, <1'8

-d"~
íI...Y H :
"~< ,.~.( ,

132 Languageand society JJ }: Wí LL t+H OOWNE;S. I.c¡qg. ~"91i.V\8(. ~A.Cl

To conclude, it is important to be clear about the methodo-, ~ I {" .. Socte.:!:'y. ~ecaf\.d Ed,n"
logicalclaimsin this section.We are not criticizingthe use od.~ ,\1 5 RhotlClty
statisticalmethods.the quantitativeparadigm,in describinglin- ~mrt, .. ~Mbr"09~ ~ Cambnctge..."
guistic variation. No other approach-to community \vide variation i;
is possible. The issués we ha ve been discussing are about hów to ,~ ...
, "

,
Uf\\Ve(1;~ ~regs.. .,
interpret and explain the quantitative results. Our main point has ''1<
been that although the statistical pattern manifests itselfin thei ;1'
(
I
variable behaviour of each individual speaker, represented by vari- ,
able rules, the socialcorrelationsencodedin such mies cannotbe:
'

said to 'causally explain' the variable performances they in Cact

represent. We don't know 'why' the individual performs the way
they do. But this is normal in statisticalstudies.As David S~ofI
writes (1988:150):' . number of éOnSommts that have begun to decay in Standard
Macroscopic sociodemographic categories or network-Ievel patterns of
relationship do not directIy affect the performance ofindividual speakers; ~I.'
implicit in any correlational study is the existence of mediating pro- l
cesses or intervening mechanisms which lead from extralinguistic
factors, through consciousintent and/or unconscious tendencies,10
actual behaviour.
The interpretative or functional issue remains to be explored
after the 'correlational' results are in. Statistical method itself is'
neutral as to why the pattern occurs, whether it is caused psycho-
logically or by internal structural pressures or socio-historically
or is intentionaUy motivated and consciously (or unconsciously)
meaningful for participants. Or a combination of these.
What makes it the case that individuals' utterances, remote in
time and space, are 'coordinated' in such a way as to produce th~
large-scale pattern that quantitative studies make visible? The
waves of change only reveal how individuals are socio-historically
placed with respect to the transmission ofthe pattcrn of frequencies.
But what is the 'coordinative mechanism' between individuals'
utterances, in the first place, such that they are generating and
diffusing waves of change? And this over and above any psycho-
logical or internallinguistic factors. In the rest of the book, some
answers to this question will be explored. In the next chapter,
however, we willlook at a single variable in some detail. We will
do this with an eye open for social factors in language change.
The genera,lIy more distinct utterance of Americans preserves a
Bnglish . . . In 1913 the late Robert Bridges belaboured the Bng-
lishcIergyforsaying 'the sawedofthe Laud'insteadof'the sword
of the Lord' ... The violent Anglophile, Henry James, revisiting
the United States after manyyears in Bngland, was so distressed
by this cIear soundiÍ1g of r, that he denounced it as a 'morose
grinding ofthe back teeth'.
Mencken (1919)
" The feature in question is postvocalic r. This is the r in words
like 'guard', 'art', 'lord' and 'fourth', and at the end of words like
'floor', 'far' and '!rider'.
In line with our popular beliefs about accents, a basic dichotomy
can be set up about how peóple pronounce words like these in
English. It is often said that English speakers of English drop their
rs, and, in contrast to this, Americáns pronounce aU the rs that
appear in the written language. The accents are r-less and r-fuU,
respectively.
Even folk perceptions are more fine-grained than this, however,
, and within Britain there is an awareness that Scottish, Irish and
West Country accents are r-full. As mentioned in chapter one, for
English English speakers r-fullness has a social meaning, not only
oftransatlantic English, but, at home, ofboth rusticity and bucolic
genuineness. Accordingly, it is sometimes used to advertise such
wholesome products as 'butter'. and 'cider'. In the United States,
133

134 Languageandsociety
on the other hand, over and against a norm of r-full pronuncia-
tion, people think of r-less English principally as a feature of South-
em accents, 'confederate' English as it is sometimes called, and' ,L
-
also ofwhat is looselycalledan 'Oxford'accent the prestige pro- '
nunciation of southem England. As we shall see in chapter seven,
stereotypesassociatedwithpeople'sattitudes towardsditIeringforros
of speech are themselves significant facts, and an important part
of the study of variability. .
After the methodological emphasis ofthe last chapter, this chap-
ter willlook at some descriptive studies, and will concentrate on a
single feature, postvocalic r. There are reasons why this is appro-
priate. This sound-feature has been extensively studied in a number
of ditIerent communities on both sides of the Atlantic. The pattem
},e-.
that emerges is very interesting, partly because of its scale in space '"...:
and time. The sound has been involved in a long-term pattem of '1;
"1
changes in many accents of English. It is currently in the process
of change in New York City and was one of the features studied by
Labov (1966). Because ofthe change in progress, we will find 00",' ~¡ stressed or not,
.:1 consonant,
interestingly ditIerent sociolinguistic structure for the variable (r),
as compared to that of (ing). We will also begin to see how social these latter environments. It may:
factors are inextricably involved, not only in language variation
at a given time and place, but in the ditIusion of linguistic change.
Before beginning, however, a note on jargon. Wells (1982) com-
ments on the use of 'rhotic', 'rhoticity' etc. in place of the more
straightforward 'r-full', 'r-fullness' etc. It seems that for some
people with r-less accents, the term 'r-full' may easily be mistaken
for 'awful'. We would then be describing awful accentsl Shades of
Henry James - see the quote at the beginning of the chapterl
Rhotic and non-rhotic l"
The situation is, of course, much more subtle and complex thoo ,
any simple dichotomy. Let us consider exactly what is meant by ,
saying an accent is rhotic or non-rhotic. The first point is that r:.
occurs in a number o{ ditIerent linguistic environments. In someI
of these, it is never or very rarely dropped. Places where either all'
or almost all speakers never drop r, are:
Rhoticity 135
1. Word initial:reed.rawetc.
2. Betweentwo vowels:arrow, borrow etc.
3. In consonant dusters beforea vowel:bread. bring etc.
Because of its appearance in these pos~tions,the sound r is part
of the sound system of English in general. It serves to tell words
apart, for example, 'raw' v. 'law', 'rUID' v. 'bum', and 'drank.' v.
'dank'.
When we talk of r-dropping in r-less accents, we are actually
taIking about its loss in two specific environments. What these
have in common is that the r follows a vowel; hence the term
'postvocalic r'. These two environments are:
4. Aftera vowel:guard, board etc.
5. Word final:floor.rider etc.
'
The environments of r-dropping can be made more delicate: in
j terms of the class of the preceding vowel. whether the syllable is
and whether
so on. the next word begins with a vowel or a
Now there are three possibilities fOrJUlacc~nt in relation to r in
(Historical Categorically
retainr (Rhotic =r-full)
direction Variabl1Tetainand drop r (Variable =has variable(r))
l ofchange) Categoricallydropr (Non-rhotic =r-Iess)
These are the three phases, noted above, that occur in the pro-
l' cess of a language change. Historically, if we look at the English
" looguage overall, the dropping of r is an innovation. In earlier
periods, English would have been rhotic throughout. That this is
. fue case can be readily seen if we think of the system of spelling
that we have been handed doWQ.,and that was regularized in the
eighteenth century. In general, thbQgh not always, the presence
, of r in the spelling tells us the positions of the historic r.
If we look at English today, all three of the possibilities relevant
to r can be found. Later on, we will be concentrating on speech
.., communities where there is a variable (r). Figure 5.1 surveys the
. overalldistribution ofrhoticity, variable rhoticity and non-rhoticity
~. among English accents (Wells, 1982). In dealing with this feature,
~
 136 Languageand society
Rhoticity 13 7
.\ ,

Rhotic Variably rhotic Non-rhotic between the two sides of the Atlantie;'V ariability and the oppos-
ite categorical pronunciations are common in both societies, in
General American Local accents
class of accents:
RP (Received particularspeechcommunities.. .
in the west of Pronunciation)
midland, north England in England and How r is made
' central. middle
Atlantic etc.
Wales

l "
Southern rnountain
accents in US, 'hill.
type' of speech
A few local accents
in the north of
England.
Local accents of
the east and north
of England
Consider,for a moment, how difIerentto the ear an r sound
appears in American, West Country, or Scottish speech - in these
three rhotic varieties. Whether-we classify an accent as rhotic or
not depends solely on the presence or absence of postvocalic r. As
NewYorkCity Most accents of
General Canadian Wales and New we saw before, this is a part of the systematic arrangement of
'Borderline'
Zealand sounds as they function in the language, permitting speakers to
rhotic/non-rhotic
Scottish accents tell words apart; for example, 'guard' Ti'om 'god', or 'board' from
areas in US,e.g. Australia

~
'bawd'. However, there are also considerable variations 'beneath'
Irish accents South, eastern
South Africa this, so to speak, on the purely phonetic leve!.There are a number of
NewEngland,black
Sorne West Indian, Englishvernacular BlackEnglish different ways in which the sound r can be made with the organs
e.g. Barbados vernacular in US ofspeech. The r sounds difIerent depending on how it is made, and
'this is irrespective ofwhether the speaker' s accent is rhotic or noto

'\
Sorne parts of
eastern New England
Southern speech area
in US, 'plantation'
type
Sorne West Indian,
e.g. Trinidad
Figure 5.1 Rhoticity In the accents of English
. friction,and the vocal chords are vibrating, giving the sound
we are talking about accents, not dialect difIerences. You will recall
that accents difTeredonly on the level of sound, and not neces-
sarily on any other leve!. Remember also that the terms in Figure
5.1 are misleading in the way that they label accents as things, as
discrete objects. In fact, we are dealing with continua, as we shall
see. So Figure 5.1 is just a general guide. One point is quite clear
from it, however.
Although prestige accents in Britain and North America - RP
and General American respectively - provide 'polar norms' of.
non-rhotic and rhotic speech,' there is not any simple dichotomy
We all have rs in some environments.
In word-initial position, most accents make r in roughly the
same way. In this position, it is a consonant sound, shaping the
beginning of an initial syllable. Usually, the tip of the tongue is
held near to, but not touching, the ridge just behind the upper
teeth. The tongue tip ~ usually turn~(L~lightly backwards. This
is called retroflexion. The sides of the tongue are touching the
molars and there is a lateral bunching of the tongue. The air
stream escapes continuously and freely out of the mouth, without
'voice'. Bronstein (1960) points out, in connection with a com-
mon American pronunciation of this initial sound, that it can be
aItematively made with the tongue-tip held low, and the central
'part of the tongue bunched and raised upwards and rearwards.
One major difIerence between accents has to do with the degree
of retroflexion in articulating the sound. Thus, RP has slight or
no retroflexion, while American and West Country accents have
more. Such movements produce the r impressions that we perceive.
There are other ways in which r can be pronounced. It is one
ofthe most various of English consonants. It was originally a trill,
I
138 Languageandsociety
a series of rapid taps by the tongue against the teeth ridge. This
I still occursin some accents and some individuals in certain styles.
It later became a fricative sound, in which audible friction can be
'\ heard in the narrowed gap between the tongue and the teeth ridge.:\
I This still occurs when r followsd, as in 'drink' in RP and other
accents. Also, in RP, in the position between two vowels, for
I
: example in 'very' or 'niarry', r is often realized as a flap, a single
1 (
.
tap of the tongue-tip on the teeth ridge. We will be mentioning
these various pronunciationscof the same abstract r sound later, as
they arerelevant.
The English pyramid
Figure 5.1 tells us that RP is non-rhotic. This accent is the
national prestigenorro ofEngland and Wales.Beforelookingspe-
cificalIyat r-Iessnessin England and Wales,we must lookat how \
various accents are related to each other within the overall
society.
RP
Highest class: RP
It is generally agreed that the relationship of accents to each
other in England takes the forro of a pyramid (Wells, 1982: 14;.
Hughes and Trudgill, 1979: 6). The pyramid represents two con-
tinua of variation. The base represents geographical variations,
local accents. The vertical dimension represents. social stratifica-
tion. At the apex is RP. ' ,
It can't be emphasized enough that RP, at the top of the pyr-
amid, is a national norm.,It is not localized.It is distributed in the
highest social classes up and down the country. At the opposite end
Rhoticity 13 9
ofthe social scale, at the base of the pyramid, we find the speech of
the lowest classes- which isalso the most highly localizedpro-
.' nunciation. It is the broad local accent of that area. Working-class
¡, accents vary markedly from region ..to region: Cockney, Scouse,
/ ~.Tyneside,West Midlands and so on. These CID1 be broadly classified,
ofcourse, in larger regional terms, as southern or northern, for
example. But the most localized accents are also those at the bottom
"
end of the social scale.
The middle social strata are distributed between these two
norms. In each geographical area, therefore, as we move up the
socialscale, speech becomes moreRP-like. Conversely, as we move
down the social scale pronunciation becomes less RP-like, and,
at the same time, more local. We saw this situation in the Norwich
(ing). In the case of this feature,consistent use of the 'ing' 'variant,
scored as 000, represented consistent RP pronunciation. This was
achieved by the middle class in two re'áding contexts. At the other
~~,
end of the scale, the lower working class scored 100, or complete
Useofthe 'in" forro in casual speech. Middle-c1assaccents, therefore,
can be 'mildly regional', in different parts of the country, in their
approximation to the national RP norro. Style-shifting tells us also
that individualscan shifttowards RP on appropíiate occasions.
A word about the RP accent itself. It is the accent particularly
associated with BBC radio and television newsreaders and with
individuals and social environments associated with the conven-
tional establishment. It has been potnted out that, like any other
accent, it is associated with and maintained by social and com-
munication networks; in this case~..~owever, networks directIy
involved in social and economic power (Milroy, 1987: 183fI. for
, example). Hughes and Trud~ill (1979: 2f.) write:
RP has. . . remained the' accent of those in the upper reaches of the
social scale, as measured by'education, income and profession, or title.
It is essentially the accent of those educated at public schools (which
are, of course, prívate, arid beyond the means of most parents). It is
largely through these schools that the accent is perpetuated. Por RP,
unlike prestige accents in other countries, is not the accent of any
regíon (except historícally: its orígins were in fue speech of London
and the surrounding area). It is .quite impossible to say Cromhis pro-
nunciation where an RP speaker comes Crom.
 140 Languageandsociety Rhoticity 141
Although RP has been extensivelydescribedin phonetic and i!
phonological terros, there is an urgent need for empirical studies
ofhow such overtly prestigious standards are created, maintained¡ j,
and convey normative pressure. There is also internal variation .
and change within RP, though not on a geographical basis. Note
also that the use of RP as a prestige accent must be interpreted as II
relative to the meaning which its use conveys in a given contexto

( There are occasions when it would be seen as 'affected'.

Tbe diffusion of r-lessness

It was in the eighteenth century in the south-east of England
that English began to lose postvocalic r. Let us look at the situ~.
ation geographically. Ir today we plot on a map of England those
areas where any rhoticity still survives, as in Figure 5.2, what are
we looking at? Ir we put ourselves at the point of origin of the,
change, the south and east, we are looking outwards to the limits,
':¡ of where the loss of r has diffused as a categorical property of all
;
accents. This is the white1area ofFigure 5.2. In this area, pronun-
ciation is r-Iess, from RP down. We are looking at the past diffu-.
sion of the change through geographical space.
But we are also, simultaneously, looking downwardstowards
the lower end of the social scale, tow,ards the base of the pyramid;
We can assume that the loss of r was a prestige innovation, and
therefore related to the norm of higher status groups. The fact thatJ
RP is categorically non-rhotic tells us this. Therefore, those areas!.
where rhoticity survives, since they are plotted in geographicalu~""
and therefore 'local' terros, also represent the situation lookingU:
down towards the bottom of the pyramid, down the scale of social{~ i

variation. We know that RP speakers are r-Iess. How far down-,¡~; '11
~i'r wards through social space has the loss of r progressed? Figure 5.2!;'
tells us something about this. The shade~ r-pronouncing areasi~1
represent the accents of the informants of the Survey01Bngl1sh~
Dialects. These are the most consentative speakers of all - older,~
working-classand ruralspeakers. f¡

The categorical area of r-Iessness, the whitc area, is large~~

if one plots regions where r is observed in urbanspeech, as in Fig11J1
ure 5.3. We can immediately see by comparing the two maps tha~l, ,

.
 142 Language and society
Rhoticity 14J
tbepressureof the non-rhotic maj6iity, now seemsto be located
totbenorth ofManchester.in placessuchasRochdaleandAccring-
ton. It remains also in the country areasaround Preston and in
tbenorth of the county' (Lancashire).In the west of England.by
contrast,rhoticityismorewidespread,evenin urban centres.Thus,
there is a postvocalic r in the speech of Bristol. with considerable

( retroflexion (Hughes and Trudgill, 1979: 47). The situation can

be generalized in this way. The further west one goes, the more
widespread is the r-fullness and the higher up the social scale
it extends. The converse is true heading eastwards. Broad local
accents are fully rhotic and thiS 'extends well up the social scale in
.,jJ CÍtiessuch as Bristol, Exeter, or (to a lesser extent) Southampton.
Plymouth and Bournemouth, large-Citieswith very mixed popula-
1,
, tions, seem to have variable rhoticity or even none. Traces of
,I,jl variable rhoticity may be found as close to London as Reading
.{Berkshire)'(Wells.1982: 341). In 'borderline'areas, r is clearly a
'¡. ,variable(r).
,It is also clearthat wearedealingwith a continuous processof
lossof postvocalic r in Englapdthat may have lasted nearly three , I

,.centuries. As Chambers and Trudgill(1980: 109) point out, the

discontinuityof the rhotic areasin Figure 5.2 shows us that we
:,aredealing with reUc areas, where the older form has persisted
in spite of the long-term spread of the 'r-dropping' innovation.
: Oneimagines that ifm~ps such as5.2.and 5.3 had beendrawn in
tbenineteenth century, the rhotic areas would have been joined
Figure 5.3 Areas (shaded) of Great Britain where Irl still occurs In urban together,rather than appearingas islands of r-full speechin a sea
accents (from Hughes and Trudgill, 1979: 33) ofr-Iessness. The diffusionof r-Iessnessmakes concrete the waves
> ofchange in socialspace.
non-rhotic. Tbis suggests, as we shalI see in chapter six, that ur-
ban areas act as fociin the diffusionof change. Its progress does Some linguistic consequences for r-Iess accents
not proceedsrn.oothlyfrom the point of origino "

.,
It is possible to infer from Figures 5;2 and 5.3 that a complex1 ,
> When an accent loses postvocalic r, it has widespread ramifica-
wave of change has passed through geographical and social space~ 'ítionsthroughout its sound system.We have alreadynoted that in
leading to a loss of rhoticity in most English accents. The procesf; non-rhotic accents we get homonyms which we do not find in
~ has gone further in the north than in the west. In the north !:
il' I rhotic ones, for example 'bawd' and -'board' in RP, or 'guard' and
urban rhoticity has practicalIy disappeared. Wells (1982: 368 .
says, 'The patch of residual urban rhoticity, ever shrinking unde¡: if ,¡god' in New York City. What lfappens is in fact very complex and
beyond the scope of this book. So, 1 will briefly sketch some basic
'J
r1'.
)~,
~¡i
~I
 144 Languageandsociety
'1
cansequences, ta give an idea af what happens when r is last,and¡i
ta shaw that changes in ane saund affectather saunds. . 'st laak at haw pastvacalic r is made in accents that have it.
The cansequences we are gaing ta cansider apply in varying!
ways ta all nan-rhatic accents, althaugh RP will be the main ex~i
ample. One majar cansequence takes place in the vawels. Therer
is a wider range af cantrastive vawel saunds in nan-rhatic thai1¡
in rhatic accents. Ta eXplain why this shauld be so.,we need ta ba
able ta describe differing vawel saunds and haw they are madef¡
We will have ta describe vawels in dealing with variables in later.
chapters in any case. Imagine the trapezium shape belaw as rep~
resenting a crass-sectian af the aral cavity (Le., the inside af the.
mauth) as viewed fram the side. The diagram is, af caurse, highly
idealized. A vawel saund is praduced by the campletely unab:"
structed autward mavement af the air-stream thraugh this chanHi
ber, with the vacal chards vibrating. The saund quality af eacb!
particular vawel saund is determined by the shape given ta the~
chamber by the pasitian af the tóngue.
(
Close
'boot'
Front \ C,) t"'
I Back
U7
'lad' 'cod'
Open
. ~~
UsingvoweIdiagramslikethis,thevawelsafa language can:!
be charted accarding ta where in the mauth the highest part o.'
the tangue is put as the saund is made. Vawels can be frant o.
back and clase ar apeno Try ta abserve the faur peripheral vawe
saunds, indicated by the faur key wards in the diagram, by natin
the pasitian af yaur awn tangue as yau make the saund. The av
at the centre af the diagram represents the central area af tb
mauth where central vawels are made. These are saunds such
the last saund in 'the', in rapid cannected speech, ar the vawel i.
RP 'bird' ar 'fur'.
Rhoticity 145
í:.Ta appreciate what happens in nan-rhatic accents, we shauld
fAbove,we laaked at haw r was made in initial pasitian.) I will
pllaw Branstein's (1960) analysis af General Ameqcan. In gen-
.:ral,the tangue is held in the pasitian af a central vawel. This
'oond is then ,r-calaured. Either the-tongue-tip is turned back
':owardsthe hard palate in retraflexian (the amaunt af retraflex-
'onvaries fram accent ta accent), ar the tangue is bunched and
!~etractedupwards, ar bath. This is a very vawel-like saund. Can-
teive af it as a canstrictian af a central vawel that praduces an
:audibler-calaur ta the saund.
. When r is drapped fram an accent, the canstrictian dis-
i~ppears.The pracess cauld be thaught af as the absarptian af the
'-calaured central saund inta the preceding vawel. In cases.ather
:thanward-final'er', where we gain a syIlablecansisting af only a
r eatral vawel, the autcame far non-rhatic accents is a system af
!~ntringglides ar diphthongs. These are vawel saunds in the
bourseaf
~' which the tangue changes-positian.In
. centring glides
lile tangue ends up in the central area af the mauth. The number
ófdistinct,
H. centring diphthangs~ and the range af pasitians fram
;;Which they begin vary enarmausly bath ~etween and within nan-
rhatic accents. Here is the RP system (Gimsan, 1962):
'lure' ,1'poor'
'Weird"\'fearQ. p
~ ~ I'force','saw'
'scarce', 'the~e' \ 0----.....
:f Other things alsa faIlaw, with much passibility far variability.
!fhus,these diphthangs can became simple vawels, so.that wards
¡ke'fear', 'paar', 'scarce' and 'farce', especiaIly-the last twa, have
(ariantswhichare not diphthangs,but a singlelengthened vawel.
!Bormast RP speakers, in fact, the 'farce' saund is always a lang
,;owel.In varieties where rhaticity also exists,variably,such as
 146 Languageandsociety , Rhoticity 147
,j
that of New York City, the situation can become very complex¡' .vfr-lessness. In fact, intrusive r never occurs in rhotic accents. But
indeed. ,} : Uis a perfectly natural phonological process for non-rhotic accents.
U followsfrom their rules.
~ In spiteof this, intrusive r is stigmatizedin Britain. Speakersof
[,inking r 1, ID.on-rhoticaccents accordingly attempt to suppress it, especially
"'t' Another consequence of the loss of r is its reappearance under: 4nsidewords (as in 'drawring'). But.think about how difficult this
, certain conditions. Thus, it is a common characteristic of non~' :wmbel In the absence of r in one's knowledge ofthe word, ¿ne has
rhotic accents that r appears in word-final position to serve as ' 1tobe guided solely by the preceding vowel's class, and the spelling.
]o
( ~
-
link with a followingword but on1ywhen the followingwor, ;{rhespelling tells us which words had r historically. In fact, the
beginswith a vowel.Thus, i !Stigmatizationis related to fue -existence of these final rs in the
"i'l
withlinkingr withoutlinkingr ':Writtenlanguage,
H and shows us the power of literacy in relation
110notions of correctness. A standard written form.in this case is a
far away fa . . . country
,,1 irehicle of normative pressure in terms of the prestige norm. This
answer it answe . . . badly iJwillensure that intrusive r will tend towards fower frequencies in
1 ear engine ea ... port
,1 }nore careful styles. In fact, suppression usually invol~es suppres-
11)1..
1 .~ionof linking r as well, or its realization by an altemative sound.
,'1 Intrusive r . set of rules, consequences of loss of r, thus opens up new
!' ossibilitiesfor variation.
'~1
:(.é! But the reappearance can go further than this. The r can ap¡¡
~; 1
pear beforevowels where(there never was an historie r and wher~ ) ,

accordingly, there is no written r. This intrusive r is common i, ,,' Scotland: rhotic accents
non-rhotic accents. Thus, ~ .
¡ A look at the map in Figure-S.3 shows that Scottish accents
intrusiver :xhibitfull-blown rhoticity; a striking contrast to English norms.
idea-r-of We can see in Scottish accents how vowels combine with r
area-r-ofagreement d contrast this wiTh the centring diphthongs of r-less speech.
Shah-r-ofIran 'he diagram below illustrates how r functions in rhotic accents
draw-r-ing adapted from Wells, 1982: 408). Ideally, each set ofwords is dis-
Thisis a veryinterestingfeaturebecauseofwhat it tellsus abod'. . guished by a simple vowel sound, numbered 1-9 in our dia-
the rules ofnon-rhotic accents. It suggests that r-less speakers, .: ,~, plus an r-sound. Ofcourse, there is much variation between
fact,don't 'know' which wordshistoricallyend in r, and that ther, ,
1. beer, fierce, weary,
fore they do not represent final r in the way they 'store' the; 9 2. stir, bird, spirit,
words. Therefore, for such words, the rule is not to drop r before " 7 3. air, searee, fairy,
consonant, but rather to insert r whenever there is a followiIi' 2
4. err, pert, ferry;
vowel. This rule applies to words where historie r occurred, anl +r= 5. bar, start, merry.
we observelinkingr. Butit aIsoappliesto similarwordswherethe 6
8 6. war, horse, sorry.
was no historie r, and we observe the so-called 'intrusive' soundi 7. ware, hoarse, story,
Since this tells us that the accent is fundamentally r-less and cann 5 8. purr, word, hurry,
have a 'dropping' rule in final position, it is an excellent diagnos~" 9. poor, guard, jury
 148 Languageandsociety
Scottish varieties, Thus, in general, Scottish vowels 2, 4 and 8 are.'
kept distinct; the pronunciation of 'bird', 'pert' and 'word' are dif..
ferent. In popular Glasgow speech, however, 2 merges with 8, and;,
'bird' and 'word' rhyme, although remaining distinct from 'pert',
By contrast, in middle-class Edinburgh pronunciation all three,
categories are merged into a constricted central vowel, and all th~
words of 2,4 and 8 rhyme. We find similar processes in Generirl~ '1'
American. .r~
Let us look now at how this systematic r is actually articulatedl'
in Scottish English. Nearly the full range of variants occurS. Weh ¡
discussed how each of these was made above. They are:
(a) trilled r (popularly,rollingyour rs)
.' (b) flappedr (a singletap on the teeth ridge)
(e) the frictionlesscontinuant (describedabove)
(d) r-lessness(newlyobservedby Romaine,1978)
. 1
1 noted earlier that the trill was the original form of English 1':'
The literature and popular opinion suggest that this persisted ~ 'emerge when sex and style are Iooked atoThe boys produced more
Scotland until this century as a major pronunciation (Wells, 1982;'
Romaine, 1978). Today, however, this sound has been largel,'
replaced by the two middle variants, the flap and the continuanil.
The rolled r is mainly conftned to special formal contexts, Tb
latter two forms are variable, and conditioned by both linguist¡'
and social factors. We thus have an (r) variable, with the flat
and the continuant as its values (although the trill couldb'
included as a third value).
Romaine (1978: 145) reports that in Edinburgh it is the con;~
tinuant form which is the prestige variant. It is more frequentl
found in Scottish standard English, is a marker of 'polite' EdinburgI,¡
.~
speech, and is endorsed for teaching. But both Romaine and Aber~
crombie had also informally observed some cases of apparentlY'1)
less speech in the city. This, in the context of complete Scottis
rhoticity, was remarkable.
Romaine (1978) investigated this r-variability by studying'
speech of twenty-four working-class Edinburgh schoolchildren ag'
from six to ten years old. Her (r) variable had the three valu
below. The (r) scores were the percentage of each variant observ,
The study was restricted (unfortunately) to r in word final positioq
Rhoticity 149
~ l. theflap I
(r) ~ 2.
3. th~co~~uant
r-Iess
\
I
The overall index scores for all ~peakers show that the flap was
the most commonly used form, the continuant next, and r-Iessness ~
~.last.Although r-Iessness did occur among Scottish schoolchildren,
I
-it did so relatively little. Remember also that it was restricted to
,word-finalposition and was most favoured by utterance-final
position.The figuresare
1. theflap 47
~
Totalindexscores(r) 2. the-continuant 38
- 3. r-Iess 15
The correlations that proved really suggestive, however, only
"Dftheflap, and ofthe r-Iess variants. In contrast, the girls produced
more of the continuant than the other sounds, and had very little
,-lessness. So, the continuant seemed t9 mark female speech, and
'fue flap and non-rhoticity, male speech. The style-shifting was even
,moreinteresting. Among the ten-year-olds, when reading passages
the boys shifted from the flap to r-Iessness and the continuant, in
;that order. But the gu.ls also shifted to r-Iessness in reading stylel
. Romaine(1978) interprets her results in this way. There seem
(to be two norms at work. The' Edinburgh prestige norm is the
'tcontinuant. This accounts for the sex difIerentiation of the scores.
GirIsare responding more positively than boys to the prestige norm.
IThegirls are closer to the middle-class variant. But, at the same
¡time,
é
there is a vemacular norm - the r-Iess innovation. The
"finnovation' seems to be Iocated among the boys. Both boys and
,irIsalso recognize the prestige of this form (a covert prestige per-
~):¡.aps)
because they both shift towards it in reading style, although
oys also shift towards the overtly prestigious continuant as well.
uch patterns suggest that a change is taking place. We will re-
iturn to such matters later. However, Romaine feels that the source
, f this r-Iessness is separate from RP influencein Scotland. The
 Rhoticity 151
150 Languageandsociety
Scottish norm is clearly r-full. ando within Scotland. this is th~,
prestige formo If the local non-rhoticity is sigoificant. it is as al
vemacular and male innovation.

Transatlantic
The rhoticity plot reálly thickens when we look to the othe
side of the water. ~
The map in Figure 5.4 sets out the main dialect areas of th~
eastem United States. The particular speech communities whicñ
we shall look at in the course of this chapter are labelled. Tb'
shaded areas show us that rhoticity is a feature of the accen ,

in the centre of America; more specifically of the north central~

central midland. middle Atlantic and mountain speech areas_..
Sometimes all these rhotic accents are grouped together under thj;
label. 'General American'. They have in common the fact thaJ~
they excludethose Bastemand Southem accentswhich are r-Ie~ ' South Carolina
It is important to realize, therefore. tlíat General American is no'
the same sort of thing as RP. Rather, it is a class of accents whiC
are geographical1y defined.
So let's be geographical. It is usually said that r-Iess speec
occursin eastem NewEngland.NewYorkCity.and in the coas
J

EASTERN BOSTON
NEW ENGLAND

)1 METROPOLlTAN NEWYORKCITY
NYC
:f Figure5.4 Major speech areas of eastem USA. wlth rhotlc areas shaded
~r: EASTERN RICHMOND/NORFOLK. Va
ain ofthe South. We can think of American r-Iessness as radiat-
~
\
1,
'
VIRGINIA
g outwards from the major ports of the eastem seaboard. (There
COASTALN. CAROIJNA a gap in the middle Atlantic region at Philadelphia.) Such a
CHARLESTON
,ilttemis suggestive of diffusion. Thus we find the eminent Amer-
COASTAL S. CAROIJNA
an dialectologist Hans Kurath writing. 'the so-called "r-Iess" type
GEORGIA SAVANNAH \.spreading as a prestige pronunciation from the old cultural
ntres within these areas. a process that in all probability has been
 152 Language
andsociety Rhoticity 153
in progress íor generations' (Kurath, 1965, in Williamsonau,
Burke, 1971: 105). .
.
,
Maps like Figure 5.4 are very frustrating because they are highi:i .,
,
,,
,,
ideaIized. This is because, as the results oí research conducted fo ,, I
, ,, ,,
linguisticatIases,they are largelybasedon the geographicaldis '
,: : Maine
bution oí variants. It ought to be obvious by now that space is onL
one dimension against which variants can be plotted. Intem
r
,
.,
,,
I
I
.'
,,
I
-,
, I
linguistic, and a multiplicity oí extemal social dimensions cou ;" , ,I
,,
tion variation - so the resulting structures are multi-dimensiona1':
Without quantitative methods and social correlations we cann t
observe these patterns. So the maps represent a static yes/no st....I
~
,'N.
\\
\

:' Hampshire,---
.1 ,
ation which hides social variation, and the processes that ha~1
happened and are happening in time. t
We find another important dialectologist, Raven McDavid, whf..
worked on the LinguisticAtlas 01the SouthAtlantic States(1947
commenting,
The conventional statement about the Southem postvocalic r is tl:t
it does not occur as constriction .'.. The fact that in every Southe
state one may find locally rootedllative speakers with constriction'
at least some of the words has been either overlooked or deliberate

'f FIgure 5.5 Malo r-pronounclng areas (shaded) oí eaStem New Bngland. clrca
'
"

,L ignored. 1930s (afterBloch.1939) .

It seems that, in íact, the íeature was variable íor at least som<
t
" 1T",i' speakers in the r-Iess areas. 1
('
:ill " ,1 t : I
1-. 'continuous relic areas, as we n'Oted above. Figure 5.5 shows
"", ",.

The sources of r-Iessness in America I number oí íairly self-contained communities bypassed by the
, lre~doí non-rhoticity. These included the islands off the coast.
The shaded areas in Figures 5.5 and 5.6 show the diStribUti i.,
.

.1'J¡¡.
.
...

oí rhoticity in the 1930s and 1940s in New Bngland and Sou ~

e shall have occasion to look at one oíthese, Martha's Vineyard,
. ater on.) There were signs in these relic areas and in certain parts
l.

.ijl: Carolina respectively. These two areas have in common the faD,~~ l.. f the boundary between the two types that younger speakers
!'
that r-Iessness seems to radiate outwards from the cultural centrl ireíerred the Boston pronunciation. This would indicate that loss
:; (
~
oí Boston and Charleston.
il; .'
:Qfr was still spreading.
jIi!.
.. . Bemard Bloch. writing in 1939, and drawing on the records, . (:Writing in 1947, Raven McDavidargues, in a classic art-
t' .
..
..~
"
,
the LinguisticAtlas 01New Bngland.reported the highest fre<iuelt, 'de, that a purely geographical account oí r~pronunciation in
'Itt
~.
.:f~ cies oí r-pronunciation in the westem part oí New Bngland. In outh Carolinacould not account for the íacts oí its distribution
t;.
. . iJ r
east, íocuSing on the Boston area, loss oí r predominated. This h,
.
, ~t
~
I .tthat date. Traditional explanatióhs of geographical data tend
' spread to New Hampshire and Maine. One sure sign that o -. be in terms oí the historical pattem of settIement. Note, íor ex-
i:~.' .tI.! is dealing with the diffusion oí an innovation is the existence
1'~fUIf'
pie, the 'mountain' speech area in Figure 5.4. The accents oí
¡
~4~~ ~:
r.r
..;)
"
 154 Languageandsociety
Figure 5.6 South Carolina: the spread ofthe plantatlon system. Shaded areas
are those where r was found In 'worm', 'father', 'bam' and 'bread', clrca
1940s (after McDavid. 1947)
the Appalachian mountains differ markedly from those of the';~:j
coastal south. This can be explained in terms of the speech of thet;
original settlers. It was Scotch-Irish (Ulster) people who mst settled\:
the mountains, and they migrated to that area from Pennsy.,
lvania. Their presence therefore explains this southward prong o~"
rhotic midland type speech.We have a reason to dividesouthe
speech into 'hill type' (rhotic) and 'plantation type' (non-rhotié~
(Stephenson, 1977: 75). f tige norm. Is the source of American r-Iessness a diffusion of the
McDavidargued that complexity of distribution in the 1940s1
could not be explained so simply. The coastal areas settled b
speakers from England, who therefore might be expected to b1:i
both rhotic and non-rhotic, was very small compared to the:
extent of contemporary r-lessness. It was, indeed, the case thl\t"
Scotch-Irisl;1had settled the two areas where rhoticity was dense~tf1
(see Figure 5.6), but both rhoticity and non-rhoticity occurred
areas where the other was expected.
Rhoticity 155
" This required a social explanation. McDavid noted that rhoticity
1" wasassociatedwith lower levelsof education,older speakers,and
rural, as opposedto urban, speech.Bis conclusion was that, in
( i SouthCarolina, r-pronunciation was a markofcultu@lisolation,
..,Variouscommunities thehistoricallydominantsocial
wereoutSide
pattem of the state. Conversely,r-Iessspeechwas associatedwith
the dominant pattem, which was the plantation system of agri-
culture, and the planter caste, focused on the culturally and com-
mercially dominant city of Charleston. Bis argument was that
r-Iessness had diffused outwards from Charleston along with the
plantation system, and had, as its social meaning, the prestige
of that city and the social system of which it was the focus.
Constriction had survived among 'poor white' speakers outside
the system. These were people living in areas wisuitable for
. plantation agriculture. .
~:' Figure 5.6 shows this quite_clI;~ly,by
.' . illustrating that r-Iess
~ areas largely coincide with those into which plantations had'
~ spread, as evidenced by black majorities in 1790 and 1860. It
¡ follows,for example, that those 'poor whites' would be those also
L most 'threatened' by blacks. Those most hostile to blacks would
: therefore be rhotic. McDavid remarks, 'It is also worthy of note
~. that almost every lynching in South Carolina in the last twenty-
,;1five years [up to 1947] occurred in counties where field work
~ for the South AtlanticAtlas has dis«losedstrong constriction of
postvocalic r.'
McDavidalso pointed out the closeties of the Charlestonelite
WithEngland, both before and after the American revolution, The
obvious question then is the relation~hip between the r-Iess areas
of the American eastem seaboard and the southem British pres-
feature across the Atlantic, or were the original settlers in these
coastal regions r-Iess?When did r-Iessnessgain prestige in America?
Or was it a prestige form in America from the beginnirig?
.~ t In order to answer these questions one haS to go back and ask
when r-Iess pronunciation firs~ appeared in England, and more
particularly when it became the prestige form in London and south-
~ east England. 1have included a note in a separate box below sum-
:marizing some arguments pebple have used to try to establish these
 156 Languageandsociety
dates, because 1think the methods and evidenceused are intrins- ,1
ically fascinating. (Ifwe are going to look at r, we might as well go"
'whole hog', right back to the beginnings of the variants.) Fo~,
those that want to skip this, however, we can conclude that both,)
forms very probably crossed the Atlantic as 'foIk' pronunciations:'
But we can also say that the period at which loss of postvocalic ri,
became a feature of the English prestige norm is quite late. This;
suggests that loss of r as a prestige form was, in all probability, ~I
diffused from British English to American English via the coastaL¡'
centres, and from there to the hinterland, as Kurath suggested.
The Kurath-McDavid theorytells the following story. PostvocaliC¡
r would have come to America, at least latterly, as a variable fea: '.,
ture. The original settlement pattems would include both rhotio,
and non-rhotic speakers from England, reflecting the state of the1
loss of r in the part of England from which they came. (Of course,)j
all Scottish and Irish settlers, like those of the mountain regíon;;
would be rhotic.) On our dating, the weight of accents would be.t
towards r-full speech the furtner back one went in time. The relip.,\
areas tell us this. r .
The r-less form acquired prestige in America, when it was
adopted from Britain by the elite classes of New England and the.
Southem aristocracy. Kurath points out the transmission mech-iil
anism: the very close ties between these groups and southerri¡
England. The planters and merchants of the South, especially, haili
their children educated at the English universities or by importe
English tutors, and were accepted as equals in London society. In~)
other words, there was a single culture and a dense network o
communication between the elites of New England, the Sou
and southem England. The change then diffused downwards ani
outwards from the coastal centres as a new 'prestige' formo W ~1'placeswhere r fluctuates in rhotic accents without indicating a gen-
saw how this process had worked itself out in South Carolina. Bu
on the borderlines, the feature remained variable.
An altemative picture is presented by both Lass (1987: 282-3\i
and Romaine (1994: 146). They argue that first settlement playedj
a crucial role in the distribution of rhoticity in eastem No
America. Lass asserts that 'the earliest colonies are in fact no
rhotic, and seem to have been at least partly so at the time o,
settlement; the history suggests that rhoticity may be a kind ofi
. Rhoticity 157
\:1
The originalloss of postvoca1icr: a note
Barly loss of r. There seems to be agreement that r-Iess speech
became the standard of London speech in the late eighteenth
century. But some writers have suggesteá'earlier dates. Archibald
Hil1(1940) describes a loss of ras early as the fourteenth century.
But this is in a single restricted environment: in stressed syllables
before just those consonants which are made by the tongue-tip and
the teeth. It looks as if r was assimilated to the dental sound which
followed it in words like 'bam' "áDd'bird' at a very early periodo
' :;,
A second view of early loss of r is that of Wyld (1920) and
Jesperson (1954). They argue that there was a very general 'weak-
ening' in the pronunciation of r from the fifteenth century. Jesperson
notes a very general movement from a trilled to a continuant r
over a number of centuries. In the seventeenth century we find
Ben Jonson writing that r was 'sounded firm in the beginning of
the words, and more liquid in middle and ends' (1639). Wyld cites
-
I~,written forms like Woseter- Worcester,and Dasset Dorset,froma
~. sixteenth-century diarist of humble origins, Henry Machyn. Sim-
-
Ik llarly, Kurath (1965) po~ts to spellings like libity liberty, and
~i .
-
,.. patchis purchase, as evidence that some American colonists were
.
'.., r-Iess, even by 1700.
~' Wyld paints the followingplcture. Th~ loss of r began in the
~: east of England by the middle of the fifteenth century, especially
Ir' before s. By the middle of the sixteenth century it had extended
,el"both to other consonants, and to the ~ndon vemacular. (The forms
. '"used by Henry Machyn are an example.) By 1650 it had diffused
)'1..
upwards to London society. In this case, both pronunciations would
V go to America, but r-Iess speech would go as a prestige formo
I~~Bighteenth-centuryorigins.There are reasons for thinking these dates
~!)'aretoo early. Stephenson (1977) poillts out that there are three
l era110ssof the sound. These are .
1. in unstressed syllables: 'adve(r)tise', 'su(r)prise' etc.
2. before a following denta1:""'ho(r)se','nu(r)se', 'me(r)cy' etc.
3. when two rs appear in the same word: 'co(r)poration',
'fa(r)ther' etc.
. These are all reported in rhotic speech. And most of the evidence
~for early
loss is based on cases like these; for example, Machyn's
},spellings above.
 158 Languageand society
Actual comments on r are more valuable, and these do not start
appearing until the eighteenth' century. We find Walker referring
toLondon speech in 1775 and 1791. He says that in England and
particularly in London words like 'bard', 'card' and 'regard', are
pronounced as 'baad', 'caad' and 'regaad'. Even more revealing is
Hill (1821), who writes that r
ought more carefully 10 be preserved for posterity, than can be
hoped, if the provincialists of the Metropolis and their tasteless
imitators are 10 be tolerated in such rhymes as Jawn and morn,
straw andJor, grass andJarce, etc. etc. 10the end ofthe reader's
patience.
Por Hill in 1821 this is clearly a new innovation as a prestige
pronunciation. r-lessness is thus not probably part of a prestige
norm till at least late in the eighteenth century. It clearly would
have been variable at this time and earlier. In this case, r-less speech
would not have originally gone 10 ~erica as a prestige formo
"secondary archaism" - Le. a larer import by speakers of moré';;
conservative Englishes (largely Ulster Scots or Irish . . . ) . . . whichi;
had no r-Ioss at all'. The r-Iessness of North America is original;:;
because eastem New England and the coastal areas from the Vir~
ginia tidewater as far south as the Garolinas were first settled fro~í
the south-east area of England where r-Iessness was most ad,
vanced. Westem New England and the mountainous interior wer~\,
settled later by fully rhotic Scots-Irish and people from northe, .'
and westem parts of Britain which would retain more r. Dillar,
(1992) is also sceptical about the claim that r-Iessness gained itl,
prestigefrom across the Atlantic.
A possible compromise view is that the Kurath-McDavid pic-l;
t 8/focus, the r-full variant from the interior. As far as r was con-
ture is correct to the extent that, if we accept a late date r-Iessne8¿~.
as a prestige feature in southem England, any r-variability of th~;.
earliest coastal colonial Englishwould have been at least COnfirme41 ' "'~
towards continuing r-Ioss by its status in England. It would havl .
been mobile persons of the travelled upper strata of American urbaq; ,
and plantation society who would have been the force behind
ditTusion.Where the feature was not variable because the origin
settlers were Scots-Irish and where the plantation-mountain or coaStt'"'
frontier cultural ditTerencesintervened, the ditTusionwas stopped~
Rhoticity 159
There is no doubt that non-rhoticity had general prestige across
. the coastal south and the are as'populated from it, for example, the
gulf states from Alabama to East Texas. Evidence for this is the
80rtofhyper-r-Iessness one sometimes finds in the Southem states,
which drops r in places where it is not normally dropped in Britain.
Some Southem accents, for example, have not only no intrusive r,
but no linking r. In some non-standard varieties, r between two
vowels, as in 'hurry', is dropped and 'Paris' can be identical to
'pass' (Wells, 1982: 544). Such features suggest attempts by suc-
cessive groups to approximate to a prestige innovation, pushing
, the change further along.
A reversal of the pattero _ h..
Much evidence has been accumulated recently that a major
'. change is taking place with regard torhoticity in the r-Iess speech
communities we have been discussing in the eastem United States.
We have been looking at the spread of r-Iess speech focused on the
coastal centres. There is evidence also of the spread of rhoticity
back into these enclaves, as if two contrasting waves of preference
were meeting each other.
Thus, Bloch (1939) noted that on the boundary between the
two forms in westem New England and in the Connecticut River
valIeyr-pronunciation was becoming more general. It was spread-
ing through westem Connecticut from the rhotic areas in the north
\\ central USA. In other words, the tWo types were 'spreading vigor-
f;ously from opposite centres': the r-Iess variant from the Boston
ocerned, New England was not a single dialect area and the speech
. ofmany individuals was variable. When Boston was re-examined
.
. by Parslow in 1967, he found that r was being widely re-introduced
even into the speech ofthe city. Comparison ofhis study with that
. of the Atlas survey of 1939-43, on which Bloch's artalysis was
based, 'demonstrates a steady progression to r-timbre for all regions
¡.and sociallevels' (Parslow, 1971: 622).
, In 1947, McDavid had noted that there were some slight hints
of a reversal of the trend in pre~!~gevalues in South Carolina;
that rhoticity, formerly associated with social peripherality, might
 160 Languageandsociety
possibly become respectable. He remarked that, even in the low
country, some girls in their late teens and early twenties were .
sporting newly acquired rs. By the time of O'Cain's study of .
Charleston in 1972, however, r had reappearedin the city itself. ;
McDavid and O'Cain (1977) write, 'Postvocalic r has advanced at
arate that surpasses almost every innovation in Charleston speech.,
Only aristocrats and older whites of other classes consistently ¡
approach fully r-Iessspeech.' In other words, it is variable for all
other groups. McDavid (1975) notes that this feature, formerly
associated with Southem poor white speech, seems to have be-
come part of the regional standard, although it is not yet used by
all standard speakers.
In 1966, Levine and Crockett, in a classic piece of research, ,
studied r-pronunciation in Hillsboro, North Carolina. This is a pied-:.
mont community at the westem edge of the coastal plain, not.
far from the Virginia border. The>community is at the confluence
of several dialect areas and near the boundary of midland and;
Southem speech types. They woridered if 'inhabitants spoke sorne'
"transitional" dialect, or, instead, one or more of the nearby dia-
lects in relatively unmixed form' (Levine and Crockett, 1966: 77).
Theyfoundtwo r-pronouncingnormswithin the community.But "
the community was not regularly stratified, from higher to lower,
according to frequency of r-pronunciation. High-status speakers
were associated either with the r-full norm, or the r-Iessnorm.
Low-status speakers fell betweenthe two norms. Levine and Crocketti
concluded thát it was the 'clarity' or 'strength' of the norm thatt
was associated with social position. So both norms could b~;::
modelsof prestige speech in Hillsboro.
The interesting point here was the direction of change observedl!~
In more formal styles of pronunciation, r was more frequent, and[a
it was also more frequent among the young. There was evidencd
that older people, men, blue-collar workers, and those who ha~
resided in the community longer favoured r-Iessness. YoungeIj
speakers, women, short-term residents, and those near the top oé
the 'white-collar' class (but not at the top of this class) favoured,
rhoticity in that they conceived it to be 'correct'. All this led to th~
conclusion that the community was transitional, and moving
towards rhoticity under the pressure of outside norms. If this tS
Rhoticity 161
right, then rhotic General American would become the prestige
norm for this community.
There is strong evidence thát rhoticity in America is spread-
:lng in white speech throughout many historically r-Iess areas in
. 'the south and northeast. Agfe~ing with this diagnosis in 1989,
~:Butters(1989: 37) quotes the res~~~.that r-Iessness lS gradually
,'disappearing among Southem whites. This is an extraordinary
~ i reversal of a historic pattem.
New York City
Labov comments that a New Yorker' s overt attitudes to his or
~,
iher own vemacular speech pattems are extremely negative; that
htis a 'sink ofnegative prestige'. Vemacular New York City features
have been stigmatized. Consider the~~Brooklynese' stereotype of
working-class New York speech: This~an be the subject of humour.
Toitydoityboidssittin' on de koib.dioipin' and, boipin'an eatin' doity
woims
Thirty dirty birds sitting on the kerb, chirping'and burping and eating
dirty worms.
In fact this 'toity-toid street' diphthong has attracted such stigma
at it is rapidly disappearing from the speech of the city, and now
'hrvives only among lower-class speakets. If the British reader
'ants to get some general idea of how the New York City ver-
~cularsounds,he can think ofArchieBUnkerin A11in theFamily.
..Perhaps connected with this general stigmatization is the
'M-contained nature of the New -York City speech community.
.~tropolitan speech pattems do not extend further than the outer
burbs. In pure geographical terms, it is by far the smallest of the
)
)eech areas on the map in Figure 5.4. In terms of r-Iessness, it
'ecame a non-rhotic island surrounded by rhoticity. Contrast this
'.th the extent of the spread of 'plantation' type speech in the
.outh. Stephenson (1977) reports r-Iess speech as characteristic
~fthe 'cultivated' classes in much of the South. Reporting on San
tonio, Texas, Sawyer (1959) notes the variability ofretroflexion.
 162 Languageandsociety
R-Iess speech characterized older informants. Even here, howeveri
r-Iess speech showed signs of decline. Middle-aged informants!
pronounceq r half the time or more, and the young educated in-~
formants were fulIy rhotic. d
HistoricalIy, New York City has folIowed the same pattem aS,~
the other coastal centres. An r-fulI area in the eighteenth century,C
it had become r-Iess by the end of the nineteenth century, unde
the influence of eastern New England and southern British normsJ:
Berger (1980) also points out a strong maritime connection wiili:
the South.
Labov (1972) writes thus about the adoption of r-Iess speech:
It seems to be one of our best examplesof a 'change from above'-
originating in the highest social group - which eventually spread,
the entire speech community and became the vemacular formo01.
first documented evidence for r-Iess proimnciation in New York eí¡,
dates to the middle of the nineteenth century; Richard Norman h~
observed that the New York poet Prederick Cozzens rhymed 'shore,
and 'pshaw' in 1856. Babbitt's study of 1896 was the first linguistid
report, and it showed tbat the r-Iess speech was the regular vemacu-~
lar pattem ofthe city. Babbitt's report as well as Linguistic Atlas inter~
views of the 1930s show a completely r-Iess dialecto
By the time ofHubbelI's study in 1950 and Labov's own stu4
in 1966, the situation had changed substantially. In the speec
of any given New Yorker in the area which Labov studied, tqE'",$
Iower east side of Manhattan, there was a great deal of apparen~i..
random fIuctuation between the production of constriction ancli,t .
absence. The two forms appeared to be variable. They were fre~~~'
substitutable for each other in the same environments. This w.J
true for the various instances where r would be possible durÍng t. '
a single utterance, as welI as between different utterances by tJ'
same speaker and, of course, between speakers. The speech be
of individuals and of the community was inconsistent. Hub~e
. . (
(1950) had assessed New Yorkers' usage as folIows (quoted ~~
Labov, 1966; 36):
The pronunciation of a very large number of New Yorkers exhibí
pattem ... that might most accurately be described as the comple}
absence of any pattem. Such speakers sometimes pronounce Irl befo:.
Rhoticity 1~3
1.
a consonant or in finalpositionand sometimesomitit, in a thorougWy
haphazard fashion.
l.-
Tbe department stores
Labov's work in New York City includes one of the best ex-
i amples of anonymous observation which we have. You will recall
'-that this was one of th~ techniques of investigation mentioned in
: chapter four. One of the'w~~_ of overcoming the observer's para-
,dox was to engineer a situation in which people's speech could be
í observed without their knowing it. Labov (1966) was able to do
this in three New York City department stores. The suspicion he
was working on was that rhoticity was peing reintroduced into
'.¡. the city as a prestige feature. Labov (1966: 64) predicted: 'If any
'two subgroups ofNew York City''SPeakersare ranked in a scale of
social stratification, they will be ranked in the same order by tbeir \
differential use of (r).'
The groups of speakers in question were the salespeople of three
Iarge department stores in Manhattan. Labov, using a series of
quite objective criteria, ranked the ~tores in the folIowing order:
, (1) Saks Fifth Avenue; (2) Macy's; (3) S. Klein. He argued that jobs
in the three stores would be socialI¡ evaluated in the same order,
¡if only in terms of the working conditions and relative prestige of
¡. the stores.
Now, how was he to observe thesalespeople pronouncing or
dropping their rs? The technique was both ingenious and amus-
" ing. He chose a department which was located on the fourth fIoor
ofeach store, and asked salespeople, 'Excuse me, where are the-?', \
'which would, of course, elicit, 'Fourth Floor', as a reply. The
investigator would then lean forward and say, 'Excuse me?' This
'"wouldnormalIy causethe unwittíng inf0rmant to repeat, in a more
. careful and emphatic way, 'Fourth fIoor.' In this way, Labov was
t able to get four instañces ofpostvocalic r, in two contrasting styles
"
and two linguistic environments, for each informant.
!" The results are below. And they are as predicted. The sales per-
( sonnel, ranked according to the three stores, could also be ranked
by their differential use of r. The overalI stratification of (r) by store
",
 1b4 Languageandsociety Rhoticity 165
is Saks 62 per cent, Macy's 51 per cent, and Klein's 21 per cent of 80 ,6-8
1'8pronounced. ""
"
Saks Macy's S.KIein v'. .... ""
"
"
"
i!

,
:;111 ¡;,," r9 ~ 60
/ ...9
,,4-5
;1..
~:
,
,1
~'I
\

11
EJ ~ ~ l'
I
I
I

" ...
..."
,,'" 2-3
(
':UJ,I.' ,~, The bottom figure in each column is the percentag~ of alI r-11 I
1,./
/"
/ /
",,/
" /
/ ,,1
J'1 l." " / " "
" "" " ,," / O
pronunciation; the top figure is the percentage of some r-pronun- :~: (r) 40
Index ..,
j ciation (Labov, 1966: 73). ~; ,,/, " / / I
(J¡i I ... '"" " " I
But the results were even more fine-grained than this. There I "'" / I
'" '" " / /
was style-shifting. In each store, the amount of rhoticity was '" ,;
" ,,/ /
greater in the more emphatic style. And, amazingly, there was a '"
20 " '"
~
~
'"
I~
diJIerence in (r) scores inside the Saks store itself. On the quieter\ ...
...
/
/

and more expensive upper flooJs of Saks, the percentage of all or'I¡j _/
some (r) was 74 per cent. In the hustle and bustle of the ground ,~!
",-",,,,-
floor the figure was 46 per cent.
O
I Sociolinguistic structure of (r)
Casual CarefuI Reading Word lists Minimal pairs
-j 0-1 = Lower class
Figure 5.7 displays the stratification of (r) by class and style
as revealed by further sociolinguistic interviews in Labov's study, 2-5 = Working class
6-8 =Lower middle class I J\
of the lower east side. The (r) index score is the percentage of!l~
postvocalic 1'8used out of number of times r couldhave been used,'.1 9 = Upper middle class
The general shape of the diagrarn is familiar to us from our studl' Figure 5.7 (r) In New York Clty (from Labov, 1966)
of the (ing) variable in chapter four.
But the structure of the (r) variable is diJIerent. On the dimen~ Labov calls this phenomenon hypercorrection. In this case,
sion of social class, each class is diJIerentiated from the other. Oq we observe hypercorrect behaviour by the lower middle class. Earl-
the dimension of style, there is a clear style-shift towards more ier on, when we were discussing the 'sharp stratification' of the
rhoticity for all social classes. This confirms that r-full speech ,' .~ Norwich (ing), in Figure 4.3, these 'borderline' classes also carne
now the prestigeformin New~ork City.The difIerenceis t4~
there is a cross-over pattem in the more formal styles. In casu ~ 1 9
,.
to our attention. In the case of (ing), howe~er, we had a stable and
regular structure, without hypercorrection. We will be making
generalizations about irregular structures later on. Suffice to say
speech only the upper middle class really has any significan,'
rhoticity. However, in the more formal styles, the r-score for the at this point that both hypercorrection and irregular structures are
lower middle class (classes 6-8) rapidly increases and, crossing.. l diagnostic oflinguistic change in progress. In particular, hypercor-
rection by the lower middle class tells us in this case that a new
over, is higher than that ofthe upper middle class (class 9) in tb
most formal styles. prestige norm is entering the community.
V)
 166 Languageand society Rhoticity 167
Let us see exactly what we find out about informants in the is really only a feature of the upper middle class. Here are the
most formal styles. Remember how these styles were defined in average (r) scores in casual speech by age and class:
chapter four. They were designed to produce relatively more self-
consciousattention to speech. They therefore measure, not what Age Lower Lower Middle
Working Upper Middle
people's vernacular is like when they are being casual and not
paying attention to how they speak, but rather what Labov (1966: 8-19 00 01 00 48
241) calls their phonic intention. They tell us the norms ofthe 20-29 00 00 00 35
speakers, rather than their everyday performance. The dotted lines 30-39 00 00 ...... 60 32
in Figure 5.7 represent these norms. 40-49 00 06 10 18
50+ 00 08 00 05
We can see this in a concrete way by looking at what the in-
formants were asked to do in the more formal styles. Here is a part
of a paragraph they were asked to read, which had the (r) variable Since 00 means a complete absence of rhoticity, these figures
concentrated in it (of course, they weren't told this). tell us that New York City, in its everyday styles, is very largely an
1remember where he was run over, not far froin our comer. He darted r-Iesscommunity. It is only the upper middle class, who are intro-
out about four feet before a car and he got hit hard etc. ducing the change, who have respectable amounts of constriction.
But note how, in the upper middle class, there is a steady de-
The most formal style involved judgements about 'minimal crease in r-scores as the informants geJ;oIder. We said earlier that
pairs'. People were asked, at the end of the interview, whether averagescoresfor classesaild stylesconceaIsignificantpatterns iñ
pairs ofwords soundedthe same or not; forexample, 'guard-god', variation on other dimensions. This is the case With age in New
'dock-dark' and 'source:::sauce'. At this stage, they are most self- York City. Speakers over forty'náve much Iower amounts of con-
conscious about Ianguage, and furthest removed from the auto- striction than do their youngercounterparts. Speakers over fifty
matic motor-production which characterizes their vernacular. All have virtually none. This suggests th~t the change entered the
their attention is focused on a pair of words which will not be city, in the upper middle classes, in the
S\
1940s.
homophones ifthe individual is rhotic, and will, if one isn't, What But notice there is no age-grading in the other classes. Rhoticity,
we observe, therefore, is precisely how people believe these words as a practice, has not reached them. But it has as a norm. We can
ought to be pronounced in the formal situation of an interview. It see this from hypercorrection and style-shifting. The younger
is the relative differences between the styles that matter. And, from
this, it seems that r-fullness was the prestige norm in New York 80 I Careful
City by the early 1960s.

60120-2~ ..,
How rhoticity is entering New York City
Yesl The norm but not the practice. Labov's hypothesis is that
rhoticíty is being introduced into the New York City speech com-
munity by the highest status group, more or less consciously, as a
new prestige formoIt replaces one borrowed norm, the r-Iess one,
40ln 3n~
20
- 40+
20-29
/
30-39
40+

with a new borrowed norm, the rhotic one; and thus reverses the o
prestige relations (Labov, 1972: 290). In casual speech, rhoticity Upper middle "
 ~

168 Languageand society Rhoticity 169

Casual Now look at the columns f6r the lower middle classin the word-

601
list style,when theyázoeself-consciousabout how they speak.The
40 age 20-29 Upper columns cross the dotted line. The shadedpart of each col.umn
(r) age30-39 :
, 1
middle .. representshypercorrect speech.What is clear is that olderlower-
index middle-class (and oldest working-class) speakers are the ones who
20 I hypercorrect. They are giving evidence, in doing so, that they
age 50-75
acceptthe new norro introduced by the younger speakers of the,
t=
O u u u u u u u u u u
class above them.
Si: ~ Si: ~ ~ ;¡: ~ Labov (1972: 136f.) has tentatively suggested that this pattern
serves to accelerate the process of change. By hypercorrecting,
80T Wordlists lower-middle-class speakers pro\Tidea model ()fthe new norro for
their own children. This l~~gs to its more rapid difIusion down-
60 age 30-39
........-..-....
wards through successive classes, iIi successive generations.
age 40-49
I
I
I
, We now have a precise picture of how an innovation may be
(r) age 50-75 cumu introduced by higher-status gro~ps into a stratified society, and a
index 40 , possiblerole for hypercorrection in its difIusion. We will be return-
,. ing to these matters later. But noteohow, in New York City, Labov's
20 quantitative methodology has made visible the way in which
( rhoticity isbeingreintroducedintothat community.Contrastthis,
( with the frustrating and very partial accounts which we have had
O inthe earlier part of our study of nthose done before or outside of
u u u u u u u ~ u u u u
~ Si: ~ ~ Si: ~ ~ ~ ~ ~ Si: ~ " the new quantitativeparadigm.
.1\

Figure 5.8 Upper middle compared 10 other classes by age group in two styles",
New York City (from lAbov. 1966)
upper middle class is the 'reference group' for this new norm. In;
fact, it is the older lower-middle-class speakers who hypercorrect;!
those over forty. And SO,for this class, average scores rise by age\
because they are using the younger speakers of the class abov.
them as their modelof correct speech.
This pattern only really becomes clear if you look at the tw0iJ.
diagrams in Figure5;S. The dotted line in each diagram is th~.
upper-middle-class score for each age group. In both styles, thi~
line steps down to the left. It is clear that rhoticity is a property of
younger upper-middle-class speakers. They are introducing tb _
new forro., " Maynor (1989, cited in Butters, 1989) argue, on the basis of con-
The columns in the two diagrams in Figure 5.8 represent tb,
average index scores of the other classes, also for each age group
, Black and white
, To conclude, we willlook at one more variety of speech.The
1,language of black people has beep the subject of research and
~controversy in America since the 1960s. We will be looking at
~Black English in more detall in chapter six. But for the moment
'
\\vhat can we say about r in black speech?
,The original home of the bléickvernacular is on the plantations'
of the south. This means that black speech is historically r-Iess.
'
So that is the norro from which we can examine black rhoticity
itoday. We saw that there seems to be a general change in progress
:whereby southern whi~e speakers are introducing r into their
"accents.Are black people participating in this change? Bailey and
1¡trasts between rates of r-Ioss among white and black childreñln
the Brazos VaUey of Texas, that the southern black community
 170 Languageand society Rhoticity 171
has not participated in the increasing rhoticity. For example, while 100
black children pronounce r 80 per cent of the time in words like
bird,white children do not pronounce it at all in that environment
and white adults have a 69 per cent score. Southem black speech
I
remains largely r-Iess. Bailey and Maynor deploy this data to sup-
port their argument that the differencebetween the speech ofblacks
and whites is increasing, as part of a wider debate about whether I
the two varieties are diverging or converging. We will examine
this debate in the next chapter. For now, suffice to say that Butters
(1989: 13-14) contests any notion of'divergence' by pointing out I
that black speech has always been r-Iess (and some 'mountain-
type' southem white accents have always been rhotic) so that any' I
new divergencecannot be at issue. Nevertheless, a change is taking
place in which blacks appearnot to be participating. But is that
true?
I
Butters (1989: 4lf.) goes on to present evidence for the oppos- I
ite. He shows that in other co~unities where there has been a I
recognized increase in ~hite rhoticity, blacks have also become '~ I
more r-full, but at a slower rateo We revisit Hillsboro, North Caro-
lina, where in 1966 Levine and Crockett found increases in white
r-pronunciation. Working in 1969-70, Anshen discovered that
blacks were also becoming variably rhotic. In Wilmington, North "

Carolina, Butters reports that his own research revealed that '
younger black speakers were style-shifting towards r in formal
contexts. We conclude that the increasing use of r, the new pres-
tige norm, is also affecting black speakers in these two commun-
ities. Butters also remarks that the normative pressure appears to
be coming from education. Rhoticity was being treated as 'correctt
in schools; starting 'in the 1930s, Hillsboro schools began stress.
ing the pronunciation of orthographic r' and this might itself'
produce style-shiftingwithout effectinga real change in the ver- ,

nacular. But both blacks and whites appear to be susceptible to

the correction.
Let's now turn our attention further northward and look at'
Wolfram's 1969 study ofblack speech in Detroit. In contrast wiili:;
New York City, r is not a variable feature in the pronunciation oí¡,
the white community in Detroit, a city which is in the north cen(7J~
ral speech area ofthe USA,and, as Figure 5.4 shows, is categOricallY,~
! rhotic. However, the feature is variable for the black community
, in Detroit.
The toP diagram in Figure 5.9 shows the percentage of times
, postvocalic r is absent in the speech of Black Detroit for four social'
, classes.Por comparison, the average score is also given for a~group
 172 Languageandsociety
of upper-middle-class white speakers. Variability of r only really
characterizes the black ethnic group. The two main groups in'l
American society whose speech is r-Iess are whites of the coastal
enclaves and speakers of the black English vemacular. These two
groups are not unconnected.
In the 1960s, the majority of black Michigan residents were:1J
not bom in the state. Olderpersons were intemal migrants from
r-Iess areas of the American South, and younger persons were
overwhelmingly the children of recent arrivals. The parents oí
53 per cent of Wolfram's informants were bom in the three deep
Southem states of Mississippi, Alabama and Georgia. Another "
21 per cent were bom in other Southem states. Thus a linguistic
boundary coincident with an ethnic boundary exists in Detroit.
But is is not a full explanation to say that r-lessness among Detroit
blacks was caused by the movement of people who speak a non- :
rhotic vemacular into a northem city in which the basic pattem i'
isrhotic. As Wolfram (1969: 25) points out, 'One of the most;
important factors accounting for speech behaviour in the Detroit,J
Negro population is racial isolation. Pattems of Northem segrega-€
tion are a main source fpr transforming many Southem speech
characteristics into ethnic and class patterns of speech in Northemi"
cities.' In fact, class, sex, age and racial isolation interact in com-"
plex ways. The migration of black people into inner cities in the,?f
north - the inner city of Detroit has very high densities of blackl!
-
peopleas a percentageofthe population has 100to the emergence"
of a 'uniform caste dialect - the black English vemacular ofHarlem " community.Those with 'more white°friends'versus 'fewer white
and other inner cities' (Labov, 1972: 299). Labov sees this as part.f
of a general tendency connected with worldwide urbanization in., '~
the past several centuries: the transformation oflocal dialects into,'~
language varieties defined in terms of stratification. Innovationlf
emerge that reflect the processes of urbanization.
Theseprocesses can lead to vemacular innovations away from'.
the norm of the 'hast' community. But with respect to r, the degred
of convergence in Detroit seemed to depend on class difference~l'
within the black community. The contact was complicated. An'
there were signs that the black working-class vemacular wasn't;
changing towards white norms. Wolfram's study showed tha~t
within the black community r reflects social class differencesamon~\~ '
.111
, 'guistic tralts are the result of real face-to-face contact,n9t the
Rhoticity 173
. blacks.Figure 5.9 shows the str~~cation of r absence by social
.'class.Each class increases its amount of rhoticity as one moves up
the social scale. The bottom diagram in Figure 5.9 shows the style-
shiftbetween careful and reading styles. We see that this is also in I
.:thedirection of rhoticity, but no hypercorrection is visible. Wolf-
ram's data on age, which show no ~ge-grading for r, suggest that
"
~working-classadults are persistently retaining a non-rhotic South- .1
fem norm (Wolfram, 1969: 118). However, the picture that emerges
'
o for r in terms of class and style is one in which, for the black
I
~ddle class at any rate, there has been an acceptance of the norm I
;;ofGeneral American. Clearly what has happened is that the pro-
:nunciationofthe host communityhas becomethe formwith overt
prestige within the black ethnic ogroup's upper strata. I
There are signs of linguistic adaptation in Detroit between I
:middle-classblacks and white rhoticity. Presumably these are the
~*)nostsocially mobile black people,'those who have the most diverse
I
~andintensive linguistic contact with whites.
00'We find evidence of the importance of such contact if we look.
!¡ltblack r pronunciation in two northeast cities, Philadelphia and
¡NewYork. Myhill (1988) studied postvocalic r in the black ver-
macular of Philadelphia. As in Detroit, white speech in that city
is rhotic (with the interesting exceptioÍl of working-class ethnic
, talians). Myhill found that there was an increasing convergence
between black and white speech.l'Blacks were becoming increas-
ingly rhotic depending on thelr-relationship to the white speech
:Jiiends'or 'more white lovers' versus'as many blackas whitelovers'
ipad significantly less r-Iessness. Commenting on this research,
~\ltters (1989: 46) writes that 'the original Irl deletion rule of
~lacks in Philadelphia .. . has been modified in the direction of r-
,'etention practices of the national standard and local Philadelphia
bite vemacular'. ~
b Butters takes the interesting further step of comparing Phila-
I'lelphiawith Labov's results for blacks in New York. As we saw,
'.e New York vemacular i.slargely r-Iess. Accordingly, New York
)ack speakers maintain a much higher degree of r-Iessness than
ose in Philadelphia. He uses-lliis to reinforce Labov' s point that
 \

174 Languageandsociety Rhoticity 175

standard norms experienced in school or the mass media. In Ne~
York, public middle class norms would be largely rhotic. But in'
New York, r-less blacks would have little opportunity of facet
face contact with a 'dominating' r-full white norm compared tú:
the experience of blacks in Philadelphia. The key fact is a con,.
,fortunes of r as a variable in the speech of coastal South Carolina.
¡Howcould this be explained? Ir a new prestige norm has replaced
~anolder one, we could look for the way institutions create and
'sustain prestige norms. Por example, in England, how is RP main-
tained as the prestige norm? The system of stratification is prob-

vergence of black speech to the rhotic norm in Philadelphia. Andi
this convergence depends on the degree of contact between blacIcl
and whites. So it is not the case, nationwide, that blacks are no"
participating in an increase of rhoticity. But they are doing so aE
a slower pace, depending on the 'dominating' dialect of each d,
and the opportunities of black contact with it.
Specificity of explanations
Our sketch ofrhoticity has been on a very wide canvas. We have:
, ,
o
,ably central 10 this; and the control of large-scale institutions by
~tbe upper strata of society, including education, public culture,
:spüe institutions and the media. But could that apply to r-less
{speech in Charleston? McDavid (1975) argues instead that the
'profound postwar changes in Southem society, including a great
,increasein educational and economic opportunity for the formerly
disadvantaged, has led to upward mobility by rhotic whites. Such
rpeoplehave modified the Charleston standard. Ir this is true, it
1.would be an example of a change introduced by a lower strata of
1~ociety. _

looked at the diffusion of the loss of postvocalic r, and the gain Qft
postvocalic r, in m~y speech communities: from Charleston, rt&
Edinburgh, to eighteenth-century London, to postwar New York
The inescapable efIects of social factors in the diffusion oflinguistiO
change were clearlyvisible.
One reason why it is possible10study r so widely is that tb,
': Whatever the case, this'explanation will clearly not do for either
;New York City or Detroit. The..point, then, is that postvocalic r is a
" ~.fferent sociolinguistic variable"m each of these communities. The
trhoticity or otherwise of speech means something different in dif-
o 'ferent speech communities.
o

innovations we have looked at have characteristically been introJ

duced by the highest status groups in the community. This was tbe:
first sort of normative pressure we discussed in the last chapterli
and is a reflex of stratification as characteristic social structure{
The processes involved are analogous, on the level of the single
variable feature, to those involved in standardization, as discusséd)
in chapter two. In.the next chapter, we shall see that social strati~
fication and its concomitant stratification of styles are just one sociál
dimension relevant 10 linguistic variation. 'f,
Even more important is the way in which r must be explaineq:
differently in different communities. In England, r-lessness is tb~1
prestige norm. In New York City, r-fullness is being introduced by;:
the highest ranked social group from outside the community. IhJ;
Edinburgh, r-less speech has appeared in one sex, in one sOciáJ}
and one age group. That looks like a vemacular innovation, whic
o"
probably will not spread because of the prestige rhotic norm 0_.
Edinburgh English. There seems 10 have been a reversal in tb~i