Journal of Sociolinguistics 15/2, 2011: 151–196
Contact, the feature pool
and the speech community: The emergence
of Multicultural London English
Jenny Cheshire,a Paul Kerswill,b Sue Foxa
and Eivind Torgersenc
a. Queen Mary University of London
b. Lancaster University, United Kingdom
c. Sør-Trøndelag University College, Norway
In the multilingual centres of Northern Europe’s major cities, new varieties of
the host languages are emerging. While some analyse these ‘multiethnolects’
as youth styles, we take a variationist approach to an emerging ‘Multicultural
London English’ (MLE), asking: (1) what features characterise MLE; (2) at
what age(s) are they acquired; (3) is MLE vernacularised; and (4) when did
MLE emerge, and what factors enabled this? We argue that innovations in
the diphthongs and the quotative system are generated from the specific
sociolinguistics of inner-city London, where at least half the population
is undergoing group second-language acquisition and where high linguistic
diversity leads to a heterogeneous feature pool to select from. We look
for incrementation (Labov 2001) in the acquisition of the features, but find
this only for two ‘global’ changes, BE LIKE and GOOSE-fronting, for which
adolescents show the highest usage. Community-internal factors explain the
age-related variation in the remaining features.
Dans certaines villes importantes de l’Europe de Nord, on observe l’émergence
de nouvelles variétés de la langue hôte dans les quartiers multilingues. Tandis
que certains ont analysé ces variétés multiethniques comme appartenant
au langage des jeunes, nous adoptons ici une approche variationniste
pour analyser une telle variété émergente à Londres, que nous appelons
Multicultural London English (MLE), en nous posant les questions suivantes:
(1) quels traits structuraux caractérisent le MLE; (2) à quel(s) âge(s) sont-ils
acquis; (3) le MLE est-il vernacularisé; et (4) à quel moment le MLE est-il
apparu, et quels facteurs lui ont permis à émerger? Nous soutenons que les
innovations dans les diphtongues et le système du discours rapporté résultent
de la situation sociolinguistique spécifique des quartiers multilingues et
déshérités de Londres, où au moins 50 pour cent de la population acquiert
l’anglais comme langue seconde au sauvage avec des amis, et où la grande
diversité linguistique entraı̂ne un réservoir de traits linguistiques (feature
pool) hétérogène. En ce qui concerne l’acquisition des traits, nous ne trouvons
le phénomène labovien d’incrémentation que pour deux changements
C Blackwell Publishing Ltd. 2011
9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden MA 02148, USA
152
CHESHIRE ET AL.
globaux (BE LIKE et l’avancement de la voyelle en GOOSE), ces deux formes
nouvelles étant les plus fréquentes chez les adolescents. Pour les autres traits
linguistiques, la variation selon l’âge des locuteurs s’explique plutôt par des
facteurs internes à la communauté. [French]
KEYWORDS: Multiethnolects, youth language, language contact,
new dialect formation, London English
1. NEW LANGUAGE VARIETIES IN THE INNER CITY:
LANGUAGE CONTACT AND GROUP SECOND-LANGUAGE ACQUISITION
In a number of European cities, the latter part of the twentieth century
has seen the emergence of new, distinct varieties of the host languages in
multilingual, working-class neighbourhoods. A considerable amount of research
in Scandinavia, the Netherlands, Belgium, Germany and Great Britain has
focused broadly on adolescent ways of talking in these neighbourhoods, applying
a range of approaches reflecting not just particular research interests, but also the
complexity and multidimensionality of the phenomenon. One major thrust deals
with the new varieties as ‘lects’ which can be described linguistically in relation
to the majority language (Wiese 2009), while others avoid reifying the varieties
by adopting a quantitative variationist methodology to the speech communities
concerned, focusing on individual features (Cheshire and Fox 2009). Others
investigate the new varieties qualitatively, preferring to see them as either
practices or repertoires, consisting of individual features which are deployed
strategically in conversation management and identity projection (Svendsen
and Røyneland 2008; Quist 2008).
How can we describe these new ways of speaking? How did they come into
existence? We are clearly talking about new contact varieties, though in doing
so we are inevitably idealising, because they are less homogeneous than either
‘dialects’ or ‘sociolects’ are assumed to be, and they are, in any case, to varying
extents context bound and (to the extent that they are ‘youth languages’)
transient. That said, within any one location these varieties are recognisable
to local people, even if naming practices are rather uncertain. For example, the
German term Kiezdeutsch (‘neighbourhood German’) was first used in print by
Wiese (2006), adopting a term used by speakers (Wiese p.c. 2011), referring
to multiethnic youth language in Germany. From this academic usage, it was
subsequently taken up by the media (Androutsopoulos 2007: 127–128). On the
other hand, the term Jafaican, used to refer to youth language in multiethnic
parts of London and beyond, most likely has media origins and is strongly
associated with hip-hop; it is likewise not essentially a ‘members’ concept’,
young people preferring the word ‘slang’ to characterise their way of speaking.
The term rinkebysvenska, referring to the Swedish characteristic of multiethnic
districts of Stockholm and named after one such district, Rinkeby, shows a third
pattern: it is an academic coinage, by Kotsinas (1988). In all three countries,
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
153
the language varieties and their labels are to differing extents part of public
discourse.
Despite the differences in naming, there are increasing indications that these
varieties have become the unmarked Labovian ‘vernacular’ for many speakers,
and that it is this that older people are reacting to when they claim that young
Londoners, for example, sound as if they are ‘talking black’. The term that has
gained some acceptance among linguists is multiethnolect, originally coined by
Clyne (2000), and we use this here, with due caution.
The reasons for the emergence of European multiethnolects at this point
in history are presumably linked to specific types of community formation in
urban areas which have seen very large-scale immigration from developing
countries. People of different language backgrounds have settled in already
quite underprivileged neighbourhoods, and economic deprivation has led to
the maintenance of close kin and neighbourhood ties. Castells (2000: 436)
writes of prosperous metropolises containing communities such as these: ‘It is
this distinctive feature of being globally connected and locally disconnected,
physically and socially, that makes mega-cities a new urban form’.
In these communities, there is often a rapid shift to the majority language
by the children of the migrants, possibly accelerated by the fact that there
are a large number of languages spoken in areas without strong residential
segregation. Because majority-language speakers may be in a minority in
parts of these districts, the availability of local, native models of the majority
language is weaker than elsewhere. This means that the majority language may
be acquired from other second-language speakers. This scenario is similar to the
kinds of community language shift which took place in Ireland and elsewhere
(Hickey 2006), with the difference that it is a minority population that is shifting
(albeit often as a majority at neighbourhood level), and that the group that is
shifting is linguistically heterogeneous. This makes the scenario an example
of what Winford (2003: 235) calls ‘group second language acquisition’, or
shift-induced interference (Thomason and Kaufman 1988: 75), where minority
linguistic groups form part of a larger host community and acquire the target
language mainly through unguided informal second-language acquisition in
their friendship groups. Thomason and Kaufman (1988: 43) argue that if the
minority group learners are well integrated into the host community, they may
initiate linguistic changes by using forms resulting from their imperfect learning
of the target language. Given suitable sociocultural conditions, these forms may
then be taken up by native speakers. However, our work in London suggests
that although speakers from minority groups may spearhead some linguistic
changes, this is not necessarily the result of imperfect learning (Cheshire et al.
2008).
Our research stems from two recent research projects, which we describe in
more detail in section 3: The Linguistic Innovators project1 and the Multicultural
London English project.2 Based on this research, this article asks the overarching
question: what are the sociodemographic and linguistic dynamics underlying
C Blackwell Publishing Ltd. 2011
154
CHESHIRE ET AL.
the emergence of new varieties of English in inner-city London at the end of
the twentieth and beginning of the twenty-first centuries? We use the term
‘Multicultural London English’ (MLE) to refer to the overall range of distinctive
language features used in multiethnic areas of London, conceptualising MLE
as a repertoire of features. We take a quantitative, variationist view, analysing
selected features from phonology, morphosyntax and discourse and thereby
avoiding unnecessary labelling (as a ‘lect’ or ‘variety’). We will argue that
language contact is an important determining factor in the emergence of MLE,
though the number of languages that are in contact in inner London makes it
unlikely that there is direct language transfer from any one language to English.
2. SPEECH COMMUNITY CHANGE:
TRANSMISSION, INCREMENTATION AND DIFFUSION
Multiethnolects arise within cities which, with respect to their native
populations, and despite substantial in-migration, can be considered ‘speech
communities’, in Labov’s sense:
The speech community is not defined by any marked agreement in the use of
language elements, so much as by participation in a set of shared norms; these
norms may be observed in overt types of evaluative behavior, and by the uniformity
of abstract patterns of variation which are invariant in respect to particular levels
of usage. (Labov 1972a: 120)
Within such a speech community, a primary mechanism of change is
transmission with incrementation (Labov 2007). It is assumed that children first
acquire the language of the primary caregiver (Kerswill 1996; Kerswill and
Williams 2000) and then, during a period of vernacular reorganisation, focus
on a new norm, driven by social forces, gradually increasing their use of the
new forms. Labov explains that ‘Children’s incrementation of the change may
take the form of increases in frequency, extent, scope, or specificity of a variable’
(2007: 346). Most studies assume that 17 is the age at which the increases level
off, leading to the ‘adolescent peak’ at this age. Much of the evidence in Labov
(2001) points to peaks in late adolescence, but in Labov (2007) he does not refer
to a specific age. Not all variables show a peak, but where there is a peak it is
around 16 years or older. Tagliamonte and D’Arcy (2009: 70) point out that 17
is the age at which people’s phonologies stabilise, and that this is reflected in the
age at which the peak occurs.
In terms of what it is that increases, most research on incrementation
investigates phonological change in the sense of gradual, sub-phonemic phonetic
shifts conditioned by phonological and morphological factors. Thus, Labov
(2001) examines the vowel system in Philadelphia, finding effects in terms of
preceding and following segments which either inhibit or promote changes.
Less straightforwardly identifiable are the ‘increases in . . . scope [,] or specificity
of a variable’ (Labov 2007: 346) which may occur as part of incrementation.
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
155
Phonological and morphological conditioning tend to be highly uniform within
each North American city, as for example the intricate, but different patterns for
the tensing of short a, as in man, bag, etc., in New York City and Philadelphia
(Labov 1994: 430, 1989), or the fronting of the GOOSE3 vowel which, in most
U.S. varieties, is strongly promoted by preceding coronals. These conditioning
factors are generally transmitted faithfully across generations, and changes are
minor – an example being the spread of Philadelphia’s irregular short a pattern
to more lexical items (Roberts and Labov 1995). It is largely only when changes
are adopted from other speech communities that radical and rapid changes in
conditioning occur – a crucial point which we return to shortly.
For vowels, the social-psychological mechanism for incrementation involves
children detecting the direction of change in the speech community, and then
carrying it forward. In the case of morphosyntactic and discourse-pragmatic
features, incrementation involves changes in frequency in relation to other
functionally equivalent forms. Tagliamonte and D’Arcy’s (2009) study of several
morphosyntactic variables and the grammaticalisation of the quotative BE LIKE
shows increases in the frequency of newer forms at the expense of other available
forms. Most of these features, including BE LIKE, show the familiar adolescent
peak. Later in this article, we address the acquisition of a new quotative, this is
+ speaker, by young Londoners, as well as their use of BE LIKE.
In his (2007) study of the spread of the New York City vowel system
to other cities, Labov distinguishes between community-internal change and
changes which arrive from another speech community by diffusion. There is
a fundamental difference in terms of how change is propagated in the two
situations:
It is proposed here that the contrast between the transmission of change within
speech communities and diffusion of change across communities is the result of two
different kinds of language learning. On the one hand, transmission is the product
of the acquisition of language by young children. On the other hand, the limitations
on diffusion are the result of the fact that most language contact takes place among
adults. It follows that structural patterns are not as likely to be diffused because
adults do not learn and reproduce linguistic forms, rules and constraints with the
accuracy and speed that children display. (Labov 2007: 349)
As a consequence, we would expect structural change, usually in terms of
simplification, to take place in cases of diffusion. Then, once a feature is accepted
into a speech community, we can assume it ‘goes native’, and is adapted by
speakers to fit the existing linguistic structure and sociolinguistic repertoire (see
Britain 2002a: 618, on the local adaptation of ‘global’ features). Labov finds
that the New York City vowel system is replicated in its most nearly intact form
in cities where the founding population came overwhelmingly from the New
York City area, in other words where an existing speech community had been
transplanted, allowing transmission and incrementation to continue to take
effect as in the original location. In cities where the system had been modified
C Blackwell Publishing Ltd. 2011
156
CHESHIRE ET AL.
and simplified, the contacts with New York City were largely the result of steady
migration leading to adult-to-adult contacts.
Both quotative BE LIKE and GOOSE-fronting are current global changes in
English, and as such are features which diffuse to ever more communities (indeed,
their spread is virtually complete today). The question arises, then, of the kind
of adaptations they undergo. We have already seen that BE LIKE behaves like
community-internal phonological features in showing a peak in adolescence,
a fact which suggests, we believe, that it is not always possible, on frequency
grounds alone, to distinguish between endogenous changes which are driven
solely by intergenerational transmission and changes which are adopted from
outside and have become nativised. Looking at the function of and constraints on
BE LIKE, however, changes have been detected as the feature spreads: Buchstaller
and D’Arcy (2009) show that linguistic and functional constraints on this
quotative differ between American English (the source) and British and New
Zealand varieties of English (whose speakers have adopted the construction). For
example, in American English it is used for mimetic re-enactment far more than in
the other two varieties. Buchstaller and D’Arcy see these differences as resulting
from the transfer of the feature, which is heavily mediated by film, TV and the
Internet, rather than arising from face-to-face contacts between young people.
Later in this article, both BE LIKE and GOOSE-fronting will be the subject of
investigation within a sociolinguistic system which is itself highly characterised
by both language and dialect contact. There have, in fact, been few speech
community studies which explicitly allow for contact by incorporating
immigrants and their descendants. An exception is Horvath’s 1985 study of
Sydney, comparing the Anglo-Irish (founder) population with people of Greek
or Italian origin. The Greek and Italian immigrants took on well-established
Australian accent features, exaggerating them. Horvath calls this variety ‘Ethnic
Broad’, but notes that subsequent generations are in fact in the lead in reversing
these features again. From this account, it is clear that there was at that time
no strong move towards a Greek or Italian ‘ethnolect’. In London, the position is
very different, for reasons which will become apparent.
The change model advocated by Labov, comprising transmission,
incrementation and diffusion, gives us a ‘toolkit’ with which to examine the
emergence of a multiethnolect. In particular, we ask: given that, in a highcontact community, transmission between generations may involve a stage of
language shift, how do features first form and then develop? Given that these
types of high-contact communities are often isolated from the mainstream for
both sociocultural reasons and the economic factor of poverty, how are global
changes adopted, and adapted there?
3. THE LONDON PROJECTS
London, the locale for this study, has seen in-migration throughout its history,
both from Britain and abroad. The population saw a steady rise from the Middle
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
157
Ages until the Second World War. However, until 1800 it also saw the highest
death rates in the country, far exceeding the birth rate. This meant that both
population replacement and increase were dependent on migration, a situation
which only changed at the start of the nineteenth century (Inwood 1998: 159,
271).
The two projects we report in this article were designed to explore the language
of working-class children and adolescents in London from two perspectives:
1. looking for innovations in relation to earlier descriptions of London English,
as well as the extent to which young Londoners’ speech forms part of
regional dialect levelling in the south-east; and
2. discovering how the London multiethnolect patterns in terms of its
acquisition, the use of the various features constituting it across ethnic
groups, and its status as an ethnically neutral variety.
In each project, young people were divided into ‘Anglos’ and ‘non-Anglos’,
with the Anglo group being composed of members of families of British origin
who had resided in the area for two or more generations. This corresponds
roughly to ‘White British’ in official terminology, and we find participants
occasionally using this term. This resulted in socially and demographically rather
homogeneous Anglo groups, while the non-Anglo groups were (intentionally)
highly mixed, roughly representing the ethnic composition of the districts where
they were recruited. In the first project, which we call Linguistic Innovators,
the inner city, represented by the multiethnic East End borough of Hackney,
was compared with outer-city Havering, which is located at the eastern
edge of London and has a very much lower rate of immigration. In the
second project, which we will refer to as MLE (standing for Multicultural
London English), all participants were from the inner city and lived in a
contiguous, multiethnic area covering parts of Islington, Haringey and Hackney.
We refer to this as ‘North London’. Figure 1 shows the locations of these
boroughs.
For the first project, the age groups were 17–19 and 70–86; for the second,
4–5, 8–9, 12–13, 16–19, c. 25 and c. 40. This last group was composed
of caregivers of some of the children; as a consequence a number had not
arrived in the U.K. until adulthood – unlike all the other participants, who
were born in the U.K. or had arrived as pre-schoolers. Each project recorded
between 120 and 130 individuals attending vocational (as opposed to academic)
courses in local Further Education colleges, as well as (in the MLE project)
children from a local comprehensive school, a primary school and a pre-school.
These figures also include 16 elderly white working-class Londoners. The older,
main interviewer was a female white Londoner, while the second, younger
interviewer, who recorded many of the younger children in the MLE project,
was a male British Asian from Birmingham. The sociolinguistic interviews
were conducted in pairs, small groups or sometimes individually. Some selfrecordings were also made, but are not considered here. In total some 2.8 million
C Blackwell Publishing Ltd. 2011
158
CHESHIRE ET AL.
words have been transcribed and stored in two databases in which the signal
is aligned with the transcription at the turn level. In the remainder of this
article, we discuss some of the innovations in London inner-city English, which
we label MLE – with all the provisos mentioned above in relation to the term
‘multiethnolect’.
4. VOWELS: INNOVATION THROUGH CONTACT
The most striking changes in the inner-city accents lie in the diphthong system.
Figure 2 shows the system of a man born in 1918.4 It manifests what Wells
(1982: 306–310) calls diphthong shift, i.e. rising diphthongs with peripheral
onsets (low for FACE and GOAT, mid-front for MOUTH and mid-back for PRICE) and
long trajectories. A number of young inner-city speakers, especially non-Anglos,
have a system similar to that shown in Figure 3. As can be seen, the trajectories
of FACE and GOAT are much shorter, with FACE now high-front and GOAT now
high-back, while both MOUTH and PRICE have been lowered and the trajectories
shortened or absent. In addition, GOOSE is characteristically very front. All the
young speakers show these changes, regardless of ethnicity, but in many cases
to a less extreme extent; predictors of more extreme pronunciations are male
gender, non-Anglo origin and Hackney residence.
Alongside the diphthong changes, there were changes in the short vowels as
well as extreme fronting of GOOSE, though in these cases the main predictor of
change was speaker generation (teenage vs. elderly) with few gender or ethnic
differences among the adolescents. The exceptions were the vowels of FOOT and
GOOSE, with FOOT being more front and GOOSE being more retracted in the outer-city
borough of Havering than in Hackney.
Together, these changes and others on the morphosyntactic and discourse
levels (some of which we discuss later in the article) constitute MLE. The questions
we need to answer are:
1. When did the changes we call MLE first emerge?
2. How are these changes acquired?
4.1 Roots of MLE: Real-time data
To answer the first question, we need to look for evidence of a situation
which is clearly prior to the emergence of MLE. Figure 4 shows the vowel
systems of 11 young London Afro-Caribbeans recorded by Mark Sebba in
1981–4 when speaking ‘Patois’ (4a) and when speaking with the interviewer
(4b). (Patois, or London Jamaican, is a recreolised variety using a number of
salient Jamaican Creole features on a London English base; see Sebba 1997:
231–233.)
To obtain the ‘Patois’ recordings, Sebba asked pairs or threes of young people
to talk in ‘Patois’ while discussing a topic given to them, upon which he left
the room. The interviews were conducted by Sebba with the same pairs/threes.
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
159
Figure 1: Map of London, with the boroughs of Hackney and Havering
highlighted (from www.cityoflondon.gov.uk/Corporation/maps/london_map.htm).
Haringey and Islington are, respectively, to the North and East of Hackney
–1.0
F*1
–0.5
0.0
0.5
1.0
2
1
0
–1
–2
F*2
Lobanov normalized means
Figure 2: Diphthong system of elderly male speaker from Hackney born 1918
C Blackwell Publishing Ltd. 2011
160
CHESHIRE ET AL.
Figure 3: Diphthong system of young male from Hackney, Afro-Caribbean origin,
born 1989
Despite the ‘set up’ nature of the Patois recordings, the conversations sounded
natural and unforced, while the interviews were a natural response to the
setting. Because the Patois recordings were relatively short, we present composite
vowel systems of the whole groups (109 tokens for ‘Patois’, 429 for ‘interview
style’). Normalisation using the Lobanov formula (Lobanov 1971) ensures
comparability across all speakers. Although inter-individual variability is
masked, the differences between the two styles are large and entirely in the
expected directions. Thus, in Patois we see a raised FACE, back GOOSE, back and
raised STRUT, open and back TRAP, and raised GOAT. In interview style, we do not
see any of these values. Instead we see a ‘modern’ London variety with an open,
but not back TRAP and a back STRUT. These are combined with a conservative front
onset for MOUTH and a back onset for PRICE (resulting in the traditional London
‘PRICE–MOUTH crossover’, Wells 1982), as well as a back FOOT, shared with Patois.
GOOSE is central, and not front as it is in the south-east today (Kerswill and
Williams 2005).
In the early 1980s, young Afro-Caribbeans could clearly code-switch between
Patois and local English. The local English itself, as evidenced by these recordings
and as discussed by Sebba (1993), is clearly very much of its time, a mainstream
variety of ‘Cockney’.
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
161
Figure 4: Composite vowel systems of 11 adolescent London Jamaicans recorded in
1984: (a) ‘Patois’ style; (b) interview with white fieldworker. (For diphthongs, only
onsets are shown)
4.2 The emergence of MLE in apparent time
The 1980s data does not suggest the existence of a multiethnolect (nor does
it show evidence of an ethnolect – given that we only have data from one
ethnic group). In order to establish whether there has actually been change
since then, we need to consider recent data collected under similar conditions
with people of similar backgrounds. In order to do this, we adopted an apparenttime methodology using several child and adolescent age groups, in a manner
similar to that of Kerswill and Williams (2000) in a study of the new-town
dialect of Milton Keynes. As already mentioned, in the MLE project we recorded
individuals representing age groups from 4 to 40. These were chosen so as to
take into consideration significant life-stages: those of the pre-school years, early
childhood, puberty, adolescence, early adulthood, and mature adulthood.
With a multiethnolect, it is difficult to talk of a ‘target variety’, because of the
high degree of variability and (quite possibly) its transient nature. However,
we can operationalise the target by first presenting data from the youngadult and adolescent speakers, on the assumption that they will already have
acquired the new set of features characteristic of this variety. In order to
make the comparison as direct as possible with Sebba’s recordings, Figure 5
shows vowel data just from adolescent participants with an Afro-Caribbean
heritage. There is a high degree of correspondence with the Hackney AfroCaribbean shown in Figure 3. We note that, as expected, GOOSE is very front. In
accordance with the south-east short vowel shift (Torgersen and Kerswill 2004),
TRAP is an open-central vowel, while STRUT is a back vowel overlapping with
START.
C Blackwell Publishing Ltd. 2011
162
CHESHIRE ET AL.
Figure 5: London inner-city vowels used by Multicultural London English project
adolescent speakers (aged 16–19) of Afro-Caribbean origin. (For diphthongs, only
onsets are shown)
To obtain a fuller picture, we show the vowel systems for the complete data set
in Figures 6a and 6b. Data for females and males is shown separately. Comparing
Figure 6 with Figure 5 shows that non-Anglos as a group (including the AfroCaribbeans) share the same distribution – in other words, there is little difference
between the non-Anglo ethnicities. Differences, however, are to be found if
one looks at ethnicity and gender together, particularly in FOOT, FACE and GOAT.
Anglo females are strongly in the lead in the fronting of FOOT, with Anglo males,
non-Anglo females and non-Anglo males following behind in that order. Recall
that fronted FOOT had previously been found in outer-city Havering, a change
which matches a general south-eastern trend (Torgersen 2002), while it was
not found among Anglos in Hackney – these participants did not differ from the
non-Anglos in backness. The second difference is the more open onset of FACE
among the Anglos, reflecting a broader, more traditional diphthong. Finally,
non-Anglo males have a fully back onset for GOAT, in this case representing the
near-monophthongal back quality seen in Figure 3. In sum, MLE as represented
in this data seems to show distinctions between:
1. conservativism (Anglo female FACE);
2. use of regional innovations (Anglo female FOOT); and
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
163
Figure 6: London inner-city vowels used by Multicultural London English project
adolescent speakers (aged 16–19): (a) Short monophthongs; (b) diphthongs plus GOOSE
and START (for diphthongs, only onsets are shown). = Anglo female (N = 5); =
Anglo male (N = 3); = non-Anglo female (N = 10); = non-Anglo male (N = 8)
3. use of inner city-specific innovations (non-Anglo male speakers’ GOAT vowel,
a change structurally parallel to the same speakers’ raised FACE).
We are now in a position to compare the modern MLE data with the 1980s
recordings. The most striking single difference is between the interview speech
of the 1980s on the one hand and that of today’s non-Anglos, as shown in
Figures 3, 5 and 6. Each set of data was recorded under roughly the same
conditions with demographically similar participants, but separated by 20–25
years. The later speakers all show unshifted (i.e. raised) FACE and GOAT, have MOUTH
and PRICE as low vowels with no crossover, and a very front GOOSE. The earlier
recordings show vowels which resemble those of the elderly, Anglo speakers,
as well as (to some extent) the Anglo females in the new data. The older data
also shows a very back FOOT, which reflects today’s non-Anglo pronunciation.
Where do the newer non-Anglo vowels come from? All the diphthongs (FACE,
GOAT, MOUTH and PRICE) have almost identical values, not to the interview style of
the earlier speakers, but to their Patois styles. Without suggesting that presentday MLE has borrowed these directly from Jamaican Patois, we can see that the
correspondence is striking. A single source is unlikely, however, because AfroCaribbeans are nowhere an absolute majority, but live alongside people of a
wide range of ethnic backgrounds. Fox (2007) finds very similar pronunciations
for FACE and PRICE in a predominantly Bangladeshi community in East London,
as does Khan (2006) for GOAT and PRICE among Afro-Caribbean and Pakistani
communities in Birmingham. The fact, too, that GOOSE is so strongly fronted,
especially among non-Anglos, when Patois has a fully back vowel, suggests
C Blackwell Publishing Ltd. 2011
164
CHESHIRE ET AL.
various origins for the MLE features. This insight is one we will develop in this
article in the context of a ‘feature pool’.
It can be argued that the data collection in the later projects did not tap into
a wide enough range of the speakers’ repertoires, compared to Sebba. However,
the phonetic changes in speech used in paired interviews with a fieldworker are
consistent across the dataset. This is not an indication that Afro-Caribbeans no
longer use Patois in particular contexts (it is still used in ‘crossing’, for example –
Rampton 2010), but rather that the vernacular base line has changed from
one which was largely Cockney (in the 1980s) to a variant of MLE today. A
further question arises as to why MLE speakers are routinely heard as ‘black’
by outsiders. It is likely that young people’s orientation towards Caribbean and
African American youth culture, especially hip-hop and rap, makes the use of
‘black’ features attractive. As evidence for this attractiveness, Rampton (2010:
138) cites his finding that young people ‘cross’ into Creole much more often than
they ‘cross’ into Punjabi. Perhaps, then, the salience of ‘black’ culture accounts
for the association made between innovative pronunciations and sounding
‘black’. A further reason may be that MLE speakers routinely use vocabulary
items such as yard, whagwan and man (the first two meaning ‘home’ and ‘what’s
up’, respectively, and the third used as an address term), which are associated
with Caribbean speakers.
Figure 7 shows the young adult speakers. Here, the sample is unbalanced,
with Anglos being represented only by two females and the non-Anglos by five
men and one woman, from a range of ethnic backgrounds. Again we note the
Figure 7: London inner-city vowels used by young adult speakers (aged 20–35,
median 23, mean 24): (a) short monophthongs; (b) diphthongs plus GOOSE and START
(for diphthongs, only onsets are shown). △ = Anglo (2 females); + = non-Anglo (5
males, 1 female)
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
165
front GOOSE vowel, though it is more back in relation to KIT by comparison with
the adolescents. The two female Anglos again have a fronted FOOT and a FACE
onset which is lower than the adolescents’. They also share a conservative feature
only weakly present in the younger group: the very marked distance between the
onset of MOUTH (front) and that of PRICE (mid-back) – the PRICE–MOUTH crossover.
Predictably, GOAT is central for the Anglos (more so than for their adolescent
counterparts) and raised back for the non-Anglos. What we see here is an adult
group of non-Anglos (all but one of whom are male) who use a mainly MLE
repertoire, and a (small) group of Anglo females whose vowels are relatively
conservative, more so than their younger equivalents. We cannot easily say
whether we are dealing with an ethnic or a gender difference; it is apparent,
however, that MLE is not just a youth variety, since these young adults use
features associated with it.
The question now arises as to when MLE features are acquired, and by whom
– given that these two Anglo adults do not appear to use them. This is the second
question listed above. We turn first to the youngest age group, the 4–5 year
olds. Unfortunately, it proved impossible to record any Anglo 4–5 year olds at
the primary school we contacted, so the display shows only non-Anglos (again
from a range of ethnic and therefore linguistic backgrounds – see Appendix).
Figure 8 displays the vowel data for this group.
Surprisingly, Figure 8 shows a more or less complete set of MLE values, with a
small amount of gender differentiation in the expected directions, viz. fronter FOOT
and lower GOAT among the females. However, this group’s GOOSE is considerably
backer than that of the adolescents. This finding is similar to that of another
Figure 8: London inner-city vowels used by 4–5 year old speakers (all speakers
are non-Anglo): (a) short monophthongs; (b) diphthongs plus GOOSE and START (for
diphthongs, only onsets are shown). = non-Anglo female (N = 8); = non-Anglo
male (N = 9)
C Blackwell Publishing Ltd. 2011
166
CHESHIRE ET AL.
(a)
1.0
F*1 caregiver
0.5
0.0
–0.5
–1.0
–1.0
–0.5
0.0
0.5
1.0
1.5
F*1 child
Figure 9a: Correlation between children’s and caregivers’
= age 8–9; = age 12–13
FACE
(F1). △ = age 4–5;
study of children’s vowels in the south-east, in Milton Keynes, where four year
olds had a relatively retracted GOOSE vowel (Kerswill and Williams 2005). The
Milton Keynes four year olds were not only conservative in GOOSE, but also in
respect of GOAT-fronting – this being a south-eastern regional variable largely
absent among MLE speakers, who tend to back and raise this vowel. In Milton
Keynes, there was a highly significant relation between female caregivers and
four-year-old children for GOAT-fronting, while there was no such correlation
for GOOSE-fronting (Kerswill and Williams 2005). Unlike in Milton Keynes, the
patterning of GOAT in London is multidimensional, involving either backing or
fronting, as we have seen, so it is instructive instead to compare GOOSE in the
two locations. Figure 9 shows correlations between children’s and caregivers’
production of another vowel, FACE, alongside GOOSE, on the salient dimensions of
F1 for FACE (9a) and F2 for GOOSE (9b).
Overall, there is no significant association in Figure 9a (r2 = 0.008, p =
0.612). There is some age differentiation, in that 4–5 year olds have a lower F1
– a more raised vowel – for FACE, though this may be associated with the fact
that all the caregivers for this age group are non-Anglo. Figure 9b shows a more
complex picture for GOOSE (axes for normalised F2 are shown). The children’s
scores cluster around a normalised F2 of around 1.0. Unlike the case for FACE,
the caregivers have a distinct bimodal distribution for GOOSE, with one group
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
167
(b)
2.0
F*2 caregiver
1.0
0.0
–1.0
–2.0
–1.0
0.0
1.0
2.0
3.0
F*2 child
Figure 9b: Correlation between children’s and caregivers’ GOOSE (F2). △ = age 4–5;
= age 8–9; = age 12–13
resembling the children in clustering around 1.0 while the other has a much
lower F2 of around −1.0, representing a back vowel. It is apparent that very
few children whose caregiver has the back vowel actually follow it – the only
exception being a 4–5 year old, near the bottom left of the chart. There is no
significant relationship between the normalised F2 of caregiver and child (r2 =
0.57, p = 0.174). From this, we conclude that, as with the Milton Keynes GOOSE
vowel, the youngest children are not influenced by their principal caregiver’s
pronunciation of these two vowels. In terms of the ‘feature pool’ of available
variants (see further below), we can see that those represented by the parents
appear to be rejected. Instead, the children already orient towards community
norms, both existing and emergent.
There are no significant differences between these three younger age groups
in the London data, suggesting no further move on their part towards the fully
front GOOSE of the adolescents’ data. If we add adolescents’ data (cf. Figure 6),
however, we find a significant difference for this age group and the youngest age
group (p < 0.05).5 The lack of correlation with the caregivers is consistent with
Milton Keynes, however (Kerswill and Williams 2005), though we can now go
further and consider why this might be. Figure 10 is a replotting of Figure 9b,
this time showing whether or not the caregiver was born in the U.K.
It can be seen that a majority of non-U.K. born caregivers have a very back
GOOSE vowel (a minus value for the transformed F2), which can be transcribed
C Blackwell Publishing Ltd. 2011
168
CHESHIRE ET AL.
2.0
F*2 caregiver
1.0
0.0
–1.0
–2.0
–1.0
0.0
1.0
2.0
3.0
F*2 child
Figure 10: Correlation between children’s and caregivers’ GOOSE (F2), by place of birth
of caregiver. × = U.K. born; = non-U.K. born
[u]. As we saw, there is virtually no tendency for the children to adopt this back
+
vowel (there is no significant relationship between the normalised F2 of caregiver
and child (r2 = 0.57, p = 0.174)). It could be argued that the caregiver vowel
is too ‘deviant’ in relation to the local norm to be copied, and the child instead
favours a more mainstream front pronunciation. In any case, in some of the
families the caregivers speak their community language at home, and if English
is used at all in the home it is between young children and their older siblings, so
the question of a caregiver model does not arise. Assuming that the phonologies
of those parents who do speak English at home are non-British, it is tempting
simply to see in this an example of what Chambers calls the ‘Ethan Experience’:
children whose parents have a non-local accent are unaware of the accent,
and filter it out even as pre-school children (Chambers 2002: 121–123). In our
view this underestimates the young child’s capacity for identity formation: some
do, after all, follow parental models for some variables, such as GOAT in Milton
Keynes. As we shall see, we prefer to see the parental variants as part of a ‘pool’
of available forms – rejected, in this case, as we have seen, perhaps because they
are too saliently non-local; we return to the interaction between salience and the
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
169
Figure 11: London inner-city vowels used by Multicultural London English project
8–9-year-old speakers: (a) Short monophthongs; (b) diphthongs plus GOOSE and START
(for diphthongs, only onsets are shown). = Anglo female (N = 3); = Anglo male
(N = 3); = non-Anglo female (N = 9); = non-Anglo male (N = 5)
feature pool below. In any case, the children’s rejection of the back GOOSE vowel
may well represent a lessening of the linguistic tie between parent and child in
comparison to majority-community language speaking families, where the local
accent/dialect is being transmitted cross-generationally in a way that is not the
norm in a highly multiethnic district like East London.
In the absence of Anglo data for the 4–5-year-old age group, we cannot tell
whether they, too, will have acquired an MLE-like phonology in the same way as
the non-Anglos. Looking at the next age group up can give us some indications.
Figure 11 shows a mainly MLE-type system for the eight year olds, with FACE
level with DRESS, and a half-close back GOAT for all groups, with no ethnicity or
gender effects. The by now predictable ethnic and gender differences for FOOT are
present (Anglos more front, male non-Anglos fully back), though the pattern for
GOOSE is not as clear. There is no trace of the conservative PRICE–MOUTH crossover,
with the onsets of these vowels overlapping to a considerable extent.
By age 4–5, and certainly by 8, children have moved firmly away from their
caregivers – except of course in those cases where they acquired English from
peers, older siblings or other people. This appears to be true especially of the Anglo
children, whose parents are highly likely to have the Anglo system illustrated in
Figure 7, but it is also true of those non-Anglos who do not follow their immigrant
caregivers’ back GOOSE, in some cases doubtless because the caregivers did not
use English in the home.
Next, we look for a further approximation to MLE by examining data for 12–13
year olds: we would expect an increase in GOOSE-fronting, at the very least. Figure
12 shows a similar basic system, but with some deviations. These are mainly
C Blackwell Publishing Ltd. 2011
170
CHESHIRE ET AL.
Figure 12: London inner-city vowels used by Multicultural London English project 12–
13-year-old speakers: (a) Short monophthongs; (b) diphthongs plus GOOSE and START
(for diphthongs, only onsets are shown). = Anglo female (N = 7); = Anglo male
(N = 5); = non-Anglo female (N = 7); = non-Anglo male (N = 8)
the result of conservative realisations on the part of the Anglo girls, who have
a low/retracted FACE vowel (p < 0.01 for F2) and a raised and backed PRICE (p
< 0.01 for both F1 and F2), giving rise to a marked PRICE–MOUTH crossover. In
this figure, the expected ethnic alignments of FOOT and GOAT are less clear: for
these, visual inspection suggests an interaction with gender, with non-Anglo
girls resembling the Anglo girls. Finally, there is no further GOOSE-fronting.
If we take our data as representing apparent time, then the 12–13 year olds’
data does not show any particular move towards the MLE pattern shown in Figure
5: the system, with its ethnicity and gender variation, is already present among
the younger speakers. The apparent ‘regression’ to an older, Anglo London norm
by the Anglo 12–13-year-old girls is in fact similar to the pattern shown by the
adolescent MLE speakers (Figure 6), but in a more marked form (though the
differences are not significant). We conclude from this that there is some gender
differentiation in the MLE vowel system, which increases as children pass into
adolescence.
The striking exception to the early acquisition of MLE is the relative lack of
GOOSE-fronting by all the pre-adolescent groups. We briefly turn now to this
variable, examining it more closely in terms of incrementation.
4.3
GOOSE-fronting: Goose The acquisition of a global feature
For GOOSE-fronting we have already seen that there is a flat distribution across
the three youngest age groups, and an apparent jump to the adolescents. We
have seen that the specificity of the BE LIKE quotative changes when it is adopted
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
171
into a new speech community; we now ask the question whether this is so for
GOOSE-fronting, too, by considering whether there is variation in phonological
constraints. Labov (2001) shows how, in most North American cities, preceding
coronals strongly favour the fronting of this vowel. A number of American
varieties lack this strong effect (Labov 2010: 262). In the London data, might
a change in constraints underlie the difference between the younger speakers
and the adolescents? In the adolescent data, there is no significant effect of a
preceding coronal on the second formant of GOOSE (Mann-Whitney test). This
is not true, however, of the 8–9 year olds and 12–13 year olds, for whom
there is a significant effect (p <.001 and p < .01, respectively; Mann-Whitney
test). This means that, besides an increase in F2, the fronting process includes
having words with non-coronal onsets join the coronal-onset words in their
frontness. It is not known if this effect is found in other varieties of British
English, but the pattern is clearly in line with Labov’s (2007) definition of
incrementation.
4.4 Summary of evidence from the vowels
Given this information, it is possible to outline some tentative conclusions from
our apparent-time investigation:
• MLE-type phonologies are acquired early on by non-Anglo preschool
children. There is little or no correlation between 4–5 year olds and their
caregiver, unlike the case for at least one variable in Milton Keynes (Kerswill
and Williams 2005) and also for a phonological variable described by Smith,
Durham and Fortune (2007) in Buckie, Scotland. This suggests that children
in multilingual communities in London attend to the speech of their peers
at a younger age than in monolingual communities like Milton Keynes and
Buckie. This point, which we will elaborate on in the next section, may be
characteristic of group second-language learning, where native models are
not necessarily available, either because the caregivers use a non-native
variety or because English is not spoken at home or, if it is, it is used mainly
between siblings.
• There is evidence of incrementation for just one variable, GOOSE-fronting,
where the peak lies with the 16–19 year olds. In the speech of the young
adults, which (at least for the non-Anglos) we take to represent (an earlier
manifestation of) MLE, the production of GOOSE is slightly less front than that
of the adolescents. This, too, can be taken as evidence for incrementation.
Incrementation is present in this global, diffusing variable, but not in the
endogenous variables.
• The age distribution for the remaining vowels appears to be flat, with
no consistent change as the children become older. There is, however,
evidence of the emergence of increasing gender differentiation with age,
and this conclusion is probably supported by the very conservative youngadult Anglo female scores in Figure 7. We will be arguing shortly that
C Blackwell Publishing Ltd. 2011
172
CHESHIRE ET AL.
incrementation does not apply when a new form (in the case to be considered,
a new quotative expression) is being generated from within a multilingual
community undergoing group second-language learning. This explanation
is also available for vowels, where we can argue that there is a distinction
between general south-eastern changes, especially GOOSE-fronting where
incrementation is present, and the other vowels, like FACE-raising and
GOAT-raising/backing, where we suspect a community-specific, endogenous
change. In these cases, we find flat age distributions.
• Comparing the 8–9 and 12–13 age groups suggests that the ethnic/gender
groups already show some of the differentiation that is characteristic of the
adolescents and adults. In particular, the Anglo females have realisations that
are relatively conservative. (Anglo data from 4–5 year olds is not available,
so we cannot easily tell at what age the convergence takes place.)
5. BEYOND PHONOLOGY
How do the preliminary conclusions just presented transfer to other levels of
language? It is not easy to decide which forms in other components of language
may form part of ‘MLE’, but some candidates are innovative forms used by both
Anglo and non-Anglo adolescent speakers in inner London but not heard in our
outer-London data. In this second part of the article, we consider three features
of this kind: first,
• an innovative discourse-pragmatic feature (a new quotative expression); then
• a morphosyntactic feature (agreement patterns with past tense BE); and,
finally,
• a morphophonological feature (allomorphy in the definite and indefinite
articles).
Investigating these features clarifies some of the questions concerning the
emergence of a multiethnolectal variety space, and helps us to build a model.
5.1 A new quotative
Examples 1 to 5 illustrate a new quotative expression, this is + speaker.6
1. this
this
2. this
3. this
4. this
5. this
is
is
is
is
is
is
them “what area are you from. what part?”
me “I’m from Hackney”
her “that was my sister”
him “don’t lie. if I search you and if I find one I’ll kick your arse”
my mum “what are you doing? I was in the queue before you”
my mum’s boyfriend “put that in your pocket now”
Table 1 shows that in our first project this is + speaker was used only by
adolescents in inner London. The adolescents in outer London never use this
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
173
Table 1: Use of quotatives in Hackney (inner London) and Havering (outer London)
by elderly and young speakers
Elderly
Hackney
N
Young
Havering
Quotative
%
%
N
Say
Think
Go
Zero
BE LIKE
This is
+ speaker
Tell
Other
Total
70.8 261
4.1 15
4.6 17
18.9 70
-
73.5 200
10.3 28
0.4
1
12.9 35
-
1.6
6
100
370
2.9
8
100
272
Hackney
%
27.4
12.8
11.7
15.1
24.4
N
351
164
150
193
313
4.8
61
1.9
24
2.0
26
100
1282
Havering
%
N
31.2
6.1
26.5
12.3
20.8
328
64
279
129
219
3.2
33
100
1052
form. The table also shows that the innovative quotative is a minor part of the
quotative system, accounting for only 4.8 percent of the quotatives used by the
Hackney adolescents. It could, therefore, be a transient phenomenon, but the
fact that several different individuals use the form suggests otherwise: seven
speakers contributed this is + speaker forms to the Hackney data set, and 14
speakers to the North London data set. Furthermore, although this quotative
expression is infrequent in our data it nonetheless occurs often enough in young
people’s speech generally for it to have been noticed by non-linguists: it has been
taken up by U.K. television personalities when they portray disaffected youth
(for example by the comedian Catherine Tate and by actors in the Armstrong and
Miller show on BBC TV).
Table 2 gives the age distribution for the different quotative forms used in
North London. The table shows that this is + speaker is not confined to Hackney,
though in North London it is still a minor part of the quotative system. Table 2
also shows that the form is used by the 12–13 and 8–9 year olds as well as the
adolescents: five speakers in the 16–19 year group, two in the 12–13 year old
group, and seven in the 8–9 year old group. Note, though, that it is not used by
the caregiver group or by the 4–5 year olds. The caregivers, we assume, have
not acquired this new quotative form, which is being introduced into the system
by younger speakers. The 4–5 year olds tell few narratives (the most frequent
context for reported speech) and they have not yet acquired the full range of
quotative forms. Say, go and BE LIKE are acquired by children in an orderly
progression, as Table 3 shows. The table shows the forms used by those children
who contribute the most tokens to the data base, arranged as an implicational
scale for say, go and BE LIKE. Say is the first form to be acquired, followed by go
and then BE LIKE. As the group figures in Table 2 show, once BE LIKE is acquired,
C Blackwell Publishing Ltd. 2011
174
CHESHIRE ET AL.
Table 2: Use of quotatives in North London by age group
Age 4–5
Quotative
%
N
Age 8–9
12–13
%
%
N
Say
93.9 46 39.5 202 25.4
Think
0.6
3
1.9
Go
4.1
2 31.1 159 23.8
Zero
2.0
1
2.0
10 14.5
BE LIKE
17.0
87 25.9
This is
+ speaker 5.3
27
2.0
Tell
1.6
8
0.3
Other
2.5
13
1.6
Total
100
49 100
512 100
16–19
Caregivers
N
%
N
163
12
153
93
166
17.0
7.2
7.3
12.5
45.7
218
92
94
160
584
%
N
50.3 174
10.7
37
5.2
18
18.2
63
10.1
35
13
3.0
38
2
2.2
28
1.2
4
10
2.7
34
3.2
11
642 100
1279 100
346
the frequency of both go and say declines. The zero quotative and this is + speaker
seem to be wild cards, used at any stage once children have become proficient in
telling stories, presumably because they serve specific pragmatic functions in a
narrative, highlighting particularly dramatic points in the story (Fox in press).
Most children do not use them until they have progressed beyond using merely
say (Uzay and Derya are the only exceptions to this pattern).
Table 2 shows that the peak in the frequency of use of the innovative this is
+ speaker quotative is in the 8–9-year-old age group. The 8–9 year olds also
use a high proportion of this is + speaker forms with non-quotative functions.
The importance of this for language innovation will become clear below. Like
the quotative form, these occurrences are all in narratives: they refer to the
protagonists’ states, feelings, actions, gestures and expressions, as in examples
(6) to (9):
6. he’s sitting on a chair this is him like he’s drunk or something
7. I been on it this is me I’m scared I’m like this . . . it goes slow and then I say
“yeah”
8. this is the this is the boy falling asleep he went “<sound effect>”
9. alright right this is this is me knocking at the door yeah and I’m knocking
at the door yeah and this is the dog “<makes gesture>”. he just went and
this is the dog “woof woof woof”
Table 4 shows that these non-quotative uses account for nearly half the this is
+ speaker tokens in the 8–9 age group, but are less frequent in the speech of the
12–13 age group and rare for the 16–19 year olds.
What is the source of the this is + speaker form? Deictics are not unusual in
quotative expressions, so as a quotative form it is unremarkable, even though
it has not been attested before for English. Macaulay (2001), for example, notes
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
175
Table 3: Implicational scale showing acquisition of quotative
expressions
Speaker (age)
Nisha (5)
Tamila (5)
Neelan (5)
Kenneth (5)
Rachel (5)
Talullah (8)
Kareen (8)
Ikram (8)
Derya (8)
Saddiki (8)
Dafne (8)
Uzay (8)
Rasgur (5)
Din (5)
Nandita (8)
Catherine (12)
Junior (8)
Stephen (8)
Louise (8)
Wahid (8)
Madeleine (8)
Mahir (8)
Dumaka (8)
Howard (8)
Lydia (8)
Ben (8)
Scarlett (13)
Christopher (13)
Sadik (11)
Barry (12)
Meg (12)
Abigail (13)
Henry (12)
Darren (12)
Say
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
Go
BE LIKE
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
This is
+ speaker
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
Zero
√
√
√
√
√
√
√
√
√
√
√
√
√
√
-
√
Table 4: Quotative and non-quotative uses of this is + speaker
Age 8–9
12–13
16–19
Type of use
%
N
%
N
%
N
Quotative
Non-quotative
51
49
27
26
87
13
13
2
93
7
38
3
C Blackwell Publishing Ltd. 2011
176
CHESHIRE ET AL.
that when BE LIKE first appeared in Glasgow it was with deictic that in expressions such as (10):
10. I was like that “on you go” (Macaulay 2001: 9)
This is + speaker is a ‘speaker-predicating quotative’ (Güldemann in press):
the speaker is foregrounded by being positioned clause-finally (assuming, albeit
simplistically, that the speaker is the predicate of this is). Identificational
quotatives are common in the world’s languages and especially widely attested
in African languages (Güldemann in press). They also occur in some varieties of
English: Irish English, for example, has here’s + speaker, as in (11).
11. <EXTREMELY HIGH PITCHED> Here was I “Then I must be hard of hearing
or something - you rapped the door and I didn’t hear you”. . . out the back and
everywhere they were. Here’s me “Have youse took leave of your senses?”
<HIGH PITCHED> He says - uh - “get everybody up, everybody up” (Milroy
and Milroy 1977: 54)
Although the form of the new quotative expression is unremarkable from a
linguistic point of view, we do need to explain its emergence. Despite the
occurrence of similar forms in Irish English and in African languages, its
emergence in London is unlikely to represent a case of direct language transfer
from Irish English or some of the African languages spoken in London. Too
many different languages and varieties of English are spoken in the multiethnic
communities we are researching for us to expect direct language contact from
any one variety to London English. Nonetheless, the type of language contact
inherent in group second-language acquisition seems to be at the heart of the
innovations that we observe in our data.
As noted earlier, we find Mufwene’s concept of the feature pool helpful in
explaining the innovations we find in the group second-language acquisition
situation in inner-city London. Mufwene (2001: 4–6) depicts both dialect contact
and language contact situations as producing a ‘feature pool’ from the range of
input varieties, with speakers selecting different combinations of features from
the pool, sometimes modifying them into new structures in the output varieties.
A crucial difference between dialect contact and language contact is that in the
first case the speakers share typologically, genetically related input varieties,
whereas in the second case there may be typological diversity and therefore
greater complexity of competition in the feature pool. Havering is a good example
of a dialect contact situation. Until the 1950s it was a small rural town in the
county of Essex, but it was transferred to the greater London administrative
area in 1965. Many families from East London were relocated there during
the post-war slum clearance and reconstruction of London. There has therefore
been recent dialect contact between London English and Essex dialects, as well as
other south-eastern English dialects spoken by other incomers. The town remains
profoundly monolingual. In Hackney, by contrast, some 95 different languages
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
177
are spoken as a first language by school children (Hackney Council 2010; cf.
Baker and Eversley 2000). Bilingual children acquire English partly at school and
partly through unguided language acquisition in multiethnic friendship groups
where a range of interlanguage varieties are spoken alongside traditional London
English, Afro-Caribbean English and indigenised second-language varieties such
as Indian English and Nigerian English. The pool of features in Hackney is
therefore more heterogeneous than it is in Havering, which we assume allows
more scope for innovation and restructuring.
What, though, determines the features that speakers select from the feature
pool? The frequency of a feature is known to be an important determining factor
in several kinds of language contact situations (for example, in the development
of pidgin and creole languages, second-language learners’ interlanguage and
indigenised second-language varieties (Siegel 1997: 139), but this can hardly
account for the emergence of this is + speaker, which is still a minor part of the
quotative system for all speakers who use it.
The available evidence suggests that it did occur but with even lower
frequencies in some of the input varieties to the feature pool in Hackney (perhaps,
then, it existed as what Trudgill (1999) terms an ‘embryonic form’). For example,
Mark Sebba’s recordings of London Jamaicans, made in the 1980s, contain three
tokens of the form, and we are aware of two tokens in the COLT corpus (Corpus of
London Teenage Speech) recorded in the 1990s, from ethnic minority speakers
of unspecified origin (Fox in press). Its more frequent use in our recordings
confirms Mufwene’s suggestion (2001: 6) that language contact can result in
more complex competition in the feature pool and that in some cases this in turn
will favour an option that was already available in some of the input varieties
but was ‘statistically too insignificant to produce the same output under different
ecological conditions’ (see also Lapidus and Otheguy 2005: 157). Use of the
form both then and now may also have been indirectly influenced by language
transfer, if some of the languages spoken by bilingual children in Hackney contain
identificational quotatives.
However, frequency is not the only factor that guides the selection of forms
from the feature pool. Siegel’s list of constraints on selection (1997: 139) includes
regularity (lack of exceptions), transparency (one-to-one relationship between
form and meaning) and, importantly for this is + speaker, salience. These factors
are rarely clearly defined in the literature, as Siegel points out, and the concept of
salience is perhaps particularly difficult to define (Kerswill and Williams 2002)
but this is + speaker meets some of the suggested characteristics of salient forms
(Siegel 1997). It tends to contain free morphemes rather than bound morphemes,
since it mainly occurs in the conversational historic present tense: compare, for
example, this is him with the inflected verb forms in he says or he goes. The
semi-fixed construction is a useful form for language learners who want to keep
the floor and maintain the pace of speech. This is + speaker is presumably easy
to perceive as a quotative expression (ease of perception is another suggested
characteristic of salient forms; again, see Siegel 1997), with the speaker put
C Blackwell Publishing Ltd. 2011
178
CHESHIRE ET AL.
into focus by the deictic this and the empty verb BE. Furthermore, as mentioned
above, the form typically occurs at a salient point in a performed narrative, at
moments of high drama (again, see Fox in press).
How does the age distribution of the new quotative expression compare to the
distribution found for the vowel changes in MLE? The 4–5 year olds are like their
caregivers in using mainly say, but this, we have argued, reflects developmental
tendencies rather than the influence of the caregivers. The 4–5 year olds, in
fact, differ from their caregivers in that they do not use BE LIKE and they rarely
use the zero form (see, again, Table 2). They are not yet at a stage of language
development where they can be expected to quote reported thought, so they do
not use think (and perhaps this accounts, to some extent, for the non-existence
of BE LIKE in their speech: BE LIKE tends to be favoured with reported thought
in the early stages of its introduction).
As with all but one of the MLE vowels, there is no adolescent peak in the use
of the new quotative and no incrementation across the 8–9, 12–13 and 16–19year-old groups. In fact, the 8–9-year-old group uses twice as many forms as the
other two age groups. They also use a large number of non-quotative this is (NP)
forms, as we saw earlier. The number of non-quotative forms shows a regular
decrease with increasing age (and they do not occur at all in our recordings of
16–19 year olds in Hackney). The 8–9-year-old age group seems crucial, then,
for acquisition of the MLE quotative this is + speaker.
Investigating the function of the form in the speech of this age group helps
explain their increased usage of the form relative to the other age groups. The
other age groups use quotative this is + speaker almost exclusively to introduce
reported direct speech. The 8–9 year olds, however, use it with roughly equal
frequency to introduce both direct speech and non-lexicalised sound and gesture,
as in (12) and (13).
12. then. then this is me “xxxxxx<making noises>xxx”
13. this is me <does an action which makes the interviewer laugh>
This age group quotes non-lexicalised sound and gesture far more frequently
than the other age groups: this use accounts for almost a quarter of the content
of the quote for them, whereas it is negligible for the other age groups (Kerswill
et al. in press). Seen in conjunction with their non-quotative uses of this is +
speaker, it seems that the narrative style of the 8–9 year olds predisposes them
to adopt this is + speaker for mimesis: to perform not only reported speech,
gestures and sounds but also actions in a way that mimics the way they actually
occurred. The animated discourse style of the 8–9 year olds is important, then,
in explaining both their more frequent usage of the form and its expanding
functions in other, non-quotative discourse contexts.
We suggest that the explanation for the age distribution we observe for the
innovative quotative expression rests on the difference between the type of
change represented by BE LIKE, on the one hand, and this is + speaker, on
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
179
the other hand. BE LIKE is a new, vigorous, rapid change affecting all Englishspeaking communities. It shows incrementation with an adolescent peak, in
a manner consistent with Tagliamonte and D’Arcy’s (2009) findings for this
feature in Toronto. Taking a phrase from Eckert (2003), Milroy refers to it as an
‘off the shelf’ feature, which is easily borrowed: ‘. . . [O]ff the shelf changes are
relatively freely available to appropriately positioned social actors as a stylistic
and social resource, regardless of the structure and location of their primary
social networks’ (Milroy 2007: 152) – though the detailed constraints on its
use might not be borrowed, as we have seen (Buchstaller and D’Arcy 2009).
By contrast, ‘under the counter’ features require relatively close-knit networks
for their transmission, and these tend to be linguistically more complex and
localised. The only other variable in our data which shows the same pattern
of increasing frequency with increasing age is GOOSE-fronting, and this, too, is
cited by Milroy as an off the shelf feature. GOOSE-fronting is widespread in the
English-speaking world and, in contrast to vowel changes generally, it is not
linked to (i.e. embedded in) changes elsewhere in the phonology of the particular
variety (see review in Mesthrie 2010: 3–4). It ‘does not appear to be sensitive to
local social indexicalities or dependent on participation in close-knit networks’
(Milroy 2007: 165). It is, then, highly ‘available’ and, linguistically speaking,
simple to acquire. As ‘global’ features, both are constantly ‘available’ in the
general linguistic community, as we pointed out above. Because these changes
are both global and rapid, face-to-face contact cannot be a necessary condition
for people to acquire them, and nor do they involve adult-to-adult diffusion (cf.
Meyerhoff and Niedzielski’s 2003 discussion of the spread of such features). With
this in mind, we can see that BE LIKE in our North London data shows exactly
the age distribution that is expected: developmental factors must be taken into
account, but once acquired, the frequency increases gradually with advancing
age until it peaks in the 16–19-year-old group. Older speakers use BE LIKE,
but less frequently than the adolescents. Although developmental factors do not
count in the same way in the acquisition of a vowel quality, a similar pattern
emerges for GOOSE-fronting.
We turn now to the particular sociolinguistic conditions of inner-city London,
and the origin and spread of this is + speaker. It is even newer than BE LIKE: it is in
the early stages and, at present at least, it is confined to inner London. Its spread is
less rapid than BE LIKE (it still represents a minor part of the quotative system) and
its origins do not fit conventional models of change. The origins of this is + speaker
lie, we have argued, in contact, but the primary agents are not adults: some of the
primary caregivers are only just beginning to acquire English themselves, and
cannot act as a target model for their children’s acquisition of English. Instead,
the agents are children growing up in a multilingual, multiethnic community
and acquiring English in an unguided informal setting from their peers. Of course,
some children may initially acquire the English of the primary caregiver: this
is likely for the Anglo children in our sample, as well as the Afro-Caribbeans
and some of the Indian and Black African children who speak English at home,
C Blackwell Publishing Ltd. 2011
180
CHESHIRE ET AL.
though their Englishes will be quite diverse. Other children, however, who speak
a language other than English at home, are likely to acquire English from their
peers or their older siblings, with varying degrees of formal language teaching
from the school.
The early English of many of these children, we assume, is a range of what
can be considered as interlanguage varieties. The 4–5 year olds in our North
London sample are very fluent but their syntax is not native speaker-like, and
the same can be said of some of the older non-Anglo children. Consider, for
example, the forms used for negation by Yeliz, a five-year-old Turkish girl,
as well as her subject pronoun form him and uninflected third singular verb
form go:
14. Yeliz:
I not got a pony .. not got it I not got the pony now
Arfaan: you haven’t? where is it? has it run away? where is he? has he
run away? has he gone?
Yeliz:
him go there
There is no clear target for acquisition, and norms are consequently fluid.
We have argued that the concept of the feature pool best fits the linguistic
situation. Perhaps the 8–9 year olds show how this is + speaker develops
from the feature pool: it is acquired first (‘created’ might be a better term) as
a high-involvement deictic form used with gestures and to act out moments
within a narrative, as well as to quote speech. This kind of animated deixis is
probably part of younger children’s style, and its use in this context may be
an example of an innovation on the part of children. As shown above, our
North London data shows that as a quotative it is more often used when the
content of the quote is a gesture or a nonlexicalised sound. This too may
be part of young children’s style. As a salient form, its emergence is in line
with the constraints found to be relevant in other language contact situations,
though we have argued that discourse function is also relevant. The uses of
this is + speaker are then refined with increasing age and increasing language
proficiency, perhaps alongside a decrease in animated deixis, until it is used
almost exclusively as a quotative expression by the 16–19-year-old group.7 This
increase in specificity as children grow older is in line with the incrementation
model, even if the peak comes earlier than expected: what we have is a link
from incrementation back to the social and linguistic maturation of the child.
This is not to say that this factor is always present in incrementation. It is likely,
instead, that it is a consequence of the particular sociolinguistic set-up under
discussion.
The way in which this is + speaker is generated, within the community as
an output from a feature pool, accounts for the different age distribution, with
its much earlier peak, of this innovation compared to the distribution that is
associated with exogenous changes such as BE LIKE. In the same vein, we can
argue that the acquisition of the MLE vowel qualities, with the exception of GOOSEfronting, is a result of the same feature pool with its specific ecological conditions
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
181
of unguided second-language acquisition: monophthongal FACE and GOAT are
characteristic of Caribbean, African and subcontinental Englishes, creoles and
pidgins, and are found in many foreign-language varieties too.
5.2 Past tense BE
We turn now to a well-known morphosyntactic variable, to explore further the
social dynamics of changes taking place in multiethnic areas of inner London.
Usage of was and were as past tense forms of BE varies immensely around the
English-speaking world. In addition to the standard English pattern, where was
occurs with first and third singular subjects, and were with second person and
plural subjects, there are two dominant patterns of (paradigmatic) levelling
involving non-standard forms:
1. Variable use of was throughout the past tense paradigm, e.g. we was wasn’t
we?; I was wasn’t I?
2. Variable use of was in positive contexts and were in negative contexts, e.g.
we was weren’t we?; I was, weren’t I?
Both the levelled patterns respond to system-internal pressures, but in different
ways. Pattern (1) results in a uniform past tense paradigm. It is said to be a
basic ‘vernacular primitive’ (Chambers 1995) and is acquired in child language
(Brown 1973), second-language English interlanguages (Schumann 1978) and
decreolising English Creoles where, according to Bickerton (1975: 115) was
occurs first as a lexical insertion, followed later by were as speakers use more
standard English forms. The was/wasn’t pattern is the most common, occurring
throughout the English-speaking world. Pattern (2) brings past forms of BE into
line with English will/won’t, do, does/don’t and am, is, are/ain’t, and responds
to the typological tendency to have a distinctive negative form. This pattern
is widespread in England (Britain 2002b) (and also occurs in some relatively
isolated mid-Atlantic coastal communities in the U.S., see Schilling-Estes and
Wolfram 1994). Although in some large urban centres in the U.K. non-standard
was seems to be declining, non-standard weren’t is increasing (for further details,
see Cheshire and Fox 2009).
Our first project found clear differences between inner-city Hackney and the
outer London borough of Havering. Table 5 shows that the Havering 16–19 year
olds have the pattern that is widespread in England: relative to the more standard
elderly speakers in Havering they show an increase in the use of non-standard
was and they have a very frequent use of non-standard weren’t.
There is no use of non-standard wasn’t in Havering. The adolescents in
Hackney, on the other hand, have both levelled patterns. In positive contexts
they use non-standard was somewhat less frequently than the older Hackney
speakers, but in negative contexts their use of both non-standard wasn’t and
non-standard weren’t is higher than that of the older speakers (again, for more
details of the analysis, see Cheshire and Fox 2009).
C Blackwell Publishing Ltd. 2011
182
CHESHIRE ET AL.
Table 5: Use of non-standard was, wasn’t and weren’t in Hackney and Havering elderly
and adolescent age groups
Havering
Age 65+
Positive contexts:
Non-standard was
Negative contexts:
Non-standard wasn’t
Non-standard weren’t
Hackney
16–19
65+
16–19
%
N
%
N
%
N
%
N
19
98
58
410
52
268
42
615
-∗
41
51
69
180
30
17
20
69
46
41
39
220
∗ numbers of contexts for non-standard wasn’t in Havering are too few to comment on
We have proposed (Cheshire and Fox 2009) that language contact explains
the difference between Havering and Hackney. In Hackney the linguistic
backgrounds of the adolescents include English-lexifier Creoles, African English
and Indian and Bangladeshi English, all of which favour the was/wasn’t pattern.
Interlanguage varieties of English also favour this pattern. Although numbers
of speakers in different ethnic groups were low, we observed high rates of
both non-standard was and wasn’t among speakers from ethnic groups where
these Englishes were found, while the Anglo adolescents tended to favour the
was/weren’t pattern. Friendship patterns were relevant in that speakers with
multiethnic friendship networks, including Anglo speakers, tended to favour
was/wasn’t. MLE differs from other varieties of English in England, then, in
exhibiting both the was/wasn’t and the was/weren’t pattern (as well, of course,
as the standard English pattern). A question that arises is whether one of these
patterns will come to dominate in London.
Table 6 shows the frequencies in North London where there is a different
ethnic mix, and where we can analyse the age distribution of the non-standard
forms. In positive contexts, use of non-standard was by the 16–19 age group
is approximately the same as that of the 16–19 year olds in Hackney (37% in
North London, 42% in Hackney). The frequencies for both older and younger
speakers, however, are higher. The peak, this time, is in the 12–13-year-old age
group. Unlike other urban centres in England, then, in London non-standard
was does not seem to be declining, though there is a falling off in late adolescence
when speakers use more standard forms.
Table 7 now shows the frequencies of non-standard was for the different ethnic
groups in the North London sample (note that numbers of speakers are quite low
in these groups). The number of tokens for the 4–5 year olds is low, so we do not
include this age group in the analysis. The peak in the 12–13-year-old group is
maintained in each ethnic group. The age distribution is the same, then, for both
the Anglo and the non-Anglo groups. The Anglo caregivers use non-standard
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
183
Table 6: Use of non-standard was, wasn’t and weren’t in North London
Age 4–5
Positive contexts:
Non-standard was
Negative contexts:
Non-standard wasn’t
Non-standard weren’t
8–9
N
12–13
16–19
Caregivers
%
%
%
%
N
%
N
N
N
78
9
56
162
77208
37357
45
157
100
33
1
6
50
25
10
32
73 15
22 54
48 27
32 89
27
22
15
59
Table 7: Use of non-standard was in North London in positive contexts – age and
ethnicity
Age 8–9
12–13
16–19
Ethnicity (speaker N)
%
N
%
%
‘Other’ (11)
Anglo (27)
Turkish/Kurdish (8)
Black African (9)
Black Caribbean (18)
Mixed race white/
Black Caribbean (4)
Bangladeshi (2)
All ethnicities
24
65
71
43
72
17
49
28
23
25
90 39
71 108
80 25
70 20
75
4
37 19
36 100
26 68
45 49
36 95
47 116
39 41
63 75
53 373
48 121
50 92
43 165
78
9
0 11
56 162
92 12
77 208
54 26
37 357
45 157
68
0
-
N
N
Caregivers All ages
%
N
%
N
47
11
-
was with approximately the same frequency as elderly speakers in Hackney (47%
in North London and 51.5% in Hackney), confirming that non-standard was is
not in decline in London.
It is, of course, the non-standard negative forms that allow us to see which of
the different levelling patterns is being followed. Table 6 showed that, for each
age group, non-standard wasn’t is more frequent than non-standard weren’t;
again we see a peak in the use of non-standard wasn’t in the 12–13-year-old
age group. What is not yet clear is whether non-standard wasn’t forms are
favoured by the non-Anglo speakers, as for the Hackney adolescents, or whether
in North London, and especially in younger age groups, they are used equally
by Anglo and non-Anglo speakers. Negative forms do not occur often enough in
the data to allow a breakdown by ethnic group, so we combine the non-Anglo
speakers into a single ‘non-Anglo’ group (for the caregivers, the non-Anglo group
consists entirely of Afro-Caribbean speakers). Table 8 gives the frequencies for
non-standard wasn’t and non-standard weren’t (the empty cell for non-Anglo
8–9 year olds indicates that there were no data for this age group).
Numbers are low for plural subjects (the site for non-standard wasn’t) even
when speakers are combined in this way, but even so it is clear from Table 8
C Blackwell Publishing Ltd. 2011
184
CHESHIRE ET AL.
Table 8: Use of non-standard wasn’t/weren’t in North London in negative contexts –
age and ethnicity
Age 8–9
Ethnicity
Wasn’t
Anglos
Non-Anglos
Weren’t
Anglos
Non-Anglos
12–13
16–19
Caregivers
Total
%
N
%
N
%
N
%
N
%
N
0
-
1
-
50
100
8
7
50
50
4
22
29
100
14
1
38
60
26
30
78
0
9
21
39
4
28
23
41
24
17
67
32
0
40
19
40
13
94
130
that non-Anglo speakers prefer non-standard wasn’t to non-standard weren’t,
and that the Anglo speakers use both patterns. For the 16–19 year olds, the
frequencies for both non-standard negative forms are roughly the same as for
the same age group in Hackney. This confirms, then, that use of both patterns
is characteristic of MLE. We can no longer, though, discern a peak in the 12–13
age group, as we could for non-standard was in positive contexts (we discount
the figures for non-standard wasn’t for which the number of tokens is low). For
non-standard weren’t, the age distribution after 8–9 (a group for which again the
number of tokens is low) is flat. The 16–19 year olds use non-standard weren’t
with the same frequency of 41 percent as their peers in Hackney (see, again,
Table 5); a much lower frequency than adolescents in Havering, who follow the
general pattern attested throughout England of high frequencies of non-standard
weren’t.
Again, the concept of the feature pool is helpful in explaining the differences
between Havering and the two multiethnic areas of inner London. This time
the main factor governing the outputs from the pool is clearly frequency. The
input varieties, we assume, contain large numbers of was forms in positive
contexts: they include interlanguage varieties, Creole-influenced varieties,
perhaps African and Indian varieties of English. Non-standard was is also used in
London English (elderly speakers in Hackney used this form with an overall rate
of 52 percent, as we saw earlier, and the Anglo caregiver generation in North
London also uses it with approximately the same frequency (47%). Furthermore,
in London non-standard were does not occur at all in positive contexts: the form
used with first and third person singular past forms of BE is always was (Cheshire
and Fox 2009). First and third singular subjects are more frequent in speech
than other subjects, and speakers use past BE frequently. In negative contexts,
frequencies of non-standard weren’t (i.e. with first and third singular subjects)
are relatively low, at least compared to outer London. This means that with first
and third singular subjects was is the more frequent form, making was the most
frequent form overall in negative contexts as well as in positive contexts. There is
enormous pressure, then, to use was, since was must dominate the feature pool.
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
185
Table 9: Use of non-standard was in positive contexts by
Abigail
Subject
First person we
Second person you
Third person they
Total
No. of tokens
11/12
1/1
8/8
20/21
% was
92
100
100
95
Although frequency seems to be the over-riding factor here, the increased use of
the was/wasn’t pattern may also reflect a drive towards economy in the system
(also relevant in the selection of certain features over others in contact situations
generally; see Siegel 1997: 139): the use of two forms, was and were, to indicate
past tense can be seen as redundant.
Like this is + speaker, changes in the use of past forms of BE constitute neither a
conventional transmitted change nor a conventional diffused change. The high
frequency of non-standard was in inner London is out of step with frequencies
outside London, and the use of two levelled patterns (was/wasn’t as well as
was/weren’t) is also different from elsewhere in England. We assume that these
uses are generated within the speech community, and owe much to the presence
of speakers from a range of different language backgrounds (the non-Anglos).
Numbers of tokens were low for negative contexts, but even so there is little
evidence of incrementation with increasing age, and there is no adolescent peak.
For positive contexts the number of tokens allows us to state with confidence
that the age distribution does not follow the incrementation pattern: if there is
a peak, it is in the 12–13-year-old age group. Indeed, rather than finding an
adolescent peak, the rates of non-standard was decrease sharply for the 16–19
age group. If we infer usage of the standard English variants from Table 8, it is the
standard English pattern that has become dominant for this age group (as well as
for the Anglo caregivers). An explanation for this apparent standardisation does
not immediately present itself. Maturation is one explanation, but can only be
a contributory factor because non-standard features are not strongly subject
to maturational change. However, the feature pool contains many features
that speakers are exposed to, including standard features; it may be that with
increasing age speakers become more inclined to select standard features from
the pool in inner London than in more focused speech communities (in the sense
of Le Page and Tabouret-Keller 1985).
The importance of the non-Anglo speakers in determining the output of
the feature pool can be seen by examining the usage of speakers who have
recently arrived in London, to see which forms they choose to use. Table 9
shows the frequency of use of non-standard was by Abigail, a 12-year-old
Albanian girl who lived in London between the ages of 4 and 7, then returned
C Blackwell Publishing Ltd. 2011
186
CHESHIRE ET AL.
Table 10: Use of quotatives – Abigail compared with other speakers in her age group
Innovators project:
Hackney young
Quotative
%
Say
Think
Go
Zero
BE LIKE
This is (subject)
Tell
Other
Total N
27
13
12
15
24
5
2
2
N
351
164
150
193
313
61
24
26
1282
MLE project:
Age 12–13
%
25
2
24
14
26
2
0.3
2
N
163
12
153
93
166
13
2
10
642
MLE project:
Abigail
%
N
26
7
37
28
2
11
3
16
12
1
43
to Albania until she was 11, the age when she came back to live permanently in
London.
Her English is fast and fluent but she is sometimes lost for a specific word
and her syntax is often non native-speaker-like. Her use of non-standard was is
almost categorical. She uses only three negative past tense forms, all of which are
non-standard wasn’t. Perhaps her usage reflects the vernacular primitive form
that is typical of interlanguage varieties of English (as mentioned earlier; again,
see Chambers 1995); perhaps, also, it reflects her desire to be recognised as a
member of her peer group – especially acute in her case because she is a recent
member whose English is not yet entirely that of a native speaker (of course,
the concept ‘native speaker’ is complex in this kind of linguistic setting). It is
noteworthy that Abigail also has very high frequencies of this is + speaker and
BE LIKE (Table 10); in fact she contributes almost all of the this is + speaker
tokens for her age group. Her production of MLE vowels is also very advanced
(see Fox and Torgersen 2009). Abigail’s usage seems a heightened example of
the contribution of speakers who have acquired English as a second language
and helps to further explain the social dynamics of the emergence of MLE: in their
efforts to be accepted in the multiethnic friendship groups that are characteristic
of the group second-language situation, some non-Anglo pre-adolescents may
overshoot the usage of the age groups immediately below them, and above them.
5.3 Simplification of indefinite and definite article allomorphy
Finally, we turn to the use of indefinite article [ə] and definite article [ðə] plus
glottal stop before word-initial vowels, instead of standard and mainstream [ən]
and [ði], and show that simplification in the allomorphy system is led by ethnicminority young people. These variables have only recently been researched
within sociolinguistic studies but Fox (2007) showed that high frequency use of
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
187
Table 11: Use of prevocalic indefinite article [ə] plus hiatus-resolving [ʔ]
Age 4–5
N
8–9
%
12–13
N
%
N
16–19
Ethnicity
%
Anglo
Black Caribbean
Black African
Mixed race Anglo/
Black Caribbean
Turkish
‘Others’
100 1
100 11
50 4/8
100 13
100 6
26 10/38
100
4
50
3/6
11 5/44
78 29/37
70 7/10
100 1
100 1
100 1
0
5
67 2/3
50 1/2
67
80
100
50 2/4
79 11/14
67 4/6
2/3
4/5
8
%
Caregivers
N
%
N
2 1/51
32 6/19
-
-
Table 12: Percentage use of prevocalic definite article [ðə] plus hiatus-resolving [ʔ]
Age 4–5
N
8–9
%
12–13
N
%
N
16–19
%
N
Caregivers
Ethnicity
%
Anglo
Black Caribbean
Black African
Mixed race Anglo/
Black Caribbean
Turkish
‘Others’
100 4
100 6
53 9/17 47 27/57 28 33/119 9 16/187
78 18/23 30 3/10 70 64/91 49 40/81
100 8
100 9
75 45/60 -
0
1
100 2
100 7
33 3/9
25 3/12
85 11/13
23 3/13 75 9/12
81 13/16 77 34/44
84 16/19 69 9/13
%
-
N
-
prevocalic [ə] and [ðə] among Bangladeshi male adolescents in London appeared
to be influencing the English of their white Anglo male peers with respect
to these features, with multiethnic friendship networks playing a key role in
their diffusion. Similarly, Gabrielatos et al. (2010) found that non-Anglo males
had the highest use of indefinite article a before vowel-initial words in their
analysis of the Linguistics Innovators Corpus (LIC). Furthermore, this change is
not restricted to London. Guzzo, Britain and Fox (2008) demonstrated the use of
these innovations among third generation immigrants of Italian descent in the
borough of Bedford where, once more, multiethnic peer group networks appear
to provide the means of transmission between users. In our analysis of the North
London data we find again that the change appears to be led by non-Anglos
with high frequencies of use among all age groups. The results are presented in
Tables 11 and 12.
The use of articles in prevocalic position is fairly infrequent (particularly the
indefinite article) so the numbers of tokens are low in some cells; nevertheless, a
distinct and similar pattern emerges in both cases. First of all, if we look at the
Anglos we see that there is a decrease in frequency of use of both [ə] and [ðə]
with increasing age. Studies which have looked at developmental trends in child
C Blackwell Publishing Ltd. 2011
188
CHESHIRE ET AL.
language acquisition of these features suggest that children acquire the standard
rules and adopt more ‘adult-like’ pronunciations between the ages of 4 and 5
years old with full acquisition almost complete at around the age of 7 (Newton
and Wells 1999: 70). The 4–5 year olds in our study have almost categorical
use of [ə] and [ðə] and support this argument. The assumption here, though, for
children acquiring these rules is that the input that the children receive is always
the standard. This may help to explain why the Anglo speakers have lower usage
of the [ə] and [ðə] variants relative to the non-Anglo speakers. There is more
competition from the input they receive – the newer forms of their non-Anglo
peers against the standard forms of their caregivers (who rarely use [ə] and
[ðə]). By contrast, the high use of [ə] and [ðə] among the non-Anglos of all age
groups seems to be consistent with other studies of contact varieties of English
where these forms have been reported in varieties such as AAVE (Labov 1972b;
Mufwene 2001), Afrikaans English (Watermeyer 1996), South African English
(Lass 1995) and Singapore English (Fraser-Gupta p.c. 2005). The much higher
frequency of use among the Black Caribbean caregivers also suggests that these
forms have arisen among speakers of contact varieties of English. Peter Patrick
(p.c. 2005) reports that Jamaican English has [ðə] whether stressed or unstressed
and whether it occurs before a vowel or not, and so this is a likely component
of the feature pool. Furthermore, we could speculate that the standard rules
may not be fully acquired among second-language learners of English. Krashen
(1987: 95) notes that users of English as a second language often fail to make the
a/an distinction. It seems likely, therefore, that the caregivers who are acquiring
English will also have the newer forms in their varieties of English. So, for the
young non-Anglo speakers the dominant variants in the feature pool are [ə] and
[ðə], both from the majority of their peers and their caregivers, with much less
competition from the traditional [ən] and [ði] variants. This can account for the
essentially flat age distribution of [ə] and [ðə] among the non-Anglos: for them,
there is little evidence of change. Change, however, is found among the Anglos,
in that the teenagers’ frequency is higher than the caregivers. We cannot talk
of an adolescent peak, however, because of the interaction between opposing
forces: maturation, leading to a gradual shift towards the standard forms with
age, and contact with non-Anglo peers, who are high users of the non-standard
variants. This pattern, we suggest, is characteristic of many contact situations
including those with group second-language learning.
We would argue, then, that this is another feature of language change which
has its origin in language contact. More generally it is a simplification of strategies
for hiatus resolution in English (Britain and Fox 2009). The clear patterns suggest
that the dominant variants in the feature pool are [ə] and [ðə], which are
being acquired by the majority of non-Anglos and to a lesser extent by the
Anglo speakers, perhaps due to the different distributions of the variants in
the input varieties that the different groups of speakers are exposed to. Again,
then, the output is governed by frequency of the forms that are available in
the feature pool (articles in prevocalic position are less frequent than articles
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
189
in preconsonantal position); it may also be constrained by a drive to reduce
redundancy in the definite and indefinite article systems, resulting in the use of
both a single indefinite form and a single definite form.
In conclusion, the analyses we have conducted for features beyond phonology
suggest that the emergence of the set of contact features we call Multicultural
London English in the present-day inner city is best seen as the result of group
second-language acquisition, and most usefully conceptualised in terms of a
feature pool. The ‘output’ varieties from the pool reflect competition within the
pool between the various input varieties (and may well involve restructuring).
Selection of features from the pool is constrained by the same factors that have
been found relevant in other language contact situations. For some features
the output reflects both the frequency of features in the input varieties and
social factors (such as multiethnic friendship patterns or the desire of non-native
speakers to be accepted by their peer group). We illustrated this with the example
of was/were variation and the non-standard indefinite and definite article forms.
For other features the output cannot be explained in terms of frequency but
by other factors that govern selection in language contact generally, such as
salience. The output may (also) reflect ‘natural’ communicative strategies, as
for the this is + speaker innovation. By comparing these innovations with more
conventional changes, such as the spread of the BE LIKE quotative, we see that
the role of children and adolescents in these contact-based changes does not fit
with conventional models of language change. Nevertheless the role of children
and adolescents is as important for these types of change as it is for others.
6. THE EMERGENCE OF MULTIETHNOLECTS AS A TYPOLOGICALLY
DISTINCT MODE OF DIALECT FORMATION
In the emergence of these new contact-based varieties, there seem to be
differences in the role of age in changes in phonology, morphosyntax and
discourse markers. In each case, a particular age group seemed to be in the
lead: for phonology, to the extent that differences between the child/adolescent
age groups were present, we see an adolescent peak for just one of the features
studied, GOOSE-fronting. However, even the youngest children appear to have
rejected local adult English models, in a way that is less true of non-contact
situations where changes are transmitted intergenerationally. Those whose
primary caregivers speak non-local varieties of English appear to have rejected
these, too. This gives rise to a flat age distribution among the child speakers, or
one in which the youngest speakers appear to be in the lead. We argued that
this was because, in this context of group second-language acquisition, adults
do not constitute a target-language model. We can see this particularly clearly
in the discussion of the indefinite and definite articles. Either this is because the
child perceives the parent as too remote from community norms (as we argued
for the high back GOOSE variant), or else the parent does not provide any model
at all. This leads to absence of any adolescent peak.
C Blackwell Publishing Ltd. 2011
190
CHESHIRE ET AL.
For BE LIKE there is a clear adolescent peak, in line with the incrementation
model. For the remaining non-phonological features, for which we claim a
community-internal origin, the peak, if present, is earlier, falling within the
12–13 age group. We argued that the child selects from a feature pool, and is
guided to some extent by frequency (as in the favouring of the was/wasn’t pattern)
or by salience and general communicative strategies (as in the use of the this is +
speaker quotative). In the case of the indefinite and definite articles, maturation
(favouring acquisition of the standard) interacts with the presence of a feature
pool in which the non-standard forms are heavily represented, leading in the
long run to convergence between Anglo and non-Anglo groups. Interaction
between maturation and the presence of a feature pool is also arguably evident
in the case of was/were variation.
However, the adolescent peak is present in the case of GOOSE-fronting, the only
phonological feature which is global and not a local innovation (as we argue
is the case for the remaining phonological features). In this respect, it patterns
identically to the global quotative BE LIKE. In its adoption of these two features,
London does not differ from other types of speech community.
What distinguishes London’s inner city from other cases of language change
generated community-internally is the presence of an exceptionally high
proportion of speakers of language varieties other than the local variety, here,
London English – 50 percent or higher in parts of some boroughs, sustained over a
considerable time by continued immigration. By using a quantitative variationist
methodology, we gain a detailed and dynamic picture of new forms of English
in this multilingual environment: we see the development of a set of innovative
linguistic features (radical diphthong changes and the discourse-pragmatic form
this is + speaker), new, divergent distributions of existing forms (past tense
forms of BE), a big increase in a latently present feature (simplification of article
allomorphy) and, finally, the adoption of two global changes (GOOSE-fronting and
BE LIKE). Individual speakers use these features variably, and we have labelled
the resulting ‘variety space’ Multicultural London English, in recognition of
the fact that the features are only loosely associated with specific ethnicities or
language backgrounds.
Real-time data allows us to place the beginnings of MLE to some time in the
early 1980s. Looking in detail at usage within very narrow age bands allows
us to see how the various features comprising MLE are differentially distributed
across the lifespan. In doing so, we can see differences and similarities between
MLE and structured variation in canonical monolingual urban communities.
Acquisition and age distributions are different for MLE, but the variation is no
less structured, both linguistically and socially. In this respect, MLE speakers are
part of a speech community.
By contrast, insights from second-language acquisition and creole studies
allow us to argue that acquisition is fundamentally different from that in
canonical speech communities. In particular, the linguistic conditions needed
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
191
for the transmission of a local dialect are only present to a limited extent. Instead,
group second-language acquisition gives rise to much of the initial set of available
variants, and concepts taken from research into language contact, most notably
the notion of a feature pool, help us model the selection of linguistic features taken
from the original set. The factors constraining the selection of features from the
pool conform to those that are relevant in other language contact situations.
Although the circumstances giving rise to this type of dialect formation are
complex, they are today far from unique. We would argue that, at least in Europe,
they are sufficiently common – and well enough understood – for the formation
of multiethnolects to be seen as a distinct and important type of community
language change.
NOTES
1. Linguistic innovators: the English of adolescents in London 2004–7, funded by the
Economic and Social Research Council, Principal Investigator Paul Kerswill, Coinvestigator Jenny Cheshire, Research Associates Susan Fox and Eivind Torgersen
(ref. RES 000-23-0680). See Kerswill, Torgersen and Fox (2008) and Cheshire and
Fox (2009).
2. Multicultural London English: The emergence, acquisition and diffusion of a new
variety 2007–10, funded by the Economic and Social Research Council, Principal
Investigator Paul Kerswill, Co-investigator Jenny Cheshire, Research Associates
Susan Fox, Arfaan Khan and Eivind Torgersen (ref. RES-062-23-0814).
3. Following Wells (1982), this word is used mnemonically to represent the vocalic
lexical set of words containing this vowel. We follow Wells’ system of mnemonics
throughout this article.
4. Here and in the remainder of the article, the vowels shown were analysed using Praat.
Measurements were taken of the onset of diphthongs (at 35 ms after onset of voicing).
The LPC pole prediction order that corresponded best with the location of the formants
on the spectrogram was used for each measurement. The data was normalised using
the Lobanov formula (Lobanov 1971).
5. The statistical results are from a univariate linear mixed model (i.e. it has speaker as a
random factor to account for dependence between observations on the same speaker).
The statistical testing was carried out in Sabre, which is a package for R that analyses
multi-process random effects data (Crouchley, Berridge and Grose 2008).
6. Transcription conventions:
All names have been anonymised
<> paralinguistic features and transcriber comments are enclosed within angled
brackets, for example <coughs> or <mobile phone rings>
.
short pause of up to approximately 0.5 second
..
medium pause of between approximately 0.5 and 1.0 second (further ‘.’
indicate further time segments of up to approximately 0.5 second)
xxxx indecipherable speech (each ‘x’ approximating to one syllable)
“”
reporteds speech including non-lexicalised sound and gesture
7. On the other hand, without more conclusive data, we have to admit an alternative
explanation whereby it is acquired, or created, first by the adolescent age group as an
C Blackwell Publishing Ltd. 2011
192
CHESHIRE ET AL.
innovative quotative, and is then taken up by younger speakers with an expansion
of functions.
REFERENCES
Androutsopoulos, Jannis. 2007. Ethnolekte in der Mediengesellschaft. Stilisierung
und Sprachideologie in Performance, Fiktion und Metasprachdiskurs. In Christian
Fandrych and Reinier Salverda (eds.) Standard, Variation und Sprachwandel in
germanischen Sprachen [Standard, Variation and Language Change in Germanic
Languages]. Tübingen, Germany: Narr. 113–155.
Baker, Philip and John Eversley (eds.). 2000. Multilingual Capital: The Languages
of London’s Schoolchildren and their Relevance to Economic, Social and Educational
Policies. London: Battlebridge.
Bickerton, Derek. 1975. Dynamics of a Creole System. New York: Cambridge University
Press.
Britain, David. 2002a. Space and spatial diffusion. In J. K. Chambers, Peter Trudgill
and Natalie Schilling-Estes (eds.) The Handbook of Language Variation and Change.
Oxford, U.K.: Blackwell. 603–637.
Britain, David. 2002b. Diffusion, levelling, simplification and reallocation in past tense
BE in the English Fens. Journal of Sociolinguistics 6: 16–43.
Britain, David and Sue Fox. 2009. The regularization of the hiatus resolution
system in British English: A contact-induced ‘vernacular universal’? In Markku
Filppula, Juhani Klemola and Heli Paulasto (eds.) Vernacular Universals and Language
Contacts: Evidence from Varieties of English and Beyond. London: Routledge. 177–
205.
Brown, Roger. 1973. A First Language: The Early Stages. Cambridge, Massachusetts:
Harvard University Press.
Buchstaller, Isabelle and Alexandra D’Arcy. 2009. Localized globalization: A multilocal, multivariate investigation of quotative be like. Journal of Sociolinguistics 13:
291–331.
Castells, Manuel. 2000. The Information Age, Economy, Society: Volume I – the Rise of
the Network Society (second edition). Oxford, U.K.: Blackwell.
Chambers, J. K. 1995. Sociolinguistic Theory. Oxford, U.K.: Blackwell.
Chambers, J. K. 2002. Dynamics of dialect convergence. Journal of Sociolinguistics 6:
117–130.
Cheshire, Jenny and Sue Fox. 2009. Was/were variation: A perspective from London.
Language Variation and Change 21: 1–38.
Cheshire, Jenny, Sue Fox, Paul Kerswill and Eivind Torgersen. 2008. Ethnicity,
friendship network and social practices as the motor of dialect change: Linguistic
innovation in London. Sociolinguistica 22: 1–23.
Clyne, Michael. 2000. Lingua franca and ethnolects in Europe and beyond.
Sociolinguistica 14: 83–89.
Crouchley, Rob, Damon Berridge and Dan Grose. 2008. An alternative package for
estimating multivariate generalised linear mixed models in R. Paper presented at
the R user conference, 12–14 August, Dortmund, Germany.
Eckert, Penelope. 2003. Elephants in the room. Journal of Sociolinguistics 7: 392–
397.
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
193
Fox, Susan. 2007. The demise of Cockneys? Language change among adolescents
in the ‘traditional’ East End of London. Unpublished PhD dissertation. Colchester,
U.K.: University of Essex.
Fox, Susan. In press. Performed narrative: The pragmatic function of this is + (speaker)
and other quotatives in London adolescent speech. In Ingrid von Alphen and Isabelle
Buchstaller (eds.) Quotatives: Cross-Linguistic and Cross-Disciplinary Perspectives.
Amsterdam, The Netherlands: John Benjamins.
Fox, Susan and Eivind Torgersen. 2009. The acquisition of Multicultural London
English: A case study. Paper presented to ICLCE 3 (International Conference on the
Linguistics of Contemporary English), 14–17 July, London.
Gabrielatos, Costas, Eivind Torgersen, Sebastian Hoffmann and Susan Fox. 2010. A
corpus–based sociolinguistic study of indefinite article forms in London English.
Journal of English Linguistics 38: 297–334.
Güldemann, Tom. In press. Speaker-predicating quotative indexes as a crosslinguistic type. In Ingrid van Alphen and Isabelle Buchstaller (eds.) Quotatives:
Cross-Linguistic and Cross-Disciplinary Perspectives. Amsterdam, The Netherlands:
Benjamins.
Guzzo, Siria, David Britain and Susan Fox. 2008. From L2 to ethnic dialect: Hiatus
resolution strategies in Bedford Italian English. Paper presented at the International
Conference on Global English, 14–16 February, University of Verona, Italy.
Hackney Council. 2010. Languages spoken in Hackney. Last accessed 30 January
2011 at http://www.hackney.gov.uk/xp-factsandfigures-languages.htm
Hickey, Raymond. 2006. Contact, shift and language change: Irish English and South
African Indian English. In Hildegard L. C. Tristram (ed.) Celtic Englishes IV. Potsdam,
Germany: University Press. 234–258.
Horvath, Barbara. 1985. Variation in Australian English. The Sociolects of Sydney.
Cambridge, U.K.: Cambridge University Press.
Inwood, Stephen. 1998. A History of London. London: Pan Macmillan.
Kerswill, Paul. 1996. Children, adolescents and language change. Language Variation
and Change 8: 177–202.
Kerswill, Paul, Eivind Torgersen and Susan Fox. 2008. Reversing ‘drift’: Innovation
and diffusion in the London diphthong system. Language Variation and Change 20:
451–491.
Kerswill, Paul and Ann Williams. 2000. Creating a new town koine: Children and
language change in Milton Keynes. Language in Society 29: 65–115.
Kerswill, Paul and Ann Williams. 2002. ‘Salience’ as an explanatory factor in language
change: Evidence from dialect levelling in urban England. In Mari C. Jones and Edith
Esch (eds.) Language Change: The Interplay of Internal, External and Extra-Linguistic
Factors. Berlin, Germany: de Gruyter. 81–110.
Kerswill, Paul and Ann Williams. 2005. New towns and koineisation: Linguistic and
social correlates. Linguistics 43: 1023–1048.
Kerswill, Paul, Jenny Cheshire, Sue Fox and Eivind Torgersen. In press. English as a
contact language: The role of children and adolescents. In Marianne Hundt and
Daniel Schreier (eds.) English as a Contact Language. Cambridge, U.K.: Cambridge
University Press.
Khan, Arfaan. 2006. A sociolinguistic study of Birmingham English: Language
variation and change in a multiethnic British community. Unpublished PhD
dissertation. Lancaster, U.K.: Lancaster University.
Kotsinas, Ulla-Britt. 1988. Rinkebysvenska – en dialekt? [Rinkeby Swedish – a
dialect?]. In Per Linell, Viveka Adelswärd, Torbjörn Nilsson and Per A. Petersson
C Blackwell Publishing Ltd. 2011
194
CHESHIRE ET AL.
(eds.) Svenskans beskrivning 16. Linköping, Sweden: Universitetet i Linköping. 264–
278.
Krashen, Stephen D. 1987. Principles and Practices in Second Language Acquisition.
New York: Prentice-Hall.
Labov, William. 1972a. Sociolinguistic Patterns. Philadelphia, Pennsylvania:
University of Pennsylvania Press.
Labov, William. 1972b. Language in the Inner City: Studies in Black Vernacular English.
Philadelphia, Pennsylvania: University of Pennsylvania Press.
Labov, William. 1989. Exact description of the speech community: Short a in
Philadelphia. In Ralph W. Fasold and Deborah Schiffrin (eds.) Language Change
and Variation. Amsterdam, The Netherlands: Benjamins. 1–57.
Labov, William. 1994. Principles of Linguistic Change. Volume 1: Internal Factors.
Oxford, U.K.: Blackwell.
Labov, William. 2001. Principles of Linguistic Change. Volume 2: External Factors.
Oxford, U.K.: Blackwell.
Labov, William. 2007. Transmission and diffusion. Language 83: 344–387.
Labov, William. 2010. Principles of Linguistic Change. Volume 3: Cognitive and Cultural
Factors. Oxford, U.K.: Blackwell.
Lapidus, Naomi and Ricardo Otheguy. 2005. Overt non-specific Ellos in Spanish in
New York. Spanish in Context 2: 157–174.
Lass, Roger. 1995. South African English. In Rajend Mesthrie (ed.) Language and Social
History: Studies in South African Sociolinguistics. Cape Town, South Africa: David
Phillip. 89–106.
Le Page, Robert B. and Andrée Tabouret-Keller. 1985. Acts of Identity. Cambridge,
U.K.: Cambridge University Press.
Lobanov, Boris M. 1971. Classification of Russian vowels spoken by different speakers.
Journal of the Acoustical Society of America 49: 606–608.
Macaulay, R. K. S. 2001. ‘You’re like “Why not?”’ The quotative expressions of
Glasgow adolescents. Journal of Sociolinguistics 5: 3–21.
Mesthrie, Rajend. 2010. Socio-phonetics and social change: Deracialisation of the
GOOSE vowel in South African English. Journal of Sociolinguistics 14: 3–33.
Meyerhoff, Miriam and Nancy Niedzielski. 2003. The globalisation of vernacular
variation. Journal of Sociolinguistics 7: 534–555.
Milroy, Lesley. 2007. Off the shelf or under the counter? On the social dynamics
of sound changes. In Christopher M. Cain and Geoffrey Russom (eds.) Managing
Chaos: Strategies for Identifying Change in English. Studies in the History of the
English Language III. Berlin, Germany/New York: Mouton de Gruyter. 149–172.
Milroy, James and Lesley Milroy. 1977. Speech and context in an urban setting. Belfast
Working Papers in language and Linguistics 2.
Mufwene, Salikoko S. 2001. The Ecology of Language Evolution. Cambridge, U.K.:
Cambridge University Press.
Newton, Caroline and Bill Wells. 1999. The development of between-word processes
in the connected speech of children aged between 3 and 7 years. In Ben Maassen
and Paul Groenen (eds.) Pathologies of Speech and Language Advances in Clinical
Phonetics and Linguistics. London: Whurr Publishers. 67–75.
Quist, Pia. 2008. Sociolinguistic approaches to multiethnolect: Language variety and
stylistic practice. International Journal of Bilingualism 12: 43–61.
Rampton, Ben. 2010. Crossing into class: Language, ethnicities and class sensibility
in England. In Carmen Llamas and Dominic Watt (eds.) Language and Identities.
Edinburgh, U.K.: Edinburgh University Press. 134–143.
C Blackwell Publishing Ltd. 2011
MULTICULTURAL LONDON ENGLISH
195
Roberts, Julie and William Labov. 1995. Learning to talk Philadelphian: Acquisition
of short a by preschool children. Language Variation and Change 7: 101–112.
Schilling-Estes, Natalie and Walt Wolfram. 1994. Convergent explanation and
alternative regularization patterns: Were/weren’t leveling in a vernacular English
variety. Language Variation and Change 6: 273–302.
Schumann, John H. 1978. The Pidginization Process: A Model for Second Language
Acquisition. Rowley, Massachusetts: Newbury House.
Sebba, Mark. 1993. London Jamaican. London: Longman.
Sebba, Mark. 1997. Contact Languages: Pidgins and Creoles. New York: St Martin’s
Press.
Siegel, Jeff. 1997. Mixing, leveling and pidgin/creole development. In Arthur K.
Spears and Donald Winford (eds.) The Structure and Status of Pidgins and Creoles.
Amsterdam, The Netherlands: Benjamins. 111–149.
Smith, Jennifer, Mercedes Durham and Liam Fortune. 2007. ‘Mam, my trousers is
fa’in doon!’: Community, caregiver and child in the acquisition of variation in a
Scottish dialect. Language Variation and Change 19: 63–99.
Svendsen, Bente Ailin and Unn Røyneland. 2008. Multiethnolectal facts and functions
in Oslo, Norway. International Journal of Bilingualism 12: 63–83.
Tagliamonte, Sali and Alexandra D’Arcy. 2009. Peaks beyond phonology:
Adolescence, incrementation, and language change. Language 85: 58–108.
Thomason, Sarah G. and Terence Kaufman. 1988. Language Contact, Creolization and
Genetic Linguistics. Berkeley, California: University of California Press.
Torgersen, Eivind. 2002. Phonological distribution of the FOOT vowel, /υ/, in
young people’s speech in south-eastern British English. Reading Working Papers
in Linguistics 6: 25–38.
Torgersen, Eivind and Paul Kerswill. 2004. Internal and external motivation in
phonetic change: Dialect levelling outcomes for an English vowel shift. Journal
of Sociolinguistics 8: 23–53.
Trudgill, Peter. 1999. New-dialect formation and dedialectalization: Embryonic and
vestigial variants. Journal of English Linguistics 27: 319–327.
Watermeyer, Susan. 1996. Afrikaans English. In Vivian de Klerk (ed.) Focus on South
Africa. Amsterdam, The Netherlands: Benjamins. 99–124.
Wells, John C. 1982. Accents of English, Vols. I–III. Cambridge, U.K.: Cambridge
University Press.
Wiese, Heike. 2006. ‘Ich mach dich Messer’: Grammatische Produktivität in KiezSprache. Linguistische Berichte 207: 245–273.
Wiese, Heike. 2009. Grammatical innovation in multiethnic urban Europe: New
linguistic practices among adolescents. Lingua 119: 782–806.
Winford, Donald. 2003. An Introduction to Contact Linguistics. Oxford, U.K.: Blackwell.
C Blackwell Publishing Ltd. 2011
196
CHESHIRE ET AL.
APPENDIX
Participants in the Multicultural London English project
Age group
4–5
8–9
12–13
16–19
Young adult
Caregivers
Total
Ethnicity∗
Sex
Total number
of speakers
Male
Female
Anglo
Non-Anglo
Mixed race
18
20
27
25
8
29
127
10
8
15
12
5
10
59
8
12
12
13
3
19
62
0
6
11
4
2
8
31
18
13
14
17
6
21
87
0
1
2
4
0
0
7
∗ With
the exception of the 4–5 year olds, ethnicities are self-defined. They include: Afro-Caribbean;
Albanian; Bangladeshi; Black African; Bengali; Congolese; Filipino; Ghanaian; Kenyan; Kosovan;
Kurdish; Jamaican; Moroccan; Nigerian; Portuguese; Somali; Turkish; Turkish Cypriot; and Zanzibar.
Address correspondence to:
Jenny Cheshire
School of Languages Linguistics and Film
Queen Mary, University of London
London
United Kingdom
J.L.Cheshire@qmul.ac.uk
C Blackwell Publishing Ltd. 2011