Universale in Linguist-
ie Theory* (New Torkt licit, Rinehart and Wilson,
1963), p# 179# to F» Enderlin, Pie ?fuar*dart von
Keast-rll In Oberthurrsm, Poltra/re mxs> achwelsser*
dontsohen Qrajwnatlk 5, (Ed, by A. Bachraannt Frauon—
fold, 1911),
16* See W, Labovi Some Sources of„ fioadln/g Problems fey
Negro Speakers of Non-Standard English, (N.C.T.E,

Spring Institute on New Directions in Elementary

English1 Chicago, !larch 1966b, Unpublished)«
17* See D, Lawton, Social Class, ^ngaare and Education,
(London* Routledge and Regan Paul, 1963),

Notes to Chapter; Twq

1, Fast .Anglla, A Study, A first report of the East

Anglift Economic Planning Council, Published for
the Department of Economic Affairs, (London* H.M.S.G,,
1963), pp, 22 and 113,
2, Ibid,, p,25. The London overspill scheme moons that
Ipswich could quite soon achieve a population
greater than that of Norwich,
For an exposition of the concept of urban hierarchy,
see J,H, Johnson, Urban geography, (London* Pergeuaon
Press, I967K pp, 92-9$• See also R*E« Dickinson,
The City Region in ^astern Europe, (London* Rout-*
ledge and ICogan Paul, 1967), pp, $7*59*
k. For an exposition of the concept of hinterland, see
J.lL Johnson, fop, ctt»), pp, 89*92# See also
J,D, Nystuon A H.F. Dacey, "A Graph Theory Inter¬
pretation of Nodal Regions", Papers and Proceedings
of the Regional Science Association, 7# (I96I),
PP# 29*$2, also published in R. J.L. Berry A
D,F, Marble (eds,), Spatial Analysis* A Ponder in
Statistical Geography, (Englewood Cliffs* Prentiee-
 530

Kail* 1968)• pp. *»07*4l8. For the similar terra

"tkalond"* see B* Eani, * Utalond Studies and Sector
Analysis" in Studies in Rural-Urban Interaction.
Lund Studies in Geography* eer* B« Human Geography
no* (Land* Sweden» The Royal university of
3»
Lund, 1551)* For the terra "Urban Field"* see
A.B, Sraniles» The Oeoracanhv of Towns. (London!

Hutchinson* 1957)» Chapter VII# See also R.E.

Dioldnson, op. elt«. p* 31. For linguistic in*
finance of cities on hinterlands* eft ff.A.K* Halli-
day* A« Mackintosh & P. Strevens, The Linguistic
Sciences- ft Language Teaching (London» Longmans*
1964), p. 85.
5« "She only remaining manifestation of the /h/ phoneme*
initial provocalio ^hj, began to be lost as early
as the 13th century in many dialects. Zte loss in
the dialects south of the Hunfcer (Vachetc should
have excepted East Anglla* where it is still very
raueh alive) has resulted in the total disappearance
of the phoneme in these dialects"* V.N. Francis*
Review of Vaohek, "On the peripheral phonemes in
Modern English** Language hZ* (1966) * pp. 142»149«
8. v* Labor* op. cit.. <1966a)* p. 328* and footnote
P* 379* Bee also V* Labor* "On the Mechanism of
Linguistio Change"* Georgetown University Mono*
graphs on Languages and Linguistics. 18. (1965)*
pp. 91*11*.
7» East Anglia. A Study. (op. clt.)
8. The publication of the complete East Midland and
East Anglia section of the Leeds University English
Dialect Survey should reveal to what extent this is
true-, (fl. Orton A P.M. Tilling* Survey of English
Dialects. Volume IH. East Midlands and East Annlia.

(Leeds1 Arnold* 1969*)#)

9* See* for example* Q.L. Brook, English Dialects.
(Londom Deuteoh* ? 963) * p. 153* inapt "English
Dialects in 1887* .

10. See Fast Anglla. A study, (op. clt.). Chapter 6*

 331

pp. 20-25, "A City Region Strategy*, and map p. 53,

The Neb city region consists of the whole of Hfk,
except for the western Loeal Authority areas of
Peeking* Hunstanton, Freebrldge Lynn, King's Lynn,
Downham, Downturn Market* and Marshland, It also
includes the Loeal Authority areas of Lothlngland,

Loweetoft* Beetles, Bungay, Wainford, and Southwold,

which are in the north-eastern part of East Suffolk,

The recent Royal Commission on Local Government has

stated that it considers most of the above areas of
western Nfk to be more closely tied to the rest of
Nfk than to Peterborough, and has proposed drawing
Local Authority boundaries very similar to ths al¬

ready existing County boundary at this point,

11. A,A. Wood, "City of Unique quality", in Eastern
Evening Hews Supplement, Norwich - The War Ahead,

(Norwich! October 22nd, 1963), pp, 1 and 2,

12, D,J. Cobbett, "Transport - the Pattern for the
Future", ibid* p. 11*
13* P, Deavin, ibid,. pp, 8-9,
lfe. Adapted from the map prepared by the Ministry of
Housing and Local Government in Fast Aivrlla, a
p. 88,
15. Ibid*, p. 22.
16* "Most of the important dialect boundaries in the
Eastern United States fall along lines which arc

natural troughs in the network of communications,"

W. Lab©v, on, elt.. (1966a), p, 499. Soe also

footnot# p, 502,
17* A,E, Smallos, "The Urban Hierarchy in England and
Vales", Geography, 29. (19Mt),
fcl-51* pp*
18. W, Chrlstaller, "Das Grund&eruet der R&ualichen
Ordnung in Buropa", Frankfurter Geo/?. Hefts. (195®)*
Map also reproduced in JUE. Dickinson, (op. cit.),
p* 166,
19. From East Anrrlla, A Study, (op. cit.)
20. For work by human geographers, see T. Hagerstrand,
The Propagation of Innovation Waves. Lund Studies

in Geography, ser, B.Human Geography, no. k, (Lund,

 532

Swedeni The Royal University of Lund, 1952)1

"Migration and the Growth of Culture Regions*
in Lund Studies in Geography ser. B. Human
Geography, no* 3, (op. cft.)i "A Monte Carlo
Approach to Diffusion*, in European Journal of
Sociology* 6. (I963), pp. 4>67| and on. clt*.
(1967)*
See alee V* ttunge, Theoretical Geography. (Lund,
Sweden1 Royal University of Lund, 1966), pp. 112*
17$ and references. For work by Unguistlo geo¬
graphers, sso A. Bach, Deutsche Mundartforachnng.
(Heidelberg! Winter» 1950), p. 136| and E# Sehwars,
Die deutschen Mundarten. (G$ttlngens Vandenliouck
and Rupreeht, 195®)» p. 7%* See also D. Be Camp,
"The Pronunciation of English in San Francisco*,

Orbla7. (1953). P. 372.

21. A trtaik road is a road for which the Minister of

Transport, rather than the Local Authority, is the

highway authority. The nam national network of
through routes is composed of trunk roads*
22. Gensral Register Office for England and Wales,
Census 1961, England and Wales. County Report!

Norfolk* (London! ii.M.8.0., 1964). General Reg¬

ister Office for England and Wales, Sample Census
England and wales. County Reportt Norfolk.

(London! H.M.S.O., 1967)*

23. Rote that the Eastern Region of the 1961 Census
differs from the EAn region of the 1966 Sample
Census* The I96I Census Eastern Region consisted
of! Bedfordshire, Cambridgeshire including the Zsle
of Ely, Huntingdonshire, Norfolk, Suffolk,ami those
parts of Hertfordshire and Essex not in the London
region*
2k* Tho 1966 figures are, of courso, not strictly com¬
parable with the 1961 figures, since they are in
fact taken from birthplace tables, and hence refer
tc all living immigrants to EA who have remained in
the region, rather than to immigration over a
 533

specified period of time* Even a large pre*

dominance of lodgrants Into EA from the SE would
not be surprising* in view of the very largo
population of the SE* and lie proximity* The
population of the SE region in 1966 was somewhat
lose than three times as large as that of tho
second largest Region* tho north West. There were*
however* between six and ten times as many immigrants
into £A from tho SE as from tho other Regions* Ses
also note 23*

25* B* Green A R*M.R* Young* Norwich * Tho Growth of a

City* (Norwicht Museums Committee* 1964)* p* 13*
26. Ibid,* and 1* Fowler* "An English Province and its
Capital*, in F* Criers (ed*)* Norwich and Its Region*
(Norwichi British Association for the Advancement
of Science* 1961)* pp* 10*17*

27* Ibid** p* 12.

28* A*A. Wood* (op* ett*)
29* Census 1961. (op* cit.)« Preliminary Report*
30* C*A* Moser AW* Soott* British Towns* Centre for
Urban Studies* Report No* 2* (London* Oliver and
Boyd, 1961)*
31. *bitl»» P- 134*
32* P.O. Thomas* "Commercial and Professional Services*,
in F* Briers (ed*)* (op* cit.). pp* 211-224.
33. I^ld*. P. 211.
34* A*R* Cartwright * A* Fitch A D.L*J. Robins * "Manu¬
facturing Industries"* (ibid*)* pp* 192-210*
35* Sample Census 1966* (op* cit«)
36* The figures in the following section are taken
partly Yromi J.H. Mabry, Norwich 1961 - An Analysis
of Censue Returns, an unpublished study of 1961
Census returns for the sixteen Nch electoral wards

obtained from the General Register Office* which

wae kindly given to me by Professor A.n. Emerson*
Professor of Sociology, university ef East Anglla*
37. See A.E* Smiles* op* clt.* (1957), p* 87* Seo
 53*

also p. 92 for remarks 0x1 intrusive features ai

sign* of residential deterioration in inner
integuments.
33* Not to be confused with the suburb of the saxao name.
39* "Enclave# are »« « reservations for special functions
»m est apart from the free interplay of forces
that make for the normal patterning of a town."
A.E. Smallest op. clt.. (1957). P. 101.
*0# Crome, moreover, does not share with the south and
south-vest in the ether pattern# of eooial dominance.
See, for example* the high figures for unemployment
Shown in Pig. 13.
*1* The indices for the soolo-eoonoisic class of each
ward were obtained by sumcalng three of the indices
developed in J.h. Mabry, fon« eit.)* and calculating
the mean. The indices vers* the socio—eoonomie
status indent the housing status indext and the un¬
employment index. The socio-economic status Index
is calculated by Mabry on the basis of the per¬

centage ef persons in each ward whose terminal' -

education age was sixteen years or ever* together

with the percentage cf seals non-manual workers in

cash ward. The housing status index ie based en

the percentage of occupied households with the use
ef four facilities (hot and cold water, fixed bath*
and W.C.)* and the percentage ef dwellings privately
rented. The unemployment index, is based on the per¬

centage of unemployed economically active persona

aged fiftoon years and ever in each ward. The
indioee are given by Mabry in terms of T-eeores*
which are eeeres converting distributions to a
common standard with a mean average ef 50 and a

standard deviation ef 10.

*2, The social class index la in fact Mabry*s socio¬

economic status index (sec note *1).
*3. ft is* however* true of many British towns that ths
south-western areas are generally more statusful than
othsr areas. This is ascribed to the prevailing
 335

south—westerly winds, which nean that south**

v«st«ra suburbs suffer lssa from sir pollution
than othsr areas.

Rotes to Chapter Thres

1* Cf. II* Orton A E. Dleth, Survey of English Dlaleots.

(Leedst Arnold. 1962**).
2. Sao 21. Kurath A R.I. McDavid, on. clt.
3. C.A. Moaer, Survey Methods in Social Investigation.
(Londoni Heinemann. 1958)# P» 51*
*• Ibid., p. 50.
5. E, Sivertsen, Cockney Phonoloar. (Osloi Oslo lbl«
varsity Press, i960).
6. Sea also the criticisms of J.T. Wright. "Urban
Dialects t A Consideration of liethod*, Zcltsohrlft
iflg ,22* (1986), pp. 232-247.
7. D. Da Camp, "The Pronunciation of English in San
Francises*. Part I Orbla 7. (1958)* PP* 373-391#
Part II Ortols 8. (1959)# PP. 54-77*
8. "The statistical reliability of the study would

naturally bo far greater if several hundred San

Franolsoans. selected by random sampling, eould all
have been interviewed. The field interview method

admittedly eaoriflooe reliability of sample in

order to gain more reliable data from each in¬
formant and to obtain kinds of data unavailable by

any other moans.* It is certainly true that De

Carap*e study could have been statistically much more
reliable. The point is. however, that if a sample
is statistically reliable, nothing like several
hundred informants is needed. It is, moreover# not
true to say that this kind of data is "unavailable
by othsr moans". D# Camp seems to have been thinking
that random sampling neoessarily implies large num¬

bers of Informants, and that large numbers necess¬

arily imply mall questionnaires•
 336

9* W* fl«rtok, Phonemtlaohe Analyse ties Dialekte

yon Ctet o3head-upon~Tyno/Co. Durban. (Haatourgf
Cram* do Grayter and Co*, 1966).
10* Ibid., p. 49.
11* Sm, for example* A,J, Dronateln, "1*1*1 take
another look at How Tork City Speech", American
Speooh 37. (1962)* pp. 13-26| T*A. Frank, The
Speech of New York City, unpublished dissertation*
(Ana Arbori University of Michigan* 1943)t
A.F* Hubbell* The Pronunciation of English In Wee
Tork Cityi Consonants and Vowela, (New fork#
Coloebia University* 1950)» 0*3* Putnam A K.M*
0*Hern, "The Status Significance of an isolated
urban Dialect", lAwmoao 31. (1933)* Supplement,
Dissertation no. 331 A* Vanvik, A HkbioUo*
Phonemic Analysis of the Dialect of Trondholg**

(oaloi Oslo university Press* 1966)*

12. V. Labor* op. oft. (1966a)* Sao also W« Labor*
"The Social Stratification of /r/ in New Tork City

department stores** Paper given at annual eonferonoo

of the Linguistic Circle of New Tork* March 19631
*
Phonological Correlates of Social Stratification* *

In J.J. (kttspera A D.Jf* Hyiaes (eda.)* The Ethnography

of Coraaunfeation* (Men&sha* American Anthropologist
Special Publication, 1964)* » American Anthropologist
66* (1964)) "The Effect of Social Mobility on
Linguistic Boharlour* in 8« Lieberson (ed»)*
Explorations in Sooiolinsulatics* (The Hague i

Mouton, 1966e)* » Sociological Inquiry 36* (1966)1

and "The Reflection of Social Processes in Linguistic
Structures* in J#A. Fishmaa (ed*)* Readings In the
Sociology of Language. (The Hague* lout on, 1963).
13. k,w. Shuy, w*A. Wolfram A W.K. Riley* Linguistic
Correlates of Social Stratification in Detroit

Speech* (M. Office of Education* Final Report*

Cooperative Research Project 6*1347* 196?)I and
Field Techniques in Urban Language Study. (Washing-,
ton D.C.* Center for Applied Linguistics* 1963)*
 \v \r .*r

c
t- 2 B €
O^ • ►2
© ac O tif >it I I

n
tf

II 1 & e a »* £ § H tr< o u
s m
m *t C 3

! i - f M
a T5 § *•"
&itf ; • 9 4#• lf I s 5© M 1 s »♦< &
,

© t© c »' *»
• *4 ^ i 1* >♦ k o 2l § *5

.ljJ
•

* 8! !° Ht ®^ « 1 >4 3 »# © 4* 4 £ ^ i .*

3I So
3 3 i* 9 »4 ©s I 9 K § BS I »4

I#
© a© ;t
r Ho
©Uit © 4* a © © 34 S
!?«
31• ll
?sI til
„

W 9U *
© M 9-
• 3 *4 9 O 4* H 9 o *4

J5 3V *4 . „ 43
»< » i* H \C OV *f tf"* •

»•« ■«© ©q O 4* O * Q H« S3 >»

H OC •
.tf tf w

•!« a

X0
isi H
© »* S4 ©V

©c
%B M«
©
tif •

L*» »4 ;■
a© © #f• ejtH

M 4r N *

* *r ©\ tf
© iff © S f % *t©I

13 •"
N s© •

»"S 1< 34

<
©©

N
|s <»
*31 91
»
P

*4
99

9©
s

0• 9 9 ©b 3 i
I

.T3 b »>
is

©q
a«

© WJ;I 1 1! i! »+ 3 H 3V 5 O

« 0 _*5 9 5 « »

:H

ilsg'"
MC O ^ | 0
_ tr® © H ©o »4 S9 " 9 &I

0 3 3 « 3 39 c >© 9 !

!11i
;

«
 533

of non*»manual worker# in a sample* for the

purposes of comparison with manual workers* see
F.M. Martin* "Some Subjective Aspects of Social
Stratification** in 0« Olaoo («4«)i Social Mobility
in Britain. (Londoni Routledge and Kegeut Paul* 195*0*
P* 52*
21. Instruction* for the use of random numbers arc

usually given with the tables*

22* The letter nowhere contained any mention of tho
sender*a address* This was to avoid the risk of

having the letter returned before a visit oould be

made*
23* Some appointments were not kept by the informant*
This meant that a further visit* and sometimes
further persuasion* were necessary*
24* if* Labov* on. clt*. (1966a), p. <538.
25* Cf* L. Levins & E.J. Crockett* on. clt*. p* 78#
26* Cue drawback which arises from taking a sample of

schoolchildren is that ws do not have in the

sample any people aged wider twenty-one who are

working rather than still at school* This may have
the effect of biasing results for this age«»group

away from standardising tendencies which may emerge

ae a result of the job situation (of* Chapter Six*)
27. k* Labov* op* cit*. (1966a), p* 638*

Notes to Chapter Fogy

1* See K*B* Mayer* Class and Society* (Mew fork*

Random House* 1955)* P*
2* T*B. Bottomore* Classes in Modern Society. (Londoni
Allen and thnrin* 1965)* p* 15*
3* U.F* Reading* A glossary of Sociological Terms*

(Athenst Contos, 1963), p. 27*

*• P« 27#
 339

3# See B. Barber, Social Stratification* (Rev York*

Harcourt, Brace and World, 1937)* Chapter 1,
6. See IC.B. Mayer, on. eft., Chapters 3 and 5,
7* Ibid.§ p. ^3.
3, "The correlation of social status and linguistic
performance first requires a careful delineation
of each** R.if* Shay et al.. op, clt*. p. 11,
9. See, for example, D.v, Class A J.R. Hall, "Social
Mobility in Croat Britain* A Study of Inter*
Generation Changes In Status", in f»V» Class (ed.),
on* cit,, pp, 177-259.
10. See J.R. Hall ft D.V. Glass* "Education and Social
Mobility*, ibid,, pp. 291*307* Sec also 8* Lawton,
on. clt.. Chapter 1.
11. See J. Goldthorpe ft D. Lockwood, "Affluence and the
British Class Structure*, Sociological Review 11.
(1963), PP* 133-163.
12. Cf. the works of Bernstein cited below (note 25),
13* See V. labor, on, cit.. Chapter VIII, for examples
of differing linguistic norrae in different social
classes.
Ibid*

13. R.w. Shuy et af., op. clt.

16. L. Loviav ft H. J. Crockett, on. cit. See also
G.R. Putnam ft B.M, OMiern, on. clt.
17* See, for example, w.L. Warner, M. Meeker ft K. Bells,
Social Class in America* A Manual for Procedure for

the Measurement of Social Status. (Chicago* Selene#

Research Associates, 19**9) I G. Lundberg, "The
Measurement of Socioeconomic Status", American
Sociological PeTlew ft. (19*10), pp. 29*%0|
R.ft. Kornhauser, "The Warner Approach to Social
Stratification", in R« Bondix & 9.M. Upset (eds.),
Class. Status and Power* A Reader ipi Social Strat¬
ification. (Glencoe* Fro® Proas, 1953)1 C. lloch*
baum, J.a, Darley, B.l>. Monachesl ft C. Bird,
"Socioeconomic Variables in a Large City", American
Journal of Soclolorrr 61. (1953), PP. 31*33.
 540

18# II# Knrath & ft# I* McDavidt emcit*

19# K. Kurath et sl.« Handbook of the Linguistic Oeo-
pmnfiv of Hew England. (New Yorkt American Council
of learned Societies# 1954)#
20* G#R* Pickford, "American Linguistic Geography * A
Sociological Appraisal*# Word 12, (1956) pp# 211-233*
21# D. So Camp# on* clt*
22, H, Orton & E. Dieth# on* clt#
23# E# Siverteen, on* clt*
24* J*L* Fischer# "Social Influences on the Choice of
a Linguistic Variant*# Ford 14* (1953)# pp, 47-56#
also published in P. Ifymes (cd*)# Language in
Culture and Society, (Nov Torkt Harper and Row# 1964)*
25* See, for example# B. Bernstein# "Social class#
Linguistic Codes and Grammatical Elements*# Language
and Speech 5. (1962), pp.221 -240t "Elaborated and
Restricted Codest An Outline*# in S. Lieberson (ed,)#
on* cit.f "Elaborated and Restricted Codest their
social origins and some consequences"# in J*J, Gump-
era A D# Hyraea# on* clt.
26, B* Bernstein# "Language and Social Class"# British
journal of Sociology 11, (1960)# pp. 271-276.
27* Sec B* Barber, on* cit*# Chapter 8.
28. O.D, Duncan, "Properties and Characteristica of the
Socioeconomic Index*# in A,J* Reiss# Occupations
and Social Status. (Gloncoet Free Press# 1961)# p* 139*
29* D« De Camp# op*cit»
30. B. Barber# or?, cit.# p. 176* Cf. G,b.H. Cole#
Studies in Class Structure. (London* Routledge and
Kegan Paul# 1955)# P»5* "no single criterion can
servo adequately as a means of assigning ©very
person in a society to his or her class"*
31* See# for example# J,A. Kohl A J.a, Davis# "A Com¬
parison of Indexes of Soclo-Economio Status"#
American Sociological Review 20. (1955)# PP* 317-3251
and F.H. Finch A A*J. Hoehn, "Measuring Soeio-
Economio or Cultural Status* A Comparison of Methods",
Journal of Social Psychology 33* (1951)• PP* 51-67*
 541

32. B» Barber* on. ctt»* pp. 184-185,

33. .Ibid,, p. 171.
34* Sao* for example* C.A, Mosor A JT,??. Hall* "The
Social Grading of Occupations® in D,V, Glass* on.

clt*| P,K. tiatt* "Occupation and Social Stratify

ication®, American ..^otimai, of Sociology 5^* (1350)*
PP» 533*3^31 O.b. Duncan* ®A Socio-Econonie Index
for all Occupations"* in A.jr, fielas, op, clt. t and
A.F, Dories* "Prestige ©f Occupations** Dritish
journal. of Sociology 3» (1952)* pp. 134-147#
35« Tb® advantages Of weighting bar© boon pointed out
by g.F. Lenskl* "Status Crystallisation! A Hon-
y#rtlenl Dimension of Social Status** American.
Sociological Per low 19.* (1954), PP* 405-413.
36, So© R*W» Mack, "Housing as an Index of Social Class®,
Social Forces 29* (1951)* pp. 391-400f and O.B*
Duncan A B# Duncan* "Residential Distribution and
Occupational Stratification®* American Journal of
Sociology €0, (1955)# PP. 493-503* (Hot© that Mack
is more concerned with tlx© condition and ©tat© of

repair of the dwelling In question than with owner*,

ship or type#) Cf. B* Barber, on, clt,, p* 144,
"The type of dwelling place and its location within
the local community are likely to be symbols of
social class position in all societies**

37, So© w, Labor* op, clt** (19660),

33, Seo P.m. Martin* op. clt.
39. General Register Office for England and Wales,
Classification of occupations* (London! !I«f,3,0«*
1966).
40, Sample Census of England and Wale®* 1966* (op# clt*).
41* D.v. Glass & J, p. flail* *A Description of a Sample

Inquiry into Social Mobility in Great Britain** in

l»,v. Glass (ed«)* op, clt.* p. 83,
42. Cf. ibid*, p* 83.
43. See* for example* n.w, Shuy et al., op, clt,, p. 13.
44. A comparison of Table 4 with the Information given
 5hZ

In Chapter Two concerning the social and economic

characteristics of the city trill show that this

subjective ranking of wards and parte of wards is
perfectly consonant with the facts and figures
illustrated there.

b5* "««< modern city neighbourhoods and suburbs them'*

selves, both as wholes and in their subdivisions*
are finely graded as symbols of superior and in*
ferior soolal class positions,* B» Barber, on.
P« 1*>5«
46. No local Authority houses, for example, were built
before 191&*

Notes to Chapter Five

1, M.A.K. Ifalliday et al.. ot>. clt.. Chapter b* See

also, for example, J, Spencer, M. Gregory & N. Enk-»
vlat. Linguistics and Style. (London* Oxford Unl-
versity Press, 1964) l and H, JToos, The Five Clocks,
(New Terki liarcourt, Brace end World, 1967)*
2. W* Labov, op. cit.« (1966a),
3, K,A,K, Halllday et al., Op, cit.« p, ©7*
4. Ibid*, p. 90. (My emphasis*)
5, See# for example, C.F. Hockett, op. cit.. p. 556.
6. See M.A.K. Halllday et al., op, clt.. p. 22.
?• Ibid., p. 93.
®* Xbld., p. 83.
9. The distinction maintained here between dialect and
accent is that of Abercrorable. See ». Abercrorabie,
Problem and Principles in Language Study. (London 1
Longmans, 1936), Chapter IVf *RP and Local Accent*
in Studies In Phonetics and Linguistics. (London*
Oxford university Press, 1965)1 and Elements of
General Phonetics. (Edinburgh* Edinburgh University
Press, 1967), P. 19.
 5*3

10. M.A.X. Halliday et aj.« pp. elt.# p. 9*.

11. Sep B« Bernstein# ot>. cit.« (1966)| on, cit.. (196*)t
end "A S6eto*LiRpistle Approach to Social Learning",
la J. Gould (ed«)a Penguin Survey of the Social
Sciences 1965* (Ilarmoadsvortht Penguin Book*# 1965).
12. V,P, Robinson# "The Elaborated Codo in Working
Class Language" # LftnTnarc and Speech 8. 1965* pp»

2*3-252*
13. J. Regan, "Southern California Ghetto language and
Roles"# Paper presented at the Autumn Meeting of the
Linguistics Association of Great Britain# University
of Leeds# (1969)*
1*. E. Goffman# The Presentation of Self In Everyday
Life. (New Yorkt Boubleday Anchor* 1959)•
15* «• Labor# on. cit.. (1966a)# Chapter IV.
16. M.A.K. llalllday et al.. op. cit.. p. 91*
17* It seems probable# however# that the effect here le
greater than that achieved by a similar instruction
in the New York City survey. Reasons for this will
be discussed later# but appear to be connected with
British "working-class solidarity" and other

ideological and psychological factors § (see Chap¬

ter Eleven).
IS. Characteristics of formal# slow reading style that
we wish to avoid are, for example, the comparative
lack of reduced forms# and the lack of certain
Juncture features, such as the "intrusive jr"•
19* The items were presented to informants in cards
which contained eighteen items each. The items
were hand-written in large block capitals to moke
for easy reading.
20. C.L. Houck# on, cit.
21, Ibid.. p. 13. Houck also states that this type of
question "would force the informant to focus on
giving the appropriate response word rather than
focusing On some particular "correct" pronunciation
he thought the investigator wanted", (p. 13). It
is of course possible for the Investigator to
 544

Instimet the informant to use his normal pro¬

nunciation, and not soma kind of "class-room*

pronunciation*
23* It* Orton A E* Ciethf ftPfcit. i

33* *. Labov, op. cit.* (1966a), p. 98.

24. Ibid., p. 100.
23. c.L. nouek, op. cit.* p. 13*
26* "Haw foa ever boon in a situation where you

thought tkore was a isrioaa danger of your being

killed? That you thought to yourself "This is it*?*
¥, jLabov, op* cit*. (1966a), p. 595.
27. Ibid.* p* 109.
28. Ibid * * p. 110*

Noteo .to Chapter Six

1* The position concerning the paradigm of the copula

be is somewhat complex* The usual forms in Hch
are those, of Standard English* There is, however,
some evidence that the same regularising .tendency
which has affected all other verb forms was formerly
also operative in this paradigm* There are, partic¬
ularly in the rural areas around ffch, traces cf
what nay be an older system which had "be* through¬
out the paradigm, although at the present time
there are only spasmodic occurrences of forms of
this type* These are especially common in locutions
such est "Iters I bo*§ "There it be*# ■ The same
tendency* moreover, Is particularly in evidence in
Nch in negative forms of this paradigm* alongside
the standard forms "I'm not# you're not, he isn't*
etc*, and the non-standard forms "I aren't, you
aren't", there occurs, almost universally in working-
class speech, the form *int" ^Xnlj* This ee-occurs
 5*5

with ail persons, ami like its geographically

□ore widespread counterpart "alnt"» also functions
as the negation of •hare* t "1 int going| you int
allowed to| they int got it" •

2. See n. Orton A P.M. Tilling, Survey of Fn~UtU

Dialects, (ft), The Basic Material. Volume XtT. The

East Midland Counties and East Anrlia. (Leedst

Arnold, 1969)* Besponses to XXX.3.7* (p* 296)
show marker-less forms for all the Nfk and Suffolk
localities where informants responded with third*

person singular present tense verb forms, and for

one such locality In North-East Essex.
5. See fl.X, McDavid, "Dialectal British and American
Standard and Ron-Standard" in A.A. (ed«)»
Hill
Linguistics Today. (New Yorkt Basic Books, 1969),
p. S7.
*. "In East Anglia, in England, the omission of this
Ithird singular presentj marker is endemic•*
R.Xi McDavld, in discussion reported in «. Bright
(ed,), on. cit.. p. 112,
5, In his Norfolk field-notes for the University of
Leeds Survey of English Dialects, Francis states
that it is of grammatical interest (in the Dllekllng
locality) that the third-person singular marker is
completely lacking, and quotes "That have been" and
"She wear* as examples, lie also states that! "This
is standard in youngest speakers". X am very grate¬
ful to Mr. S.F. Sanderson of the Institute of Dia¬
lect and Folk-Life Studies for giving mo permission
to examine these as yet unpublished records. X am
also grateful to Mr. Philip Tilting, Mr. Stanley
Ellis, ami Professor W.N, Francis himself for their
help and advice.
6. "In grammar, the Chicago-born Negro who grows up in
an environment of poverty and limited cultural
opportunities — as moat Chicago Negroes grow up —

has a tendency to use forms that identify him easily

 5*6

and to his disadvantage, in writing: as weli as

in speech* Host of these are forma of common
verbs * absence of the third-#insular present

marker, as in he do. it make *♦." R,I* McDavid,

"Dialect Differences end Social Differences in an

Urban Society", in Bright (ed*)» om cit** p* 80*

See also W* Labov, on* cit*. (1966b)t "The third
person singular marker //-s// does not exist in
the particular grammar being used here"* (The
grammar referred to is that of Negro children in
Harlem, Nov York City*) Hnny white Southerners
apparently also have this speech characteristic,
but in the Northern cities it has come to ho

associated with Negroes*

7* under the heading "marker-loss" are included all

occurrences of the form have for has*

8* Only relatively few instances of relevant verb

forms were obtained from each informant * and taking
a mean of Individual percentages would have intro¬
duced a bias for this variable in favour of those

informants who provided only one or two instances

and whose percentage scores were therefore less re¬
liable* An overall group percentage score in this
ease gives a much more accurate reflection of group
behaviour*
9* It is perhaps interesting to note that where the
evidence of occupational status is over-ridden by

the other parameters of the social class index, so

that, for example, a non-manual worker is classified
as working-class, ths interviewer*# subjective
impressions concerning the informant's social class
in all eases confirm the classification produced
by the index* This fact is of course of no scient¬
ific value*

10* This classification of non—manual workers as working-

class stresses the Importunes and usefulness of a

multiple-item social class index* The linguistic
 547

behaviour of these informants* and their sub¬

jective attitudes* have definite working-class

characteristics•

11. To avoid confusion* Claea 1 will be referred to

as "MMC* (• "Kiddle middle-class") and Clasa XV
aa *lfc?C* (• "middle working-class*)« The term
"middle-class* ean then be used to apply to both

Close X and Close XX. and "wor?ting-class" to

Claseee XXX* XV and V.
12* Those figures* together with the social class
characteristics of tho schoolchildren illustrated
in Chapter Four, suggest that it is possible to
consider that the sample of schoolchildren forms a
natural extension of the adult sample* and that
its inclusion should Introduce no significant bias
into any results obtained*
13. Tho implications of the approach adopted here* and
the problems of "age-grading" and "apparent tine*
will be discussed at greater length in Chapter
Eleven.

14* tfe must* of course* assume the presence of Standard

speakers in the community before this time*
15* Anecdotal evidence suggests that* although this
stigmatised form is not often the object of overt
critical comment in Keh Itself* it is one of the
main objects of derision in* for example* the armed
forces outside KA.

16* These facts point to the relative ineffectiveness

of schools as normative influences» at least as
far as speech-forms are concerned * and etresses the
importance of face-to-face contacts * such as those
that occur in Job situations* in the dissemination
of linguistic features* Cf* "Dissemination through
private or group conversation easily outbalances
ether means of communication* Even today we are

very neolithic in that respect* X am sure*"

T. U&geratrand* "Quantitative Techniques for Anal¬
ysis of the Spread of Information and Technology"*
 543

in C#A* Anderson A W# J, Bowmn (eds# )# rdttcation

and Kconotnio Development« (Chioagsi Aldlne, 1965)»
p. 263.
1?» St Lftbovj IffP*n dX^#.# (1966&}i p. 325*
i®* ibid## p. 325.
19* IMd«, p. 323. ***t the evidence from the present
study indicates that the greatest degree of stylistic
fluctuation, &ad linguistic insecurity, 00euro in
tniddlo—aged speakers fro® the middle-ranking groups#*

Rote® to Chapter Seven

1# The distinction between alteration or distortion

and suppression is one that ha® escaped at least
one linguist# J.T. Wright has written in his dis¬
cussion of Labov*s work* *Ths variables should also
be salient in order to facilitate studies of the

subjective reactions of the respondents to tho form®»

bat this requirement conflicts with the further
criterion of Immunity from conscious distortion*#

J.T, Wright, op, cit, tabov, on the other hand#

quite clearly statest "Immunity from conscious
distortion is not required*» On# elt## (1936a)#
p. 31#
2# Post-veealie /r/ does, in fact# occur as a relic
forts in the speech of some older speakers#
3* This# of course# i® also a feature of tetany American
Jnglish accents, at lsast following alveolar con¬

sonants#

4# In the speech of some older, rural CAn speakers#

these items do not have iJj# These speakers also
have, for example, education as (. .eda'k^etjnj# as
opposed to Hch (, .edjs'k^WifnJ.
5# See W,3, Avis, "The Nsw England short £i a re-
oessivo Phoneme*# Language 37. (1961), pp* 544-558#
 5*9

Sm Also K* Kurath ft R«Z. Me&avid, op, clt,. p. 12*

"A regional f eatsire not found elsewhere in the
eastern States pother than Eastern Wow Englandj
is a checked vowel /#/ In each words as coat, road,
homo, contrasting with the free vowel /o/ of know,
rode, etc,"
6. In Neh this shortening Is much more comimn in some

Items than in others. Zn hows, aerodrome. it is

very widespreadi in don't, won't, it is quite

frequenti it is also quite often heard in road.
stone, broke.

7* It is not clear whether the shortening of room, roof.

boot, is, diachronically or synclironical 1 y, part of
the saie process as the shortening of boat, homo.
This will be discussed in Chapter Ten.

8« This merger is not a surprising one in view of tho

small functional load of the distinction* p.B. Fry

has calculated that /*»/ and /la/ are respectively
seventeenth and eighteenth in order of frequency of

occurrence of the twenty English vowel phonemes§ and

between them account for only 0*55 por cent of the
total number of phonemes. See "The Frequency of
Occurrence of Speech Sounds in Southern English",
Archives Soorlandalaes de Phonetlnue Experlmentale

20, (19*7)» PP* 103-106, (see Chapter Eleven). This

merger also appears to have token place in tfev Tork
City, and in South Carolina, (see ft. Kurath a
P.I. h©David, op.clt.).
9, A secondary articulation of palatalisation fre¬

quently accompanies this vowel. This point will be

discussed in Chapter Ten.

10, ??o information appears to be available on the speech

of other urban centres in BA.

11, information from these onrces has in seme cases

been supplemented or confirmed by evidence from

personal observation.
 550

12, II* Kokerlt>« The Phonology of the nffolfc ufnlact,

(Uppsalai Appelberg, 1952), (ttUppsala Universitets

Arsskrlft 1932*4), Kokeritz also includes notes
© is the dialect of Essex.

13* Lowmn'a records are a part of those he made in a

surrey of a large area of toe .South of England, The

records are unpublished, and are in the custody of
the University of Chicago. X am extremely grateful
to Professor U.I. HcUavid who was kind enough to
take a great deal of trouble to permit mo to see
these records,
14. II. Orton <% P.M. filling, op* clt*
*

15* Kofcerltz's information shows that there are some

traces of this distinction in Essex, but that t»Uj
is usual In both sets of items.

16, lawman** records are written in the phonetic

transcription system used by the Linguistic Atlas
of the Eastern United states. Details of this can
be found in »1, > urath, op* cit., pp. 122*146, and
IS. Karath A R.I. BeDavld, op* cit,, p. 1. Mete that
Lowraan uses the symbol ( to indicate unrounding.
The symbols and [Vj indicate vowels more open
than I.Uj and (.«j respectively.
17, One qualification must be that, as flap 1 shows, the
SKJD Suffolk localities are further south than the

northernmost localities studied by Kokeritx and

Lowman.

18, This, and the subsequent quotations from francis,

are taken from unpublished material found under

*phonetic notes® in the fiold*work note*books,

19* n. Kokerits, on* clt.. p. 46,
20. A clear example of this is toadstools, which Lovmn
writes ito-dst
21. It can perhaps be assumed that the conflicting /ui/»
/u*/ and /U/ forms in Kch represent dialect mixture
resulting from the absorption of forms froa all three
areas•
 351

22* The notations hero* (like Kokeritz's i,Uj| and tut J),
would appear to Indicate a pre-merger state be¬
tween ll.I*, /ui/ and EAn /ui/*
23. This evidence would appear to suggest that the
two sets of items (which are close but distinct in
Kokeritzj have, in Lowman and Francis, overlapping
areas of articulation* but not a complete merger*
Francis* transcriptions* which do show sons very
minimal contrasts* e.g., l§d J / ttfujl l«*J / l»*Jt
may tend to obscure this state of affairs as far as
his records are concerned* The complete merger can
be assumed to be a ffeh feature*

2k, H* R&keritss, on* cit.* p. 159*

23- Kokerits also reports that older informants recall
sure" and stvrar being pronounced with initial taj*
26. Note that Martlesham is very close to the urban

centre of Xpswleh. (See Hap 1).

27* It is also interesting to note that three of the
Esses localities have L&J* and the two Cambridge
localities
23, 31. K&koritz * on. cit*« p. 71* Kokerits also cites
some evidence of the merger in Esses*

29, Lowman*s records have been taken by fturath 4 McDavid

as evidence of the merger of /la/ and /ea/ as /ea/
in tho "eastern counties" of England* H» Kureth 4
n.I. McDavid* op. cit** pp. 118-119*
30. This suggests that the merger was once usual through¬
out liA* but that the distinction is being intro¬
duced from the Home Counties in the South*

31* Personal communication*

32. It* Koteeritz, op. clt*. p. 55*
55* Kokerits also mentions another An feature* which
la also found in Mchi the -oaophthongisatlon of this
vowel in final position* e.g. say [sit* ■— awl J.
3^. Rokerltz writest "My observations ... have convinced
me that today 01 is rapidly disappearing frera the
duff, dialect* for the children hardly over used
et *.. but replaced it by a broad diphthong (si — al)"»
II* ftftkerltz* op. cit.« p. 19*

t
 552

35* P. 13.
36. Phonological variables are symbolised by enclosure
in parentheses*

37. dee also fi. J. Oregg# "Notes on the Phonology of a

County Antrim Scotch-Irish dialect", Qrbia 7. (1955)#
PP» 392-406.
33. This method of computing indices una initially
developed by Labov in his New York study. The terra
"norm" is perhaps not accurate here* The mode. It
Could be argued, might perhaps be a better indic¬
ation of the pronunciation norm than the mean.
One fault of Labor's system of calculating indices
is that only the rac-aii score is given, and that the
range of variable-types employed by a given speaker
is concealed. For example, a speaker who has 50
per cent (t)-l and 50 per cent (tj—3 will have the
same index as a speaker who has 100 per cent (t)-2«
This fault can be rectified by the inclusion of

cone measure of consistency of performance, such as

the standard deviation.

39* See. for example, JT.L. Fischer, on. clt.

40. Intrusive /r/ is rare after a preceding /r/ in the
same syllable! e.g. extra eggs (.'{testae *«gaj.
41. In Labor's work, this problem of phonetic tran¬
scription is not adequately dealt with, he says,
for example, "The height of the vowel which occurs
in words t,of the class bad! forms a continuous
scale. This may be codified into several discrete
units with the help of other word classes that are
relatively fixed". W, Labov, op, clt.. (1966a),
p. 52. There is no discussion of exactly how this
continuous scale is to be divided precisely into
discrete units, or of any of the problems that
arise from this.

42. (a)-5 indicates taut there is obviously a further

possibility of a merger in the front vowels Involving,
e.g., bad and bared* There is also a further pro-

f
 533

nunciation typo t&j. This centralised vowel la

most common la items of the typo natter. Alan, and
would therefore appear to be Involved In the same

type of development aa (e), (eee Chapter Eleven).

It is alee possible that transcriptions like
Lowman*s hawser {.hama— h>«tadj Indicate the same

phenomenon.
43» In many types of R.P. this vowel is in fact long*
thened to (a)»2, especially before /b, d,
dj. a. g,
a/. In Itch this lengthening occurs in all environ*
stents. Cf• A.C. Gists on, Ait Introduction te the
Pronunciation of English. (London* Arnold. 1962).
p. 100.
44, Piphthonglsation is particularly common before
/e* k.y /. This fact is not taken into consider¬
ation in the calculation of index scores, since any
possible bias should bo minimised by the largo
amount of material.

43« Francis has observed a distinction in the distrib¬

ution of these types, he has*
ME o > {.St j / f e.g. off
> t«« J / voiceless C e.g. box
>l»J elsewhere e.g. collar

Notes to Chapter Elirht

1. e. (1966a).
Labov. op. cit.t
2. Sec ibid., pp. 39<K)99| J.L. Fischer, op. clt.i
A.C. aimson. op. cit.. p. 194| S* Sivertson, op.
elt.. p. 129f G.L. Erook. op. cit.
3. J.L. Fischer, op# cit.
4. Labor*o terms *change from below" and "change from
above" refer respectively to changes from below and
above the level of conscious awareness. Usually,
 55*

however* changes froa above involve the downward

dissemination of prestige features, l.«« they are
socially change® "from above" ae well* Change®
from below* moreover* very often start among lower
class group®* (See Chapter ! leven.)
J. See a.C. Gimson, on. cit.« p. 131,
6. See <1* Pellewe* "Establishing Speech Varieties of
Conurbations"* Journal of Linguistics* (forthcoming)«
7* fl.M. Blalock* Social Statistics* (New Yorki McGraw-
Hill, I960)* p. 66.
8. w. Avis* on. clt.
9* See A.C. Cimsen* oi>» cit.» p* 128.
10. w* Labov* on. clt., (1966a)* pp. 520-52*. See also
T.H, Hetmore* The Lou—Central and Low—Back Vowels
in the fastern United States. American Dialect

Society Publication 32* (Alabamat University of

Alabama Press* 1957)«
11. for different types of argument in favour of a

theory of phonology which allows for "complete over¬

lapping"* see P.M. Postal* Aspects of Phonological

Theory. (New York* Harper and How* 1968)* Chapter 9-
12. C.A* Ferguson* "Diglossia"* Word 15. (1959)* pp.

325-3*0.
13. H. Weinreich, ¥. Labov & M.X. Hergog* "Empirical
Foundations for a Theory of Language Change"* in
w.P, Lehtaonn & T. Malklel (eds.)* Directions for
Historical Linguistics. (Austin and London1 Uni¬
versity of Texas Press, 1968)* p. 13*.
1*. For a discussion of the unmarked nature of /t/ as

opposed to marked /p/ and /k/ see K* Kohler*

"Modem English Phonology", Lingua 19. (196?)*
PP. 1*5-176.
15« H. Kokerita, or*, clt.. p. 10.
16. k. Forby, The Vocabulary of East Angtla. (London,
1830).
17. (er) and (ir) provide a true ox«ra?>le of "complete
 555

overlapping"* in contradistinction to (o) and

(on) where there is a tendency for the realisation
of one unit to predominate in one particular section
of the phonetic area* and vice versa*
18. "Another form of advanced RP uses a long pure vowel
(.•iji often somewhat centralised* especially in a
non-final syllable* «,g, careful* scarcely ^*keifij*
i'skeisllj*" A.G* Gimaon, on. ctt., p. 138*

Rotes to Chanter Fine

1* Xt is a matter of some interest* for example* that

speakers who* to take a rather trivial example* do
not normally distinguish between the vowels of name
and nail are perfectly well able to do so without
error for humorous or other purposes*

2, The term dinars ten trill be used ambiguously to

describe both tho linguistic competence of the
native speaker as a member of the speech community
and to tho model of this competence which will be
developed in Chapter Ten*
3* w* Labor, op* oit*« (19&>a) •
4. 9* Koinreich* "la a Structural Dialectology Post*
lblet"* word 10* (1934), pp. 388-400.
5. D, Do Camp, op. cit*
6m Vm Weinreich, op. cit*
7. 0*1#* Trager A H.JU. Smith* An Outline of English
Structure* (a Studies in Linguistics Occasional
Papers 31* (Washington* 0*0.» American Council of

Learned Societies* 1957)*

8. G»rt, Cochrane* "The Australian English Vowels as a
Diasystem"* Word 15. (1959)* pp. 69-88.
9* D« Jones* The Phonemet Xts Mature and Ose* (Cambridget
Goffer, 1950)* p* 195*
 556

10. G.B. Cochrane, on. ctt.

11. l* ,a, Moult on, "The Short Vowel byatem of Northern
Switzerland", Iford 16. (I960), pp. 155*182.
12. i;« Pulgram, "Structural Comparison, Diasysteas and
Dialectology*, Ltn-nlotlco 4, (1964), pp. 66*82•
*3. Bid,
14. See for example tf. Viereck, or?, cit.
15• O.R-. Cochrane, on. elt.
16. £. sAvertsen, op. clt.
17. Ibid.« pp, 115-117.
18. P. Do Coop, op. cit.
19. C.C. Fries & K.L. Pike, "Co-existent Phoneraio
Systems", Lanrninge 25. (19^9), PP. 29*50.
20. Cf. C.F. Hockott, op. elt., pp. 332*333.
21. S.J. Keyser, Review of Kurath & McBavid (on, cit.).
Language 39, (1963)1 PP. 303-316.
22. n. Kurath A n.I. HcDavid, on. cit.
23. S* Saporta, "Ordered Rules, Dialect Difference®, and
Historical Processes", Language 41. (1965), pp. 218*
224.
26. IS. Vaailiu, "Towards a Generative Phonology of Daco-
Rumanian Dialects'', Journal of linguistics 2. (1966),
Pp. 79-98.
25. E* Vaailiu, "Transornafcional versus Piuniquo Phon¬
emic Typology" in J. Datum (ed«) Phono1ogle c?er
Gegcnwart. (Vienna 1 B&hlaua, 1967)* PP. 254-261.
26. R.D. King, Historical Mnnalatlcs and Generative
Grammar. (Englewood Cliffst Prentice-Hall, 1969).
27. 8* Lamb, "Prolegomena to a Theory of Phonology",
Language 42. (1966), pp. 536-573.
28. R.D. King, on. clt.. p. 38.
29. Ibid., p. 39.
30. uf «A, 0*Nell, "The Dialects of Modern Faroeaei A
Preliminary Report", orfals 12. (1963), PP. 393-397#
31. D»'". Klltaa, "^elatedness between Gramsictlcal Systems",
language kOm (1964), pp. 1-20.
32. a.r. Thomas, "Generative Phonology in Dialectology",
Transactions of the Ihllological Society, (1967),
pp. 179-203.
 557

ffotes to Chapter Ten,

1. E.C. Fudge# "The Mature of Phonological Primes"#

Journal of Lin mist lea 3# (1767 )# PP. 1-36.
2. Ibid.

3# R.T Harms# Introduction to Phonological Theory*

(Englevood Cliffs* Prentiee-fiall, 1963}# Chapter 4.
4. E.C. Fudge# on* J5lt.# P. 3»
5* Ibid## p# 4.
6# O. Jones# "Concrete and Abstract Sounds"# Proceed-
in/?s of the Third International Congress of Phonetic

Sciences# (Ghent* laboratory of Phonetics of the

University# 1937}# pp. 1*7»
7. B.C. Fudge# op. clt.# p# 3.
3. If#- Chomsky A M. Halle# The Sound Pattern of English.
(Hour York* Harper & Roar, 1963}#
9. M.$*« J. Tay, A- Phonological study of Hoklciea# (uni¬
versity of Edinburgh* Unpublished Doctoral Thesis#
i968)| and "flokklen Phonological Structure"# Journal
of Linguistics 6# (1970), pp. 81-38.
10. B.C. Fudge# "Syllables*# Journal of Linguistics 5.
—
mwmweww »iiin*m«»i imi.wiiiminini.nnnwflma* #>—n# 1 HIMWIIBMn—mm* w

(1969a), pp. 253-286.

11. Ibid,

12* R.L, Cheng# "Mandarin Phonological Structure*#

Journal of Linguistics (1966), pp. 133-133.
13* For criticisms and comment3 on this approach# see
J. Anderson# "Syllable or Non-syllabio Phonology*#
Journal of Linguistics 5. (1969)# pp. 136-142|
6. Broun# "Syllables and Redundancy Pules in Gener¬
ative Phonology*# Journal of Linguistics 6# (1970)#
PP. 1-171 K.J. Kohler, "la the Syllabi© a Phono¬
logical Universal?"# Journal of Linguistics 2#
(1966J# pp. 207-208.
14. B.C. Fudge, op. cit,# (1969a).
15. Ibid., p, 269.
16. Ibid., pp. 269-270.
17. See# for example# ■!. Chomsky ft M. Halle# op. clt.#
Chapter 4.
 353

13* B.C. Fudge, op, clt.. (19^7)» p< 3-

19* See, for example, n.Q. King, on. clt.# Chapter 4*
20. A.R. Thomas, op. clt.
21. if* Weinreich et al** op,clt..
22. Xbld. * p* 170*
23* V* Labov, The Aims of boclolinratlatle Research*
(Columbia Unlversitxi Unpublished Working Paper, 1964)*
24* s.A. Schane, "Natural Rules in Phonology" in Pro*,
oeedlnga to an Historical Lirvmistics Conference*

UCLA. 1969* (Forthcoming).

25. N* Chomsky & M* Halle* op* elt*
26. See N* Chomsky* Current Issues In Lln/pilstlc Theory.
(The Hague 1 liouton* 1964)* p« 72*
27* B.C. Fudge* "notation Rules and Ordering in Phono*
logy"* Journal of Linguistics 5. (1969b)* p» 37*
23. E.C. Fudge* on. clt*. (1967)*
29. E.C. Fudge* on. clt.* (1966b)* pp. 23*38*
30* P. Ladefoged* Linguistic Phonetics. (UCLAi Working
Papere in Phonetics 6* 1967a)*
31* E.C* Fudge* op. oit.* (1969b).
32. E.C, Fudge* op. Pit.* (1967).
33# P.M* Postal, op* elt** p. 273.
34. Note that the introduction of the performance phon*
etio level does not mean that wo are attempting to
construct a performance model as such* of the type
discussed in V. Fromkln* "Speculations on Perform*
ance Models"* Journal of Linguistics 4. (1963)*
pp. 47*63# A performance model proper would be eon*
eerned with factors such as motor commands* muscle
activity and neural Impulses in a way that is not
possible in a work of this nature* We are concerned
with features of actual speech* but only in so far
as they characterise Hefr speech and in that sense
form part of the competence of the native Nch speaker*
as a member of the Nch speech community* Our phon*
etic realisation rules* for example* are competence
rules it they are rules relating to the whole speech
community* The exact way in which these rules are
 559

actually used in performance, and the physic*

logical form they take* is a topie for <tl9ous$ion
in a perfawaooe model*

35» B.C. Fudge* op* clt*. (1966b).

36. S.M. Lamb* "On Alternation, Transformation* Heal*
issation and Stratifieatlon" * Monograph Series on
Languages and Linguistica 17* (Washington D*C. t
Georgetown University Press« 1964), pp. 103*122*
37* In the speech of some older informants« /«»/
results from //a// ♦ //f//* //«//. and //o//* giving,
for example* half /hsetf/* (See rule 36*)
33* The necessity for this rule was pointed out to me
by 0*0* Knowles*
39. E,C, Fudge, op, clt*. (1969a), pp. 275-276.
40. P. Klparsky, on. cit.* p* 133*
41. A. Martinet* Fconomi® dea Changemonts Phonettones*
(Bemi Franoke» 1955) I and "Function, Structure and
Sound Change"* Word 8. (1952), pp. 1*32.
42. V.G. Moulton, "Dialect Geography and the Concept
of Phonological Space** Word 18. (1962)* pp. 23*33.
43* A* Martinet* op. clt.* (1955)* P» 54.
44. Ibid., p. 54.
45. D.r*. Fry, op. clt.
46. A. Martinet* op. cit** (1955)» P* 93.
47. B.II, Robins * A Short History of Linguistics* (London1
Longmans* 1967)* P. 224.
43. V, Labov* op* clt.. (1964* unpublished),
49. P* Ladefoged, Three Areas of Experimental Phonetics.
(London* Oxford University Press, 1967b), p* 163*
50. W. Chomsky A M. Halle* op. clt.
51. P. Ladefoged, op. cit.* (1967a).
52* S, 1;.G. Oilman, "Coarticulation In VCV Utterancest
Speetographie Measurements"* Journal of the Aeonst*
leal Society of America 39. (1966)* pp. 151*163.
53* See n. Honikraan, "Articulatory Settings", in
D* (eda *)* In Honour of Paniel
Abercronbie et al.*

Jones, (London* Longmans, 1964).

 360

54. J.P.M. Lairor, "Voice Quality and Xndexical la*

formation" , firttish Jtmraal of Pisorders of
CortrmniCftt ion 3* (1963), pp. 43*54, (.Note that
Laver9* model of voice quality is baaed on D,

Abererombie, on. cit.. (1967)# Chapter 6. I am

extremely grateful to John Laver for his help
with this section.J
53* Ibid.. p. 44.
56. See J.C. Catford, "Phonation Types 1 the Classifies
ation of some Laryngeal Components of Speech Pro*
auction*, in ft. Abercrombie et al.» op. cit.
57. J.P.H. Laver, op, cit.. p. 46.
5s* Ibid., p. 46.
37* Ibid.. p. 47.
69. On this point, see C.*tf . Kim, "Two Phonological
Notest A* and Bk", (unpublished1 University of
Illinois, 1770).

Notco to Chapter Eleven,

1. V. Labor, or>. cit.. (1966a), p. 319»

2. Ibid., p. 325. Compare the discussion of the age
differentiation pattern of the verb marker in

Chapter Six.
3. P. Forby, op. cit.. p. 90.
4. O.Y. Shevelov, *0n Predicaments, Predictability
and Futurism In Phonology*, in J. Hamm (ed.),

5. A.C. Oitason, op. cit.. p. 114.

6. A.C. Glstson, "Phonetic Change and the P.P. Vowel
System*, in t>. Abercrooble et al.. on. cit.. p. 133.
7« P.O. King, 00. cit.. pp. 90*91*
8. ft. Forby, on. cit., p. 96.
9# W. tteinreich, Lanrninr-es In Contact. (New Yorki
Linguistic circle of New York, 1933)* P* 97*
10. A.C. Glcison, on. cit.. (1962), p. 197-
 561

11. Ibid*# p. 100.

12. £. Sirertsoti, op. cit.* p. 62*
13. A.C. Gtaaon, op. clt., (1962)* p. 125*
14. See It* Orton ft P.m. Tilling, op. elt. * XI.6.2
"TO BIND", pp. 207-203.
15. J. Anderson ft B. Lass, (forthcoming)t a book on
Old English and historleal linguistics. References
to Anderson ft Lass are based on a seminar given by

Lass In the Department of Linguistics* university

of Edinburgh, May 1970* entitled "Sound Change".
They nay therefore require revision upon public¬
ation of the book.

Roto that the exceptions to the metarule mean

that this most probably needs to be presented la a

somewhat more sophisticated form.

On metarules, eee alsot C.-V. Slim, op, clt.
16. C.F. Hockett, on. clt., p. 440.(Soo also C.F.
Rockett, "Sound Chang©"* Language ,41, (1965).
pp. 185-205.)
17. Ibid., p. 440.
18. Ibid., p. 443.
19. Ibid., p. 443.
20. L. Bloomfield* Language. (London* Allen ami Unwln,
1935)* P. 369.
21. 17. Velnreich* V. Labov ft H.I. Hersog, on. clt..
p. 187.
22. P.M. Pcstal* cp. clt.. p. 237.
23. Ibid., pp. 297-298.
24. R.D. King* op. clt.. p. 190,
25. See Ibid;, p. 79.
26. for example* cany of the major changes that led to
the distinctive characteristics of Middle English
(as compared to Old English) occurred during the
period after the Roman Conquest* when French was
the official language*
27. R.D. Sing* op. clt., p. 199.
28. J. Anderson ft R. Lass, op. clt.
 56a

29. P.M. Postal, Q!>, cit.* p. 284.

30. Ibid,* p. 283.
31. See J, Lyons, op. clt., p. 90.
32* J.i.M, Tries has reported hearing the utterance,
made by one London boy to another* (.gl&XzeosJ,
which, out of context. would be totally ineoapre*
hensibis as *giv# hits something else** Koto the
parallels la the relationship between this utter*
once and Ite more formal equivalent and the re*
latleashlp between, say, Early Modern French and
Latin.

33. »*D. King, op. clt.. p« 189.

3**. Ibid.* p* 64*
33. According to King, "spontaneous* changes do inks
place In the child's grammar* These, however, are
usually of the ease of articulation type* See Ibid.*
p« 79i "child Innovations probably are often (if
not always) asslsllatory In nature**
36. Ibid* * p* 66.
37. See «f* Vachek, A Prague School Reader In Linguistics,
(bloomington* Indiana U.P,, 1964)*
33. P.M. Postal, op* clt.* pp. 153-193.
39. n. Chomsky A H, Halle, op. clt.. pp. 400-433.
40. »,», King, op. cit., pp. 192*193.
41. T* lingerstrand, op. clt*, (1933), p* 8*
42. J.A. Ever son & B,P. Fitssgerald, Settlement Patterns
Concepts In Geography 1, (London* Longmans, 1969)*
P. 97.
43. See B»S. Andres©n, Pre*glottallsatton In English
Standard Pronunciation. (Oslo* Norwegian Universities
Press, 1963), pp. 12*24*
44* The gravity model formula we have used here Is mors

properly expressed as* PI Pi

(dij)a
where the value of a depends on the nature of the
terrain and communications* Where transport facll*
itlee are poor, for example, the formula can include
 563

the expreaion (dij)**. In the measurement of

linguistic Interaction, however, the indices
arrived at are purely relative* and this ie
therefore of little importance*

Oai gravity models in geography* see P* Raggett*

op* cit*. Chapter 2*
45. See l,A, Crown* diffusion Processes and Location*
(Philadelphla« Regional Science Research Institute*
1963)* pp* 2»J.
46. V* Labov* "Hypereorrection toy the Lover Middle Class
as a Factor in Linguistic Changs** in W, Bright (ed.)*
op* cit«» p* 83.
47* Xbid*. p* 101.
48. Ibid., p. 102.
4p. M. Joos, "The Medieval Sibilants*» Langua/re 23.
(1952), PP. 222-231.
50. A.C, Ginson# on* clt.* (1962)* p« 128.
31* V* Labov* "Tho Sooial Motivation of a Sound Change**
Mord 19. (1963)i PP. 273-309.
52. W. Labov, op. ©it, * (1966a), p. 455.
53. Ibid*, p* 455.
54. Ibid.* p. 480.
 56k

APPENDIX

Th» Questionnaire

X* "First of fill, we'd Ilk* to ask you a few questions

about your local background,"

1* (1) Whore in your time have you lived

apart froa Nch, » end for how long?
(ii) What different parts of Nchi have you
lived in, and for how long?

(ill) Where did your father* mother* grand*

parents* huaband/wife come from?

2, (1) What Job do you (and/or your husband)

do now? (Ascertain exactly, ) '
(ii) What other Jobs have you done previously?
(ill) What is/was your father*s (last) Job?

3. (i) Which schools did you go to?

(ii) How old wero you when you loft echool?

<iii) Do you have any o- or A-level passes?
(or equivalentt e«g( "Cambridge*•)
(iv) Do you have any further night-school*
college or university education?

II. (word Liet) (Cards 1 * 12)

"Now we'd like you to read out the words on these

cards as naturally as you can*"

**** ( Lexical)

"A fow questions about some Nets# or Norfolk words,"

i. <i) Do you know what a dwilo is?

(ii) Have you ever heard anybody say this word?
(ill) Do you ever use it yourself?
 565

(Similarlys taardlo, mayther, sciult, swad.)

2* Witat do you call that »tuff you can buy on

tho fish stalls on tho market and fry up for

your toa - it*o fish eggs really, some of it

hard, some soft? (jfllta and roes.)

3* Bo you know any other local words?

XV. fRcadtn/r Paasoffo)

•W«*d like you now to read this short story. Please

don*t read it as if you were in the class-room at
school, but as naturally as you possibly con. Ws«d
like to seo Just hew naturally you can read it#*

(isizSsii)
"Some questions about Noh. itself**

1* (i) that do you think of Jfch, as a place to live?

(11) What do/don*t you like about it?
(iii) Would you rather live somewhere else/In
some other part of the city?

(iv) Since you oan remember, has Nch, changed

very much? For the bettor/worse?

2* (i) Do you take muoh interest in what the city

council do?

(li) What do you think of the council?

(Hi) Zf you were a councillor, what would you
have done in the elty?

3. (i) Do you take an Interest in football at all?

(ii) bo you ovsr watch Nch. City? (Zf so, intro¬
duce some topic of Interest, e.g. promotion,)

k. (1) bo you find there•s very much to do in tfeh,?

 566

(ii) Would you aay to someone who was thinking

of" moving here that you could enjoy your¬

self hero O.K. if you wanted to?

(iii) Have you been in a situation, recently or

some time ago, where you had a good laugh,
or something funny or humorous happened to
you, or you saw it happen to someone else?

YI. (panid Word list) (Cards A - C)

1, (i) "Would you now read out the words on theso

cards as rapidly as you can but without
being incomprehensible**
2* (!) "And would you now please say for me the
daye of the week**
(ii) *•*• count from 1 to 20,"

VII, (Pairs) (Cards 1 - iv)

"There"s one last set of words we'd like you to read

out, at a normal speed* Would you please read these

out in pairs**

YTII* (Subjective Attitudes)

"Some questions on the way people speak in Hch«*

1* (1) Co you like the way people in fteh* speak?

(ii) What in particular do/don*t you like?
(iii) Is there anything yon doa*t like about the
way you speak yourself/your children speak?
(4v) Have you ever tried to do anything about it?
(v) Would you like to hear local radio or T*V*
announcers with Nch. accents? Why (not)?

2* (i) Has anybody outside fteh, or Kfk* ever

laughed at you for tit© way you speak?
(ii) lias anybody ever recognised/made a mistake
about where you come from by the way you
apeak?
 507

(ill) !>o you think people outside Hoh. or Nfk#

like lbs way we talk here?

3# (l) Do you think people in Neb# speak different¬

ly from the country people in Kft.t If so#
how?

IX# (Self Evaluation) (Chart One)

"You're now going to hear some different pronun¬

ciations of twelve different words# All these pro¬
nunciations are used in Hoh# We would like you to
mark which pronunciation of each word is nearest to

the one you generally use yourself# If you use

more than one# mark more than one# Zf you can hear
no difference, leave a blank#*

X# (Lify^istlo Zqse<nirltg)
"This is the last thing we're going to ask you to

do# You're going to hear two different ways of

saying ten different words# tfe'd like you to tick
which way you think is correct# and then to under¬
line the way you say it yourself# either if it's
the same or different#*

XX# 1# Age Group#

2# Income Group#
 568

BIBLIOGRAPHY

Aborcromble, D», Problems and Principles In Language

study. London* Longmans. (I956).
Abereromble, D«, Studies in Phonetics and Ilnffit9tlc9«
Londont Oxford University Press. (1965).
Abercrorable, D., Elements of General Phonetics.
Edinburgh* Edinburgh University Press. (1967).
Abercromble, D«, D.U. Fry* P.A.D. MaoCarthy, N.C. Seott
and J.L.M, Trim, ©da., In Honour of Paniel Jones.
London* Longmans.(1964).
Anderson, C.A., and M.J. Bowman, ©ds», Education and
Eeonomio Development. Chicago* Aldine.(1965).
Anderson, J., "Syllabi© or Non-Syllabio Phonology".
Journal of Linguistics 5.(i960). 136-142.
Andresen, U.S., Pre-Glottallgatlor* in English Standard
Pronunciation. Oslo* Norwegian Universities Press.
(1968).
Avis, U.S., "The »New England short o*i a Recessive
Phoneme"« Language 37. (1961). 544-358.
Bach, A*. Boutsche Mundartforschung. Heidelberg*
Winter. (1950)*
Bach, E., and R.T. Harms., Universals in Linguistic
Theory. Mew York* Holt, Rinehart and Winston.(1963).
Barber, B*, Social Stratification. New York* ifareourt,
Brace ami World.(1957)*
Bendlx, R., and S.M, Lipset, eds., Class. Status and
Power* A Reader in Social Stratification. Glencoet
Free Press.(1933).
Bernstein, B«, "Language and Social Class". British
Journal of Sociology 11. (i960). 271-276.
Bernstein, B*, "Social Class, Linguistic Codes and
Grammatical Elements". Language and Speech 3.

(1962). 221-240^.
Bernstein, B«, "Elaborated and Restricted Codes* their
Social Origins and Some Consequences". In Gumpera
and Hyraes, eds. (1964).
 569

Bernstein, B»» "Elaborated and Restricted Codest An

Outline"• In Lieberson, ed. (1966).
Berry, B.J.L., and D.F, Marble., Spatial Analysis» A
Reader in Statistical Geography. Englevood Cliffst
Prentice Hall. (1963 ) •
Berry, 3,j.l., and a. Prod,, Central Place Studies,
Philadelphiat Regional Science Research Institute.
(1961),
Blalock, B.H.. Social Statistics# hew Yorkt McGraw*ttill.
(I960).
loomfleld, L..-Language. Londoat Alien and Unwin.(1935)*
*
ottoroore, T.B., Classes in Modern Society* Londont
Allen and Unwin.(1965)*
Briers, F., ed., horwlch and its Pegion. horwiohi Brit*
ish Association for the Advancement of Science.(1961),
Bright, V«, "Introduction» The Dimensions of Socio*
linguistics". In Bright, ed# (1966).
Bright, V#, ed# Sociollnculstlee. The Haguei Mooton.(1966).
Iironstein# A.J., "Let's take another look at hew York City
Speech". American Speech 37. (1962).13*26.
rook, O.L., English Dialects. Londont Deutech.(1963)•
'ruwn, O., "Syllables and Redundancy Rules in Generative
Phonology". Journal of Linguistics 6# (1970)# 1*17•
Brown. L.A,, Diffusion Processes and Location. Phila*
delphla t Regional Science Research Institute. (1963).
Bunge, W., Theoretical geography, Ltsndt Royal University
of Lund. (1966)*
.apell, A., Studies in Soclolingulatlca. The Haguet
Houtoa. (1966).
Cartwright, a.r., a. Fitch and B.L.J. Robins., "Nana*
factoring Zndustrlce". In Briers, ed. (I96I).
Catford, J.C., "Phonation Types» the classification of
seme laryngeal components of speech production".
In Abercrorable et al., eds. (196%)*
Cheng, R.L., "Mandarin Phonological Structure"* Journal
of Linguistics 2.(1966). 137*159*
 570

Chomsky, If*, Current Issues ir» Linguistlo Theory. The

Hague* Mouton. (1964)*
Chomsky* If* , and H. Halle* The Sound Pattern of English.
New York* Harper and How. (1963)*
Chrlstallor, k», "has Grundgerust dor R&umlichen Ord- i

nung in Europa" • Frankfurter Qeog. ffcfte. (1950)♦

Cobbett, D.J., "Transport- the Pattern for the Future"*
In Norwich - The Way Ahead . (1963)»
Cochrane, G.R., "The Australian English Vowels as a
Dlasystem"• Word 15.(1959)* 69-88*
Cole* G.D.H., Studies in Class Structure. London•
Routledge and Regan Paul. (1955)*
<navies, A*F.1 "Prestige of Occupations"* Pritlsh
Journal of Sociology 3. (1952)* 134-147.
Do Camp* O* f "The Pronunciation of English in San
Francisco* Part I"* Orbis 7* (1953). 372-391.
Do Camp, D.» "The Pronunciation of English in San
Francises* Part II". Orbla 8. (1959)• 54-77•
Dickinson, S,£, , The City Region in Western Europe.
London* Routledge and Regan Paul. (1967).
Dunoan, 0*D«, "Properties and Characteristics of the
Socioeconomic Index"* In Reiss. (1961)*
Duncan, 0*D*, "A Soclo-Economio Index for all Occup¬
ations". In Reiss* (1961)*
uncan, O.D*, and R* Duncan, "Residential Distribution
and Occupational Stratification". American Journal
of Sociology 60* (1955). 493-503*
East Anglla Economic Planning Council, Fast Anglla* A
Study. London* H.H.S.O. (1963).
Eastern Evening News, Norwich - The Way Ahead. Norwich*
Supplement, October 22nd, (1963),
Enderlin, F*, Die Mimdart von Resswli in Oberthurgau.
Frauenfeld 1 liachmann. (1911) •
Everson, «f*A., and B.P, FitzGerald, Concents in Geo¬
graphy* 1# Settlement Patterns. London* Longmans.
(1969)*
Ferguson, C.A., "Diglossia"» word 15. (1959)* 325-340.
 371

Pinch, F.A., and A.J. Hoehn, "Measuring Socio-

Economic or Cultural Status i A Comparison of
Methods". Journal of Social Psychology 13. (1951).
51-67.
Fischer, J.L., "Social Influences on the Choice of a

Linguistic Variant". Word 14. (1958). 47-56.

Flshroan, J.A., ed., Readings in the Sociology of Langu¬
age. The Hague* Mouton. (1963).
Forby, R., The Vocabulary of East Arurlla. London.(1830).
Fovler, R.§ "An English Province and its Capital". 1st
Briers, ed. (1961).
Francis, Revise of Vachefci "On the peripheral
phonemes in Modern English". Language 43. (1966).
142*149.
Frank, Y.A., The Speech of Rev York City, Ann Arbort
university ef Michigan, unpublished dissertation.
(1948),
Fries, C.C., and K.L. Pike, "Co-existent Phonemic Sys¬
tems". Language 25, (1969). 29-5O.
Fromkin, V., "Speculations en Performance Models".
Journal cf Linguistics 4. (1968)* 47-68.
Fry, 2>.B«, "The Frequency of Occurrence of Speech Sounds
in Southern English". Archives Meerlandaises do
Phonetiqus Experlmentale 20. (1947)* 103-106.
Fudge, B.C., "The Mature of Phonological Primes".
Journal of Linguistics 3. (1967)* 1-36.
Fudge, B.C., "Syllables". Journal cf Linguistics 3.
(1969a). 253—286*
Fudge, B.C., "Mutation Rules and Ordering In Phonology",
Journal of Linguistics 5. (1969b). 23-38*
Oiason, A.C., "Phonetic Change and the R.P. Vou«l Sys¬
tem". In Aborcrombie et al*. eds. (1964).
-lass, D.Y., ed., Social Mobility in ffrltaiq. London*
Routledge and Regan Paul. (1954).
alas*, 1>.V., and J.p. Hall, "Social Mobility in Great
Britain* A Study of Inter-Ooneration Changes in
Status". In Glass, ed# (1954).
 572

(•lass* D.V, t and Jf.Jt. Hall, "A Description of a Sanplo

Inquiry into Social Mobility in Oreat Britain"# In
Glass, ed. (1954).
loffman, £i , The Presentation of Self in Everyday Life#
Mow York« Doubloday Anchor. (1959),
Qoldthorpe, J,# and 0, Lockvocxl, "Affluence and the
!British Class Structure** Sociological Review 11.

(1963)# 133-163,
oede, W, and P,K» Hatt* Methods In Social Research,
Nov YorkI McGrav Hill, (1952),
Green, B.« and Young, Morwich — The Growth JSJLSl
City. Norwich! Museums Committee, (1964),
Oregg# R.J., "Notes on the Phonology of a County Antrim
Scotch-Irish Dialect*, Orbls 7. (1959), 392-406.
Gumpers, J,J,# and D,H, Itytnes» eds,, The Ethnography of
Communication. flenashat American Anthropologist

Special Publication* (1964),

I iSger strand, T,, "Migration and the Growth of Culture
Regions"• Lund Studies in Geography Series p.
Human Geography 3. Lundi Royal University of Lund.
(1951).
H&gerstrand, T., The Propagation of Innovation Waves.
Lund Studios in Geography Series B, Hunan Oeo-

rrraphy 4. Lundt Royal University of Lund. (1952).

lingerstrand, T,, "A Monte Carlo Approach to Diffusion",
European Joturaal of Sociology 6. (1965), 43-67,
Hagerstrand, T,, "Quantitative Techniques for Analysis
of the Spread of Information and Technology", In
Anderson and Bowman, eds, (1965),
It&ger strand, T,, Innovation ?>lf fusion aa a Spatial Pro¬
cess* Chicago! University of Chicago Press, (1967).
Hoggett, P., Locatlonal Analysis in ffrtman Geography.
Londont Arnold, (1965).
Hall, J,R,# and D.V. Glass, "Education and Social
Mobility*, In Glass, od, (1954),
Ilalllday, M.A.K., A, Mcintosh and P, Strevens, The Lingu¬
istic Sciences and Language Teaching, London* Long¬
mans, (1964).
 373

Hatam, J* * ed,, Phonologic der Ctegemmrt. Yienaai

Bohlaus. (1967).
Harms, n.T.« Introduction to Phonological Theory*
Engi«uood Cliffs« Prentice-Hal1» (1963)*
Harris* Z*S,, Structural ttnrmlsttcs* Chicago« Uni*
veralty of Chicago Press, (1951)*
llatt* P.K., "Occupation and Social stratification*•
American Journal of Sofltologf 53, (1930), 533*5^3»
Hill* A,A, * ed,, Linguistics Today, New York* Basio
Books, (1969)*
Hill* T,* "Institutional Linguistics** Orbis 7, (1956)*
44i-4 55*
■ochbaum* 6*| <1,0. Barley* E,D, Monachesi and C« Bird*
"Socioeconomic Variables in a Large City*,
American Journal of Sociology 63(. (1955)* 31*38•
liockett* C,F** A Course in Modern Linguistics, Mew
York* Macraillan. (195®)«
Iloijer, IS** "Anthropological Linguistics** In MoSirmann
* * edo, (I96I),
Honlkiaan* S.* "Articulatory Settings** In Abereronbie
e*. al*. eds* (1964),
douck* C«L* Methodology of an Urban Speech Survey.
Leedst University of Leeds* Institute of Dialect
and Folk Life Studies* unpublished* (196?)*
Uubbell* A»F.* The Pronunciation of English in Mew York;
City1 Consonants and Vowels, New York* Columbia

University, (1950),
Johnson* Urban Geography. Loudont Pergaraon, (1967)*
Jones* &** "Concrete and Abstraot Sounds", Proceedings
of the Third International Congress of Phonetic
Sciences. Chent1 Laboratory of Phonetics of the
University, (1939)*
Jones* D*« The Phonenet Its Mature and Use, Cambridget
lief for* (1950),
Joos* M,* "The Medieval Sibilants** lanCTB^e 28, (1956),
222*231,
Joos* M,* The Five Clocks. New Yorki Harcourt* brace
and World. (I967)*
 57*

Kahl. J,A,, and J.A, Iiavios. "A Comparison of Indexes

of Socio—Bconoraio Status"• American Socioioffieal

Review 20. (1955). 317-325.

Kantt L. . "Uraland Studies and Sector Analysis". Studies
in Rural—Urban Interaction. Lund Studies in
Geography Series P. Human Geography 3. Lund*

Royal University of Lund*

Keyser. S,J, . Review of Kurath and McDavidt The Pronun¬
ciation of English in the Atlantic States* Language

22* (1963). 303-316.

Kim, C.-W.» "Two Phonological Motes* B4". Un¬
A#" and
published paper. University of Illinois. (1970).
King. II.D.. Historical Linguistics and Generative
Grammar. Lnglewood Cliffs* Prentice-Hall, (1969).
iiparsky, P.. "Linguistic Universala and Linguistic
Change"• In Bach and Harms» eda. (1963).
Klicsa. U.S.. "Ilelatedncss between Grammatical Systems".
Language *0. (1964). 1—20.
Kohler. K.J., "Is the Syllable a Phonological Universal?"
Journal of Linguistics 2. (1966). 207-208.
Kohler* K.J.. "Modern English Phonology". Lingua 19.
(1967). 1*5-176.
Kokerltss. H». The Phonology of the Suffolk Dialect.
Uppsala1 Appelberg. (1932).
Korahausor, P.P., "The Warner Approach to Social Strat¬
ification". In Bendix and Lipset, eds. (1953)*
Kurath( H.. Handbook of tho Liivrulstic Geography of Mew
England. Providence* Brown University. (1939).
Kurath. I!.. and H.I. HcDavid. The Pronunciation of
English in the Atlantic States. Ann Arbor* Uni¬

versity of Michigan Press. (1961).

habov. W». "The Social Motivation of a Sound Change".
uord 19. (1963). 273-309.
Labov. W#. "The Social Stratification of /r/ in New York
City Department Stores". Paper given at the annual
conference of the Linguistic Circle of New York.
March. (1963).
 573

Labor, W. , "Phonological Correlates of Social Stratif¬

ication.'*. In Guraperx and Hytaes v eda. (1964).
Labor, ¥., "On tho Mechanism of Linguistic Change* •

OeoriTotot'/n University Monographs on Languages and

Linguistics 13, (1965)* 91*114 •

Labor, V«, Tho Social Stratification of English in Now
York City. Washingtont Cantor for Ai»plied Linguistics.
(1966a).
Labor, VM "Some Sources of Problems for Kegro
Speakers of Non-Standard English*. Unpublished
paper. N.C.T.r. Soring Institute on Next Directions
In Elementary English. Chicago, March. (1966b)*
Labor. Vff "The Effect of Social Mobility on Linguistic
Behaviour*• In Lioberson. ed. (1966c).
Labor. V«. "Hyperoorrection by the Lower Middle Class
as a Factor in Linguistic Change*. In bright, ed.
(1966).
Labor, V., "The Reflection of Social Processes in
Linguistic Structures". In Flshmon, ed. (1963).
Ladefoged, p.. Three Areas of rxoorintontal Phonetics.
London* Oxford University Press. (1967).
Ladefoged, P., "Linguistic Phonetics". Working Paoera
In Phonetics 6. UCLA. (1967)*
Lamb, S.M., "On Alternation, Transformation, Realisation
and Stratification*. Georgetown University Mono—
graphs on Languages and Linguistics 17. (1964).
105-122.
Lamb, s.M., "Prolegomena to a Theory of Phonology*.
Language 42. (1966). 536-573.
Laver, "Voice equality and Indexical Information"•
British Journal of disorders of Corrmnicatton 3.

(1963). 43-54.
Lavton. D., Social Class. Language and Fdncatton.
London! Routlodge and Kegan Paul* (1963).
Leltmann, V.P., and Y. Malkiel, eds,, Directions for
Historical Linguistics. Austin! University of Toxas
Press. (1963).
 576

Unski, , "Status Crystallisation* A Non*Vertlcal

Dimension of Social Status"* American Sociological

Review 19- (195*0* kQ3»k13.

Levins* L*« and H.J, Crockett* "Speech Variation in a
Piedmont Community* Foot vocalic r" • Zn Lieberson,
ed* (1966)*
Lleberson* SM ed«» Explorations in Soclollramistica*
The Hague* Mouton. (1966)*
Lundberg, G** "The Measurement of Socio-Eeommi© Status"•
American Sociological Review 5. (19fcO)» 29**»0.
Lyons* J*» Introduction to Theoretical Liwmistics*
London* Cambridge University Press* (1963)*
Mabry* J.H.* Norwich 1961 • An Analysis of Census
Returns* University of East Anglia* unpublished.
Undated*

MoDavid, R.Z,* "Dialect Differences and Social Differ*

eases in an Urban Society** Zn Bright* ed* (1966)*
MeDatid* R,Z** "Dialectst British and American Standard
and Nonstandard"* Zn Hill* cd* (1969)*
Mack* R*V«* "Housing as an Index of Social Glens"*
Social Forces 29. (1951)* 39l-*00.
Martin* F.M.* "Seme Subjective Aspects of Social
Stratification". Zn Glass, ed* (195*0.
Martinet* A«, "Function* Structure and Sound Change*•
word 8. (1952)* 1-32*
Martinet* A** Boonomle dea Changeraents Phonetlcmea.
Borne1 Francke. (1955).
Mayer* S.B.* Class and Society. New York* Random House.
(1935).
Mohnaann* C#» A* Somraerfelt and J* Whatmough, Trends in
European and Amerloan Linguistics 1930*1960.
Utrechtt Spectrum. (1961)*
loser* C*A. * Survey Methods in Social Investigation.
London* Rtiamann. (1953),
Moser, C.A.* and J.R, Hall* "The Social Grading of
Occupations*♦ Zn Glass* ed. (195*0*
Moser, c.A* * and w, Scott, British Towns. London* Oliver
and Boyd, (1961).
 577

doulton, W.G# » "Th® Short Vowel SystMi® of Northern

Bvltsserlandi A Study in Structural Dialectology*•
Word 16, (I960). 155-133.
loulton, w.G. » "Dialect Gwgraphy and the Concept of
Phonological Speeo*# Word 18# (1962), 23«33»
Nystuen* J#i># ( and M.F# Dacey, "A Graph Theory Inter¬
pretation of Nodal Regions*• Papers and Proceed-
tnrrs of the Regional Science Association 7. (1961).
29-42.
&hman, S.E#G, * *Coarticulation in VCV Utterances t
Spootographio Measurements*• Journal of the
Acoustical Society of America 39. (1966). 151-163#
O'Nell* W#A.. "The Dialects of Modem Paroeaei A Pre¬
liminary Report*• Orbls 12# (1963). 393-397.
Urtoa* R«» and E# Dleth. Survey of Fn/yllsh Dialects*
Leedsi Arnold# (1962-)#
Orton, H# * end P#M, Tilling* Stirrer of Fnrrlish Pialeets,
Volume III East Midlands and Fast Atvrljn# Leedst
Arnold. (I969-).
Pellowe# J#* "Establishing Speech Varieties of Con¬
urbations"# Journal of Linguistics. (Forthcoming)
Pickford* G#n.« "American Linguistic Geography - A
Sociological Appraisal*# Word 12. (1956). 211-233.
Postal# P.M#, Aspects of Phonological Theory. New York#
Harper and Row# (1963)#
Pulgram* £** "Structural Comparison* Diasyatema and
Dialectology** Lin -mistics 4. (1964). 66-32*
rutnam, G#N. * and E«M# 0#flern, "The Status Significance
of an Isolated Urban Dialect*# Language Dissertation

22- (1955).
Heading* H.F.* A Glossary of Sociological Terns. Athens«
Contos. (1963).
ogan, J.* "Southern California Ghetto Language and
Poles*# Paper presented at the Autism Meeting of
the Linguistics Association of Great Britain, Uni¬
versity of Leeds# (1969)*
Meiss* A#J# * Occupations and Social status# Glencooi
Free Press# (1961)#
 573

Bobina* R*!I. , A Short History of Linguistics. Londont

Longmans. (19*>7).
Robinson* W,P. , "The Elaborated Code In Working Class
Language"* Language, and Speech 8, (1963). 243-232*
tiaporta* S* » * Ordered Bales * Blaise t Differences* and
Historical Processes** language 41. (1963)* 218-
224.
Sehano# S.A, * "Matural Hales In Phonology** (To appear
In the Proceedings to a historical linguistics
conference* UCLA* February 1st and 2nd* 1969.)
(Forthcoming)
Schwara* E* * Die dent,achen Muivlnrten. Gotiingem
Vandenhouck and Huprocht. (1950).
Shevelov, G*Y.» "On Frodicanonts* Predictability end
Futurism in Phonology*. In Hocus* ed* (1967) •

Shuy, R*w*« W.A. Wolfram and W#K« Riley* Llurruiatic

Correlates of Social Stratification in Detroit

Speech* U.S. Office of Educationi Final Report*

Cooperative Research Project 6-1347. {1967)*
snuy*
niqijos In Tfrhnn .Umflnnge Study* Washington!
Center for Applied Linguistics. (1963).
Sivortsen* E« * Cockney Phonology. Osloi Oslo Uni¬
versity Press. (I960).
Smailos* A.E* * "The Urban Hierarchy in England and
Wales** Geography 29. (1944)* 41-31.
Smalles* A *18** The Geography of Towns. Londont Hutchin¬
son. (1957)*
Spencer* J** M, Gregory and N* Enkvlst* Linguistics and
Style. London1 Oxford University Press. (1964)*
Tay* M.W.J.* A Phonological Study of Uokklen. University
of Edinbtirght Unpublished Doctoral Thesis. (1963).
Tay* H.W.J., "ftofctclen Phonological Structure". Journal
of Linguistics 6. (1970). 81-88.
Thorns* A.R.* "Generative Phonology in Dialectology*.
Transactions of the Philological Society. (1967).
179-203.
 379

Thomas, P.O., "Commercial and Professional Services"»

In Briers, ed, (1961),
Trager# G,L«, and 1I.L. Smith, An Oat line of English
Structure. Washington* American Council of Learned
Societies. (1957)*
Vachek, J«, A Prague School Pender in Linguistics.
Jjlooningtoni Indiana University Press. (1964),
Vanvlk, A,, A Phonetlc-Phonenlc Analysis of the Dialect '
of Trondhoin. Oalot Oslo University Press* (1966)*
Vaslliu, E», *Towards a Generative phonology of Daeo-
Uumanian Dialects*• journal of Lin'mlstica 2.

(1966). 79-93.
Vaslliu, E«, "Transformational versus Btunique Phonemic
Typology*. In llamm, ou, (1967),
Viereck, W., Phonenntlache Analyse des Djalekta von

Oateshead-upon-Tyne/Co. Purfiam. Hamburgt do

Gruyter. (I966)•
Earner, W.L. , H, Meeker and K. Lolls, Social Class in
America« A Manual for Procedure for the Measurement

of Social Status. ChicagoI Science Research Assoc¬

iates. (19^9).
Weixtreloh, U», Languages in Contact. New Yorki Public¬
ations of the Linguistic Circle of Hew York. [li£2>)
Volar#1ch, U., "Is a Structural Dialectology Possible?"
Word 10. (195^). 333-^00.
weinreich, U,, W, Labov and M.I, flersog, "Empirical
Foundations for a Theory of Language Change". In
Lehmann and Malkiel, eds. (1966).
Wetmore, T.H,, The Low—Central and Low-hack Vowels in the
Eastern United States. Alabama1 University of
Alabama Press. (1959).
wood, A.A., "City of Unique Quality". In Harwich - The
i£2JL Ahead. (1963).
Wright, J.T,, "Urban Dlaleots - A Consideration of Meth-
od". Zeitachrift ffir Utmdnrtforschun/r 33. (1966),

232-2h6.