The Acrolect in Jamaica: G. Alison Irvine-Sobers

The acrolect in
Jamaica
The architecture of phonological
variation
G. Alison Irvine-Sobers
language
Studies in Caribbean Languages 1 science
press
Studies in Caribbean Languages
Chief Editor: John R. Rickford

Managing Editor: Joseph T. Farquharson
In this series:
1. Irvine-Sobers, G. Alison. The acrolect in Jamaica: The architecture of phonological

variation.
The acrolect in
Jamaica
The architecture of phonological
variation
G. Alison Irvine-Sobers
language
science
press
G. Alison Irvine-Sobers. 2018. The acrolect in Jamaica: The architecture of
phonological variation (Studies in Caribbean Languages 1). Berlin: Language
Science Press.
This title can be downloaded at:

http://langsci-press.org/catalog/book/203
© 2018, G. Alison Irvine-Sobers
Published under the Creative Commons Attribution 4.0 Licence (CC BY 4.0):
http://creativecommons.org/licenses/by/4.0/
ISBN: 978-3-96110-114-6 (Digital)
978-3-96110-115-3 (Hardcover)
DOI:10.5281/zenodo.1306618
Source code available from www.github.com/langsci/203
Collaborative reading: paperhive.org/documents/remote?type=langsci&id=203
Cover and concept of design: Ulrike Harbort

Typesetting: Felix Kopecky
Proofreading: Amir Ghorbanpour, Amr Zawawy, Andreas Hölzl, Dietha Koster,
Eitan Grossman, Ezekiel Bolaji, Felix Hoberg, Gerald Delahunty, Ivica Jeđud,
Jean Nitzke, Jeroen van de Weijer & Rosetta Berger
Fonts: Linux Libertine, Libertinus Math, Arimo, DejaVu Sans Mono
Typesetting software: XƎLATEX
Language Science Press

Unter den Linden 6
10099 Berlin, Germany
langsci-press.org
Storage and cataloguing done by FU Berlin

Contents
Abbreviations iii
1 Ideologies of the acrolect and English in Jamaica 1

1.1 The (Jamaican) acrolect . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Sense 1: The acrolect as defined by structure . . . . . . . . . . . 4
1.3 Sense 2: The acrolect as the outcome of language acquisition . . 7
1.4 Sense 3: The acrolect as defined by its speakers . . . . . . . . . . 13
1.5 Sense 4: The acrolect as an idea of Jamaican English . . . . . . . 18
1.6 Aims and methods of the study . . . . . . . . . . . . . . . . . . 23
1.7 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2 Phonological variation in the Jamaican acrolect 29

2.1 Why consider phonological variation? . . . . . . . . . . . . . . . 29
2.2 The phonological variables and their variants . . . . . . . . . . . 32
2.3 Features commonly identified in the literature (Group A) . . . . 33
2.3.1 Basic JC and JE phoneme inventories . . . . . . . . . . . 35
2.3.2 Word initial glottal fricative /h/ . . . . . . . . . . . . . . 35
2.3.3 The interdental fricatives /θ/ and /ð/ . . . . . . . . . . . 39
2.3.4 The low back stressed vowel /ɔ/ . . . . . . . . . . . . . . 41
2.3.5 Pre-consonantal/pre-rhotic mid tense vowels /e/ and /o/ 43
2.3.6 The word initial velar stops /k/, /g/ . . . . . . . . . . . . 45
2.4 Features not widely discussed in the literature (Group B) . . . . 47
2.4.1 The word final unstressed vowel in words that end in -er 47
2.4.2 The vowel in the final syllable -tion . . . . . . . . . . . . 48
2.4.3 The alveopalatal affricate . . . . . . . . . . . . . . . . . . 49
2.4.4 Post-vocalic rhoticity . . . . . . . . . . . . . . . . . . . . 52
2.4.5 Word-final clusters . . . . . . . . . . . . . . . . . . . . . 54
2.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3 Sociolinguistic variation in Jamaican English 63

3.1 Level of education . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.1.1 Group A variables and education . . . . . . . . . . . . . 66
Contents
3.1.2 Group B variables and education . . . . . . . . . . . . . 71

3.2 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2.1 Group A variables and gender . . . . . . . . . . . . . . . 81
3.2.2 Group B variables and gender . . . . . . . . . . . . . . . 86
3.3 Parent’s background . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.3.1 Group A variables and background . . . . . . . . . . . . 95
3.3.2 Group B variables and background . . . . . . . . . . . . 98
3.4 Speaker age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.4.1 Group A variables and age . . . . . . . . . . . . . . . . . 103
3.4.2 Group B variables and age . . . . . . . . . . . . . . . . . 106
3.5 Hypercorrection in the JAMPRO sample . . . . . . . . . . . . . 109
3.6 Constructing the acrolect, Standard Jamaican English . . . . . . 114
4 Sociolinguistic Variation in JAMPRO 121

4.1 JAMPRO: One site of promoting a SJE ideology . . . . . . . . . 121
4.1.1 Jamaican institutional discourses on language . . . . . . 122
4.1.2 The JIDC/JAMPRO: Their construct of the Jamaican lan-
guage situation . . . . . . . . . . . . . . . . . . . . . . . 128
4.2 Frontline staff – The public face of JAMPRO . . . . . . . . . . . 132
4.2.1 Group A variables in frontline staff . . . . . . . . . . . . 140
4.2.2 Group B variables in frontline staff . . . . . . . . . . . . 142
4.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
5 Conclusion 149
5.1 The architecture of phonological variation in the Jamaican acrolect 150
5.2 An endonormative definition of SJE phonological features . . . 157
Appendix A: Prepared script for all interviews of JAMPRO informants 161

A.1 Personal data collected . . . . . . . . . . . . . . . . . . . . . . . 161
A.2 Data on patterns of workplace interaction . . . . . . . . . . . . 162
A.3 Data on working at JAMPRO . . . . . . . . . . . . . . . . . . . . 162
Appendix B: Profiles of JAMPRO informants 163
Appendix C: Parent’s Occupation 167
References 169
Index 183
Name index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
ii
Abbreviations
AJE Acrolectal Jamaican English
CXC Caribbean Examinations Council
EBE Educated Belizean English
IAE Internationally Acceptable English
JAMPRO Jamaica Promotions
JIDC Jamaica Industrial Development Corporation
JNEC Jamaica National Export Company
JNIP Jamaica National Investment Promotions
JSA JAMPRO Staff Association
JTB Jamaica Tourist Board
KMA Kingston Metropolitan Area
MSE Metropolitan Standard English
PJS Putative Jamaican Standard
SJE Standard Jamaican English
1 Ideologies of the acrolect and English
in Jamaica
1.1 The (Jamaican) acrolect
The Jamaican language situation was described very early on by linguists such as
Le Page (1960) and DeCamp (1961; 1971) as a linguistic continuum, formed primar-
ily from the historical sociolinguistic contact of speakers of various Niger-Congo
languages and several dialects of English. Similar continua have also been iden-
tified in Guyana, Belize, Trinidad and Barbados (see the discussion in Winford
1997: 233–236). The polar varieties of this theoretical construct, the continuum,
are basilectal Creole and “standard” English (discussed below), which have been
treated as separate linguistic systems by some linguists (for example Bailey 1971
(Jamaica); Devonish 1978 (Guyana), and more generally Alleyne 1980 and Win-
ford 1997) and which are linked by a seamless range of intermediate lects identi-
fied as the mesolect.
These intermediate varieties have been explained in a number of ways. They
have been attributed to: decreolization (Bickerton 1973; DeCamp 1971), due to the
targeting of English by Creole speakers with varying levels of access to the pres-
tige or high variety; or to basilectalization brought about by increasing numbers
of West Africans arriving over time and acquiring approximations of approxi-
mations of English (Mufwene 1996; 2001)1 ; or to the social stratification of plan-
tation communities from the earliest stages of language contact (Alleyne 1980).
This continuum is depicted as
… a continuous spectrum of speech varieties whose extremes are mutually

unintelligible, but which also includes all possible intermediate varieties
(DeCamp 1971: 28), ranging from the speech of the most backward peasant
or labourer to that of the well-educated urban professional (DeCamp 1961:
82).
1 Chaudenson (2001) describes a similar process of basilectalization for French-lexicon Creoles.

1 Ideologies of the acrolect and English in Jamaica
It is to this sociolinguistic discussion that W. A. Stewart (1965: 15–16) added

the terms acrolect and basilect, (the label mesolect came later). In his description
of “Washington Negro speech” he wrote,
I will refer to this topmost dialect in the local sociolinguistic hierarchy as

acrolect (from acro- “apex” plus -lect as in dialect). In most cases what is
meant by “Standard English” is either acrolect or something close to it. At
the other extreme is a kind of speech which I refer to hereafter as basilect
(from basi- “bottom”). (…) In between basilect and acrolect, there are a num-
ber of other dialect strata, and it is in this middle range that the majority of
adult Washington Negroes probably belongs.
Acrolect, the focus of this dissertation, is the name for the upper end of the
continuum and, while in many cases the term is not explicitly used, it has been
defined in the literature in a number of clearly related ways. General descriptions
of acrolect speak of,
• the “topmost dialect in the local sociolinguistic hierarchy” (W. A. Stewart

1965: 15);
• the stereotype of the educated standard, focussed by institutions of educa-

tion, the media, and white collar employment (Le Page 1988: 34–35);
• a “high, prestigious language or variety of a language” in any speech com-

munity (McArthur 1998: xvii).
Acrolect here is essentially a community of speakers’ idea of the speech pat-

terns of those positioned at the top and centre of their social space. It is in par-
ticular speakers, in those perceived to be of relatively high social status, that
we initially locate the acrolect in a community and idealise their ways of speak-
ing as “good” or necessary. This then makes the acrolect coterminous with “the
standard” for members of communities with that type of metalinguistic labelling.
This view of the acrolect is also reflected in many of the definitions applied to
the Caribbean generally and Jamaica specifically:
• the educated counterpart of the lexifier in the Creole speech community

(Chaudenson 2001: xi (n1));
• the English of educated nationals of the Caribbean used in formal social

contexts, bound to a common core of morphology and syntax shared with
“Internationally Accepted English” (Allsopp 1996: lvi);
2
1.1 The (Jamaican) acrolect
• the speech of those with high education, typically urban residence and
the higher socio-economic status that comes with certain (non-manual)
occupations (DeCamp 1961; 1971 who used the speech of Philip Sherlock
and Norman Manley as examples).
Central to these views of the acrolect is the speaker, as it is a particular set

of speaker variables that are typically associated with and used to identify the
acrolect in specific communities.
In another approach, the acrolect has been defined by its structures or in rela-
tion to the structures of other varieties that coexist with it in the community:
• the local standard, similar to other standard Englishes except in “a few

phonological details and a handful of lexical items” (Bickerton 1975: 24);
• the variety or varieties used in Jamaica that are structurally farthest re-
moved from the Creole, i.e. basilect (Akers 1981: 73; Bailey 1971: 342 for
example).
Finally, the acrolect is seen as the outcome of Creole speakers’ acquisition of

an idealised superstrate that is (and was) the target in language contact:
• a non-native version, derived from spontaneous or guided learning, of the

standard language (Escure 1997: 67).
We can abstract from the above that the acrolect has been defined,
1. in terms of its structure, and this includes not only its relationship to En-
glish, say, but also its relationship to the Creole basilect;
2. as a consequence of language acquisition, as it is suggested that the acro-

lect is the learned English of Caribbean peoples who are not themselves
vernacular speakers of English;
3. sociolinguistically, in relation to its speakers and the situational contexts in

which these speakers use language, i.e. the formal, public speech of those
with high social status;
4. ideologically, an idea of English that is legitimised by institutions that con-

struct norms by their own expectations and practice.
3
1.2 Sense 1: The acrolect as defined by structure

The focus of research in Caribbean continuum situations has, understandably,
been the Creole basilect, the less well-known variety to linguists. This has gen-
erally meant that the speaker’s central role in the characterization of lects along
the continuum has been overshadowed by the search for what Bailey (1971: 341)
called “neat, clearly defined patterns of behaviour” and the analysis of idealised
basilectal linguistic systems. This process of erasure or selective disattention to
“often unruly forms of variation” (Kroskrity 2000: 23) is no doubt driven by the
requirements of linguistic description (J. Milroy 1999: 33); but it is also, ironically,
a reflection of the way speakers themselves construct language and language use,
particularly in the process of social differentiation (J. Irvine & Gal 2000: 38). Ja-
maicans, for example, tend to see the local language situation in terms of what
we Jamaicans call Patwa and English, and to produce speech in a socio-functional
relationship that would accurately be described as diglossic (Akers 1981: 8; Win-
ford 1985, Devonish & Harry 2004: 256). This has had consequences for the study
of the continuum generally, but in particular the acrolect, because the underly-
ing system that generates acrolectal speech is thought by both the speaker and
the linguist to be no different from that of other Englishes. Therefore, much of
the literature assumes, but has not verified, the features associated with the acro-
lect (Akers 1981: 73; Bailey 1971: 342; Mufwene 2001: 209 n6; Patrick 1999, Patrick
2000: 11; Winford 1991) even when making passing reference to “a local standard”.
With few exceptions (see below), this local standard has not itself been the focus
of research as most linguists presumably already know the structural properties
of English and are seeking to uncover those of the Creole. In this respect, these
studies are following in an earlier ideological tradition, described by J. Irvine &
Gal (2000), which in practice meant that
“once a variety has been declared to belong to the ‘same’ language as an-
other, already-described variety, there was no reason to investigate it…” (57).
Additionally, and indeed because this “local standard” is putatively not so dif-
ferent, metropolitan Standard English (MSE hereafter) has been used not only to
compare but to typologize structures found along the continuum2 , particularly
the morphology and syntax but also the phonology (J. C. Wells 1982c: 575). As
Alleyne pointed out over two decades ago, Creole languages are compared with
Standard Written English of the most formal registers (1980: 16).
2 J.
Milroy (1999: 33 in particular) discusses the effect of a standard language ideology on the
investigation and description of “less well-known languages”.
4
1.2 Sense 1: The acrolect as defined by structure
A few examples can be used to illustrate the approach. Firstly, the negative past
particle neva, in the sentence Jan neva tiif di moni ‘John did not steal the money’,
has been analysed syntactically as Creole (Bailey 1966) or as non-acrolectal (Es-
cure 2000: 141; Schneider 1998: 217) even though it is used as a negative preterite
marker throughout the Jamaican continuum (as in John never stole the money or
in the following: Speaker A – Give Damian back his book; Speaker B – What? I
never borrowed it). This type of analysis is not limited to contact situations like
Jamaica. As Cheshire (1999: 133) points out, the construction never + V-ed when
used like this (in MSE) is “…frequently label[led] (…) as ‘non-standard’, despite
the fact that the contexts occur in educated speech and writing”.
Secondly, Meade (2001), in his study of phonological acquisition in Jamaican
children, uses MSE phonology to determine what he labels acrolectal phonolog-
ical features. Some of his Jamaican children are said to be acquiring at least one
feature either late or at the mesolectal level (162) – in this case the interdental
fricative – because they continue to vary them with dental stops up to ages 4-
6 in words that would have [ð] or [θ] in MSE. Both explanations of his results
are based on an idea of acrolect that reflects both an external model as well as
an idealization of that model of English. One must therefore ask what children
in Jamaica are likely to actually hear around them from local models of English,
and should that English be treated as “falling short”, i.e. non-acrolectal, when it
does not resemble the external ideal. Meade does compare the children’s output
with that of their caregivers in his study and concludes that frequency of use by
caregivers of a particular feature is crucial to the rate of acquisition of the chil-
dren (161). However, Meade treats interdental fricatives and their variants as if
they pattern as they do in MSE, i.e. he does not distinguish voiced from voiceless
when describing the significance of the feature to Jamaican English which, as he
puts it, is “ranked very high on the implicational scale of the Creole continuum”
(162). It may well be that the local use of interdental fricatives and their variants
differs from that of MSE (see §3.6).
Finally, Akers (1981: 8) argues that the speech of British, American and Cana-
dian tourists provides one prestige model for Jamaicans. This suggests that for
Akers these three phonetically different varieties are idealized as one “English”,
notwithstanding the distinctiveness the speakers of the varieties seem to claim
when labelling even their Standards. Here, arguably, metropolitan varieties of
spoken English are also being characterized as “Standard Written English of the
most formal registers”, notwithstanding the problems of doing so at the phono-
logical level.
Akers’s (1981: 73) description of the acrolect as a rhotic variety was based
largely on his analysis of the Creole basilect as non-rhotic, and not on the ac-
5
tual production of acrolectal speakers. His pool of informants consisted of 10

relatively young people (the oldest was 33 years old), only one of whom could
have been characterized as having the kind of background typically associated
with the acrolectal speaker, a female college graduate. As such, his conception
of the acrolect is a sort of anti-Creole that stands in opposition to the basilect
structurally. Patrick’s (2000) is too. While he does not explicitly characterize the
acrolect in his analysis of variation in Veeton, he also proceeds like Akers, Meade
and Wells when he states that,
… variables are here counted in their presence or absence so that maximum

usage approximates to the acrolectal standard, i.e. absence of (KYA) [kja ~
kh a] … testing the assumption that higher status speakers favour acrolectal
speech more often. (Patrick 2000: 11)
What seems to have emerged is a kind of circularity in the definition of the po-
lar varieties on the (Jamaican) continuum that, typically, takes this form: a) Since
the acrolect is not unlike other standard Englishes, phonologically as in [kh ] (in
a word like cat) or morpho-syntactically as in did not take; then b) the basilect
reflects structures that are maximally divergent from the acrolect – [kj] or no
ben ~ neva tek; as such, c) acrolectal speech is maximally non-basilectal and can
therefore be expected to produce the forms [kh ] or did not take. Moreover, this
circularity extends to the identification of so-called basilectal speakers, so that
samples of speech collected in the field are rejected when they are “too English”
and therefore “noncreole” (Escure 1997: 74 commenting on the general practice
of creolists), because the linguist already knows what is supposed to be basilectal.
One way of accommodating the above conception of the acrolect, is, on one
hand, the catch-all term “upper-mesolectal” (Bickerton 1975: 161–162 for example)
for those forms that do not pattern like those of MSE but have features found
elsewhere in the continuum. Patrick (2002: 17) tells us that,
Speaky-spoky is associated with lower mesolectal and basilectal speakers

because of their distance from the standard; and it is associated with ‘mis-
takes’, failed attempts to speak a metropolitan prestige variety that is not
native to these speakers. (Acrolectal and upper mesolectal speakers are viewed
as using Standard Jamaican English appropriately, and not making linguistic
errors in its use, despite the variation in their spoken and written speech)
[My emphasis].
In this description “the standard” is MSE, the prestige target variety, and not
Standard Jamaican English (SJE hereafter), the variety used appropriately by both
6
1.3 Sense 2: The acrolect as the outcome of language acquisition
acrolectal and upper mesolectal speakers. It is therefore not clear why this dis-
tinction is being made, though this could be interpreted as defining acrolectal
and upper mesolectal as labels of presumed social group.
The irony of this conception of the acrolect is that Jamaican speech is being
analysed in terms of a variety that is not actually spoken in the speech commu-
nity and the structures used by Jamaicans are being defined in terms of norms
external to the community of speakers. Moreover, this speech is being analysed
more in relation to written metropolitan structures than to those used in spo-
ken MSE. When Lalla & D’Costa (1990: 89) write, for example, that in the 18th
century “basilectal features occur in the usage of all classes sampled, though to
varying extents”, they beg the question of why a feature is labelled “basilectal” if
it is also found elsewhere in the continuum of varieties. There would seem to be
no Jamaica-based structural criteria for singling out as particularly “basilectal”
features attested to in all lects.
1.3 Sense 2: The acrolect as the outcome of language

acquisition
The acrolect has also been conceived of as the product of decreolization (Escure
1997), the outcome of (post) Creole speakers’ acquisition of English as a result
of access to education or increased exposure to an external linguistic model (in
Belize, British or American Standard English) available in the society. Escure
takes issue with the traditional notion of decreolization only because it suggests
replacement of, rather than addition to, Creole varieties. I will use her analysis of
acrolect, as it is explicitly discussed by her and includes many of the issues that
are relevant to the discussion in this book.
Escure (1997: 67–68) conceives the acrolect to be:
A non-native version of the standard language, which is acquired through

spontaneous or guided learning, functions in formal contexts, and extends
its speaker’s repertoire without necessarily leading to loss of the speaker’s
vernacular. The acrolect is not necessarily a dominant or prestigious dialect,
although it is usually associated with education. The definition of the term
acrolect implies that there is an available standard that provides a linguistic
model instrumental in the formation of acrolects, even in the absence of any
guided learning.
She goes on to add that the label “English” is:
7
extremely confusing to Belizeans, primarily because English is an external

standard, never physically present, although it is officially proclaimed to
be the language of Belize. English is not spoken in Belize by any particu-
lar group, excepting a handful of immigrants (…) The only direct access to
native renderings of a standard English variety appears to be through ra-
dio programs, BBC news and American Evangelical broadcasts. Standard
English, then, has practically no spoken presence within the country of Be-
lize….
The above quotes suggest the following:
a) Acrolect and standard are not the same – the former is non-native English,
the result of speakers targeting the latter.
b) Acrolect is the formal, educated Belizean variety of English (EBE hereafter),

but it is not Standard English which is only spoken by immigrants and on
foreign radio.
c) What distinguishes Standard English and the acrolect is the issue of being
a native speaker and, more particularly, the phonology. In the absence of
MSE phonological patterns then no Belizean is a native speaker of (stan-
dard) English, though they may write it. Elsewhere in the book (1997: 66)
Escure offers a definition of standard that incorporates the ideas of edu-
cated and prestige varieties and says that the term English clearly denotes
for Belizeans a local standard variety, because they are not motivated to
learn the remote (MSE) standard (1997: 73).
As in the discussion above, of Sense 1, the properties of MSE are used as ref-
erence for defining Standard English in Belize. Escure, however, goes further by
saying that no one who is Belizean speaks Standard English because what they
produce, i.e. acrolect, differs from MSE:
…acrolects develop independently because the officially prescribed standard

is not present in the country; there is simply no linguistic exposure to a
consistent British or American standard of English, either in daily activities
or in school contexts (Escure 1997: 73).
Nevertheless, Escure tells us that for Belizeans, what she calls the acrolect is
standard English as the British or American standards are sociolinguistically re-
mote, and therefore unimportant in the speech community. Moreover, Belizeans
8
feel that they should speak standard English as it is the official language and in-
dexes education (see also Young 1973), and they therefore target what they con-
sider it to be, there being no spoken MSE presence in either the school system
or the society generally. Escure also suggests that this local EBE is not the super-
strate (1997: 74), even as it is the variety that lives in the community of speakers
and is the variety of English Belizeans (will) target and/or acquire. In this respect
Escure is not unlike Patrick (quoted above), as both suggest that the target (for
Belizeans and Jamaicans) is metropolitan English.
In her discussion of acrolects as innovations, Escure questions why, in light of
the goals and choices a community or individual might make in the acquisition of
a language, “acrolectal speakers would choose, albeit unconsciously, to produce
varieties distinct from the official standard” (1997: 66). But EBE is the standard,
the spoken variety of educated Belizeans that has its own phonological, lexical
and idiomatic forms.
Such an exonormative approach to defining the standard in Caribbean territo-
ries like Belize and Jamaica is curious, given the history of English in the Carib-
bean. It speaks as much to a view of what is “native” English as it does to who is to
be called a native speaker of English. It is never suggested, for example, that Gen-
eral American represents the acrolect, imperfect targeting of British English by
speakers because of ties of history, even though much of the U.S. population his-
torically were Europeans acquiring/learning a foreign language (Romaine 2001
passim). Certainly, the percentage of L1 English speakers was/is greater in the
U.S. than in Jamaica or Belize. However, that suggests that the characterization
of a variety as native is a quantitative matter – with an arbitrary decision made
as to numbers of L1 speakers – rather than an issue of acquisition per se.
Escure’s idea of acrolect necessarily takes into account the historical social con-
text of language contact, but seems to ignore the existence of vernacular speakers
of English in continuum situations like Belize. Judd (1998: 148–151) suggests that a
vernacular English speaking population in Belize existed since at least the 1900’s.
And in Jamaica, Lalla & D’Costa (1990: 98) carefully show
the existence, from the eighteenth century, of extensive variation (among

speakers and within individual usage) in features of basilectal and acrolectal
models of Jamaican speech [and that] (…) the data also confirm considerable
mixing of such features.
The anonymous author of Marly, A planter’s life in Jamaica, (cited in Lalla &
D’Costa 1989: 44) seems to make a distinction, as does Lady Nugent (writing be-
tween 1801 and 1805), between white Creoles who have been educated in England
9
and those wholly “educated” in Jamaica.3 The latter description applied mostly,
but not exclusively, to women. These reports, by speakers of metropolitan va-
rieties, indicate use of (a local variety of) English, with occasional “lapses” into
Creole in “moments of excitement” or when “not on guard”. The exposure to MSE
for some Jamaicans does not negate the vernacular status of English in Jamaica.
It only suggests the nature of the variation to be found in Jamaican English. As
Lalla & D’Costa (1990) conclude from Long’s (1774: 90–91) description of late 18th
century Jamaica,
of the landowning upper class alone (…) we see (1) a British-educated creole
élite speaking RP; (2) others of the same socioeconomic class speaking a
Jamaican version of RP; (3) others, less educated and less exposed to urban
influence, speaking RP “with much difficulty” and creole with ease; and (4)
monolingual, rural creole speakers, who were usually women and brought
up on plantations. (…) Among blacks (…) (3) creole blacks (“the better sort”)
acquiring some RP and altering their JC accordingly; (4) locally educated
free blacks and coloureds using the Jamaican form of RP in schools such as
Francis Williams’s; and (5) British educated blacks and coloureds, such as
Francis Williams himself, speaking RP.
Moreover, as Michele M. Stewart (2002: 14) notes,
many [free coloured men] attended Wolmer’s, the only secondary school
in Kingston, alongside Jews, and some were sent to public schools … By
1788, free coloureds comprised 12.4% of the population. A few of the English
educated men had been admitted to the highest ranks of society by the turn
of the century.4
In addition Bryan (1996) describes the emergence of the black middle class in
19th century Jamaica, whose institutions included the Jamaica Union of Teachers,
the Artisan’s Union and the Advocate newspaper.
3 Lalla & D’Costa (1990: 90), citing Long (1774), state that most Creole white women were home
tutored if educated. This home tutoring was in many cases carried out by British governesses,
though the average governess was “an impoverished and ill-educated female down on her luck”
(Brereton 1995: 83, citing Carmichael 1833).
4 While free coloured men are specifically mentioned here, publications like Mary Seacole’s
autobiography, Wondeful Adventures of Mrs. Seacole in Many Lands, refers to a class of what
she called “Creole” and “yellow” women who were also clearly English speakers (Busby 1992).
10
The acrolect then, as a macrosociolinguistic phenomenon, cannot be learned

English alone, as not only were local norms of English identified some two cen-
turies ago, displaying the expected variation that factors such as gender, ethnic-
ity and education would influence, but there was also an idea of Jamaicanness,
which for some excluded English born persons and the ascriptive label English-
man (Patterson 1973: 34). Nor can the model for those acquiring English have
necessarily been exclusively the metropolitan variety. For most colonies then,
contact with the metropole would have been limited to the élite few.
I argue that Jamaican English is not non-native English, as defined by Kachru
(1982: 31) (cf. Moag 1982: 270). And I have difficulties with the tacit view of the
researchers described above (discussed in §1.1–§1.2) who use an external model
to identify the acrolect. The term “non-native” is itself problematic, as it is not at
all clear what it means. It is applied to the English of SE Asia, India, the Carib-
bean and Africa – all of which have different historical contexts of development.
It is not, however, applied to the English of North America, Australia or New
Zealand. This suggests that ethnicity may be one of the considerations in cate-
gorizing regional varieties of English. The distinction seems to be part of what
Mufwene (2001: 107) calls the “disfranchising of particular varieties as illegitimate
offspring”. English in the Caribbean is not really an external colonial language
grafted on to indigenous, living languages5 as it is in Africa or Asia, even if it
does not have the continuity that is argued for, say, North America (for example
by Fisher 2001, but cf. Görlach 1987, Mufwene 2001). Jamaica was a settlement
colony, its primary reliance on the sugar plantations producing a non-European
majority. And if non-native means an “institutionalized variety” (Kachru 1982:
38–40), or a “localized form” (Strevens 1982: 24), then English in Jamaica, the ac-
rolect, has nearly always had multiple status for different groups of speakers. It
has, however, been the vernacular of a section of the population for over three
centuries, and can only be called a learned variety for some, maybe even most,
speakers.
The guided learning of English in the school system, which would serve to
focus and legitimize an educated norm, has, for most Jamaicans with access, been
“localized” (Moag Ibid.,278) since the mid 19th century, if not before. Bryan points
out that,
As far as the racial factor is concerned, most of the elementary school teach-
ers were black and coloured … Teachers clearly regarded themselves as an
upwardly mobile section of the black population (Bryan 1996: 287–288).
5 I am aware of the indigenous languages of places like Belize and Guyana. I am therefore refer-
ring to the general context of language contact between West Africans and Europeans.
11
And Wilmot’s (2002: 317) description of early local government politics in Ja-
maica notes that
In order to provide more prestige for this public meeting [in 1854], individ-
ual black and coloured teachers from the Beckford’s Free School in Span-
ish Town and the Wolmer’s School in Kingston, respectively, addressed the
gathering.
For high schools, a “limited number of elementary school children [were]

given the opportunity to attend” (E. Miller 1989: 209, citing B. H. Easter, the Di-
rector of Education, Jamaica 1946).
These high schools, some of which had been administered by British (trained)
teachers, were almost completely localized and feminized by the last decade of
the colonial period, the 1950’s (E. Miller 1989). This produced what Le Page (1968:
440) called the gap between “what is supposed to be happening in the schools
and what is actually happening”.
Local educated norms of English were identified in Jamaica at least a century
ago, if not before. The acrolect was then being added to or altered, not created,
as these norms became generalized for more educated Jamaicans, as fewer and
fewer people sent their children to school in England. The speakers of this Ja-
maican English would have increased the range of stylistic and social variation
in the acrolect and perhaps generated different models of acceptable English. The
acrolect as an aspect of a linguistic and sociolinguistic continuum, however, must
have existed and been contemporary historically with a creole basilect.
The issue here, then, is not the theoretical validity of decreolization. It is that
without a clear idea of what the phonology or syntax of the acrolect is, it be-
comes difficult to discuss what it is speakers are targeting when they acquire or
learn English. For example, Rickford (1987: 275) gives instances of the speech of
a Guyanese teacher and of a barrister delivering his summation in a Guyanese
court. In both cases he identifies aspects of the phonology of these speakers that
he labels “non-standard” – [d ~ ð] as in ‘them’, [fʌŋ] ‘found’, non retroflexion in
words like culture or render. And as Rickford goes on to caution, “their classifi-
cation as “non-standard” in the Guyanese context is open to question”. I suggest
that Guyanese speakers are more likely to target these local features than RP or
GenAm. variants.
Discussions of the acrolect, in both Sense 1 and 2, represent English as if it
was and is a thing apart, foreign to places like Jamaica and “a more or less well-
defined and discrete” layer that has little relation to the continuum (Winford 1997:
241). If English is for its speakers locally defined, then continued analysis of that
12
1.4 Sense 3: The acrolect as defined by its speakers
English in terms of MSE will result in paradoxical conclusions. Some of these

conclusions reflect what Kachru (1982: 50) calls “linguistic schizophrenia … [an
inability] to decide whether to accept a mythical non-native model [MSE], or to
recognize the local functioning model [acrolect] instead”.

A smaller set of work on the (Jamaican) continuum has looked at the acrolect, in
terms of its speakers and the contexts in which they use language. The speaker
variables that are generally used are socio-economic class, residence and level
of education, particularly the latter. DeCamp’s original characterization of the
acrolect, for example, was the speech of “the well-educated urban professional”
(1961: 82), although the implicational scales he eventually produced were a struc-
turally defined linear spectrum of features. Beckford‑Wassink (1999a) also uses
education, urban residence and social class to locate acrolect-dominant speakers.
The situational context of use is generally described as formal, as a distinction
is made between formal speech and the production in relaxed, informal interac-
tions.
Allsopp’s (1996: lvi) definition of Caribbean Standard English in the introduc-
tion to the Dictionary of Caribbean English Usage is typical, focussing on the two
variables of education and formality - “the literate English of educated nationals
of Caribbean territories and their spoken English such as is considered natural
in formal social contexts”.
Thaxter (1977), in his study of what he calls Standard Jamaican English, defines
it as, “the kind of language which most educated Jamaicans speak on formal
occasions in business transactions, legal affairs, in educational institutions and
in public life” (2). F. Miller (1987: 48,182) describes Acrolectal Jamaican English
(AJE hereafter) as “the actual spoken variety [of English] which the educated
élite uses in formal circumstances”.
The focus on education may be explained in a number of ways, although, as
Pollard (1998: 179) points out, SJE is the “first language of the man at the top of
the social ladder”. Firstly, it is the case that many, if not most, Jamaicans learn
what Allsopp calls Internationally Acceptable English (IAE hereafter) in school
and English has been a compulsory exam always required by the school system.
Secondly, the school system is one of the means of legitimising and standardizing
norms (Lippi-Green 1997: 65). It is in the classroom that, according to Bourdieu
(1991: 61–62), “…mastery of the legitimate language may be acquired through
familiarization, that is, by more or less prolonged exposure to the legitimate lan-
guage…”.
13
And of course, education as a defining characteristic of “standard” speech is

not peculiar to the Jamaican social context and therefore does not entail external
origin of the standard. McArthur (1998: 119–137) cites 52 examples of definitions
of Standard English from a variety of sources, and of these 18 mention educated-
ness of speaker as the important criterion.
The school system then links a sociolinguistic notion of acrolect, the speech
of a social group called “the educated”, with an ideological one, the construct
called “Standard English” legitimated by institutions like itself. By analysing the
product of the school system, the researcher can identify the individuals in the
speech community who have best adopted the sanctioned linguistic forms and
been rewarded for doing so. And this identification of speakers, using language
in formal contexts, is important so as to minimize assumptions about the forms
to be found in the acrolect.
The other important variable in such studies is formality. Thaxter’s study (1977)
sees formality as the style used in specific kinds of physical and psychological
contexts (after Hymes 1974: 55). He recorded students of a Teacher’s College dur-
ing classroom debates, speech training exercises, election speeches and votes of
thanks. F. Miller conducted interviews designed to elicit certain self-conscious
styles of speech (after Labov 1972: 79). And the Guyanese data presented by Rick-
ford (1987: 275–276), discussed in the previous section, also illustrates speech in
certain formal settings – the courtroom and the classroom – as produced by ed-
ucated informants.
In all these studies the assumption is that the informant’s attention to speech,
because of where and when language is being used, is likely to produce greater
use of prestige variants and the most appropriate variety for the context (Labov
1966; 1972; Wolfson 1976). This is not unrelated to the issue of education and the
idea of a legitimised variety, which is, in part, the focussing of a set of norms
(Le Page 1980) and the reduction of variation in the language that is to be used
in such circumstances.
In his study of SJE, Thaxter looked at productions by informants as well as
their evaluations of speech along a series of dimensions that include perceptions
of acceptability, suitability for classroom use, and preferences of identity. Among
the phonological variables he examined were – [ð/θ ~ d/t], [ɪŋ ~ ɪn], CC## ~ C##,
[ɔ ~ a], [o ~ uo] and [e ~ ie]. One of his conclusions is that,
The speakers have a wider range of language usage in phonology than

in morphology and syntax probably because they spent more time learn-
ing more morphological and syntactic than phonological features … [and]
14
probably because morphology and syntax decreolize faster than phonology

(1977: 215).
What Thaxter explains as an aspect of the rate of decreolization, can be ex-

plained in another way. The “standard” that speakers learn in the school system
is essentially a written variety of English, with prescriptive rules about “correct
grammar”, i.e. morpho-syntactic use. Gupta (2001: 370) makes the important ob-
servation that,
The remarkable thing about Standard English, especially the Standard En-
glish found in edited, printed documents, is its uniformity across the world
(…) Even our areas of uncertainty are shared. Countries with contact va-
rieties of English (such as Jamaica, Nigeria, Singapore) participate in this
agreement on what Standard English is like (…). We must not forget that
the concept of Standard is very weak in lexis and phonology. There is no
Standard accent of English (though there are prestige accents…).
The English of the classroom is pronunciation neutral, based as it is interna-

tionally on a literary standard, with few prescriptions on segmental features and
fewer, if any on prosody (see Ho & Platt 1993: 7; J. Milroy & L. Milroy 1985: 66–67;
Trudgill 1999: 118). Moreover, Cheshire (1999: 147) argues, at least for the curricu-
lum in English schools, that it is stylistically neutral, as typically little distinction
is made between formal and informal styles of speaking in the English taught in
the classroom.
What distinguishes the English of an educated Jamaican from an educated
Australian or Nigerian is most clearly the phonological patterns of the speaker,
his or her accent. One can identify features of morphology and syntax that are
peculiar to communities, but these are relatively few in the educated speaker of
formal English. Rickford’s examples from Guyana cited above led him to com-
ment that the Guyanese sociolinguistic continuum is gradient really only with
respect to the phonology (Rickford 1987: 278–279). The standard/non-standard
use of syntactic forms he found much more sharply distinguished socially.
On the evaluative dimension, Thaxter’s informants found the “local” voice
preferable to the “foreign” one, that of an RP speaker; and the “librarian’s speech”
the variety they would most encourage their pupils to use. What distinguished
“her” speech from the English person was the phonology, specifically a greater
use of initial [d] in words such as “this” and the use of mid-vowel diphthongs
in words like [fies] face and [buot] boat. Following the tradition outlined in the
previous sections, Thaxter labelled this phonology “mesolectal” (239).
15
Miller’s study (1987) of AJE described aspects of the phonological and morpho-
syntactic use of post-secondary educated, urban (Kingston) dwellers. She select-
ed and stratified her 24 informants based on their socio-economic class (indexed
by occupation and income), gender and age (18–35 and 45–65 years old). Her
conclusions, summarized in the final section of the dissertation, are confusing
at times because Miller tends to use the terms “standard” and “prestige” inter-
changeably, assuming them to be the same thing. For example, she contends
that the typical pattern of female speech, the use of a greater number of pres-
tige forms, is not borne out in her study because they differ from metropolitan
forms –
On the whole, the females in this study use more non-standard phonological
forms than the males … Only UMC/UC [upper middle class/upper class]
women exhibit the tendency to produce more prestigious pronunciation …
(1987: 112–113)
Miller does not present support for her assumptions about the prestige of MSE
phonological forms. She does say later that the speech of women and younger
informants is possibly “a pattern which is emerging where educated speakers are
expressing themselves by using linguistic forms which are not part of a British or
American model” (177). We do not, however, know if any forms are entering the
acrolect, and there is enough historical evidence to suggest that its phonology
has displayed variation for some time.
The phonological variables F. Miller studied were i) TH Stopping (θ/ð ~ t/d), ii)
[ɪŋ ~ ɪn], iii) Vowel laxing [tek ~ tɛk] or [onli ~ ɔnli], iv) [kj ~ kh ] and v) [jʊ ~ jǝ ~
i] in words like education. (See Chapter 2 for a more detailed discussion of many
of these phonological features).
She found, in general, that
The high frequency of occurrences of palatal realizations (44.3%), a 30 per-

cent occurrence of TH Stopping, a 50.4 percent realization of the [i] variant
of the variable /jʊ/ etc. behoves us to realize that we are indeed dealing with
‘inherent variation’ (1987: 183–184).
This inherent variation, she suggests, is due to substrate Creole influence and
its relationship to the emerging Jamaican norms of English – for Jamaicans see
MSE patterns as stilted, pedantic and unnatural even in formal circumstances
– where aspects of “popular speech” (178) are used to signal this local English
language identity.
16
In a previous study of acrolectal varieties (A. Irvine 1988; 1994), I was inter-
ested in exploring the speech of informants who lived in two neighbouring af-
fluent suburbs of Kingston, one characterized by its high level of ethnic minor-
ity residents (Whites, Syrians, Chinese for example). My 32 informants were all
educated (post-secondary) and were born into households of relative affluence
(indexed by father’s occupation). The segmental phonological features looked at
were post vocalic rhoticity and [kh a ~ kja], in addition to aspects of intonation
and vowel quality. It was found, for example, that the informants from the com-
munity with the greater number of ethnic minority residents were much more
likely to produce [kja] than [kh a], and women generally also followed this pat-
tern. The other community, with a predominantly black educated population,
tended to use the [kh a] variant more frequently, particularly the men. It was not
at all clear, however, whether this finding was to be explained by the influence/
avoidance of the “Creole” [kja] or by a distancing from the patterns of the other
community.
One could take the extreme position that MSE, and certainly the phonology,
is not especially relevant to any synchronic discussion of acrolectal speech or
variation in the continuum generally; indeed, its relevance diachronically is also
somewhat problematic. Winford (1997), in his discussion of Caribbean English
continua, suggests that “synchronic style shifting from creole vernacular to more
formal standard usage has much in common with historical language shift from
creole to more ‘standard’ or ‘acrolectal’ targets” (271). But terms like “standard-
like” to discuss the production of EBE speakers (268) do, in a sense, presuppose
what is standard in the particular social context and therefore what is being tar-
getted.
It is necessary at this point therefore to try and distinguish use of the terms
acrolect, Jamaican English and Standard Jamaican English. The first label, acro-
lect, is more typical of studies that seek to describe formal properties of systems
along the continuum (as in the previous sections). In doing so an idealization of
the speech at the apex of the continuum is assumed, while variation is explic-
itly acknowledged and disregarded in the analysis. Beckford-Wassink uses the
term acrolect-dominant to describe the linguistic characteristics of certain types
of speakers, i.e. urban, well educated and of relatively high income. She does anal-
yse variation and argues that no such speaker produces the acrolect all the time,
but it does represent the dominant lect in their repertoires. Implicit in the label,
however, is the assumption that forms which deviate from MSE represent shift
away from the acrolect in speakers, even when produced in formal/self-conscious
contexts. Moreover, her later discussion of the changing indexicality of a form
like [kja] as in cat for example, which she suggests is “now welcome” (Dyer &
17
Beckford-Wassink 2001: 31) in the speech of the upwardly mobile, speaks to a

presumption that non-MSE forms are somehow becoming part of the acrolect.
Lalla and D’Costa’s comments cited earlier would suggest otherwise.
If acrolect is meant to be the idealized variety at one extreme of the Jamaican
continuum, essentially a device to describe a basilect, then it cannot be coter-
minous with “the local standard”, as used in the discussions of some Creolists
(see §1.1, 5). The latter label suggests forms that are or have been normalised by
speakers in the community as “correct”, notwithstanding the lack of corrobora-
tion for this process in many of the discussions. A shift then “from Creole to
more ‘standard’ or ‘acrolectal’ targets” as Winford states above is problematic if,
for example, a form is used along the continuum or if the discussion centres on,
say, phonology. The latter, acrolect, suggests the linguists’ construct of Jamaican
English, informed as it is by the linguist’s knowledge of formal written English;
the former, local standard, is an idea held by speakers of what that English should
be like, particularly, but not exclusively, when spoken.
I argue that the two should be the same. Any discussion of language change,
or targetting of forms, or prestige of forms must incorporate the speakers’ ideas
of what is English or Creole in their social context. As Shields (1987: 119–120)
points out,
though some phonological features are shared by mainstream RP and Ja-

maican Creole these are often eliminated from the speech of [English speak-
ing] informants because of their obvious associations with Creole.
1.5 Sense 4: The acrolect as an idea of Jamaican English

Two examples can be used to reinforce why it seems necessary to resituate the
idea of acrolect in the speaker and in the (Jamaican) social context. Firstly, in
Jamaica, the Revised primary curriculum 1999 (Ministry of Education and Culture
1999: 14) states:
The language programme seeks to sensitize pupils to the richness and va-
riety of language. A major objective is to assist them to acquire the target
language Standard Jamaican English. [My emphasis]
For the Jamaican government, and specifically the Ministry of Education, there
is such a thing as a SJE, the official and prescribed target for Jamaicans in the
school system. The current phonics text selected for use in primary/preparatory
education in Jamaica contains the following lesson:
18
The words which describe the pictures below have vowel pairs (sic) which
make the same sound - bowl cow towel out couch round mouth (Gbedemah
1995: 14). [My emphasis]
This is the 7th edition of the textbook, and it is normalizing or has normalized
the above pronunciation of bowl (on the model of other [aʊ] words) as SJE. An
informal survey of most of my own students at the University of the West Indies
(Mona, Jamaica) suggests that they do distinguish ‘bowl’ [baʊl] ‘a ceramic dish’
from ‘bowl’ [bol] ‘the delivery of the ball in cricket’.
Secondly, the teachers at one preparatory (and therefore private) school in
Kingston consistently instruct children to say [brekfa:sth ] breakfast and [lIth ǝl]
little, but seldom ever comment on the structure never + V-ed (for the negative
preterite, as in never borrowed) unless children say neva borrow. The teacher’s
preoccupation with passing on [lIth ǝl] and [brekfa:sth ] to her students cannot
be explained as the targeting of MSE, but is a response to an idea that Creole
speakers say [lɪkl] and “leave off word endings”. Moreover, her actions provide
us with insights into her idea of SJE, one obviously connected to her idea of
“Creole”.
As such, it is the sociolinguistic patterns within communities, fuelled by the
ideas speakers hold about language and language users in their community, that
explain linguistic behaviour (over time). In that respect, language situations like
that found in Jamaica are not likely to be different from other communities of
speakers, where it is generally assumed that foreign norms are peripheral.
Le Page & Tabouret-Keller (1985: 191) believe standard, as applied to language,
to have two, sometimes indistinguishable, meanings,
that of norm, (…) and that of a prescriptive yardstick against which people
and things are measured … norm often becomes converted into a prescrip-
tive standard used as a yardstick; through the education and examination
systems as a test for admission to various occupational elites, and through
social convention as a test for admission to social elites.
Le Page mentions, in particular, the emergence of the Caribbean Examinations

Council (CXC), the Caribbean Lexicography Project (realized in Allsopp’s Dic-
tionary of Caribbean English Usage), the civil service and educational institu-
tions that operate as “focussing agencies” (Le Page 1988: 34–35) for these norms,
thus converting them into yardsticks. Institutions such as the electronic media
or the education system, their effect on the requirements for employment, and
the stereotypes that they reinforce do inform speakers’ judgements of standard
usage.
19
In media, for example, the British Broadcasting Corporation (BBC) has tended
to choose a certain type of voice/broadcaster to sell a particular idea of Britain
and the British, particularly in its historical role as the model for the English
language community. Lippi-Green (1997) cites the then (1924) Managing Director
of the BBC as explicitly supporting the prescriptive role of media:
One hears the most appalling travesties of vowel pronunciation. This is a

matter in which broadcasting may be of immense assistance … We have
made a special effort to secure in our stations men who … can be relied
upon to employ the correct pronunciation of the English tongue (1997: 137).
She provides similar documentation from National Broadcasting Corporation

(NBC) urging use of General American, as it is “most readily understood” (Lippi-
Green 1997: 138). Pronunciation is the salient feature of language mentioned in
directives from media like the BBC or NBC, not only for the hiring of staff, but
also because broadcasters are expected to provide a model of “good” English
diction for the public.
According to McArthur (1992), the BBC policy on language use in broadcast-
ing continued to favour RP up to the 1950’s, using non-RP speakers typically for
weather, entertainment (gardening, sport, drama) and less serious topics. The
implicit message is clear – regional voices do not have the authority of the RP
Standard. The more inclusive language policy of the 1960’s did not extend to the
World Service, the branch of the corporation aimed at external markets, which
continued to prefer RP speaking broadcasters (at least until 1990). The British En-
glish heard by outsiders, as reflected in this distinction between local and foreign
broadcast requirements, is still, to a great extent, RP. And Sharwood-Smith (1999:
59) notes that this is itself the wish of the World Service target audience, who,
like the bosses of NBC in relation to American English, hold to an ideology that
RP is the English “everyone understands”.
In education, Lippi-Green (1997: 109) cites documents published by the Na-
tional Council of Teachers of English (USA), and Cheshire (1999: 147) the National
Curriculum for English (England and Wales), which suggest that language edu-
cation policy is also founded on the belief that the school system can and does
(re)direct pupils’ language in the appropriate way. And appropriate means the
legitimised “standard”, the variety
at the neutral (hegemonic) top and center of stratified society; any devia-
tions can only be interpreted as marked variants that index (…) the producer
as also being a correspondingly removed one (Silverstein 1998: 412).
20
Jamaican teachers and pupils are told in the curriculum that there is such a
standard locally – SJE – the target language for all Jamaicans in the education
system. However, examination of the material in the curriculum suggests that SJE
is, at least in morphology and syntax, little different from the IAE in use in the
school system of countries in which English is in general use. There are, of course,
Jamaicanisms that occur, even in formal written Jamaican English (Christie 1989;
1998b; Craig 1982; Shields-Brodber 1997) but these are not generally accepted in
exams.6 SJE in texts is little different from MSE, though students are drilled in
the specific aspects of structure that are deemed to be problematic because of
Creole. The syllabus for secondary schools is designed to equip the student with
(among other things) “the ability to use the language with precision, clarity and
grammatical correctness”. As Christie (1989: 256) notes, however, “it is not usu-
ally the case (…) that conscious attempts are made to depart from the traditional
model, British Standard English”.
The examples of teaching practice given at the start of this section, and the
data from sociolinguistic studies of Jamaican English, suggest that phonology is
therefore going to be the locus of any construct of a Jamaican norm of Standard
English. It is the phonology of standard English that is simultaneously the car-
rier of national identity - transforming members of the English language commu-
nity into participants in a speech community (Silverstein 1998: 407) – and one
of the indices of what Bourdieu (1984: 228) called “cultural capital”, aspects of
behaviour that reproduce the local social structure and privilege the holders of
the legitimised patterns of these behaviours. Thaxter’s and Escure’s works (cited
above) show that foreign models of spoken English are not used or targetted in ei-
ther Belize or Jamaica. Arguably, they can tell the listener only that the speaker
is foreign or pretentious. But when a speaker of SJE has options such as, stan-
dard [standa:rd], [standʌd] or [standǝd]; or only [onlɪ], [ʌnlɪ] and [ondlɪ] in the
social context, each will be associated with particular groups, length and type
of schooling, and other such factors that can locate the speaker socially in the
speech community for the hearer.
6 Christiecites as examples Today, when the eyes of the world is on us . . . or The water entering
the reservoirs were extremely muddy. Additionally, in formal Jamaican English (written and
spoken) sentences such as Here we have yet another of the anti-drug rally being held or A suspect
charged with possession of firearm are more and more common. In both cases, speakers seem
to be assigning number based on a logic that a) another = one = rally (SINGULAR) and b) the
suspect had one, therefore SINGULAR, firearm. In addition, the use of prepositions (as in I’ve
been in school from I was six) needs further investigation. She points out, however, that many
of these forms are not necessarily peculiar to Jamaica, but occur in MSE also (Christie 2003:
17–18).
21
Notions of standard, then, are fundamentally informed by perceptions of who

is using the language in the speech community. Thakerar et al. (1982: 235) report
that listeners presume “standard” forms in speakers they expect to use them. They
suggest in their discussion that, “it may well be that speech stereotypes exist such
that high status speakers are expected to talk with a standard accent (…) low sta-
tus speakers are expected to talk with a more non-standard accent”. Informants
in a previous study (they cite Thakerar & Giles 1981) heard more standard phono-
logical forms in the speech of a voice they were told belonged to a man who was
doing well in his university exams; the group who was told he was doing poorly,
heard more non-standard forms in the same voice (236). Accent is the term used
to define the supposed difference in speakers, with an assumption that success
in education is naturally associated with “proper” use of English.
Dyer & Beckford-Wassink’s (2001) study shows how problematic the issue is in
their discussion of Jamaican respondent’s judgements of other Jamaican’s speech.
19 informants were asked to listen to three guises of Jamaican speech and judge
who spoke better English or used more Patois. To summarize the data (25), the
particular structures that occurred (usually once or twice) in two of the guises
were as in Table 1.1.
Table 1.1: Features in two Jamaican guises (from Dyer & Beckford-
Wassink 2001: 25)
Speaker 1 Speaker 2
[kja] yes yes
[kh a] no yes
[tʌn] (turned) yes yes
tur[nd] (turned) no yes
fi im (his) no yes
his yes no
copula (attr.) yes no
noun + dem yes yes
noun + s yes no
Speaker 1 cannot be said to use fewer Creole forms than speaker 2, but she was
judged to be a Patois speaker by more listeners than Speaker 2. Speaker 2 was also
the only voice considered to be possibly someone from Kingston, “stoosh” and
“mixing” English and Patois. Speaker 2, for example, was presumed by listeners
to be the younger guise and therefore her use of Creole forms did not suggest
22
1.6 Aims and methods of the study
rural origin as it would in older speakers. Moreover, Speaker 2 was the only
one perceived to be a speaker of proper English by any of Beckford-Wassink’s
informants. It is, however, not clear if informants were interpreting the question,
who “uses” Patois or English, as meaning a) “is likely to be a user of” or b) “is
here using”. The single instance of “turned” in Speaker 2 might have been enough
to index the former in relation to English, notwithstanding whatever else the
speakers said. What may be important then is not how many English forms the
speakers uses, but which English forms and which Creole forms.7 Additionally,
given a particular social profile in the listener’s judgement, here in Beckford-
Wassink’s data being perceived as young and urban, occasional use of a particular
variant like [tʌn] turned or [fɪ ɪm] his may not index the same things as if those
same variants are used by another type of speaker.
What seems to be at issue here is that the expectation of a particular phonolog-
ical pattern, informed in large part by the idea of the speaker held by the hearer,
affects judgements of speech. And these judgements, of standardness or intel-
ligibility or group membership, are not necessarily altered by what is actually
produced. Indeed they colour perceptions in spite of it. So one identifiable influ-
ence on a speaker’s construct of standard speech is the forms they perceive to
be used by successful or high status persons in the community, notwithstanding
the circularity inherent in the idea that standard pronunciation is more likely to
be heard in speakers we expect to produce it.
1.6 Aims and methods of the study

I wish to explore aspects of the phonology of educated Jamaicans, speaking in
formal circumstances in the workplace, as I argue that SJE is essentially going to
be distinctive for its phonology. The study will have two main sections. In the
first (Chapters 2–3), the phonological characteristics of the staff of a government
agency, called JAMPRO, will be described and subject to a number of sociolin-
guistic correlations. The aim of this first section is to provide a description of
the use of select linguistic variables and their variants by a sample of educated
Jamaicans. I am especially interested in providing data on the extent to which
these features are in use in actual spoken formal Jamaican English. In the sec-
ond section (Chapter 4), the agency in question will be explored as an example
7 Mühleisen (2002) reports that for young black Londoners “the actual competence is of sec-
ondary importance, it is the symbol that matters. (…) phatic uses of Creole like the greeting
“wha‘appen” is often enough to stake one’s claim of “talking black” and belonging to a certain
group” (169). In much the same way, interviews on Jamaican television often start out in En-
glish (thus establishing the speaker’s competence) and then move into more Creole varieties.
23
of one of the mechanisms for normalizing SJE phonology, as it functions in the

Jamaican social context as a government marketing agency set up to promote
Jamaica to both local and international investors.
I take acrolect to be the following:
• The acrolect is the English normalized in Jamaica through public/formal

use by speakers at the top of the sociolinguistic spectrum, which evolves
with reference to that local sociolinguistic context. The evolution of course
is more constrained by the prescriptions of written English and the greater
access to education.
• The acrolect, SJE, is therefore typically going to be phonologically distinc-

tive from other varieties of Standard English. Moreover, the patterns that
occur in the acrolect or SJE must be informed by those that occur in the
speech perceived to be Creole. This is why some forms that occur in MSE
are avoided by SJE speakers when they are also aspects of Creole.
• While some speakers in the Jamaican speech community are vernacular

speakers of a variety of JE, and many more are vernacular speakers of a
variety of JC or both, SJE represents a situationally defined and defining
variety that is sanctioned and reinforced by institutions and agencies of
the state for formal/public discourse.
1.7 Data collection

Data was collected in 1994 at Jamaica Promotions (JAMPRO). JAMPRO has a to-
tal staff complement of 188, in various offices in both urban and rural Jamaica.
The principal JAMPRO building, from which all data was collected, is a five
storey structure located in New Kingston – the financial\business district in the
Kingston Metropolitan Area (KMA hereafter). It housed 153 members of staff at
the time data was collected. The layout of the building is depicted in Figure 1.1.
Of the 153 staff members in the Head Office, 104 were interviewed. Of this
104, 82 had their interviews recorded. The results presented here reflect the or-
ganization as it was at the time of data collection. This sample of the JAMPRO
population was then grouped in a number of ways, based on information gath-
ered from the questions asked of each of the 104 informants.
All interviews were done in an office on location loaned to me by the com-
pany. All informants were given appointments, usually a day before the actual
recording session, and told that I was conducting research on the company itself
24
1.7 Data collection
FLOOR 5 – President’s Office, Legal Services, Corporate Services

No. of staff, 12
President (Female)
Vice Presidents (1 Female, 1 Male)
Group Director (1 Male)
Directors (2 Female)
Officers (3 Female)
Secretaries (2 Female)
President’s Driver (1 Male)
FLOOR 4 – Research, Documentation Centre, Library, Market Development,
Policy & Projects
No. of staff, 26
Group Directors (2 Female)
Officers (8 Female, 2 Male)
FLOOR 3 – Manufacture, Tourism/Film, Productivity Centre
No. of staff, 29
Group Directors (2 Female)
Directors (3 Female, 1 Male)
FLOOR 2 – Agriculture/Agribusiness, Public Relations
No. of staff, 28
Group Directors (1 Female, 1 Male)
Graff
Directors (3 Female, 2 Male)
Ancillary Staff (1 Male)
FLOOR 1 – Human Resources, Office Administration, Registry, Intl. Trade
No. of staff, 53
Group Directors, (1 Female, 3 Male)
Ancillary Staff (9 Female, 9 Male)
LOBBY/RECEPTION
No. of staff, 5
Secretaries/Receptionists (4 Female)
Security (1 Male)
Figure 1.1: JAMPRO, Layout of Company by Sex, Rank and Floor
25
and the morale of staff. All were tape recorded with a notebook in plain sight in
which I would, from time to time, make jottings.
Wolfson (1976) and Bell (1984) both discuss the effect of the audience on speak-
er style. Both advance a notion that speakers design their speech style in response
both to the situational context and to their audience. Wolfson focusses more nar-
rowly on the interviewer’s effect on the subject, particularly when background,
variety and/or gender are seen to be different. In addition, she argues that in-
terviews, of the type conducted here with a question and answer format, consti-
tute a specific speech event with its own norms of interaction and language use.
This is, however, precisely the situation that an employee of JAMPRO is likely
to face when dealing with clients – providing answers to questions about do-
ing business in Jamaica. These interviews were all conducted by me, a female, a
stranger and one expressly from the local University. Moreover, all informants
knew they were part of a study and that they were being recorded. The audience
in these interactions is therefore a local one which includes not only me, the
interviewer, but, potentially, others at the University. I was interested in elicit-
ing formal speech, and was attempting to set up a situational and interactional
context that would discourage informality.
An informant can, of course, in contexts like this, “exploit available linguis-
tic resources to construct for herself a particular persona or to construct an en-
counter as intimate, distant, friendly or otherwise” (L. Milroy & M. Gordon 2003:
206). The informant, like all speakers, is an active initiator of use of particular
styles of speech (Rickford & McNair-Knox 1994), and might choose to depart
from the SJE that would be considered, certainly among educated Jamaicans, un-
marked in a formal, taped interaction with a stranger. As such, this study of
speech at JAMPRO rests on the assumption that for most, if not all, informants,
the speech produced during the interviews reflects their use of this unmarked
variety.
Each interview lasted, on average, 20–25 minutes. Each interview was tran-
scribed phonetically for subsequent analysis. These are, of course, relatively short
interviews, as I did not wish to give informants time to relax, but still wanted to
record a reasonable sample of each speaker’s production.8 This, of course, means
that for some variables the number of tokens recorded per speaker was small.
For many commonly occurring forms like the interdental fricatives or the vowel
/e/ I was able to collect adequate data. For example, on average each informant
8 Young (1973: 188, 191) shows that speakers in Belize used different frequencies of variants of
the same variable in earlier and later parts of interviews. Speakers who started with high
frequencies of “Standard English” variants, used fewer later on.
26
1.7 Data collection
produced roughly 50 tokens of the voiced interdental fricative variable; or 27 to-

kens of the mid-vowel variable. However, for other variables the data set is small,
and below the minimum recommended amount (10) for sociolinguistic analysis
(L. Milroy & M. Gordon 2003: 164, for example).
The collection of this data was therefore affected by two competing necessities.
The first, getting enough data for good statistical analysis, requires a fairly long
interview with each subject. The second, eliciting formal speech from an infor-
mant who is neither relaxed nor becoming more accustomed to the interaction
with me, required that interviews be kept short and business-like. In my favour
methodologically, the literature shows that judgements about the social place of
a speaker in the speech community, informed by their language use, are likely
to be based on the occurrence of a few forms rather than how many times that
form occurs (see results in Graff et al. 1986: 57; and other matched-guise studies
such as Dyer & Beckford-Wassink 2001). Indeed, the use of a single token, which
has significance to others in the speech community, can be enough to signal a
claim to a particular identity (Bell 2001: 167).
It is not unreasonable for me to assume that in cases where a few tokens of
a particular variable were collected from an informant, these not only reflect
appropriate use for the context of interaction but are also the basis on which
other Jamaicans can and do make judgements about the speaker.
All informants, whether recorded or not, were interviewed using the same set
of questions. None were required to fill out a questionnaire, nor were they shown
the questions. However, I asked all informants questions from a prepared script
(Appendix A).
The issues that I wish to explore, through use of this social and linguistic data,
are therefore:
• What is the construct that is SJE, as reflected in the patterns of phonologi-

cal use found in this sample of educated Jamaican speakers when in formal
interaction?
• How do the various groups identified differ in their use of these phonolog-
ical variables?
• Does JAMPRO select speakers of a particular sociolinguistic type for high

status positions?
• Do speakers with mobility aspirations pattern the phonology of their suc-

cessful colleagues?
27
• Is JAMPRO, in its practice and expectation, legitimizing certain speech pat-

terns?
• What is JAMPRO’s role in reflecting or promoting a Jamaican Standard

English?
28
2 Phonological variation in the
Jamaican acrolect
2.1 Why consider phonological variation?
This study is concerned with the formal, spoken language of a sample of edu-
cated Jamaicans, the Jamaican acrolect as explained in Chapter 1. The informants
used in this study all work for JAMPRO, an agency that has explicit language re-
quirements for staff, and this is reflected in the various language qualifications
prospective employees are expected to have. This agency, as outlined in Chap-
ter 4, also describes in its publications a construct of the language situation and
of the socio-cultural milieu in which it operates in Jamaica. In both descriptions,
the use of Standard English is stated to be an imperative. As such, there are a num-
ber of reasons why the analysis of language variation in the speech of JAMPRO
informants is necessarily going to focus on phonology, the essential concern of
this book.
In general, qualifications in English are required for non-manual employment
in Jamaica. Certainly, nearly all advertisements for white-collar jobs require ap-
plicants to have “a good command of the English language” or “excellent written
and oral skills” (in English, by implication, since JC is not normally written) or
“proficiency in English”. Additionally, announcements of these positions are ac-
companied by phrases such as “the ideal candidate should have…” – typically
a university degree, a credit in CXC English, or other such stipulations about
educational qualifications. These qualifications are, for the most part, locally or
regionally attained and success is judged through these examinations.
The CXC (and the national) exams, designed to assess proficiency in English,
have no equivalent test of oral communication on leaving the school system.1
Employers, therefore, rely on interviews, particularly, but not exclusively, for
those candidates to supervisory or management positions.
A number of comments by senior management at JAMPRO can be used to
reveal this focus on the interview and its importance in selecting staff.
1 The Caribbean Advanced Proficiency Examinations did introduce an oral English examination
in 1995, for students at the sixth form level.
2 Phonological variation in the Jamaican acrolect
…there was the question of why is it that we could not find any [men] at
the interview and selection. We always came up with women and we were
doing it strictly on the criteria that we set for recruitment…one of the first
criterion (sic) for employment is a first degree from CAST [now the Univer-
sity of Technology] and any other tertiary institution. I think they were so
much better at the interview. The women are qualified…am…I wouldn’t say
better qualified, wouldn’t say better qualified, in fact…at the stage of the
interview I ask is this person somebody I could send away next week to go
and talk to an investor or to go and sell Jamaica. (F55).2
…the women did better, they were more reliable, more conscientious in their
work. And even in interviews the women seem to shine more than the men,
so invariably they ended up employing more women than men (F87).
It seems to me that they [women] seem to do better at the interviews…(F56)
These comments clearly demonstrate the perceptions held by senior manage-

ment about male and female suitability for JAMPRO employment (see §3.2). But
they also reveal that the interview, face to face and spoken interaction with
the would-be employee, is crucial. Candidates for JAMPRO jobs are short-listed
based on their paper qualifications; but it is in the interview that the selection is
made, other qualifications notwithstanding.
In the interview, and in other contexts of spoken interaction, attention usu-
ally focusses on phonology and lexis when standards (in both the linguistic and
denotative sense) are being assessed. Hudson (1980: 44–45) proposes that within
communities alternatives in phonology are more likely to mark social differen-
tiation than those in syntax or morphology, particularly as mass education has
increased the spread of a (written) standard (see also Lippi-Green 1997; Sahgal &
Agnihotri 1985). It is true that broadly speaking in communities like Jamaica, dif-
ferences in morphology and syntax will sharply distinguish social groups across
the continuum, as in im ben a sliip ~ shi did a sleep ~ she was sleeping. However,
Hudson’s proposition would apply in the context of a formal interview, particu-
larly for a white-collar position, where the expected variety is SJE.
Salikoko Mufwene (pers. comm.) relates an example in support of this. Stu-
dents were asked to listen to two speakers, one an elderly woman and the other a
middle-aged man, both of them comedians, and to judge which seemed more non-
standard. There was general agreement that the man was the more non-standard
2 The code here is to be interpreted as M = male, F = female; the numbers represent the sequence
of interviews. This was then the 55th interview. Appendix B has profiles of all informants.
30
2.1 Why consider phonological variation?
of the two, even though he used a more standard morphology and syntax. His
phonology (especially the prosody), however, seemed to inform the listener’s
judgement of his standardness, as little attention was paid to his grammar until
it was brought to the students’ attention.
Further, Rickford’s (1987: 275) examples of Guyanese English related in the pre-
vious chapter, involving the speech of a teacher and of a barrister, are distinctive
only for the phonology. The informants used in this book are Jamaicans who
have mastered this written standard English well enough, as all but four have
completed at least a secondary level of education, and many have a tertiary or
graduate level education. JAMPRO would not have otherwise considered them
for employment.
As suggested above by Gupta (2001: 370), and as already shown in sociolinguis-
tic research, particularly for English speaking communities (Hewitt 1986; Labov
1972; Mugglestone 1995; Trudgill 1978), there are some accents that are considered
prestigious, as markers of identity, or social class membership or level of educa-
tion. While certain accents, such as RP, may have historically been given high
status socially, it is not clear that they still universally carry such associations. L.
Milroy (2002) suggests, for example, that in the UK it is apparent that RP “does
not constitute the general model of careful or educated speech” (9) for many
regions whose populations favour local patterns. Outside of the UK, as shown
in Thaxter’s (1977) Jamaican study, speakers also favour the local voice, the RP
accent assessed as foreign and/or pretentious (239). And Nair-Venugopal (2001:
47) reports that in Malaysia exonormative models of English might be viewed as
nonconventional or unexpected in corporate business contexts.
The idea of prestige accent(s) is therefore locally developed and particular to
the specific speech community in which speakers operate. It is within the speech
community that the shared indexicality underlying communicative acts of iden-
tity and groupness evolves (Silverstein 1998: 407). This would necessarily seem
to be most focussed around the phonology, especially for speakers who pass
through institutions that disseminate a near universal literary standard of En-
glish.
There is evidence that Jamaicans perceive accent and vocabulary also to be the
primary difference between Creole and English, at least as suggested by Beckford-
Wassink’s data (1999b: 66). Of 51 informants in her study, only 9 (18%) identified
any aspect of morphosyntax as distinguishing Creole and English, while 42 (82%)
mentioned either accent (“how they sound their words”) or accent and vocabu-
lary. And, Beckford-Wassink adds,
31
All the professional domains about which they were questioned (those
where the auditor might be a non-familiar or subordinate, as when answer-
ing the telephone, addressing an employer, or teaching) were deemed inap-
propriate for Patois usage. (Beckford‑Wassink 1999b: 72)
Mühleisen (2002) reports a similar functional distribution of language use

among Trinidadians. What respondents referred to as “Standard” (Trinidadian
English) was mostly used in formal contexts (classroom, workplace, church) and
with strangers (40). The speakers I recorded at JAMPRO did find themselves in
such a context, interacting with a stranger in a formal interview at their work-
place. There is every reason to expect them to be using the appropriately formal
style, with the focussed morphosyntactic norms typical of the acrolect. It is their
phonology, therefore, that is of primary interest here.
2.2 The phonological variables and their variants

This study, which takes an endonormative approach to defining the acrolect in
Jamaica, will therefore analyse use of a set of segmental phonological features in
the JAMPRO sample. The phonological features are as follows:
a) the word initial glottal fricative [h]
b) the voiced and voiceless interdental fricative, in word initial, middle and
final environments.
c) the low back stressed vowel, in words like not and possible.
d) the mid tense vowels [e] and [o] as they occur pre-consonantally, in items
like face and goat, and before [r] in beer and poor type words.
e) the word initial velar stop [k] before the low central vowel [a].
f) post vocalic rhoticity, before [+coronal] consonants in words like party and
forty.
g) two word final phonological stop clusters, specifically [st] and [nt].
h) two word final morphophonemic clusters, -n’t when used as not (can’t,
won’t, don’t); and past tense marking.
i) the word final unstressed vowel in words that end in -er (butter, teacher).
32
2.3 Features commonly identified in the literature (Group A)
j) the vowel in the final syllable -tion, as in education type words.
k) the voiced and voiceless alveopalatal affricate, in words like culture and
soldier.
I attempted to include all features that showed phonological variation in my

data. Most of these features were present in my data in adequate frequencies.
Where they were not, the data that can be discussed is presented. For example,
rhoticity is analysed before certain kinds of consonants, [+ coronal], because
there was, in part, better data in that phonetic environment than before [−coro-
nal] consonants.
The literature on Jamaican language varieties has identified some of these
phonological features as having variants that are described as either typically, if
not unequivocally, basilectal or mesolectal and has not included them as part of
the acrolect. These are presented in Group A, i.e. features a–e above. Additionally,
there were some features that I was interested in exploring, and which I have not
seen specifically discussed elsewhere in the literature. These are categorized as
Group B, i.e. f–k above. They might prove interesting in this sociolinguistic anal-
ysis. As a Jamaican, and as someone who considers herself a vernacular speaker
of Jamaican English, I recognize all of these features in my speech community.
While phenomena like h-dropping and voiceless TH stopping are possibly now
stereotypes (in the sense used by Labov 1972: 248) in Jamaica, pronunciations
such as [kʌltjʌ] culture are perhaps less in the consciousness of speakers. I wish
to explore how widely they are in use, particularly in the formal speech of ed-
ucated Jamaicans. The middle three features, in particular (f–h), have been de-
scribed elsewhere for Jamaica. However, there has been little focus on their use
by acrolectal speakers (but see Shields’s 1987; 1984 study of consonant clusters
in JE).
The results for the total JAMPRO sample, all 82 informants, will therefore be
presented for each feature discussed, so that some comparison can be made with
the descriptions in the literature and generally.
2.3 Features commonly identified in the literature (Group

A)
Generally, the discussion of these features across the Jamaican continuum is typ-
ified by the following description by J. C. Wells:
33
At the upper (acrolectal) end of the social and linguistic scale, Jamaican
consonants are phonetically much as in the standard accents (. . .) Further
down the scale, in the mesolect and basilect, the characteristics emerge:
TH stopping, Cluster reduction, avoidance of [ʒ] and some [v], H Drop-
ping, semivowels in words such as /kjat/ cat (. . .) face acrolectally [fes] and
basilectally [fiɛs]… (1982c: 575–576)
Acrolectal varieties in this description are characterized by very low frequen-

cies, if not an absence, of certain variants; it is, in fact, this absence that distin-
guishes them from mesolectal and basilectal varieties. Interestingly, Wells says
that the phonetic properties of acrolectal varieties are much as in other standard
accents. Clearly, an American, Canadian and British speaker (or an Indian, an
Australian and a Jamaican) will not sound the same. This analysis suggests, there-
fore, that Wells is either speaking of an idealization of Standard English that
disregards even regional variation in spoken norms or is locating much of the
distinctiveness of the acrolect in prosody. As discussed above, there is no stan-
dard English accent. Spoken English is generally outside the normative pressure
exerted by the literary standard and it is typically the phonetic characteristics of
the educated speaker that distinguish varieties from different speech communi-
ties. My concern here is then phonetic, the distribution of the variants of these
variables in formal spoken JE. As was pointed out in the previous chapter, very
little investigation of what has been called “the local standard” or acrolect, as
used by speakers, has been done. I will therefore use the term Putative Jamaican
Standard (PJS) to label an unverified but presumed local standard in order to
distinguish it from the SJE which is the subject of the research in this book.
As point of reference I include the following phonemic inventory of JC and
JE in Tables 2.3–2.4. The JC vowel inventory is based on Meade (2001: 35–42).
Meade’s inventory discusses a number of previous texts3 and his own work done
on language acquisition in Jamaica. The JE vowel inventory is based on Beckford-
Wassink’s data (2001: 150–151), which discusses earlier work by J. Wells (1973).
The consonants of both systems are taken from Meade’s work. As such these in-
ventories can provide some indication of what phonological variation may occur
in this JAMPRO sample as well as an example of what is idealized at the po-
lar ends of the Jamaican continuum. While both Meade and Beckford-Wassink
use similar symbols for most of the phonetic representations of phonemes, one
or two differences did occur. Meade, for example distinguishes [a] and [ɐ] in
3 Meade’s work discusses, for example, Akers (1981); Alleyne (1980); Cassidy & Le Page (1967);
Beckford‑Wassink (1999a); Devonish & Seiler (1991).
34
JC, which is more typically presented as [a] in the literature and by Beckford-
Wassink. I have therefore taken the common symbols from both accounts in or-
der to facilitate comparison.
2.3.1 Basic JC and JE phoneme inventories
Table 2.1: JC and JE vowel inventory
Item JC JE
bit /ɪ/: [ɪ] /ɪ/: [ɪ]
beat /i/: [i] /i/: [i]
bet /ɛ/: [ɛ] /ɛ/: [ɛ]
bait /ɪɛ/: [ɪɛ], [ie] /e/: [e]
bat /a/: [a], [ɐ] /a/: [a]
bath /a:/ [a:] /a:/ [a:]
Bob /a/: [a] /ɔ/: [ɔ]
but /ʌ/: [ʌ] /ʌ/: [ʌ]
boat /ʊʌ/: [ʊʌ], [uo] /o/: [o]
book /ʊ/: [ʊ] /ʊ/: [ʊ]
boot /u/: [u] /u/: [u]
bout /ʌʊ/: [ʌʊ], [au] /aʊ/: [aʊ]
bite /aɪ/: [ɐɪ] /aɪ/: [aɪ]
noise /aɪ/: [ɐɪ] /ɔɪ/: [ɔɪ]
Essentially, we can identify the mid-vowels and the low back vowels as areas
of difference between the two systems. JC typically does not distinguish /a/ and
/ɔ/; and the JE mid vowels /e/ and /o/ are diphthongs in JC.
Most of these consonant features are discussed below in more detail. Briefly,
JE is typically described as having interdental fricatives, the voiced alveopalatal
fricative and the glottal fricative while these features are not usually identified
for JC. Additionally, JC has contrastive palatalization of the velar stop in some
environments.
2.3.2 Word initial glottal fricative /h/

/h/ is not phonemic in many varieties of Jamaican Creole (Akers 1981: 32; Devon-
ish & Seiler 1991: 7; Meade 2001: 40), though there is evidence that in Western
Jamaican varieties (e.g. Trelawny, St. Elizabeth) it is (J. Wells 1973: 12). Hyper-
correct use of the fricative is common in Jamaica (Cassidy & Le Page 1967: lxii),
35
Table 2.2: JC and JE consonant inventory
Item JC JE Item JC JE
bat /b/: [b] /b/: [b] fat /f/: [f] /f/: [f]
pat /p/: [ph ] /p/: [ph ] zap /z/: [z] /z/: [z]
dab /d/: [d] /d/: [d] sat /s/: [s] /s/: [s]
tap /t/: [th ] /t/: [th ] shat /ʃ/: [ʃ] /ʃ/: [ʃ]
that /d/: [d] /ð/: [ð] rouge /ʤ/: [ʤ] /ʒ/: [ʒ]
thing /t/: [th ] /θ/: [θ] jack /ʤ/: [ʤ] /ʤ/: [ʤ]
kit /k/: [kh ] /k/: [kh ] chat /ʧ/: [ʧ] /ʧ/: [ʧ]
give /ɡ/: [ɡ] /ɡ/: [ɡ] lap /l/: [l] /l/: [l]
cat /kj/: [kj] /k/: [kh ] wag /w/: [w] /w/: [w]
gab /ɡj/: [ɡj] /ɡ/: [ɡ] rat /r/: [r] /r/: [r]
mat /m/: [m] /m/: [m] yak /j/: [j] /j/: [j]
nab /n/: [n] /n/: [n] hat /h/: [h]
ing /ŋ/: [ŋ] /ŋ/: [ŋ]
vat /v/: [v], [b] /v/: [v]
particularly in Kingston where my sample was recorded. In local theatre, for

example, this hypercorrect use is often played for comic effect in characters un-
derstood as “speaky-spoky”. According to Patrick (1997) this style of Jamaican
speech “is clearly recognised and labelled by Jamaicans, and is realized regularly
by hypercorrect [h] and [ɔ]” (48). He gives examples like [hɔlon] alone and jokes
such as “hemphasize your haitch you hignorant hass” which are heard locally.
In my total sample of speakers, eleven informants produced this hypercorrect
pattern, as in F52 [honiŋ] owning or M27 [haʊt] out, and typically with single
attestations. The more general pattern was h-dropping, a term that clearly as-
sumes that speakers are using an English phonological system, which does have
/h/. Evidence for this is furnished in the pattern of use of aitch in the texts col-
lected. Except for the 11 whose usage suggests non-contrastive [h], as in [haʊt ~
aʊt] out, most informants produced /h/ in the same way as in IAE, distinguishing
lexical items like ‘hand’ and ‘and’, without inserting [h] where it is not lexically
specified in IAE. It is entirely possible that some speakers have idiolects that have
neither /h/ nor a hypercorrect usage of [h], while having other English phono-
logical features. I did not find such a speaker in my sample.
36
Outside of Jamaica, [h]
…has become one of the principal signals of social identity, its presence in
initial positions associated almost inevitably with the ‘educated’ and ‘polite’,
while its loss commonly triggers popular connotations of the ‘vulgar’, the
‘ignorant’ and the ‘lower class’ (Mugglestone 1995: 107).
For some Jamaicans, h-dropping is also overtly stigmatized, though it is not at

all clear that this is universally the case or that h-lessness is even perceived by the
hearer. It is a stereotype held by others in the Caribbean of bad Jamaican speech
(J. C. Wells 1982c: 569). This stereotype is confirmed in Roberts (1988: 90) and in
Allsopp’s dictionary. As Allsopp points out, “/h/, especially in initial position, is
lost noticeably in Jamaica, even in the speech of educated persons” (1996: xlvii).
Historically, variable use of [h] was a feature of some of the varieties of En-
glish that came to and were used in Jamaica during the early years of British
settlement in the mid to late 17th century. H-dropping became salient and the
shibboleth of vulgar speech in the metropole by the 18th century (Beal 2002: 103).
The more consistent pronunciation of [h], largely due to its presence in the or-
thography, indexed literacy; and therefore it was the “illiterate speaker” of the
1850’s in Britain who was perceived to drop [h]. Mugglestone (1995) documents
that:
While ‘literate speakers’, at least in terms of the prevailing prescriptive ide-

ology, thus made plain the facts of their superior education by matching
grapheme with appropriate sound, or <h> with [h] in their speech, so there-
fore did the non-appearance of [h] take on the values (. . .) of the ’lower
class’ (117).
As such, hypercorrect use of [h], due to the stigma attached to [h] dropping,
developed as a stable feature in sections of the British population over two cen-
turies ago, a function of the changing prescriptions on [h] articulation and the
focussing of attitudes to “h-less” speakers. Mugglestone (1995) cites a number
of examples, one of which is from Poor Letter H (1866): “…he would persist in
saying that the habbey was his ’obby” (124).
However, J. C. Wells (1982b: 432) tells us that h-drop is not a feature of Irish
English; moreover /h/ is more widely distributed in Irish than it is in English.
This suggests that variable (but not yet hypercorrect) use of [h] would proba-
bly have been more typical of founder English varieties (Cassidy & Le Page 1967:
lxii) than an aspect of the speech of the Irish indentured servants who were also a
37
presence in Jamaica at the time (see for example Beckles 2000: 228). Moreover, it
is possible that given the regional associations in Jamaican Creole with variable
use of [h] (Eastern Jamaica and Kingston), and the early history of English settle-
ment in Eastern Jamaica (Sherlock & Bennett 1998: 85) particularly in urban areas
like Port Royal and Kingston, the pattern of aitch distribution in Jamaica reflects
to some extent the pattern of white settlement in the early years of the colony.
In that respect, the linguistic data may, as Devonish argues (2002: 180), provide
historians with information that points to particular demographic patterns that
require exploration.
Lalla & D’Costa (1990) cite the example from the early 19th century of a Ja-
maican planter who is reported as saying “hedicating the negroes is the only
way to make them ’appy (…) hedication is hall in hall” (142). This suggests that
hypercorrect use of [h], specifically h-insertion, has also been a feature in some
Jamaican varieties for at least the past two centuries, no doubt also influenced
by varieties of Creole coexisting in the speech community. It is interesting that
neither h-drop nor h-insertion has been identified in US varieties, though occa-
sionally “an” occurs before words beginning in aitch, suggesting possible absence
(Krapp 1925: 206 gives the example an halfe from 1653). It has been described in
some Canadian (Newfoundland) varieties of English (Kirwin 2001: 447). It ap-
pears, therefore, that some ecologies were more favourable to the persistence of
variable/hypercorrect [h] than others. Certainly, speakers acquiring or learning
English in Jamaica would have been exposed to and selected these hypercorrect
forms, options existing in the Jamaican “feature pool” (Mufwene 2001: 4), espe-
cially if such forms were used by some speakers at the top of the society.
The JAMPRO data show the pattern in Table 2.3 for h-dropping word initially4 .
Table 2.3: h-drop in the total sample
h-drop h
184 (10.4%) 1592 (89.6%)
The data here points to a fairly low incidence of h-drop in these speakers,
with many not doing it at all. The hypercorrection discussed above does suggest
a sensitivity to use of this variable as marking a speaker of “good” English. Of the
total sample, 15 of the 82 informants account for the majority of cases of h-drop
4 This data does not include occurrences of typically unstressed items like has, as when the
auxiliary of a participle (he [hǝz] gone) or have (we [hǝv] seen).
38
(111 of the 184 or 60%), the more typical pattern being one or two instances of
the feature in speakers’ recordings. Of those 15 informants, 8 – two men and six
women – had no productions of hypercorrect h-insertion. A fuller discussion of
hypercorrection in this sample is presented in §2.4 of this Chapter, and Chapter 3
and 4 present the results of a number of sociolinguistic correlations.
2.3.3 The interdental fricatives /θ/ and /ð/

The interdental fricatives [θ] and [ð] vary with the alveolar stops [t] and [d] in
Jamaica, as in [θɪk ~ th ɪk] thick and [ðat ~ dat] that. The alveolar stop variants are
identified as a Creole feature in this social context, though it is found in varieties
of non-standard English such as Irish or some US varieties. The literature on
Jamaica has generally suggested that the stop variants are not part of the acrolect
(Akers 1981: 33; Meade 2001: 23; Thaxter 1977: 239) or the speech of the highly
educated (J. C. Wells 1982c: 575). However, Allsopp (1996: xlvi) sees this as a
general feature of Caribbean speech; and Young (1973: 188) found relatively low
frequencies of the fricative variants in even formal Belizean speech. Similarly,
F. Miller (1987) in her study of AJE found that her speakers, in formal contexts,
used the alveolar stop variants about 30% of the time (183).
Miller’s analysis does not distinguish patterns of use for the voiced and voice-
less stop, though the tables she presents suggest that the voiced stop is pro-
duced more frequently by speakers (57). In Thaxter’s (1977) study, the “librar-
ian’s speech”, identified by his sample of Jamaican teachers as the type of speech
they would pass on to their students (249), was characterized by no voiceless TH
stopping, slightly more use of [d ~ ð]. It seems necessary then to distinguish the
patterns for the voiced and voiceless variables in our analysis. Certainly, I have
heard the use of [θ] where most speakers would have [t] in items like [θrut]
truth and [θɛrǝrɪzm] terrorism in some Jamaican speech, including the JAMPRO
sample (F76 [θʌg] tug). This suggests that contrastive /θ/ is an aspect of the ac-
rolect and that some informants perceive the interdental fricative to be “correct”.
A parallel use of [ð] did not occur in my sample and is rare, even though J. C.
Wells (1982c: 565) says that [bɛð] bed is possible. It is not clear from his text,
however, whether he is suggesting a theoretical possibility, as his analysis does
not differentiate patterns for the voiced and voiceless stop either.
TH stopping is a general feature of most dialects of British English (Cassidy
& Le Page 1967: lviii), and in some with an additional variant [f]. This latter vari-
ant is also found in AAVE and Bajan but not in Jamaica. The pronunciation of
the interdental fricative, influenced by spelling changes prior to the 15th century
(Pyles & Algeo 1993: 168), was normalised at the time of language contact in
39
Jamaica. I find no references to hypercorrect use except in varieties of Hiberno-

English. Baugh & Cable (1993: 313) cite 19th century forms such as [θru] true and
[bʌtθǝr] butter, pronunciations that occur today in southern Irish English (Trudg-
ill & Hannah 1994: 105). There is evidence of an Irish population in early Jamaica,
particularly in the early colonial period before the 18th century. According to
Cassidy & Le Page
Thus, whilst it is almost certain that [interdental fricatives tended to be pro-

duced as alveolar stops] due to the lack of /ð/ and /θ/ in many West African
languages, the possibility remains that in the speech of [early white settlers]
these processes had already begun without African influence (lvii).
I find no reference to hypercorrect use of interdental fricatives in early Ja-

maica. It is possible that the forms came from the speech of Irish indentured
servants, but the citations I can find are all from the 1960’s, when there was
more widespread access to education and greater exposure to the standard usage
for speakers. Moreover, hypercorrect [θ] is not identified in American varieties,
notwithstanding the Irish influence in some regions.
I will analyse use of these two variables in different positions in the word. For
one, it is possible that the most common occurrences of word initial /ð/ (words
like the, them, that) are less consciously monitored by speakers, even in formal
contexts, because they are typically unaccented and weak in discourse (see Gim-
son 1980: 185). In addition, the type of word that will have a middle TH is, at the
very least, disyllabic and the variant use here may better reflect the patterns of
formal speech, with its greater proportion of longer and more erudite words.
The data show the pattern in Table 2.4 for the interdental fricatives and vari-
ants in all speakers.
Table 2.4: TH stopping in the total sample
d ð t θ
Word Initial 2290 (54.3%) 1924 (45.7%) 117 (14.4%) 693 (85.6%)
Word Middle 76 (25.0%) 228 (75.0%) 22 (18.8%) 95 (81.2%)
Word Final 33 (27.0%) 89 (73.0%) 20 (24.0%) 63 (76.0%)
The data reveal two things. Firstly, the pattern for the word initial voiced frica-
tive is noticeably different from all the others in the table. It could be that con-
trastive /ð/ is not a feature of some idiolects in my sample, while /θ/ is; however,
40
as I have no instances of forms like [ðɔɡ] dog, it is more probable that it is and
that its phonetic realizations are less focussed in word initial contexts. Crucially,
in the same context, speakers produce significantly fewer voiceless stops – and
the hypercorrection mentioned above again indicates that use of [θ] particularly
is a salient feature of “correct” speech. The relatively low incidence of TH stop-
ping in all other environments also suggests that use of fricatives is the more
general pattern for these speakers in formal contexts, with some variation with
the alveolar stop.
2.3.4 The low back stressed vowel /ɔ/

In “not” words the possible variants are [nat ~ nʌt ~ nɔt]. The “Creole” variant
in this set is the low central vowel [a], as Jamaican Creole has one low vowel
phoneme /a/, with [at] produced for both hat and hot. J. C. Wells (1982c) says
that, “block is acrolectally [blɒk] (or sometimes with an unrounded back vowel
[blɑk), but basilectally homophonous with black [blak]” (576).5 Similar analyses
of the basilect can be found in Akers (1981: 25) and Devonish & Seiler (1991: 5).
Hypercorrect use of [ɔ] is the other feature that Patrick (1997) associates with
“speaky-spoky”. Forms like [sɔlɛrɪ] salary (M47), [rɪlɔks] relax (F53) and [fɔ:mʌ]
farmer (F80) produced by some of my informants would suggest the kind of sen-
sitivity to the feature discussed by Patrick and a belief that [ɔ] is the “correct”
reflex of [a]. Like Wells, Beckford‑Wassink (2001: 151) and Meade (2001: 42) de-
scribe acrolect (dominant) speech as clearly distinguishing /a/ and /ɔ/ in “not”
words. However, words like mother, government and colour can be, and often are,
pronounced as [mʌðʌ ~ mɔðʌ], [ɡʌvǝmɛnt ~ ɡɔvǝmɛnt] and [kʌlʌ ~ kɔlʌ] in Ja-
maican English. This suggests that the relationship between /a, ʌ, ɔ/ is a complex
one at the upper end of the construct called the continuum and that, for some
speakers, [ɔ] is a reflex of /ʌ/ as well. In Beckford‑Wassink’s (2001) analysis, her
sample of speakers from rural St. Thomas in Jamaica did not have a distinct /ɔ/,
though they did distinguish the other two sounds (150).
Prescriptive texts for British English from 1673 are cited by Freeborn (1998: 356)
as listing the following as homophones: chaps and chops; band and bond; knots
and gnats. By the end of the 17th century there is evidence of variation between
[a] and [ɔ] – in words such as quality, watch, what – the latter variant initially
regarded as vulgar but becoming more standard by the 19th century (Beal 2002:
128). Mugglestone (1995) gives an example from 1697 of an “affected” literary
5 Wellsuses the symbols [ɑ] and [ɒ] to depict what I will identify as the [ɔ] sound. I find the
Jamaican vowel to be higher and rounder than he suggests (see also Beckford‑Wassink 2001:
142).
41
character, Lord Foppington, who is depicted as speaking in this way: “Now it

is nat passible far me to penetrate what species of fally it is thau art driving
at (. . .) I must confess, I am nat altogether so fand of” (216). The rounding of
the vowel seems to have spread from pre-rhotic environments (warm, quart),
gradually becoming the prestige variant generally (Beal 2002: 128). Essentially,
the varieties of English that came into Jamaica would have had this variation
between [a ~ ɔ], and, like the “problem” with [h], it persisted for some time.
A number of commentators on Jamaica remark on the language of white Cre-
oles, in particular women’s speech. Lady Nugent writing in the early 1800s re-
marked that “…many of the ladies, who have not been educated in England, speak
a sort of broken English” (Wright 2002: 98), a complaint echoed by Marly, More-
ton and other contemporary writers (Lalla & D’Costa 1989: 131). They cite the
forms “haut” (heart) and “knaum” (nyam) in a 1793 publication, which I interpret
as [hɔ:t] and [njɔ:m], both of which would be called hypercorrect in current JE.
It is not clear whether they should be so labelled in the cited female speaker, but
the forms have therefore been in the Jamaican speech community at least since
then. Crucially, the speaker has taken a clearly African lexical item ([njam] eat)
and adapted the phonology to what she considers a prestige pronunciation. Ja-
maican ideas of correct pronunciation must have been informed by a Creole that
did not have the /ɔ/ vowel, and it is even possible that its selection as prestigious
was an autonomous Jamaican occurrence, and not one that necessarily followed
from metropolitan norms of use. Table 2.5 presents the JAMPRO data.
Table 2.5: The low back vowel and variants in the total sample
nat nʌt nɔt

385 (19.8%) 325 (16.8%) 1230 (63.4%)
In these speakers there is a low incidence of the “Creole” variant, in contrast to

the relatively high incidence of [ɔ]. There is also a low occurrence of the [ʌ] vari-
ant. However, close to a third of the time the [ɔ] variant is not produced. Again,
this is one of the variables for which we find speakers producing hypercorrect
forms, which may indicate what is perceived to be correct and to be stigmatized
in the general speech community. For some speakers at least, [a] is to be avoided,
even where an item is lexically specified in IAE to have this phoneme.
42
2.3.5 Pre-consonantal/pre-rhotic mid tense vowels /e/ and /o/

In words like boat [buot ~ bot] and face [fies ~ fes], speakers vary the mid tense
vowels with a diphthong (described by Devonish & Seiler 1991: 8 as a syllable
nucleus preceded by either a palatalized onset in the case of [fjes] or a rounded
one in the case of [gwot]). Beckford‑Wassink (2001) found that in her acrolect-
dominant speakers, production of this pair of features was related to factors like
variation in style (more informal) and gender (male speakers). Generally, she
found that all acrolect-dominant speakers showed a predominance of monoph-
thongal forms in all test contexts, but particularly when called on to produce care-
ful speech in a word list (153). However, her informants did not use the diphthong
variants with the same frequency, [uo] being less likely to occur than [ie], specif-
ically for men (154). Alleyne (1980) also suggests that [ie] is a more widespread
feature in Jamaica (41). Meade (2001: 113) identifies the diphthongs as “variants
in most Jamaican varieties, but to different degrees”. Use of the diphthongs is de-
scribed as mesolectal in Thaxter’s study (1977: 239) and basilectal by Akers (1981:
25) and J. C. Wells (1982c: 576).
In my analysis, I make the distinction between the mid-tense vowels as they
are produced in a pre-rhotic environment and when occurring before other con-
sonants. For some Jamaicans sets of words like beer, bare and bear or where and
wear are homophonous either as [be:r]/[we:r] or [bie r] / [wie r]; and this even
in speakers who vary for example [e ~ ie] elsewhere before other consonants.
Christie (2003: 19) describes the former pronunciations as “hypercorrect”, an
overuse of the [e] vowel in order to avoid the Creole diphthong, as a distinction
between words like beer and bear is the pattern in MSE.
Historically, the diphthongs (both before other consonants and [r]) were as-
pects of some of the varieties of English that came to Jamaica (Cassidy & Le Page
1967: xlvi). In addition, Freeborn (1998: 356) cites late 17th century prescriptions
that indicate that bare and bear or chair and cheer are to be homophones. To la-
bel similar contemporary Jamaican pronunciations “hypercorrect” is, of course,
to stipulate that the changes in the Standard in Britain are necessarily to be re-
flected in Jamaica.
With the back vowel, speakers can be distinguished by the surface vowel of
an underlying /o/ in poor type words – having either the vowels [o] or [ɔ]. For
some speakers the /o/ can become phonetically [o], [uo ] or [u o]; for others the
/o/ is realized as [ɔ]. Indeed, in some Jamaican speakers a word like our can be
realized as [or], and possibly as a consequence is sometimes pronounced as [ɔr]
in formal contexts such as that of a radio broadcast. Pre-rhotic back diphthongs
occur in Jamaican speech therefore in variation as follows: poor, more or court-
43
house is produced as [puor ~ pɔr], [muo r ~ mɔ:r], [ku othaʊs ~ kɔrthaʊs]; note
that JE /ɔr/, as in forty, horse, north, also varies with the expected [a:] described
for the previous variable ([fa:ti], [ha:s], [na:t]). It is possible, therefore, that the
pre-rhotic context is one linguistic constraint on the type of variation identified
above for the production of diphthongal allophones in speakers.
Table 2.6: Mid tense vowels and variants (all environments) in the total
sample
uo o ie e
246 (11%) 2041 (89%) 748 (26%) 2145 (74%)
The data here does support Alleyne’s view and Beckford-Wassink’s findings
about the higher incidence of [ie] in Jamaican speech when all linguistic contexts
are analysed together. However, when pre-rhotic occurrences are excluded from
the data, there is a noticeable difference in the way the diphthong patterns. As in
Beckford-Wassink’s data on careful speech, in these JAMPRO interviews we ob-
serve a predominance of monophthongal forms generally. However, I find there
are some speakers who do allow diphthongs, and typically only in the pre-rhotic
environment.
The data in Table 2.7 shows how the pre-rhotic diphthongs occur at JAMPRO.
Table 2.7: Comparison of mid-tense vowels and variants in the total
sample
[ɔ] [uo ] [o] [i e] [e]

pre-rhotic 453 (73.2%) 116 (18.7%) 50 (8.1%) 485 (45.5%) 579 (54.5%)
(court) – (beer)
pre-consonantal 130 (8.0%) 1538 (92.0%) 263 (14.0%) 1566 (86.0%)
(coat) – (babe)
For the back vowel set, most speakers produce the [ɔr] in poor type words,
with a little over quarter of the sample using the other variants. The pre-rhotic
environment for the back vowel then seems to inhibit diphthongization in ways
that it does not for the front vowel. And the /o/: [uo] allophone is also very infre-
quent pre-consonantally in my sample. Close to half of my informants, however,
use the diphthong variant of the front vowel before [r]. Essentially these speak-
ers tend not to produce [uo] in any contexts, selecting either forms like [bot] boat
44
or [pɔr] poor in formal speech. In the same stylistic situation front diphthongs
occur more frequently in all environments, but particularly pre-rhotically.
2.3.6 The word initial velar stops /k/, /g/

In Jamaican speech, before the low central vowel /a/, the /k/ and /ɡ/ have the vari-
ants [kh ] and [kj], as in cat [kh at ~ kjat] or card [kh a:d ~ kja:d]); or [ɡ] and [ɡj], as
in gap [ɡap ~ ɡjap] or guard [ɡa:d ~ ɡja:d]. The /kj/ and /ɡj/ are contrastive in Cre-
ole before /a/ - [ɡja:dn] garden and [ɡa:dn] Gordon; [kja:f] calf and [ka:f] cough;
but *[ɡjaspl] gospel, *[kja:k] cork . It is non-contrastive in JE, as the items garden
and Gordon are distinguished instead by the vowel a/ɔ contrast. The palatal glide
in words such as garden or card was, up to the end of the 18th century, regarded
as “smooth and elegant (…) sufficient to mark the speaker as either coarse or ele-
gant, as he adopts and neglects it” (Mugglestone 1995: 233 citing Walker 1781). By
the middle of the next century its status had changed (Cassidy & Le Page 1967:
lviii) and descriptions such as “affected”, “old-fashioned” and “rapidly dying out”
were used for this feature, though its association with “proper” female speech
seemed to have persisted somewhat longer (Mugglestone ibid.).
Current analyses of Jamaica tend to label [kj] a “Creole” or basilectal variant
(Akers 1981: 33; F. Miller 1987: 89; J. C. Wells 1982c: 569; but cf. Patrick 1999:
96 and 2000: 11). However, in her AJE sample, F. Miller found that the [kj] was
produced for 44.3% of items and more by upper middle class men than by other
groups. She states that, “no linguistic constraints on this variable were observed”
(pg. 118). Interestingly, she cites a letter to the local newspaper by a reader who
wished to point out to Jamaicans that the “correct” pronunciation of a word like
Canada is [kjanɪdʌ]. This possibly indicates that some speakers hold a competing
idea of the standard pronunciation that differs from modern MSE norms.
In a previous study (A. Irvine 1994: 69), I found that educated, affluent speak-
ers avoided the [kj] before the long vowel, but more consistently produced /kj/
before the short [a] (see Table 2.8).
Table 2.8: The incidence of the palatal glide in eight male speakers
kj + a kj + a: kh + a kh + a:
61 (94%) 10 (6%) 4 (9%) 41 (91%)
Patrick does acknowledge the importance of vowel length for the incidence of
the palatal variants, but argues instead for -ar words as a separate class for anal-
ysis from -at type words. So, for example, the prestige pattern of [kj] use (found
45
in the urban, upwardly-mobile, middle class speaker) will permit [kjat] cat but
inhibit [kja:r] car. I have found in my sample, however, that speakers do produce
forms like [kjarɪ] carry, [kjaraktʌ] character, [kjarIbiǝn] Caribbean; and there is a
prominent insurance company locally that advertises its [gjaranti] guarantee of
good service. Moreover, the -AR class he identifies (pg. 107) of lexical items like
cards, car, guards and garbage all have long vowels, and it is therefore not clear
why they are analysed as an exceptional word class. Dyer & Beckford-Wassink
(2001) conclude in their study that: “where the status of the upwardly mobile
speaker was once marked by the absence of Creole forms, certain Creole forms
such as (KYA) are now welcomed” (31). They find (pg. 23) palatal glide insertion
to be constrained by the number of syllables in the word (occurring typically
in mono- and disyllables). I have also not found this pattern in my sample, as
attested to above.
The pattern of use among my informants was as described in Table 2.9.
Table 2.9: Word initial velar stops and variants in the total sample
kja kh a kja: kh a:
119 (56.9%) 90 (43.1%) 19 (19.8%) 77 (80.2%)
In this data, can (the modal auxiliary) was excluded in the many instances it
was produced as [kɛn] and therefore was not comparable with other items in
the data set. This reduced the token count for this feature in the data. I have
therefore collapsed the voiced and voiceless velar stop in this analysis under the
general heading of /k/. The palatal velar before the short vowel occurs frequently
in the informants’ speech. Notably, the incidence of the feature before the long
vowel is much lower, and comparable to the more stigmatized variants [nat] and
voiceless TH stopping. The pattern in the sample, which shows variation before
the short vowel, suggests that [kh ~ kj] is linked to /ɔ/ use in JE and may be a
phonetic feature that draws attention to possession of distinct /a/ and /ɔ/ sounds
(see Cassidy & Le Page 1967: xlix). Arguably, use of [kj], which is prohibited
before [ɔ], more sharply distinguishes [a] from [ɔ] in a sociolinguistic context
where the two vary in many speakers’ production.
The data on TH stopping, [e ~ ie] and [kj ~ kh ], raises an interesting question
about sociolinguistic variation in pairs of variables. It is possible that speakers
pay “attention” to only one member of the pair of related phonological variables
when prestige/stigma are at stake. While the incidence of word initial voiceless
TH stopping is very low in the sample with a 14% rate of occurrence, the voiced
46
2.4 Features not widely discussed in the literature (Group B)
counterpart was produced 54% of the time for the variable. And the incidence
of the front diphthong before /r/ is more than four times that of its counterpart
elsewhere. Here again we find one of a pair of variables being singled out and
being used to index prestige/stigma, as [kja:] is seldom produced in a sample in
which [kh a ~ kja] freely vary. This issue will be explored fully in the final section
of Chapter 3.
2.4 Features not widely discussed in the literature (Group

B)
Brodber (1989) identifies at least two types of speakers of Jamaican English – one
group “generally displays a command of a variety of formal and informal styles
[of JE]”, while the other, an “adherence, in speech, to spelling pronunciation, and
careful, somewhat measured, articulation” (46). The former tend to be vernacular
speakers of English who acquire Creole later, the latter learn English in the school
system.6 Mugglestone (1995: 158) and J. C. Wells (1982a: 229) describe a similar
distinction in RP English, with speakers of adoptive RP i.e. “those who did not
use this accent as children” tending to avoid features that occur naturally in the
speech of “mainstream” speakers. I take the first three variables dealt with below
to be instances of the kind of articulations Brodber discusses for adoptive JE.
This section on Group B variables is completed by an examination of rhoticity
and final cluster simplification in the JAMPRO sample.
2.4.1 The word final unstressed vowel in words that end in -er
In some speakers, pronunciations such as [ʤʌmekʌr] Jamaica and [bʌtʌr] butter
occur. This may be explained in two ways. Firstly, there seems to be a generalized
sensitivity to the (perceived as) Creole [a] discussed above, which occurs at the
end of words like [mada] mother, [tiʧa] teacher and [tiela] tailor. Some speakers
use a spelling pronunciation to avoid the [a] ending, and this has been extended
even into words that do potentially end in [a] and have no -er in the orthography
(like Jamaica).7 This kind of spelling pronunciation is a consequence of avoid-
ing a stigmatized item, and in that respect is different from the phenomenon of
6 According to Hernández-Campoy & Jiménez-Cano (2003) the implementation of the standard
variety usually follows the same route. As such, the greater the frequency of standard forms
in informal/familiar styles, the greater the degree of standardisation.
7 However, the kind of hypercorrection being discussed here is attested to elsewhere in the
English language community. Mugglestone (1995: 100) cites texts from the late 19th century
which have this sort of complaint: In a young author’s first volume I found “Italy” made to
47
speakers in some (principally L2 ) communities saying for example [dɛbt] debt

and [θɪŋg] thing (Trudgill & Hannah 1994: 123), which does not typically appear
in Jamaican English varieties. Secondly, retroflexion of an -er ending is an aspect
of rhotic varieties of English – for example General American (Giegerich 1992:
64f) – which JE is said to be (see Table 2.10). The [ʌr] variant may reflect either
this general rhoticity or the adoptive JE spelling pronunciation or both. No one
in my sample produced forms like [ʤʌmekʌr] Jamaica, which suggests that the
retroflexion for the JAMPRO informants is spelling pronunciation rather than
the targeting of a generalized rhoticity.
Here I am looking at the production of lexical items that do have a final spelled
-er in the written form (illustrated by the word butter). See Table 2.10 for what
the informants in my sample produced.
Table 2.10: Final articulation of -er words in the total sample of infor-
mants
butt[a] butt[ʌ] butt[ʌ r]

237 (18%) 958 (72.9%) 118 (9.1%)
Very few occurrences of the [ʌr] variant are produced in these informants. In-
terestingly, this feature (and the one in the following discussion) is one aspect
of a register that has been called “teacher English” by some Jamaicans (see also
Christie 2003: 19). The label carries with it a number of associations: overly care-
ful speech due to linguistic insecurity, an unvarying classroom formality, and
femaleness.
In the sample, there were six informants with a high frequency of [ʌr] in their
speech, but only two speakers who used this variant more frequently than any
of the other variants (M15, F52).
2.4.2 The vowel in the final syllable -tion

In words like education four possible variants occurred in the sample for the
final syllable – [ʃan] ~ [ʃʌn] ~ [ʃɔn] ~ [ʃǝn]. Shields-Brodber (1996: 4) mentions
this [ʃɔn] pronunciation in her article on the changing voice of public, formal
rhyme with “bitterly”. Now “Iterly”, in the mouth of a public speaker, would condemn him as
a thorough Cockney (Carpenter 1868).
She concludes that in England, and as suggested by the quote, there was a widespread stan-
dard ideology that “literate speech” meant a rhotic articulation, notwithstanding the reality of
evolving mainstream RP production.
48
discourse in Jamaica. Her impression is that [ʃɔn] is also an aspect of what she
calls “adoptive JE” (Brodber 1989: 46), use of which indexes both the learning of
English and length of stay in the school system. I suggest that [ʃɔn] is an example
of distancing from the form associated with Creole ([ʃan]). Just as the stigma
attached to voiceless TH stopping and non-contrastive /h/ generates production
which is regarded as hypercorrect by some users of JE, the association of [a]
with its “correct” reflex [ɔ] in stressed syllables has thrown up a variant which
exaggerates backness and rounding in syllables that are typically unstressed in
MSE. To avoid saying something like educa[ʃan], or being perceived as someone
who does, some speakers produce educa[ʃɔn] even in unstressed syllables, thus
over-extending the conversion of /a/ to /ɔ/.
The JAMPRO informants produced the data in Table 2.11.
Table 2.11: Articulation of final -tion in the total sample
educa[ʃan] educa[ʃʌ n] educa[ʃɔ n] educa[ʃǝn]

149 (20.9%) 377 (53%) 120 (16.8%) 65 (9.3%)
The [ʃɔn] variant, like the retroflex variant in Table 2.11, was used by a small
subset of speakers (11 informants) with high frequency. However, this group of
11 has only one speaker in common with the group of 6 who produced [ʌr]
(F52). The sociolinguistics of these sub-groupings in the JAMPRO sample will
be explored in the following chapters. Generally, there seems to be much less
focussing around a particular norm for this feature (unlike the data in Table 2.11),
with close to half of the attestations of -tion varying among the possible phonetic
options. What is noticeable, however, is that the [ʃɔn] variant occurs with greater
frequency than the MSE (schwa) variant.
2.4.3 The alveopalatal affricate

In words like creature, culture, soldier and procedure, some informants produce a
palatalized alveolar stop variant and not an affricate, e.g. [kritjʌ], [kʌltjʌ], [sol-
djʌ] and [prǝsidjʌ] procedure.8
I propose it is partially the form of the written word that influences the pro-
duction of [tj] and [dj], which appears to be a fairly widely used pronunciation
for culture type words in my sample. Speakers would seem to be using English
8 This last item is frequently pronounced with syllable initial stress, as in [prɔsɨdjʌ] procedure.
49
Table 2.12: Articulation of culture type words in the total sample
dj ʤ tj ʧ
34 (75.5%) 11 (24.5%) 73 (35.4%) 133 (64.6%)
orthography as a guide to “correct” pronunciation. Words such as teacher or

bleacher, that pattern the phonemic shape of creature in MSE, but with a written
symbol <ch> that suggests affricate pronunciation, never exhibit this variation
between affricate and palatalized stop. However, the spelling pronunciation it-
self may function in another way in these cases. It also distances speakers from
a stigmatized JC feature, the use of an affricate in words such as pleasure [plɛʤa]
(F91, F23), usually [juʤǝli] (F71) and division [dɪvɪʤʌn] (F57). Notably, these JC
pronunciations seem to occur typically with the voiced affricate, and this may
explain the greater frequency of palatalized voiced alveolar stops used by in-
formants in my sample. The phoneme inventory described at the start of this
chapter identifies /ʃ/ in JC but not /ʒ/. Arguably, the stigmatized affrication in JE
items that would have the voiced fricative /ʒ/, has perhaps resulted in the higher
incidence of [dj]. Moreover, two of my informants (F6, F10), on more than one oc-
casion, produced the form [djunjʌ] junior, an extension of /ʤ/ → /dj/ even when
the orthography suggests the former. Interestingly, Devonish (pers. comm.) also
cites [tju:] chew as occurring.
Additionally, it is possible, based on Marshall’s study (1983), that Creole is
associated with the use of affricates generally. He found spirantization, in forms
such as [ʒǝmeka] Jamaica and [ʒɛnǝrʌl] general, to be increasingly a feature of
some formal speech, an avoidance of the affricate even when the orthography
requires it. Another two informants at JAMPRO (F58, F59), produced [wɪt] which
and a third item [ʧa:rd] charge was recorded (F52).
In at least these speakers, I suggest, “good English” is not producing the affri-
cate, notwithstanding its use in MSE. (Note again that in -tion words, the MSE
schwa variant is also not as widely used as [ʃɔn]). For the variable in creature type
words, the Creole form and the MSE form are similar. These informants, in favour-
ing [tj]/[dj], are responding to local patterns of language use and to local notions
of what is Creole and what is English, not to a remote standard of English that is
external to their speech community. Discussions about the “metropolitanization”
50
of speech in social contexts like Jamaica (Trudgill 2002: 70, for example9 ) seem
to disregard the local ideologies of language that must mediate any changes in
language use that speakers are seen to make. Shields (1987) bears repeating,
though some phonological features are shared by mainstream RP and Ja-

maican Creole these are often eliminated from the speech of [English speak-
ing] informants because of their obvious associations with Creole (119–120).
Historically, in MSE, the literature does point to pronunciations such as

[kritjʌ], [kʌltjʌ] and [soldjʌ] as variants of 18th century British pronunciation,
though its status in terms of prestige was unclear by then (Lehmann 1973: 178).
Certainly, in the late 17th century the recommendation was that pastor and pas-
ture, pick’t her and picture should be pronounced the same (Freeborn 1998: 365).
Beal (2002: 146) discusses a number of prescriptions on the matter from a number
of contemporary sources and concludes that, for some:
This pronunciation [as in ne:tju:r nature] would have been beyond reproach
in the late eighteenth century (…) This is almost certainly the style that
Spence would have heard in the rhetorical style used by the clergymen who
seem to have provided his model of “correct” speech.
Moreover, the recommendation from Sheridan’s dictionary that favoured the

affricate was criticized as follows: “…if a foreigner or native be ambitious for
passing for an English gentleman, let him avoid with utmost care, Mr. Sheridan’s
-SH-” (Beal 2002: 147, citing a source from the 1790’s).
This historical data may point to the reputed conservative nature of colonial
Englishes (Fisher 2001: 84, for example), so that the variation in my sample may
represent continuity from the patterns discussed above for 18th century English.
But that explanation would possibly have more traction if this kind of variation
was observed in other such speech communities; I have not been able to find any
reference to it, though my own impression is that it occurs in Trinidad English
as well. Interestingly, the association of the [tj]/[dj] feature with clergymen may
suggest an early model of educated JE, given the historical connection between
education and the church in Jamaica (see Campbell 1996: 262, for example). It
9 Hancock (1994) uses the term to mean “the replacement of creole features by those from the
coexistent metropolitan language” (97) – which is initially confusing because it would be diffi-
cult to argue that Jamaican Creole and any metropolitan variety of English coexist. However,
Hancock’s definition of metropolitan English (“any native variety of non-creole English”, 97)
suggests that his use of the term metropolitanization may be interpreted as the replacement of
Creole features by those from JE.
51
is probable that this pattern in JE is a coincidence of all three influences: an as-

pect of what Coye (1998) calls “orthoepic piracy” – when a spelling form is more
frequently used (181) – reinforced by the avoidance of the stigmatized “Creole”
affricate, and an earlier prestige norm that also disfavoured the affricate.
Nearly all my informants varied the palatalized alveolar stop with the affri-
cate when voiced. 17 informants used only the [dj] variant and 22 the [tj]. Inter-
estingly, informant F52, whose speech patterns are singled out above in the dis-
cussions of [ʃɔn] and [bʌtʌr] as one informant who produced these consistently,
used only the affricate and not the spelling pronunciation that would perhaps
have been predicted by her language use in those cases.
2.4.4 Post-vocalic rhoticity

The seminal work on rhoticity in the Jamaican continuum is Aker’s implicational
scale (Table 2.13).
Table 2.13: Rhoticity in the Linguistic Continuum, adapted from Akers
(1981: 73)
Stage
0 1 2 3 4 5 6
Intervocalic x x x x x x
Word Final x x x x x
Before Coronal liquids/nasals x x x x
Before other Coronals x x x
Before Consonants x x
After Schwa x
Aker’s description idealizes a rhotic acrolect (stage 6), largely an extrapolation

of a system maximally divergent from a non-rhotic basilect (stage 0), and is not
based on the actual production of acrolectal speakers. There is general agreement
that Jamaican Creole varieties are generally non-rhotic pre-consonantally (Beck-
ford‑Wassink 1999a: 184; J. C. Wells 1982c: 577). Non-basilectal varieties show
more variation, as typified by Well’s description:
…in start and north words, sporadic pre-consonantal rhoticity is characteris-

tic of many mesolectal and some acrolectal speakers: thus short, basilectally
[ʃa:t], may be pronounced either [ʃɔ:rt] or, more commonly, [ʃɔ:t].
52
Post-vocalic rhoticity does function as a sociolinguistic marker in Jamaican

speech. Beckford‑Wassink (1999a) found that male informants in her acrolect-
dominant urban sample were less likely to be rhotic than their female counter-
parts. In an earlier study (A. Irvine 1988: 142, see §1.4 in this book), I found factors
like parent’s background and speaker’s level of education to also show correla-
tions with rhoticity. For example, speakers were more likely to be rhotic when
they were highly educated; those whose parents were relatively affluent were
typically non-rhotic, unless they had spent a long time in the school system. My
conclusion then was that education in Jamaica had normalized a rhotic variety
of English. Indeed, r-insertion was observed in two informants, who produced
forms like [wɔrn] one and [ɔrpǝtjun] opportune, which suggested that rhoticity
is perceived to be “correct” JE.
In the JAMPRO sample, speakers were almost categorically rhotic word finally
where the syllable is stressed (as in car or more), and following the vowels in
words like bird and beard. The data for sociolinguistic variation here concerns
rhoticity before coronal consonants and after the vowels /a/ as in party, and /ɔ/
as in forty.10
Table 2.14: Rhoticity in the total sample of informants
party type words forty type words

[partɪ] [pa:tɪ] [fɔrtɪ] [fɔ:tɪ]
114 (44.4%) 143 (55.6%) 210 (66%) 108 (34%)
r-insertion did occur in one informant (F16), who produced the forms [ɔrtɔrɪtɪ]
authority and [masarʤ] massage; and according to F. Miller (pers. comm.) one
of her informants produced the form [mɔrðʌr] mother. In light of the findings in
the earlier study, for some JE speakers at least, such as those with considerable
formal education, “correct” English is rhotic. Generally, however, as in Wells’ de-
scription above, speakers at JAMPRO varied in relation to rhotic and non-rhotic
post-vocalic productions after the low central vowel. However, there is a much
higher incidence of rhotic productions after [ɔ] in the same sample.
10 Historically, loss of rhoticity in this phonetic environment is first attested in 1640, but is more
widely a feature of 18th century British prescriptions (Beal 2002: 164f; J. C. Wells 1982a: 218).
Indeed, according to Beal, such rhoticity became as stigmatized as h-dropping in the 19th cen-
tury. Wells suggests that the pattern of rhoticity in YS varieties reflects the different historical
influences on American English, with the rhotic varieties reflecting earlier contact with norms
from Britain and the non-rhotic varieties later contact with new prestige pronunciations.
53
2.4.5 Word-final clusters

In this study I wish to compare the patterns for two word-final clusters, [st]
and [nt], distinguishing for the latter between its use when only a phonological
cluster (as in government, rent, important) and when it also has morphological
function (as in don’t, can’t, aren’t). I am not including [nt] clusters that have
past reference (as in sent) in this set, as analysis sometimes proved problem-
atic. For example, in the sentence “when I sen’ for the report…” it is not always
clear whether [d] or [t] is the absent consonant in the cluster, particularly in a
speaker’s longer narrative text. I cannot necessarily assume past tense marking
as it is very possible that on some occasions speakers produced an invariant verb
form as is done in Jamaican varieties. I will also analyse separately the extent to
which clusters with past tense reference (as in looked or kept) occur in my sample.
In my data I have collapsed regular bi-morphemic past tenses (passed, believed)
and semi-weak forms (felt, meant) together to furnish enough data for analysis.
The literature would suggest that in varieties of English, the former will be more
consistently marked than the latter (Guy 1980: 5; Neu 1980: 47, for example); but
speakers of educated (standard) varieties are said to generally delete grammatical
clusters less than they do purely phonological clusters. Additionally, a preceding
nasal in a phonological cluster is one environment where it is not clear whether
clusters tend to be promoted or inhibited in varieties of English (the table in
Patrick 1999: 131 summarizes a number of findings in other studies). In contrast,
a preceding sibilant (as in fast or most) is a more favourable context for simpli-
fication. In that respect, it will be interesting to see what occurs in the acrolect
and compare it to findings from other varieties.
I am interested in three specific aspects of cluster production in the acro-
lect: whether it is more frequent than has been found in studies of basilectal/
mesolectal speech, as would be expected if we accept the notion of continuum;
whether it is socially distributed in the acrolect and associated with factors like
education, gender and the like; and whether it is a salient feature in identifying
one as a speaker of “good” JE.
The literature on final clusters in varieties of English is detailed and rich, sug-
gesting that there are two important phonological constraints on the rate of pres-
ence/absence of CC##:
1. the following segment, so that clusters are less likely to be produced when
the following segment is a consonant (Labov 1972: 217);
2. the preceding segment, so that clusters are less likely to be produced after
a sibilant (Santa Ana 1991, cited in Patrick 1999: 131).
54
In addition, the data for (mesolectal) Jamaican also suggests that clusters with
morphological content are least likely to be produced when they function as a
negative -n’t (Patrick 1999: 150). The data in Table 2.15 summarizes Patrick’s find-
ings for mesolectal11 Jamaican; his analysis assumes that mesolectal Jamaican is
a variety of English with underlying CC## that can have morphological content.
Table 2.15: Percentage of (TD)-absence in mesolect (Patrick 1999: 140,

150, 157)
Preceding segment Absence rate

Nasal 74%
Sibilant 85%
Grammatical function
Regular past 56%
Semi-weak 59%
Negative n’t 87%
Patrick’s data on mesolectal Jamaican varieties suggest that absence of clusters

is the general tendency, overall only produced on the surface about a quarter of
the time. Moreover, the cluster is typically absent in words like can’t or don’t,
most often heard as [kjã:] or [duon]. Importantly, the rate of past tense marking
in the mesolect is similar for regular and irregular forms, which Patrick interprets
as substrate Creole influence on the morphology. Akers’s (1977: 130, 140) results
for tense marking in Jamaica, with a 53% absence on regular verbs in the word-
list test, suggest that clusters with morphological content are more likely to show
t/d absence than purely phonological (i.e. mono-morphemic) ones.
Shields (1984) provides data from interviews with Jamaican teachers in formal
contexts (the classroom primarily), and therefore her findings are specific to the
acrolectal pattern for the general treatment of consonant clusters. She notes that,
[i]n SJE (…) it is not uncommon to find [cluster] reduction [before a fol-
lowing vowel] at a very high rate even in very formal contexts. In SJE, the
11 Patrick1999 defines the mesolect as “an intermediate [between the acrolect and basilect] variety
or range of varieties” (16); the mesolect is therefore a necessary variety or set of varieties in a
continuum, but varieties intermediate to an idealized JE and JX do not require a continuum to
exist. Indeed, Meade (2001: 25) argues for the mesolect as a separate mixed system, given that
speakers may have “a narrow linguistic range [of competence] that does not extend to include
the acrolect or basilect”.
55
(…) following vowel only partially rather than totally inhibits simplification
(…) resulting in acceptable variation between SJE last evening & las’ evening,
with [the former] being preferred in most formal contexts (3).
Additionally, speakers are conscious of the need to preserve the clusters and
therefore produce exaggerated forms like [la:sth ] last or [se:nth ] saint or [handǝ ]
hand. Shields explains both the high rates of cluster simplification (45% of 1389
tokens, p. 128) and the exaggerated forms by relating them to a fundamentally
Creole open syllable structure in most of her informants, notwithstanding the
relatively superficial acquisition of English forms (14). Indeed, Devonish (1992: 2)
has argued that perhaps the term TD insertion might more reflect the realities of
the Jamaican situation, suggesting that speakers acquire rules that insert the final
element in the cluster rather than delete (as happens in English) in the production
of an underlying CC##. This would perhaps explain the following hypercorrect
forms that I have heard in JE: 1) the government put a [band] ban on cigarettes;
2) I have a pain in my [mɪdrɪft] midriff ; 3) Chomsky discusses the under[laɪnd]
underlyin’ level of representation; 4) [and] an analysis of population movement;
and 5) the staff are trying to underm[aɪnd] undermine me. Such forms do not oc-
cur in a variety like AAVE (Labov 1972: 217), nor have I seen them reported else-
where. Interestingly, these are all hypercorrections that generate well-formed
English items, as I have rarely heard hypercorrect non-English forms as seen in
the examples from TH stopping or avoiding affricates. One informant, F16, did
produce [ondlɪ] only once in her text, which may suggest the hypercorrect inser-
tion of a stop. If an insertion rule is operating in some speakers, then it seems
to be a phonological rule that is also lexically constrained, typically substituting
already heard/existing well-formed English morphemes like drift and band, and
not merely converting C## → CC##. With the exception of F16’s [ondlɪ] none of
my informants produced any of the forms discussed above.
Notably, hypercorrect insertion of stops word finally was also a feature of some
17th –18th century varieties of British English, particularly in London (Pegge 1814:
57–73).12 He cites the following examples: attackted, sermont, drownded, paragraft
and sinst. Mugglestone (1995: 238) also cites the use of “oust” house as an exam-
ple of how vulgar speakers were depicted in the contemporary literature of that
period.
The data for all speakers shows the following. I have excluded following alve-
olar stops, as in pay rent to, but have not excluded interdental fricatives from the
data as the typical pattern in this sample is production of [θ].
12 Le Page (1960: 11–12) gives London and the ports of London as one source of early Jamaican
settlement, particularly the deportation of prisoners from Newgate.
56
Table 2.16: Phonological clusters in the total sample of informants
Before a following vowel C(C)##V

-nt -nth -n -st -sth -s
166 (68.3%) 34 (14%) 43 (17.7%) 99 (41.5%) 14 (6%) 125 (52.5%)
Before a following consonant C(C)##C

-nt -nth -n -st -sth -s
149 (45.5%) 29 (9%) 149 (45.5%) 44 (11%) 12 (3%) 356 (86%)
The data for -nt clusters in Table 2.16 does not include -n’t negative clusters.
That data is presented separately.
As the literature would predict, the frequency of cluster simplification is much
higher before a following consonant (e.g. χ2 = 48.48, p < .001 for -nt clusters).
Speakers are more likely to produce clusters before a vowel. Also, perhaps not
unexpectedly given his sample of mesolectal speakers, Patrick’s absence rate of
74% for [nt] is higher than the rate of 34% in my sample. However, for [st] clus-
ters, the absence rate in my data is 74% and is much closer to his mesolectal
sample and lower than what Shield’s described in her teachers. One possible ex-
planation for these teachers’ productions is the “teacher English” referred to in
an earlier section of this chapter (§2.4), characterised by its overly careful speech
and recourse to spelling for “correct” pronunciation. For clusters with morpho-
logical function, the data for all speakers is as follows. I will use the label -ed in
this study to refer to the cluster final [t]/[d] produced for the past suffix.
Table 2.17: Morphological clusters in the total sample of informants
Before a following vowel C(C)##V

-n’t -n’ -ed Ø
31 (51%) 30 (49%) 156 (77%) 47 (23%)
Before a following consonant C(C)##C
-n’t -n’ -ed Ø
89 (16%) 480 (84%) 86 (44.5%) 107 (55.5%)
57
Before a following vowel segment, 40 informants produced -ed categorically

and 5 speakers not at all. Before a consonant, the results were strikingly differ-
ent, with only 8 informants producing -ed all the time in their interviews and 17
never doing so. Most informants varied in their production of clusters. Compar-
ison of rates of deletion of bi-morphemic past tense clusters with other studies
suggests that JAMPRO informants are less likely to consistently mark past tense
than speakers of Trinidad English (Winford 1997: 269) or any North American
varieties (Labov 1972: 222; Neu 1980: 43), but are more likely to do so than the
mesolectal speakers in Patrick’s sample. And the data would indicate that phono-
logical constraints on cluster production are important even for tense marking in
JE, a phenomenon noted for a variety of non-standard varieties of English such
as Puerto Rican, Tejano, Appalachian and the like.
For words like can’t, don’t and the like, arguably the general pattern is to sim-
plify the cluster. Indeed, these two specific items, even in the formal speech here,
are frequently produced as [kh ã:(n)] and [dõ(n)]. It is possible that these two
items have become or are becoming lexicalized without [t]. This would explain
the consistent production of non-morphological [nt] before a vowel (82.3%) but
the variation that occurs when it functions as a negative.
2.5 Discussion
The lists below show the rates at which all features are absent/present in the
speech of informants. The number of informants varies according to the presence
of a particular variable in the recorded interview.
The shaded cells in Tables 2.18–2.19 show where a considerable number of
speakers produced a particular variant of a variable, and they suggest an idea of
good Jamaican English that can be positively defined in terms of having certain
features: having (voiceless) interdental fricatives, pronouncing /h/, using [kh ]
before the long vowel, articulating past tense clusters and -nt before a following
word that starts with a vowel. Notably, in spite of the reputed status of the back
vowel [ɔ] and the hypercorrection that has been identified with it, most speakers
show some variation between it and [a].
The distribution of Group B variants is much more diffuse than for certain
Group A variants and speakers cannot be said to consistently use any variant
here except the two aforementioned consonant clusters. This is well illustrated in
the distribution of -tion pronunciations in the sample. At best, the data suggests
that most speakers will vary [ʌ] with some other possible variant. I suggest that
some features seem not to be salient in producing JE – for example [st] clusters
58
2.5 Discussion
Table 2.18: Group A variables and distribution in the sample of infor-

mants
Variant
Feature Informants taped Never use Always use
/h/ 82 11 (13.0%) 31 (38.0%)
Initial [ð] 82 2 (2.0%) 0
Initial [θ] 82 2 (2.0%) 40 (49.0%)
[ɔ] not 82 0 1 (1.0%)
[o] boat 82 0 32 (39.0%)
[e] face 82 1 (1.0%) 27 (33.0%)
[ɔ r] poor 81 0 28 (34.5%)
[ber] beer 82 4 (5.0%) 3 (4.0%)
[kh a] 64 13 (20.0%) 14 (22.0%)
[kh a:] 48 7 (14.5%) 33 (69.0%)
and -n’t clusters. More clearly, JE speech as reflected in these informants is not
having certain JC features, such as an invariant low vowel [a]; for other variables,
and particularly [kja] and [d ~ ð], speakers of JE seem to be characterised as not
having only JC pronunciation.
Most of these areas of variation were identified elsewhere in the English lan-
guage world, and most from the time that English was also being established in
Jamaica. So, for example, the ‘illiterate speaker’ of the 1850’s in Britain was said to
use hypercorrect ‘h’ and to drop ‘h’ (Mugglestone 1995, Ch. 2). And the refined
speaker produced [dj] in soldier words or [kj]. Further, prescriptions on post-
vocalic rhoticity or the distinction between chap and chop words were chang-
ing at the time. Such ideologies would have also arrived and evolved in Jamaica,
given the phonology of Creole varieties occupying the same social space and the
desire for social differentiation. A good illustration of this is the pronunciation
of culture type words. The pattern of use in my sample includes the palatalized
stop variant, notwithstanding the affricate (now) in use in MSE. Though I do not
necessarily attribute this to historical continuity, it is clear that the metropoli-
tan norm has changed. The use in my sample is as much an avoidance of “bad”
English, i.e. JC affricates, even when the feature is an aspect of MSE, as it is a re-
sponse to spelling and therefore a signal of being literate and educated. Clearly,
there is considerable variation for most features in this sample of speakers. As
such, the social distribution of variants must be important for any discussion of
59
Table 2.19: Group B variables and distribution in the sample of infor-

mants
Variant
-er butter 82 35 (43.0%) 0
[ʌ] butter 82 0 7 (8.5%)
[a] butter 82 27 (33.0%) 0
[ʃan] -tion 81 39 (48.0%) 6 (7.0%)
[ʃʌn] -tion 81 6 (7.0%) 7 (9.0%)
[ʃɔn] -tion 81 41 (51.0%) 0
[ʃǝn] -tion 81 43 (53.0%) 0
[tj] culture 65 6 (21.0%) 17 (61.0%)
[dj] soldier 28 13 (20.0%) 14 (22.0%)
[r] party 79 29 (37.0%) 21 (26.5%)
[r] forty 78 8 (10.0%) 29 (37.0%)
-nt## V 71 3 (4.0%) 41 (58.0%)
-nt## C 77 11 (14.0%) 14 (18.0%)
-st## V 71 19 (27.0%) 12 (17.0%)
-st## C 81 47 (58.0%) 2 (2.0%)
n’t## V 33 12 (36.0%) 10 (30.0%)
n’t## C 78 34 (43.5%) 0
-ed before V 71 5 (7.0%) 40 (56.0%)
-ed before C 62 17 (27.0%) 8 (13.0%)
the acrolect, as the association of variants with factors like education, or gen-
der, can suggest one type of normalisation of features that index “good” English
among or for these speakers.
Certainly, as Devonish (2003) points out, “with reference to the Jamaican situ-
ation, in the face of two abstractions, Standard Jamaican English and “basilectal”
Jamaican Creole, the first is accepted (…) by the society at large as real” (164).
But this abstraction is fundamentally defined in terms of written SE, as it is else-
where in the English language community. Data on the spoken norms of this
idealization called SJE, as reflected in use that Allsopp (1996: lvi) calls “consid-
ered natural in formal contexts” can reveal received pronunciations of SJE, in
the sense of (an) institutionally disseminated accent of an influential minority
(Yallop 1999: 31). What is it that determines which of two prospective qualified
60
2.5 Discussion
employees will be selected during an interview? What judgements are being re-
flected when someone is said to be “well spoken” or “having a good command
of English”? The abstraction that is SJE is realized for members of the speech
community in the language actually used in these formal contexts. Recourse to
external norms of speaking for a description do not reflect the reality. In this
sample, use of [h] is widespread and the norm in JE. But, equally, use of [d] there
and [kj] carry is also widespread. The [d], no doubt, would be excluded if most
speakers were asked which of the two, [d] or [ð], is standard. It is not as clear
that norms about the latter feature [kja] are as focussed. Moreover, when others
in the speech community hear successful and/or influential people using some
of these features and not others; when people who use some of these features
are employed and successfully so, then the idea of what is spoken SJE must be
informed by such practice.
61
3 Sociolinguistic variation in Jamaican
English
3.1 Level of education
As noted in Chapter 1, DeCamp’s original characterization of the acrolect was the
speech of “the well-educated urban professional” (1961: 82). And in any reading
of the literature on the Caribbean, education is the single most important social
factor used to locate the speaker of the standard variety of English.1 This is not
peculiar to the Caribbean, however, as educatedness is important for locating the
standard speaker elsewhere. Trudgill (1999: 118) can be used as an example:
…it is the variety associated with the education system in all the English-
speaking countries of the world, and is therefore the variety spoken by those
who are often referred to as “educated people”.
This, of course, implies that the perception of a speaker as “educated” can in-
fluence the way we evaluate their language use as standard, possibly as much as
the structures used in speech (Thakerar et al. 1982).
As stated earlier, the Revised primary curriculum 1999, published by the Min-
istry of Education in Jamaica, conceives of a locally legitimised target variety for
the school system, SJE.
Historically, English has been the one compulsory exam taken in Jamaica, and
the data indicate that, on average, roughly 40% of CXC candidates achieve grades
acceptable for tertiary education (E. Miller 1989: 222). A similar percentage of
students, according to the Ministry, passed in 2000:2
1 Other factors like social class (typically indexed by occupation and income), urban provenance
and, to a lesser extent, race/skin colour have also been used (see the discussion in §1.4 of this
book). However, with the possible exception of race/skin colour, factors like class, residence
and occupation are themselves inextricably linked to the speaker’s level of education.
2 These results are controversial as they reflect the percentage of students allowed to sit the exam,
not students eligible to sit. When students eligible to sit are analysed, then the percentage of
students achieving the appropriate English Language grade is halved. For example, in 2003, by
reason of pre-selection, 16,000 students were not allowed to take the CXC exam in English, an
average exclusion rate of about 50 percent (The Daily Gleaner, November 21, 2003).
3 Sociolinguistic variation in Jamaican English
In English Language, the passes moved from the 41.2 per cent in 1999 to 47.9
this year [2000]. Grades 1, 2 and 3 are the acceptable grades for entry into
tertiary institutions (The Daily Gleaner, September 2, 2000).
The SJE described by the Ministry of Education and Culture (1999: 17) has only
a handful of phonological prescriptions for the teacher/student:
• distinguish between false homophones in JC and SJE e.g. at/hot, an/on,

doze/those
• clarify JC/SJE confusion of words such as file/foil
Interestingly, in light of the discussion in the previous chapter, these pinpoint

for the teacher and the student – phonemic /h/ usage, control of the [ɔ] vowel and
TH stopping. This might suggest some predictable outcomes in any correlation of
language use and education in this study. We might, for example, expect a greater
degree of uniformity in the use of the variants [h], [ɔ] and [θ/ð], focussed by
the prescriptions of the school system. Additionally, the informants with higher
education and more prolonged exposure to the school system might produce
fewer non-standard variants of these variables.
Within JAMPRO itself, education was the most important of the criteria men-
tioned for employment and success at the agency. Of the 104 informants in this
study, 77 (74%) specifically said that having at least a (University) degree was
essential, and they suggested the following reasons why:
• [it gives you] an ability to see the bigger picture and think a little differently
(F50)
• if you’re well educated and everything your social class tends to be middle
class (F56)
• it equips you to have a good command of the English language (F7)
• it gives you an edge (F21)
• it impresses (F46).
One consequence of this agency policy is that JAMPRO staff is highly educated
relative to the wider society (STATIN 1991: 15:22).3
3 The category “post-secondary” refers to those employees who have completed high school
and gone on to do courses, diplomas or certificates at some institution; “tertiary” then refers
specifically to a university degree. The Jamaica data does not add up to 100%, as I have not
included the uneducated (i.e. unschooled) in the table.
64
Table 3.1: JAMPRO and Jamaican educational attainment compared
JAMPRO (All) (%) Jamaica (%)

Primary 4 50
Secondary 10 29.4
Post-Secondary 42 18.6
Tertiary 21 1.3
Postgraduate 23 0.7
Most of the staff has post-secondary as minimum level of education, with close
to half of the total sample being university educated. All but the primary level
informants have therefore achieved Grades 1, 2 or 3 or equivalent in English ex-
aminations at least at the CXC or GCE “O” level standard. Those with a university
degree, will also have completed tertiary level English language courses such as
UWI’s English for Academic Purposes.
Of those informants whose speech was recorded, the distribution is as in Ta-
ble 3.2.4
Table 3.2: JAMPRO educational attainment and informants compared
JAMPRO (All) (%) Jamaica (%)

Primary 4 (3.9%) 4 (4.9%)
Secondary 10 (9.7%) 9 (11.1%)
Post-Secondary 43 (41.7%) 36 (44.4%)
Tertiary 22 (21.4%) 16 (19.8%)
Postgraduate 24 (23.3%) 16 (19.8%)
It should be mentioned that for many members of staff, the importance placed
on “having a degree” was a contentious issue. Two attitudes were observed. The
first was a resentment that JAMPRO placed more stock in the “piece of paper”
than it did on actual performance. A few informants expressed the perception
that the requirements for promotion, in particular, stressed further education
over an ability to do the job. This was a decidedly minority view from 5 in-
formants. A more common feeling was that JAMPRO favoured candidates with
foreign degrees, earned outside of the Caribbean. Twenty-three (23) informants
4 One informant did not give a response and therefore the total for JAMPRO adds up to 103.
65
specifically said that having a foreign degree meant a greater chance of being
hired and better prospects in the company. Of the 46 informants with tertiary
level qualifications, 12 (26%) had degrees from either the USA or the EU and, not
surprisingly, neither they nor senior management shared this perception.
Level of education will of course be an intervening variable in a number of
the other social categories discussed in this study, such as status in the company
and frontline position. In this section, however, I am only interested in any cor-
relations that can be made with education alone, and where it is relevant in sub-
sequent sections it will be discussed there. The educational categories have been
collapsed into “primary”, “secondary+” and “tertiary+”, for in many correlations
tokens were not enough to support a more nuanced stratification. The statistical
analyses that follow exclude primary educated speakers; by inspection their pat-
terns of use are very different and, when included in the chi-squared tests, they
tend to distort the results. Moreover, as there were only four such informants in
the agency, sufficient data could not be collected for many variables.
In the tables below the data for the 4 primary educated informants will be
displayed so that their patterns can be compared to that of the other informants.5
3.1.1 Group A variables and education
Table 3.3: h-drop by education level (p < .001, χ2 = 59.72)
n h-drop h
Primary 4 18 (37%) 31 (63%)
Secondary+ 45 127 (17%) 636 (83%)
Tertiary+ 32 39 (5%) 729 (95%)
There is a clear correlation between frequency of h-drop and level of educa-

tion, with less of the feature used as the level of education rises. University grad-
uates almost never h-drop, though all informants generally produce relatively
5 All correlations were done using a chi-squared test of association. My interest here was not
to weigh the relative influence of several factors on the production of informants or groups of
informants. Indeed, the ultimate purpose of the study is to describe the speech of a particular
subset of the JAMPRO sample, who constitute a specific, select set of employees (see Chapter 4).
For that, and because of the type of data collected (frequencies and not scores), I used a non-
parametric statistical procedure. I accept as statistically significant any association that yields a
probability value of less than 5% (<.05). However, I also comment on data that produces results
close to that value.
66
low frequencies of the feature. Of the four primary informants, one (M40) never
dropped [h] (see Table 4.16 on page 147 on frontline staff). The few attestations
of hypercorrect use, – that is, “incorrectly” adding [h] – were most typical in the
“secondary+” cohort (6 of the 9 informants who produced it), along with one “ter-
tiary+” informant (M101), and two primary educated speakers. Seemingly, there
were no h-less lects in my sample, though one informant (M88) produced [h]
only once in his discourse.
Table 3.4: TH stopping by education level
Word Initial d ð t θ
Primary 111 (92.5%) 9 (7.5%) 13 (59.0%) 9 (41.0%)
Secondary+ 1205 (56.0%) 940 (44.0%) 78 (18.4%) 345 (81.6%)
Tertiary+ 920 (49.0%) 965 (51.0%) 25 (7.0%) 341 (93.0%)
(p < .001, χ2 = 21.86) (p < .001, χ2 = 23.24)
Word Middle d ð t θ
Primary 10 (71.4%) 4 (28.6%) 2 (100.0% ) 0
Secondary+ 36 (23.0%) 120 (77.0%) 13 (18.6%) 57 (81.4%)
Tertiary+ 27 (21.0%) 100 (79.0%) 7 (15.6%) 38 (84.4%)
(p > .70, χ2 = 0.117) (p > .50, χ2 = 0.17)
Word Final d ð t θ
Primary 6 (100.0% ) 0 0 0
Secondary+ 12 (20.0%) 48 (80.0%) 11 (29.0%) 27 (71.0%)
Tertiary+ 15 (27.0%) 40 (73.0%) 8 (18.0%) 36 (82.0%)
(p > .30, χ2 = 0.8) (p > .20, χ2 = 1.27)
Word finally and in middle position there is no statistically significant differ-

ence between “secondary+” and “tertiary+” informants. Both sets of speakers
have similarly low levels of TH stopping, whether voiced or voiceless, though the
data set per speaker is small and generalizations must be cautiously made. Word
initially, the more educated the speaker the lower the incidence of TH stopping,
particularly when voiceless. Only one informant (M14) produced a high propor-
tion of the voiceless stop variant. He alone accounted for 7 of the 25 tokens of [t],
suggesting that for most speakers in this group a more accurate characterization
would be absence of voiceless TH stopping. Importantly, tertiary+ informants
67
who almost never use the voiceless stop variant, have close to half of their pro-
ductions of the voiced variable as stops. While we can say that tertiary+ educated
speakers, who almost never vary [t ~ θ], produce much less of the Creole form,
we cannot say that their speech is characterized by more consistent use of the
PJS form [ð] in this case.
Table 3.5: The low back vowel and variants by education level
nat nʌt nɔt

Primary 23 (46%) 10 (20%) 17 (34%)
Secondary+ 263 (25%) 184 (17%) 610 (58%)
Tertiary+ 96 (12%) 127 (16%) 573 (72%)
In productions of not type words, primary educated informants show a great

deal of variation in their use of the low back vowel variants – with [a] used most
frequently – though it cannot be said that this use is focussed around the Creole
variant as more than half their attestations of the vowel are not the JC variant. At
the other educational levels, what distinguishes the groups is a greater frequency
of use of the [ɔ] as level of education increases, and less use of the [a] (p < .001,
χ2 = 53.5). The more central vowel [ʌ] is produced by all levels of informants at a
similarly low rate.
Table 3.6: Mid-vowels by education level
uo o ie e
Primary 3 (5.6%) 51 (94.4%) 11 (27%) 30 (73%)
Secondary+ 89 (9.0%) 918 (91.0%) 174 (16%) 883 (84%)
Tertiary+ 37 (6.0%) 545 (94.0%) 76 (11%) 623 (89%)
(p < .10, χ2 = 3.08) (p < .001, χ2 = 15.75)
The statistical difference between speakers when it comes to the back diph-
thong is below the critical level. The tertiary educated do not use significantly
fewer diphthongs than the secondary educated; and importantly, the primary
educated informants also follow the general pattern of these speakers and more
consistently use the monophthong. In this formal context all speakers produced
few [uo], though as discussed earlier, in Beckford-Wassink’s study, the pattern in
informal speech is somewhat different. It is use of the front [ie] that distinguishes
68
speakers from the various levels of education, and as with initial voiceless TH
stopping, h-drop and the low vowel [a], the higher the level of education the
lower the incidence of the Creole variant and the more the [e] is used. However,
it appears that not all Creole features are evaluated in the same way. Speakers
generally are more likely to say [fies] than they are to say [guot]; and it is the
more frequently used Creole form that does distinguish educational groups at
JAMPRO. If the [ie] is in wider use (see the previous section) then it may be
more likely that judgements of social place will be linked to that variant, given
a general avoidance of [uo] in the data. This avoidance demonstrates a consen-
sus across all educational groups here that [o] is the JE form. Contestation about
what is JE takes place with [e ~ ie], with education one factor determining fre-
quency of use. It is also just as likely that use of [uo] may index some other social
difference, as in Beckford-Wassink’s (2001: 155) study which showed a link to
gender.
Table 3.7: Pre-rhotic mid-vowels by education level
poor type words beer type words

[ɔ] [uo ] [o] [ie r] [er]
Primary 6 (40%) 5 (33%) 4 (27%) 17 (81%) 4 (19%)
Secondary+ 246 (77%) 47 (15%) 26 (8%) 241 (51%) 234 (49%)
Tertiary+ 192 (70%) 62 (23%) 20 (7%) 222 (40%) 332 (60%)
Higher education was correlated with fewer front diphthongs overall, and in
pre-rhotic environments we find a similar pattern (p < .001, χ2 = 11.71). However,
the incidence of diphthongization is noticeably higher in beer type words, with
educated speakers varying almost freely between the two variants. In poor type
words, educated speakers overwhelmingly select the [ɔr] variant and therefore
have fairly low frequencies of dipthongization (p < .02, χ2 = 5.92). However, those
with tertiary+ education produce more of the diphthong than the secondary edu-
cated. Higher education then does not necessarily disfavour use of the [uo ] vari-
ant, though we can identify which of the three is the JE variant. The secondary
educated, in particular, are more likely than any other group to use the [ɔr] vari-
ant, perhaps approximating more to a prestige pattern that, from the data, seems
to be use of the /ɔ/ rather than /o/. Arguably, the stigma attached to using [uo],
which speakers tend to avoid, as well as the prestige attached to the [ɔ] vowel in
JE, have reinforced the selection of the [ɔr] variant.
69
Table 3.8: Word initial velar stops and variants by education level
kja kh a kja: kh a:
Primary 5 (50%) 5 (50%) 2 (100%) 0
Secondary+ 65 (61%) 41 (39%) 13 (28%) 34 (72%)
Tertiary+ 46 (52%) 43 (48%) 4 (9%) 41 (91%)
(p > .30, χ2 = 1.8) (p < .05, χ2 = 5.36)
The data set for this feature is small, at best 2 to 3 tokens per speaker recorded.
The discussion of these results is done with a full awareness of the problems
of generalizing from such a small sampling. The creation of word-lists to test
informants, however, would not have been useful here to add to the data set.
Word-lists do not reflect the way people normally use or hear spoken language,
even in formal contexts. Moreover, they test what the informant perceives to be,
for example, the standard form when it is isolated and made the focus of attention.
This does not necessarily reveal how the form is used when it is just one of a
number of variables that the speaker uses/varies when producing discourse.
The relatively high incidence of [kja] in the speakers with higher education is
in sharp contrast to the incidence of [kja:], in particular for university graduates.
It is the production of the latter feature that distinguishes the two groups statis-
tically. Indeed, it is possible that the secondary educated pattern, with 61% use
of [kj] before the short [a], may reflect the type of quantitative hypercorrection
typically described for the LMC (Labov 1980: 254). This suggests either that [kja]
is not perceived by speakers as Creole, or that it is a feature that is found across
the continuum, even in acrolectal speech (see §3.4 on age). If so, in a linguistic
culture in which perceptions are binary, then this feature cannot be regarded
as Creole, even though it occurs in JC. When the tertiary educated are analysed
in terms of presence/absence of features, the following obtains. In all cases the
speakers had the opportunity to produce a particular variant.
For the “tertiary+” speakers, [kja:], voiceless TH stopping and h-drop are the
three features that most do not produce in formal speech. To a lesser extent,
diphthongs (except the pre-rhotic [ie]) are also not an aspect of many informants’
texts. But voiced TH stopping, [kja] and [nat] all occur in spoken JE, though the
data above suggest that in tertiary educated speakers the latter is infrequent in
the individual’s speech even as it is a universally present variant.
Notably, the pattern for primary educated speakers is clearly different (with
a caveat about high percentages in a small sample). Only h-dropping and [uo]
70
Table 3.9: Group A variables and distribution in the highly educated

(“Tertiary+” speakers)
Variant
/h/ 32 0 17 (53%)
Initial [ð] 32 0 0
Initial [θ] 32 0 19 (59%)
[ɔ] not 32 0 0
[o] boat 32 0 15 (47%)
[e] face 32 0 15 (47%)
[ɔr] poor 31 0 11 (35%)
[ber] beer 32 0 1 (3%)
[kh a] 27 3 7 (26%)
[kh a:] 20 2 17 (85%)
were absent from the speech of anyone with primary education. Interestingly,
informant F96 produced no [uo] in her recording (for 15 instances of the variable),
but produced [ie] for 6 of 13 instances of that variable (46%).
3.1.2 Group B variables and education
Table 3.10: -er word ending and variants by education level
butt[a] butt[ʌ] butt[ʌr]

Primary 38 (76.0%) 12 (24.0%) 0
Secondary+ 150 (22.0%) 473 (69.0%) 60 (9.0%)
Tertiary+ 48 (8.5%) 463 (81.5%) 57 (10.0%)
The suggestion made earlier in this section, that it is possibly only in the fea-
tures perceived to be along the Creole/English dimension that level of education
is indexed, can be examined in relation to this set of variables. Here, variation
is not only between Creole/English options but also includes alternants of JE as
well, though in many cases the phonetic options themselves can be explained in
terms of avoiding certain Creole forms (notably [a ~ ɔ] and affricates). The data
71
in Table 3.10 show that production of the JE butt[ʌr] form is not what distin-
guishes the tertiary from the secondary educated; rather it is use of less [a] (the
stereotypically Creole variant) and more [ʌ] (p < .001, χ2 = 42.38). This would
seem to confirm the importance of the Creole/English relationship in defining
“good” or standard English in Jamaica, as reflected in the speech of the most ed-
ucated. Retroflexion of the word ending does not distinguish groups with higher
education.
Table 3.11 below further illustrates the effect of Creole avoidance strategies on
speakers’ productions, but also lends support to the earlier comment that MSE
norms are not necessarily the only target of speakers either.
Table 3.11: -tion word ending and variants by education level
educa[ʃan] educa[ʃʌn] educa[ʃɔn] educa[ʃǝn]

Primary 11 (85.0%) 2 (15.0%) 0 0
Secondary+ 102 (30.0%) 171 (50.0%) 39 (11.0%) 31 (9.0%)
Tertiary+ 31 (9.0%) 201 (59.0%) 76 (22.0%) 34 (10.0%)
The frequency of use of the schwa is similar for the two higher education levels.
What accounts for the value of the χ2 is the difference in the use of [ʃan] and [ʃɔn]
(p < .001, χ2 = 52.23). Tertiary educated speakers use less of the Creole variant, and
more of the JE alternant [ʃɔn]. Educatedness, arguably, is more clearly indexed
by use of the locally evolved variant, not the MSE schwa.
Interestingly, if the 7 informants with foreign degrees are isolated for discus-
sion, their patterns of variation are, by inspection, different from the locally ed-
ucated (see Table 3.12).
Table 3.12: Articulation of word endings by place of education

Foreign 3 (4.0%) 48 (68.0%) 11 (15.0%) 9 (13.0%)
Local 28 (10.25%) 153 (56.5%) 65 (24.0%) 25 (9.25%)
Foreign 4 (4.0%) 87 (83.0%) 14 (13.0%)
Local 44 (9.5%) 376 (81.0%) 43 (9.0%)
72
For these variables, the variation evident in the foreign educated informants
is phonetically more akin to MSE. In the first table, there is a marginally greater
frequency of schwa; the locally educated produce more [ʃɔn] and [ʃan] (p > .10,
χ2 = 5.93). And in the second table, “locals” produce fewer retroflex endings, and
more [a], than those with overseas education (p < .10, χ2 = 4.9). Among infor-
mants whose education has been entirely in Jamaica we find use of the Creole
variant and use of the local JE variant(s) to be the main areas of difference be-
tween them. Higher education in Jamaica then cannot necessarily be correlated
with an increased use of the metropolitan norm, but, as the data here would sug-
gest and as would be expected, such norms are more likely in those who have
been exposed to them.
Table 3.13: Articulation of culture type words by education level
dj ʤ tj ʧ
Primary 1 0 2 (40.0%) 3 (60.0%)
Secondary+ 19 (76.0%) 6 (24.0%) 36 (51.0%) 34 (49.0%)
Tertiary+ 14 (70.0%) 6 (30.0%) 33 (26.0%) 93 (74.0%)
(Foreign) 2 (40.0%) 3 (60.0%) 8 (33.0%) 16 (67.0%)
(Local) 12 (80.0%) 3 (20.0%) 25 (24.5%) 77 (75.5%)
The difference between the secondary+ and tertiary+ speakers use of the voice-
less palatalized stop variant is statistically significant (p < .001, χ2 = 12.56). Speak-
ers with secondary+ education typically vary between affricate and palatalized
stop variants, with no evidence that one or the other is favoured here. Tertiary+
educated informants clearly produce more of the affricate, and this whether with
a foreign education or local education (p > .30, χ2 = .77)6 . When the sample is en-
tirely made up of the locally educated, the results show a similar association
between use of the palatalized stop and level of education, with less of the fea-
ture in informants with higher education (p < .001, χ2 = 11.73). The data for the
voiced variable are inadequate for more than a tentative comment on production.
Speakers here generally do not use the affricate, and this at both levels of educa-
tion. Discussion of this feature will focus on the voiceless variable. There seems
to be consensus that JE has [dj] in items like [ɡradjʊɛt] graduate or [ɪndɪvɪdjǝl]
individual, and it is noteworthy that one of the hypercorrect forms in my sample
6 Because the data for the voiced variable is small, I have not run any statistical tests on it. I
include it only so some impressionistic comparison can be made.
73
involves the replacement of [ʤ] with [dj] as in [djunjʌ] junior. The other replaces
[ʤ] with [d] as in [dʌs] just. However, it is with the voiceless variable that social
differences here can be correlated.
Typically, hypercorrection suggests that a speaker perceives X in her vernac-
ular to be Y in the prestige variety. All Xs are converted to Ys, even when not
required. But if the speaker has X in her vernacular and X in the prestige variety,
and X is stigmatized, then Y becomes the option merely because it is not X, and
not necessarily because it has prestige. In the process items are converted from
X to Y, even though X is the PJS form. Moreover, the data here raise an interest-
ing question. Given that some speakers may perceive affricates to be “Creole”,
as demonstrated in the discussion in the previous chapter, what happens when
conversion of the stigmatized item takes place in a context where there is no
focussed prestige reflex? In this study the following variants occurred in words
like culture or structure: [ʃ], [ʧ], [t] and [tj]. So all these pronunciations were
produced, either by a few or by many of the informants – [strʌkʃʌ], [strʌkʧʌ],
[strʌktʌ] and [strʌktjʌ] structure. The variation that is evident for this variable
suggests that not only is there competition among features for selection by speak-
ers (Mufwene 2001: 143), but the preferred variants are either the one used by the
highly educated speaker [ʧ] or the one which indexes education perhaps because
of its clear association with literacy [tj].7
A final comment that can be made concerns this and some of the other fea-
tures discussed so far: the interdental fricative, the mid-vowels and the palatal-
ized velar stop. In all these cases, a pair of related variables has been analysed and
distinguished, for example, by voicing or phonetic environment. What emerges
is that speakers seem to pay more attention to one of the pair, in the sense that
its distribution in the sample is sociolinguistically significant, while the other
tends to be produced with a very similar pattern by speakers. Informants freely
vary [d ~ ð], but voiceless TH stopping distinguishes speakers’ levels of educa-
tion, with those of higher education using more of the fricative. All informants
produced the [o] monophthong, but [ie ~ e] can also be correlated with level of
education. In much the same way, there seems to be some consensus about use
of the voiced palatal stop, but the voiceless stop in variation with the affricate is
again distributed in the sample in relation to speakers’ levels of education.
For both variables, higher education can be correlated with more rhotic artic-
ulation. In addition, the hypercorrect rhoticity identified in the previous chapter,
7 My impression is that a lexical item like nature seems to be more consistently produced with
the affricate than items like structure. The former item is part of the Creole lexicon ([nieʧa]
libido), and therefore the [tj] variant might be more frequent in words that are thought of as
“English”.
74
Table 3.14: Rhoticity by education level
poor type words beer type words

Primary 1 (11.0%) 8 (89.0%) 6 (50.0%) 6 (50.0%)
Secondary+ 57 (39.5%) 87 (60.5%) 105 (73.0%) 39 (27.0%)
Tertiary+ 54 (53.0%) 47 (47.0%) 94 (61.0%) 61 (39.0%)
(p < .05, χ2 = 4.61) (p < .05, χ2 = 5.03)
does suggest that speakers hold the idea that good English is rhotic. However the
pattern for the back [ɔ] is somewhat different from the pattern for [a]. Firstly,
speakers tend to be more rhotic after the former vowel, in particular the sec-
ondary+ educated cohort. Their production is similar to the data for poor words
and [kja], possibly showing the quantitative hypercorrection of the linguistically
insecure who tend to produce “too much of an indexically good thing” (Silver-
stein 2000: 138). However, when the foreign educated are removed from the ter-
tiary+ count, there is a clear association between high levels of locally attained
education and rhoticity after [ɔ] as well (p < .02, χ2 = 5.67).
Table 3.15: Rhoticity after [ɔ] by place of education
[fɔrtɪ] [fɔ:tɪ]
Local 62 (87%) 9 (13%)
Foreign 32 (38%) 52 (62%)
This may also be another example of the effect of the JE vowel [ɔ] on acrolectal
phonology – for I suggest that a rhotic articulation makes clear that the speaker
does have this vowel, in a context where not being rhotic, as in [fa:tɪ ~ fɔ:tɪ] forty,
the speaker may not be as clearly distinguished from the Creole speaker. As such
it is the local ecology, and not an increased exposure to a (rhotic) North American
model of speech (A. Irvine 1994: 67), that explains the pattern of rhoticity in JE.
While there is a general increase in rhotic articulation as education increases,
speakers (in particular the locally educated) more consistently rhoticize [ɔ].
The data for the production of word final consonant clusters are shown in
Table 3.16.
75
Table 3.16: Phonological clusters by educational attainment
C(C)##V:a -nt -nth -n -st -sth -s

Secondary+ 74 (70.0%) 7 (6.5%) 25 (23.5%) 52 (44.0%) 4 (3.5%) 62 (52.5%)
Tertiary+ 89 (67.0%) 26 (19.5%) 18 (13.5%) 46 (40.0%) 10 (9.0%) 59 (51.0%)
(p < .01, χ2 = 10.95) (p > .20, χ2 = 2.8)
C(C)##C: -nt -nth -n -st -sth -s
Primary 5 (62.5%) 0 3 (37.5%) 0 2 (11.0%) 16 (89.0%)
Secondary+ 58 (39.0%) 11 (7.0%) 79 (53.0%) 27 (13.0%) 6 (3.0%) 180 (84.0%)
Tertiary+ 85 (50.0%) 18 (11.0%) 66 (39.0%) 17 (10.0%) 4 (2.5%) 148 (87.5%)
(p < .05, χ2 = 6.6) (p > .70, χ2 = .7)
a No data were available for the primary educated.
The data here reveal that [nt] clusters can be correlated with level of educa-
tion; the same is not the case for [st] clusters, whether before a following vowel
or consonant. The tertiary educated produce more [nt] phonological clusters, and
more exaggerated articulations of these clusters, than do those with secondary
qualifications. We would expect such a result, given the relationship between
speaker’s level of education and language use revealed in the data above and in
other studies. However, for [st] clusters the two groups are very similar, suggest-
ing perhaps that the two clusters are not sociolinguistic equivalents for speakers.
An alternative explanation is also suggested by LaCharité (1996) who argues that
[st] is not to be analysed as a cluster in Jamaican (non-acrolectal) phonology at
all, but as an illicit segment that undergoes repair by having its [−cont.] feature
removed.8
The data show that use of morphological clusters does not really distinguish
groups either, though as education level increases, past tense t/d before a follow-
ing vowel segment is marginally more consistently used (p > .10, χ2 = 2.31), see
Table 3.17.
Most speakers recorded use -n’t before a following consonant infrequently,
and like past marking, rates of presence/absence are sensitive to following seg-
ment. The surface forms of -n’t clusters are similar in my informants, either [n]
or a nasalized vowel like [õ]. What distinguishes speakers is, in some lexical
8 LaCharité’s arguments for the status of [st] in Jamaican Creole are interesting, but [st] does
occur, at least medially, in (newer?) Creole words like [mɛstIko] mexico or [fɛstIval] hush puppy
(YS) or [rasta] Rastafarian (the latter syllabified ras•ta). Her argument suggests that [st] cannot
be syllabified heterosyllabically (5).
76
Table 3.17: Morphological clusters by educational attainment
C(C)##V: -n’t -n’ -ed ∅

Primary 0 1 3 0
Secondary+ 16 (52.0%) 15 (48.0%) 65 (71.0%) 26 (29.0%)
Tertiary+ 15 (52.0%) 14 (48.0%) 87 (80.5%) 21 (19.5%)
C(C)##C: -n’t -n’ -ed ∅
Primary 0 14 1 3
Secondary+ 41 (13.0%) 274 (87.0%) 35 (45.5%) 42 (54.5%)
Tertiary+ 42 (18.0%) 189 (82.0%) 49 (44.5%) 61 (55.5%)
(p = .10, χ2 = 2.7) (p > .80, χ2 = .017)
Table 3.18: Group B variables and distribution in the highly educated
Variant
-er butter 32 10 0
[ʃɔn] -tion 32 10 0
[tj] culture 25 10 6 (24.0%)
[r] party 30 10 9 (30.0%)
[r] forty 30 2 8 (27.0%)
-nt## V 31 1 19 (61.0%)
-nt## C 32 4 5 (16.0%)
-st## V 29 7 2 (17.0%)
-st## C 32 17 1 (7.0%)
n’t## V 15 5 5 (33.0%)
n’t## C 30 11 0
-ed before V 29 1 16 (55.0%)
-ed before C 32 6 4 (12.5%)
77
items, the preceding vowel, i.e. don’t as either [duõ(n)] or [dõ(n)]; or the onset
consonant – can’t as either [kjã:] or [kh ã:].
Most tertiary educated speakers produce [nt] clusters and past marking, at
least before a following vowel. In addition, most speakers do not produce [st] clus-
ters when followed by a consonant segment. For all other Group B variables the
situation is diffuse, with much intra-idiolectal variation. A number of patterns
though are interesting, such as the rhoticity after [a] and [ɔ], with more speakers
being less likely to rhoticize the former; and the general relationship between ar-
ticulation of any type of cluster and the phonological environment in which it oc-
curs. An overall comment that can be made is that there is much more focussing
around certain features stereotyped as Creole; and more focussing around not
using particular variants rather than trying to produce others.
3.2 Gender
There are a number of reasons why the dynamics of gender and language use at
JAMPRO are important areas of study in this book. Firstly, this agency is over-
whelmingly staffed by women. In total, there are 37 men and 116 women em-
ployed in the New Kingston office, with males comprising roughly one in four of
the staff population. Of the 104 informants interviewed for this study 22 are male
and 82 female (85%), and therefore a proportion comparable to the population at
the agency was reasonably maintained. This, of course, in no way reflects the
male:female ratio in the national or regional (KMA) population statistics, which
is 49:51 and 47:53 respectively (STATIN 1991). Nor does it reflect the percentage
of the national population of employed persons who are female, which stands
at 53% (PIOJ 2000: 86). When analysed with reference to the statistics available
from previous studies (D. Gordon 1986; E. Miller 1991), the gender distribution at
JAMPRO becomes less remarkable.
Table 3.19: Sex and white collar employment: 1943–1994 (Male:Female
of 100)
Type of Occupation 1943 1984 JAMPRO (1994)

Senior Management/Professional 96 : 4 68 : 32 43 : 57
Junior Management/Professional 78 : 22 59 : 41 26 : 74
Clerical/Secretarial 28 : 72 25 : 75 1 : 99
78
3.2 Gender
Table 3.19 shows the population by sex and relevant occupation in the island
over a 40-year time span and the distribution in JAMPRO at the time of data
collection. What is apparent is that more women over time, and fewer men, are
being employed in clerical, administrative and managerial positions. Clerical/
secretarial work is now almost exclusively female, while middle and senior man-
agement positions have become increasingly feminized. Moreover, roughly 74%
of graduates from the University of the West Indies (Mona) are female, making
a high level of education stereotypical of women in the wider Jamaican society.
The high proportion of female informants is then not inconsistent with the pro-
portion of women employed in similar occupations in the wider society. More-
over, as was discussed earlier, the interview process of weeding out “unsuitable”
candidates, seems to result in a greater number of female staff hired into JAMPRO.
It may well be that factors like the expectation of better education in women and
the already overwhelmingly female workforce predispose those making the se-
lection to opt for the female candidate.9
When gender and status in the company is looked at, it becomes clear that this
issue is complicated by factors like education, access to mobility and representa-
tion in senior management and other supervisory positions.
Table 3.20: Gender and education in JAMPRO (One informant did not
give a response)
Primary Secondary Post-Secondary Tertiary Postgraduate

Female 2 (2%) 8 (10%) 38 (47%) 17 (21%) 16 (20%)
Male 2 (9%) 2 (9%) 5 (23%) 5 (23%) 8 (36%)
Proportionately, men at JAMPRO generally tend to be more educated (close

to 60% with at least tertiary level qualifications compared to 40% of women).
Men do not, however, hold a proportionately higher number of positions with
authority (see §4.2). It therefore bears repeating that JAMPRO is an essentially
female agency, but that this is not atypical in the Jamaican workplace context,
particularly in the public sector. JAMPRO is, therefore, an agency that is female-
9 However, Cameron’s discussion (2000) of gender and the commodification of language, sug-
gests that in service industries where interaction with a client is crucial, female norms of lan-
guage use tend more and more to be favoured. She attributes this, in part, to the type of work
being done, “emotional labour – the management of feelings” (338); in part, it is a reflection
of a popular adoption of some of the discussions of gender and language in academic circles
(333).
79
dominant. This is not merely a matter of distribution of staff in terms of num-

bers. In its recruiting and promoting practices, indeed in its culture generally,
the agency is not gender neutral.
The effect of gender on language use in a number of societies has been widely
discussed in the sociolinguistic literature. Traditionally, sociolinguists have re-
marked on women’s greater (reported) use of the standard form, explained in
terms of their lower status (Labov 1972: 243; Trudgill 1972) and the pragmatics of
politeness and face (Deuchar 1988). In these studies, conducted in societies that
consider themselves to have clear norms of good English reflected in a widely
used standard, prestige variety and standard variety were assumed, with some
justification, to be synonymous. Indeed, the style shift that occurred in Labov’s
studies of New Yorkers (1972 passim), for a number of phonological variables, il-
lustrated the extent of the generalization of this idea of the standard. This was
an idea which, it was shown, was most clearly held by women. Women, it has
been argued, are more sensitive to the distribution of power in the society and
to the norms that index membership in the groups that have this power.
Subsequent work showed that this sensitivity is not a function of gender per
se, but of sociological factors like mobility aspirations, access to employment and
education or social-psychological issues of identity and group membership. For
example, Nichols (1983) concluded that it is patterns of employment that explain
a greater use of standard forms by young women in Gullah speaking areas of
South Carolina. Young women in the community she studied are more likely to
stay in school longer, and use more standard forms, as they are the group finding
employment in the wider society that requires ability in Standard English. Male
patterns of employment in the same study were typically found to be in more
self-employed, blue-collar jobs, such as trucking. Earlier work by L. Milroy (1980:
184) also showed that more standard use, in both men and women in Belfast,
could be related, in part, to employment outside of communities in which non-
standard forms were normative and the consequent network structures in which
men and women operated.
Walters’ (1996) discussion of English speaking women newly arrived in Tuni-
sia revealed that choice of language use is also closely related to the linguistic
markets in which one is operating and the identities that are indexed by the
use of particular varieties. These women, finding themselves in a sociolinguistic
context that uses French, Standard Arabic and “kitchen” Arabic, are discouraged
from using the latter variety because of its association with lack of education and
sophistication, even as it is the variety of Arabic to which they have most access
and which, arguably, is of most use to them in their daily routine. French and
Standard Arabic more reflect the status positions of their husbands, returning
80
3.2 Gender
university graduates and professionals trained in the US/UK, and membership in

the groups to which they aspire.
In Caribbean continuum situations like Jamaica, the acrolect is the label tradi-
tionally used in most studies to describe the unmarked prestige variety, the local
standard. In brief, command of this variety indicates higher levels of education,
socio-economic class, and the like. In the workplace, where an ability to speak
English is necessary at nearly all levels, the issue is about the precise linguistic
characteristics that constitute for speakers “proper” English. As was suggested in
the previous sections, there are no general pronunciation prescriptions for Stan-
dard English and reference to metropolitan norms cannot shed much light on
local speech patterns.
F. Miller (1987: 112), who has to date done one of the few descriptions of the Ja-
maican acrolect, concludes that women in this social context are not as sensitive
to the prestige pattern as men seem to be. She attributes this to an insecurity in
men who are more careful to project “being educated”, given its lack of associa-
tion with being male, by more frequent use of MSE forms.
F. Miller uses British and American norms as reference points for both what
she calls the standard and the prestige variety in Jamaica (177), and, as noted in
Chapter 1, this is problematic in her analysis. Local norms of English have been
found to be more apparent in the speech of women in the Caribbean (Le Page &
Tabouret-Keller 1985, for St. Lucia; F. Miller 1987, A. Irvine 1994 for Jamaica). Two
examples from Jamaica can be used to illustrate this: a) Use of [kjat]-type words
has been correlated with socio-economic class. They are found with greater fre-
quency in the formal speech of informants labelled UC/UMC (F. Miller 1987, A.
Irvine 1994), and in the speech of women (A. Irvine 1994). b) A rhotic variety of
English, which has been correlated with level of education, is more common in
acrolect(-dominant) women than men (Beckford‑Wassink 1999a: 184; A. Irvine
1994: 72) and there is some evidence to suggest that the local educated norm is
rhotic (see the previous chapter). The choices women make in the Jamaican so-
ciolinguistic context will not therefore be said to be approximating the prestige
pattern, if prestige is synonymous with metropolitan standard. However, if we
reinterpret the acrolect as a legitimate national standard, Jamaican English, with
its own norms, then we can re-examine the analysis of the speech of women and
its relationship to prestige varieties in the Jamaican society.
3.2.1 Group A variables and gender

15 men and 67 women were recorded during interviews, roughly proportionate
to the 1:4 ratio of men to women in the agency.
81
Table 3.21: h-drop by gender
n h-drop h
Male 15 56 (19%) 239 (81%)
Female 67 128 (10%) 1169 (90%)
Men in the sample were much more likely to produce forms such as [ɛvɪ] heavy
than their female counterparts (p < .001, χ2 = 19.64). For this feature, women were
more likely to use the JE form, though the generalized pattern of [h] discussed
for the total sample in the previous chapter does suggest that hypercorrect use
of [h], i.e. insertion of [h], is the more socially diagnostic variable. Moreover,
most informants regardless of speaker sex used [h] most of the time. Of the nine
informants who did show a hypercorrect pattern of [h] use two were men, a
distribution which does not suggest a particular tendency to this in either men
or women in the sample here.
There is no statistically significant difference between men and women when
it comes to word final TH stopping. Both groups of speakers generally use more
of the interdental fricative variants. For word middle [t ~ θ] there is no statis-
tically significant difference between men and women either, with both groups
favouring the fricative. Where the groups differ is in their frequencies of, primar-
ily, word initial TH stopping, with women more likely than men to produce the
fricative variants. For women, greater use of the fricative is clearly more consis-
tent with the word initial voiceless TH. Even if we place less importance on initial
[d ~ ð] for the reasons already discussed (see §2.3.3), women’s greater use of [ð]
in word middle position does support the generalization made about women’s
greater use of standard/prestige forms.
Table 3.23 shows that both men and women have similar patterns of [a] pro-
duction, with both having a low incidence of the Creole variant. Where the two
groups differ is that women are more likely to use the [ɔ] variant than men, who
use greater frequencies of [ʌ] (p < .001, χ2 = 15.23). Four informants, 2 males and
2 females, could be described as using hypercorrect [ɔ].
M47 – [rIlɔks] relax and [strɔp] strap, F53 –[fɔ:ma] farmer and [sɔlǝrɪ] salary,
F11 – [strɔktʌ] structure, and M89 [ʤɔs] just. Again, it cannot be said that gender
correlates with this hypercorrect pattern. Arguably, if we take hypercorrect use
as an indication of what speakers perceive to be standard/prestigious, then here
again women were more likely than men to use the standard [ɔ] variant. Greater
use of the standard form, however, must be clearly distinguished from avoidance
82
3.2 Gender
Table 3.22: TH stopping by gender
Male 510 (66.0%) 259 (34.0%) 41 (30.6%) 93 (69.4%)
Female 1780 (52.0%) 1665 (48.0%) 76 (11.0%) 600 (89.0%)
(p < .001, χ2 = 54.33) (p < .001, χ2 = 33.96)
Word Middle d ð t θ
Male 25 (45.0%) 30 (55.0%) 2 (22.0%) 7 (78.0%)
Female 51 (20.0%) 198 (80.0%) 20 (18.5%) 88 (81.5%)
(p < .001, χ2 = 14.96) (p > .70, χ2 = 0.073)
Word Final d ð t θ
Male 7 (37.0%) 12 (63.0%) 4 (31.0%) 9 (69.0%)
Female 26 (25.0%) 77 (75.0%) 16 (23.0%) 54 (77.0%)
(p > .30, χ2 = 0.99) ( p > .80, χ2 = 0.36)
Table 3.23: Low back vowel by gender
nat nʌt nɔt

Male 69 (21.4%) 76 (23.5%) 178 (55.1%)
Female 316 (19.5%) 249 (15.4%) 1052 (65.0%)
Table 3.24: Mid-vowels by gender
uo o ie e
Male 26 (11.0%) 215 (89.0%) 61 (19.5%) 251 (80.5%)
Female 104 (7.0%) 1323 (93.0%) 202 (13.0%) 1315 (87.0%)
(p < .10, χ2 = 3.6) (p < .01, χ2 = 8.2)
83
of the Creole form. For this variable both men and women tend not to use the
Creole variant [a].
Women use fewer of the diphthong variants than men, and it is the front vowel,
in particular, that distinguishes male from female speech. This finding is in keep-
ing with Beckford-Wassink’s data (2001: 153–155) on this feature. Moreover, in
her study of two sets of Jamaican speakers, one from St. Thomas (rural-basilect
dominant) and the other from Kingston (urban-acrolect dominant), she found
that
[a]cross groups, males produced more downgliding variants than females

did in the word list and conversational sessions. It is interesting that males
and females pattern together, regardless of group.
(Beckford‑Wassink 2001: 154)
Greater use of diphthongs was a male feature, notwithstanding the differences

in background between the Kingston informants and those from St. Thomas. In
the previous section on education it was shown that it is use of this front vowel
[e] that distinguished speakers from differing levels of education as well. Use of
the back vowel [o] here is similar between men and women, as it was among
primary, secondary+ and tertiary+ speakers.
This pattern is maintained in the pre-rhotic environment, with men produc-
ing more diphthongs than women. However, for both groups the incidence of
diphthongization is higher than it is elsewhere.
Table 3.25: Pre-rhotic mid-vowels by gender
poor words beer words

[ɔ] [uo ] [o] [ie r] [er]
Male 66 (57.0%) 37 (32.0%) 13 (11.0%) 100 (54.0%) 84 (46.0%)
Female 387 (77.0%) 79 (16.0%) 37 (7.0%) 385 (44.0%) 495 (56.0%)
Women are more likely to produce monophthong variants of both variables.

The logic of the data on gender so far suggests that female speech is more likely to
reflect standard/prestige norms, here less use of the diphthong. Women are much
more likely to produce the [ɔ], while the men’s speech is more likely to show
some variation with [uo ] (p < .001, χ2 = 20.06). The use of diphthongs is generally
more widespread across speakers of both sexes in beer type words. Again, what
84
3.2 Gender
Table 3.26: Word initial velar stops by gender
kja kh a kja: kh a:
Male 35 (59.0%) 24 (41.0%) 4 (23.5%) 13 (76.5%)
Female 84 (56.0%) 66 (44.0%) 15 (19.0%) 64 (81.0%)
(p > .50, χ2 = 0.18) (p > .50, χ2 = 0.17)
distinguishes male from female speech here is women’s more frequent use of the
[e] than men (p < .01, χ2 = 6.89).
No statistically significant difference was found between men and women and
their use of either of these two variables. Both men and women pattern the same,
with a much lower incidence of [kj] before the long vowel. It is not possible to
say what is the standard or prestige form before the short vowel. Neither can
we say that [kja] is the Creole variant, as suggested by avoidance of a feature;
though arguably, this label is much more plausible for the [kja:] variant.
Table 3.27: Group A variables and distribution in women
Variant
/h/ 67 0 26 (39.0%)
Initial [ð] 67 0 0
Initial [θ] 67 0 36 (54.0%)
[ɔ] not 67 0 0
[o] boat 67 0 23 (34.0%)
[e] face 67 0 22 (33.0%)
[ɔ r] poor 66 1 24 (36.0%)
[ber] beer 67 3 3 (4.0%)
[kh a] 51 12 11 (21.5%)
[kh a:] 39 6 27 (69.0%)
Table 3.27 describes female use of Group A variables. What can be generalized
from the above data is that many of the women interviewed consistently used
voiceless interdental fricatives and [kh a:]. And as with the data for the tertiary+
educated, there are a number of speakers who always used the PJS variant, but
85
many more who show much variation. On the whole, unlike in F. Miller’s data, I
find women to be more likely to use PJS forms than men at JAMPRO. This may
be due to the specific context of working in this agency, with its overwhelming
female majority. Certainly, the comments detailed earlier (in the introduction to
Chapter 2) specifically state that women’s language use tends to make them more
acceptable candidates for JAMPRO employment.
Additionally, it is important to mention that the interviews for this study were
all done by a woman – an interaction context that might favour more non-stan-
dard male speech and/or language use that asserts masculinity through non-stan-
dard forms (Trudgill 1978). Holmes (1997: 38), citing a number of studies done in
the framework of accommodation theory (Giles & Powesland 1975), suggests that
men are less likely than women to converge to their addressee’s style of speech.
Moreover, the pattern for inter-sex and other kinds of talk in Jamaica, even in
situations of asymmetrical status, is for speakers to use prestige forms in the
opening phases of interaction. However, a shift in code to more non-standard or
Creole forms is more likely in male speakers than in women (Shields-Brodber
1998: 200).
3.2.2 Group B variables and gender

In butter type words, both sets of speakers tended to produce the variant [ʌ] with
no retroflexion. However, the patterns for men and women differed (p < .01, χ2 =
11.3). Men had a less focussed selection of variants and were also more likely to
use the [a], typical of Creole varieties, and [ʌr] word finally. The association of
this “overly precise” articulation with women, mentioned in the previous chapter,
is not apparent in this sample.
Table 3.28: Final articulation of -er words by gender

Male 58 (24.0%) 157 (64.0%) 29 (12.0%)
Female 179 (17.0%) 801 (75.0%) 89 (8.0%)
For the other feature, [ʃɔn] in education type words, there was no statistically
significant difference in the speech of men and women at JAMPRO (p > .30,
χ2 = 2.22). Men and women showed similar distributions.
86
3.2 Gender
Table 3.29: Final articulation of -tion words by gender

Male 29 (21.0%) 76 (54.0%) 19 (13.5%) 16 (11.5%)
Female 120 (21.0%) 301 (53.0%) 101 (17.5%) 49 (8.5%)
Arguably, women showed a slight tendency to favour the more back and round
variant [ʃɔn], while men produced more of the schwa.10
For the third variable, the results were as shown in Table 3.30.
Table 3.30: Articulation of culture type words by gender
dj ʤ tj ʧ
Male 18 (69.0%) 8 (31.0%) 14 (38.0%) 36 (62.0%)
Female 16 (84.0%) 3 (16.0%) 59 (28.0%) 97 (72.0%)
The correlation between gender and use of the [tj] shows no statistically signif-
icant difference between men’s and women’s speech (p > .20, χ2 = 1.57). A more
detailed breakdown of informants’ productions presents a different picture of
variant use (p < .001, χ2 = 15.83); see Table 3.31.
Table 3.31: Articulation of culture type words in informants who vary
n tj ʧ
Male 7 10 (24.0%) 31
Female 22 50 (62.5%) 30
Among speakers who tended to use both variants in their recorded speech,
women are more likely than men to use palatalized stops more frequently. Recall
that use of palatalized stops was more frequent in the secondary+ educated in-
formant, a class of informants overwhelmingly made up of women. It is perhaps
this association with education that may explain the pattern of variant use here.
However, among speakers who were categorical in their use of either affricates
10 Le Page & Tabouret-Keller (1985: 145) describe a pattern of use for boys in St. Lucia, with fewer
of the local forms studied occurring in their formal speech sample.
87
or palatalized stops, more women can be classified as [ʧ] speakers – 21 of 30

women (70%) as opposed to 2 of 6 men (33%). I can therefore make no conclusive
statement about this feature and its relationship to gender, except that the coin-
cidence of lower levels of education and female speakers in this sample may be
apparent in the greater use of palatalized stops in female informants.
For nearly all features analysed so far in this section on gender, where variation
is more clearly along a Creole-English dimension (Group A), women have been
more likely than men to produce PJS variants of variables, to use the forms also
associated with the highly educated. In that respect, this would be in line with
traditional sociolinguistic descriptions of women’s speech and its relationship to
prestige forms, at least in this more formal workplace context. However these
three Group B variables also have variants that are not associated with Creole,
but represent JE phonetic variation. One could simply argue, of course, that the
women’s patterns here also reflect what is to be considered “proper” Jamaican
English, i.e. non-retroflexion on butter type words, perhaps [ʃɔn], both of which
are also associated with higher levels of education. But the data on affricates show
a difference between female speech in general and that of the highly educated.
Among women who vary in their use of the [tj ~ ʧ], the tendency is to produce
the [tj] variant, typically more than men do.
When education level and [tj] use are examined in more detail the data in
Table 3.32 emerge.
Table 3.32: Speakers with categorical articulation of culture type words
by education
Categorical use of
[tj] [ʧ]
Secondary+ 15 13
Tertiary+ 6 14
Roughly equivalent numbers of the secondary+ use affricates only and palatal-
ized stops only. Given that I have a sample of 45 secondary+ informants, a similar
number vary between [tj ~ ʧ]. Affricate use then is more focussed in, and an as-
pect of, the tertiary+ educated speaker. Among speakers that do vary [tj ~ ʧ],
17 secondary+ and 12 tertiary+ speakers, there is no statistically significant as-
sociation between use of variants and differences in level of education (p > .95,
χ2 =.001). Secondary+ speakers’ patterns are therefore diffuse and show much
variation both inter-idiolectally and intra-idiolectally.
88
3.2 Gender
The results here suggest [tj] use may be more an aspect of female speech, as
29 or 43% of female informants use [tj] either all or more of the time. It is not
that women at JAMPRO proportionately have lower levels of education and this
is why they use more of the palatalized stop variant. Rather it is that proportion-
ately more secondary+ informants are women and this explains the frequencies
found in this educational cohort. Palatalized stops represent a local variant that
coexists with the affricate. Certainly the hypercorrect example, [djunjʌ] junior,
cited earlier does suggest this. Men, on the other hand, show a different pattern
– 9 male informants, or 69% of men, use affricates most of the time.
The results for rhoticity after the vowel are shown in Table 3.33.
Table 3.33: Rhoticity by gender

Male 19 (28.0%) 49 (72.0%) 39 (62.0%) 24 (38.0%)
Female 95 (48.0%) 103 (52.0%) 171 (67.0%) 83 (33.0%)
Beckford-Wassink in her study found women to be more rhotic generally than

men in their productions (1999a: 184). In my data, there was no statistical differ-
ence in the frequencies of variants used after the back round [ɔ] (p > .30, χ2 = 0.6).
Rhoticity after this vowel seems to be a normalised feature of JE, given a gener-
ally rhotic articulation among the highly (locally) educated speakers. However,
I find men are significantly less rhotic after [a] than women (p < .01, χ2 = 8.3).
Women generally, and regardless of level of education, tend to be rhotic –
11 of the women are rhotic in all phonetic contexts tested here and 43 of them
are rhotic in one of these contexts, typically after [ɔ]. Men generally are more
likely to be non-rhotic, as only 1 of the male informants is fully rhotic in both
vowel environments (M100) and 11 of them are non-rhotic in one of the contexts,
typically after [a].
The results for gender and phonological cluster production are interesting, in
light of the association of higher education with use of nasal + stop but not with
[st] clusters. Here, there is little correlation between gender and [nt] (in any envi-
ronment), but production of final [st] does distinguish male from female speech.
Note that the frequencies and proportions of [nt] clusters are very similar for
the educated and for both men and women. For [st] clusters, men produce very
few, regardless of following segment. Women, however, pattern the tertiary+ ed-
89
Table 3.34: Phonological clusters by gender
C(C)##V: -nt -nth -n -st -sth -s

Male 34 (61.0%) 9 (16.0%) 13 (23.0%) 7 (16.0%) 5 (11.0%) 33 (73.0%)
Female 132 (71.0%) 25 (13.0%) 30 (16.0%) 92 (47.5%) 9 (4.5%) 92 (47.5%)
(p > .30, χ2 = 2.06) (p < .001, χ2 = 16.31)
C(C)##C: -nt -nth -n -st -sth -s
Male 32 (40.0%) 10 (12.0%) 39 (48.0%) 3 (5.0%) 0 59 (95.0%)
Female 117 (47.0%) 19 (8.0%) 110 (45.0%) 41 (12.0%) 12 (3.0%) 297 (85.0%)
(p > .20, χ2 = 2.46) (p < .10, χ2 = 4.75) (-st and -sth collapsed)
ucated in their use. It must be remembered that women make up the overwhelm-
ing majority of this sample, and therefore the norms identified for the educated
are essentially going to be educated female norms, with the possible exception
of [tj] use. One could therefore ask whether it is that women approximate more
to standard/prestige forms here or whether women’s language use is more of an
influence on what is considered the standard/prestige form, particularly where
there is a less focussed norm. This is a sample in which men are proportion-
ately better educated than women, at least in terms of formal qualifications. It
is women in this sample, even those with lower levels of education, whose vari-
ant production is more similar to that of the highly educated as a group. And
certainly it is women who are stereotyped in this speech community as “edu-
cated”. Moreover, it is how women present themselves during the interview that
is explicitly preferred by management at JAMPRO.
The data show that production of morphological clusters, specifically past
tense marking, is more a feature of women’s speech. The female pattern also
correlates closely with the rate in those with higher levels of education. Simi-
lar findings are described in Neu (1980: 52). This finding suggests that women at
JAMPRO do tend to produce more standard variants, when they function in the
speech community as such, assuming that more educated speech represents the
standard. The results for -n’t clusters, for both education and gender (and the me-
solectal pattern in Patrick 1999), do not suggest that production of [don’t] don’t,
say, is necessarily salient for the production of spoken JE in Jamaica.
A significant number of women are rhotic after the back vowel and produce
both phonological and past tense clusters. It is women, at the interview stage
of applying for JAMPRO employment, that management reported “seemed to do
better”. And when the patterns of the tertiary+ woman are analysed, the features
90
3.2 Gender
Table 3.35: Morphological clusters by gender
C(C)##V: -n’t -n’ -ed ∅

Male 6 (54.5%) 5 (45.5%) 25 (66.0%) 13 (34.0%)
Female 25 (50.0%) 25 (50.0%) 131 (79.0%) 34 (21.0%)
(p > .70, χ2 = .07) (p < .10, χ2 = 3.32)
C(C)##C: -n’t -n’ -ed ∅
Male 9 (12.0%) 65 (88.0%) 11 (29.0%) 27 (71.0%)
Female 80 (16.0%) 415 (84.0%) 75 (48.0%) 80 (52.0%)
(p > .30, χ2 = .73) (p < .05, χ2 = 4.68)
Table 3.36: Group B variables and distribution in women.
Variant
-er butter 67 28 0
[ʃɔn] -tion 66 31 0
[tj] culture 52 21 9 (17.0%)
[r] party 64 20 17 (26.5%)
[r] forty 65 5 26 (40.0%)
-nt## V 58 3 36 (62.0%)
-nt## C 63 11 13 (21.0%)
-st## V 58 14 11 (19.0%)
-st## C 67 35 2 (3.0%)
n’t## V 24 7 6 (25.0%)
n’t## C 64 25 0
-ed before V 60 3 35 (58.0%)
-ed before C 51 13 5 (10.0%)
91
shown in Table 3.37 are consistently used by most. For the purposes of compari-
son, relevant tertiary+ male patterns are in brackets.
Table 3.37: Selected variables and distribution in tertiary+. For the pur-
poses of comparison, relevant tertiary+ male patterns are in brackets.
Feature Informants taped Always use variant

[h] 25 (7) 56.0% (43.0%)
Voiceless TH 25 (7) 68.0% (0.0%)
[o] in boat 25 (7) 10.0% (71.0%)
[e] in face 25 (7) 48.0% (43.0%)
[ɔr] in poor 24 (7) 42.0% (28.5%)
[kh a:] 15 (5) 87.0% (80.0%)
[nt]##V 24 (7) 67.0% (43.0%)
– ed ##V 22 (7) 64.0% (28.5%)
Tertiary+ educated men are more likely than their female counterparts to have
TH stopping and consonant cluster simplification in their speech here. Addition-
ally, these men are more consistent users of [o] before a following consonant.
Most educated speakers tend to produce [kh a:] as in card, [h] in words like help,
[e] in face and [nt] in words like decent. For all other variables used in this study,
there is less consensus and variation can be correlated with either level of edu-
cation or gender. Specifically, women tend to be rhotic after vowels, notably [ɔ],
and more likely to produce palatalized stops in culture type words. It is in the use
of these two variables, rhoticity and affricates, that we see the clearest examples
of gender differentiation in the sample.
3.3 Parent’s background

Informants were asked to state the occupation of the breadwinner in the house-
hold in which they were raised. In spite of the problems of using occupation as an
index of class and status, it remains a useful tool. It allows, implicitly, some com-
parison of income and status levels in societies and therefore relative standards
of living and access to opportunity (Crompton 1993: 120). A person’s occupation
also indicates their level of education, to some extent, and therefore their expo-
sure to the norms of the school system. An informant whose parent is/was a
teacher is more likely to have been raised in a JE speaking context than an infor-
mant whose parent is/was a peasant farmer or casual labourer. Moreover, that
92
informant would have had more access to the facilities and environment that
would support success in the school system, such as equipment, physical space
and parental assistance.
As an example, Table 3.38 shows patterns of final level of attained education in
two neighbouring communities, separated only, to some extent, by the University
campus (STATIN 1991: 2–150).
Table 3.38: Level of education and residence in two selected communi-
ties
Population Primary Secondary University

Mona Heights 4644 948 (20.0%) 1927 (41.0%) 1062 (23.0%)
August Town 7359 2955 (40.0%) 2875 (39.0%) 71 (1.0%)
The first, Mona Heights, was built in the 1950s as a government housing
scheme for middle income families headed by occupational groups like civil ser-
vants or teachers; August Town is a much older low income community. Bryan
(2000: 43) cites evidence that suggests it was established well before 1890. Not
unexpectedly, residents of Mona Heights have attained higher levels of educa-
tion, even as the same numbers of schools and the university are (geographically)
available to both communities. Undoubtedly this is partially a function of house-
hold income, but also less obvious factors like the ability to assist children with
their education. Occupation of parent, therefore, can suggest the type of access
to education each informant would have had in both their formative years and
at higher levels of the school system.
A number of answers on parental occupation were given by informants, too
many for any successful correlation, and I therefore initially abstracted 5 occu-
pational categories from the raw data:
(1) “Cleaner”: This group included all informants who said their parents had low
paying, unskilled work which enabled them to live at subsistence level
(small farmer, household helper, gardener, manual labourer). These occu-
pations are also typically low status ones, either because of their link to
peasant agriculture or servitude and, of course, poor remuneration (Net-
tleford 1970: 138–140; Stone 1980: 20).
A group of skilled workers who are self-employed but whose in-

(2) “Artisan”:
come and status in the society, while higher than “cleaner”, is still relatively
low. Generally, skills are acquired through apprenticeship rather than in
93
any formal system of education. Included here are such occupations as

carpenter, plumber, mechanic, dressmaker and so on.
(3) “Teacher”: Members of this group (also nurse, secretary, junior civil servant),
while not necessarily very high income earners, are afforded more status.
These occupations are dependent on having access to higher education, as
they all require(d) at least post-secondary schooling, even at a time when
access to schooling was more limited. Historically, for example, the effect
of their education among community members was to make people like
the teacher typically a leader in communities as well. Importantly, many of
the occupations in this group are in the public sector and receive payment
from government. This has meant relatively low pay as well as less control
over the means to increase their income.
(4) “Doctor”: In contrast, this group is made up of self-employed professionals in

the private sector who can choose to increase income in response to the
possibilities in the market, or better paid employees in the public sector. In-
formants placed in this group said their parents were lawyers, accountants,
business managers, senior civil servants. These occupations therefore com-
bine the relatively high status given to group 3 and relatively high income
levels as well.
A fifth category, “Business owner” was identified and excluded from the data. Six
informants gave this as their parent’s job, mentioning specifically occupations
such as contractor, shopkeeper, or developer. In the Jamaican context these labels
cannot be consistently associated with any measure of personal income. Though
some owners of business do earn more than the other groups mentioned above,
many do not. Nor do the occupations reflect a shared level of education or status
either. (See Appendix C for the full list of occupations mentioned and how they
are grouped in this study.) I have therefore not included these 6 informants in
the data presented below.
Parent’s occupation was eventually analysed as two categories, which attempt-
ed to combine likely income and level of education. Category 1, cleaner/artisan, is
made up of parent(s) with relatively low paying jobs that have no or few require-
ments of formal education (46 informants). Category 2, teacher/doctor, is made
up of 28 informants with higher income and well educated parents. By compar-
ing the speech of informants from these differing backgrounds, I am assuming
a comparison of informants more likely to be from a bilingual JE/JC speaking
94
household with those from households more likely to be monolingual JC speak-

ing. The cleaner/artisan group would therefore typically have much less exposure
to JE at home.
Table 3.39: Informant’s level of education and parent’s occupation
Cleaner/Artisan Teacher/Doctor
Primary 4 (100.0% ) 0
Secondary+ 36 (75.0%) 12 (43.0%)
Tertiary+ 15 (36.5%) 26 (93.0%)
Table 3.39 shows the extent to which informant’s background correlates with
their level of education (p < .001, χ2 = 13.2). There is close to a 40% association
between the two social variables, so that a tertiary education is typically more
likely in informants with educated and better-off parents (ϕ = 0.38). This would
predict a certain pattern of variant use among the teacher/doctor group, one very
much akin to the pattern discussed in the section on education, such as lower
frequencies of h-drop or TH stopping.
3.3.1 Group A variables and background

A number of the variables do show the kind of pattern that would be predicted by
the literature and the data in the previous sections. Speakers with parents from
a teacher/doctor background used fewer “Creole” forms in their interviews than
those from the cleaner/artisan background. The results of the correlation for h-
drop (p < .001, χ2 = 24.7) are reflected in an avoidance of TH stopping (p < .001,
χ2 = 47.7), use of the low vowel [ɔ] (p < .001, χ2 = 34.29) and a low incidence of
mid-vowel diphthongs (p < .05, χ2 = 5.78).
Table 3.40: Parent’s occupation and h-drop.
n h-drop h
cleaner/artisan 46 123 (15.5%) 668 (84.5%)
teacher/doctor 28 43 (7.0%) 575 (93.0%)
The results for gender and level of education as well suggest consensus on
which variants are considered “good” English. This discussion will therefore fo-
95
cus on the one result that did not follow this pattern, and requires closer exami-
nation and analysis.
Table 3.41: Parent’s occupation and velar stops
kja kh a kja: kh a:
cleaner/artisan 67 (58.0%) 48 (42.0%) 14 (28.5%) 35 (71.5%)
teacher/doctor 47 (57.0%) 35 (43.0%) 5 (9.0%) 48 (91.0%)
(p.>.50, χ2 = 0.17) (p.< .02, χ2 = 6.08)
I found no correlation between speakers’ background and production of [kj]

before the short vowel. What distinguishes the two groups is use of [kja:], which
is also the result yielded by the correlation with education as well. Informants,
when stratified along a number of social dimensions seem to vary [kh a ~ kja]
but tend to produce [kh a:] most consistently, particularly those groups tradition-
ally said to use standard forms. My analysis so far has shown that along the
Creole/English dimension of Group A variables, level of education, gender and
background of household are indexed by more or less use of Creole forms. This
would suggest that [kja:], which also follows this pattern, can be analysed as an
aspect of this Creole/English paradigm. However, use of the voiceless velar vari-
able here is characterized by [kja ~ kh a] variation, and this may suggest one of
the following possibilities:
a) [kja] is a feature that is generalized throughout the Jamaican continuum

and is therefore not a basilectal/mesolectal feature;11
b) [kja] is reflective of an emerging norm of Jamaican speech, a Creole fea-

ture that is entering JE. This latter explanation has been suggested by Beck-
ford‑Wassink (2001: 31) and F. Miller (1987), who describes “a pattern which
is emerging where educated speakers are expressing themselves by using
linguistic forms which are not part of a [standard] British or American
model” (177). One would assume, therefore, that older educated speakers
may produce lower frequencies of [kja] if it is indeed an emerging JE norm.
This is discussed in the following section.
11 Mugglestone (1995) in fact mentions the “delicate palatal glide” in words like [kjaind] kind as
an aspect of proper speech for women (206). [kja] may well have been an aspect of acrolectal
speech in the same time period (the 19th century).
96
Patrick (1999: 109) demonstrates that
[i]n all cases, the incidence of the palatal glide is higher in informal speech
than in test [reading passage, word list] situations, supporting the claim
that (KYA) is not a marker of prestigious speech but a vernacular variable.
This is no doubt the case, but in this study I wish to examine a somewhat dif-
ferent aspect of language use than Patrick’s study. Firstly, when the informant
is cued to a particular feature under investigation, as in word lists, we do have
some access to the ideal that speakers believe ought to be used in correct speech.
However, J. Milroy & L. Milroy (1985: 19) suggest that informants may interpret
such tasks as tests of their knowledge of the correct pronunciation. I imagine, for
example, that all informants here would report that correct JE has word initial
interdental fricatives and not alveolar stops. However, in actual recordings of for-
mal speech the voiced and voiceless fricatives are not sociolinguistic equivalents
-saying [derfɔr] therefore appears to be much more acceptable than saying ‘ting’
thing, at least in Jamaica.12 I wish to make a distinction between the speaker’s
reported ideal of SJE, whether explicitly stated or implicit in test results, and
the idea of SJE that members of the speech community can abstract from spo-
ken “model” JE – “the speech patterns of the teachers, religious leaders, media
personnel and other high status groups in the society” (Brodber 1989: 47).
Secondly, word lists and reading passages do not necessarily reflect the way
language is used in even formal interaction. When a stranger comes into an office
and asks for help, when a teacher is explaining a subject, or when an address
from Parliament is broadcast, these are formal contexts of language use which
other members in a community are more likely to hear than, say, the reading
of a list of words. The formal mental construct of idealized Jamaican English
is perhaps revealed by the latter. But it is from the former situations, and who
is speaking, in their social context that members of the community form their
ideologies of acceptable language use and which sets of variants may be selected
in certain stylistic and social contexts. And it is in one such context, unplanned
formal interaction, that the data here are collected. Speakers in this sample, with
differing levels of education, gender and background do produce [kja]; but [kja:]
is much less frequent in the sample, which suggests that only the latter feature is
not perceived to be permissible in SJE. Arguably, use of [kja] is an acceptable form
in spoken SJE, heard in the actual formal language use of educated Jamaicans
12 The impression I have been given from colleagues from other parts of the West Indies (Guyana,
Trinidad) is that Jamaicans focus on this voiceless TH stopping as an aspect of bad English more
than is done in their communities.
97
likely to be linguistic models. Certainly, at this point in the analysis of speech

produced here, neither variant, [kja ~ kh a], emerges as more typical in the formal
context in which data were collected; nor is either variant more typical in the
speech of those groups said to be more likely to produce standard/prestigious
forms.
3.3.2 Group B variables and background
Table 3.42: Parent’s occupation and butter type words

cleaner/artisan 159 (25.0%) 423 (66.0%) 61 (9.0%)
teacher/doctor 55 (10.0%) 442 (81.0%) 47 (9.0%)
Speakers can be distinguished here again by more, or less, use of the Creole
variant butt[a] and the JE butt[ʌ] (p < .001, χ2 = 44.68). There is no difference in
the groups’ productions of the retroflex ending. The results for -tion words show
a very similar pattern, with the statistical significance largely the result of the
difference between those who produce [ʃan] and [ʃʌn] (p < .001, χ2 = 38.51). For
nearly all other Group B variables there was no statistically significant difference
in variants used. Speakers in this sample, regardless of background, tend to pro-
duce similar frequencies of affricates/palatalized stops and post-vocalic rhoticity.
The presence of word final stops in phonological consonant clusters also pat-
terns similarly in the two groups, though for [st] clusters before a following
vowel those with higher status parents tend to produce more of the final stop
(p < .10, χ2 = 5.4). Interestingly, it is this variable that also distinguished male
and female productions, with women more likely to articulate the final stop.
Table 3.43: Phonological -st clusters by parent’s occupation
C(C)##V: -st -sth -s

cleaner/artisan 42 (36.0%) 6 (5.0%) 70 (59.0%)
teacher/doctor 47 (51.0%) 5 (5.5%) 40 (43.5%)
It is entirely possible that background of speaker can be correlated with gen-

der, and that women tend to come from more affluent households than their male
98
counterparts at JAMPRO. This would go some way to explaining some of the pat-
terns of variant use, either among women in the sample or among informants
from teacher/doctor backgrounds. Women’s general tendency to use more stan-
dard forms of Group A variables, say, might be due to them being more likely to
come from backgrounds with a greater JE presence.
Table 3.44: Parent’s occupation by gender at JAMPRO
Male Female
cleaner/artisan 12 (86%) 34 (57%)
teacher/doctor 2 (14%) 26 (43%)
Table 3.44 shows that females are fairly evenly distributed in terms of back-
ground. Men, however, tend to come from less well-off households, backgrounds
more typically monolingual in JC. Interestingly, when [tj ~ ʧ] (in creature type
words) is looked at, taking into account the possibility of gender as an interven-
ing variable, the results suggest another possibility. In Table 3.45, only female
informants from the two types of households are compared (with relevant male
results in italics for comparison).
Table 3.45: Palatalized stops in cleaner/artisan informants by gender
(with relevant male results in italics for comparison)
[tj] [ʧ]
Female cleaner/artisan 23 (43.0%) 11 (24%) 30 (57.0%) 35 (76.0%)
Female teacher/doctor 32 (30.5%) 73 (69.5%)
There is no statistical difference in production of [tj] between women from dif-

ferent backgrounds (p > .10, χ2 = 2.58). And women, among those from cleaner/
artisan households, vary [tj ~ ʧ] more than men (in italics in Table 3.45). Signifi-
cantly, [tj] is not the variant of the highly educated in this sample, who typically
produce the affricate. This would support the earlier analysis (see Table 3.31) that
this variant, [tj] in culture words, is more a feature of female speech generally.
Before a following vowel, those with parents classified as teacher/doctor are
more likely to produce a morphological word final cluster.
Again the difference between rates of -ed before a vowel and before a conso-
nant, as well as the high levels of cluster presence in the cleaner/artisan group
99
Table 3.46: Morphological clusters by parent’s occupation
C(C)##V: -ed ∅
(p < .05, χ2 = 4.1)
C(C)##C:
(p > .10, χ2 = 1.95)
(74%), suggest that this is a phonological difference, rather than a difference

in the use of English morphology. Generally, both sets of informants tend to
produce the tense inflection when it occurs before a following vowel; and both
vary considerably when the following segment is a consonant. This suggests that
the use of the morphological cluster is very dependent on phonological context,
notwithstanding the difference in production that distinguishes teacher/doctor
from cleaner/artisan. The pattern for -n’t clusters is similar, though not statis-
tically significant, with simplification of the cluster much more likely before a
consonant.
Generally then, informants’ backgrounds can be correlated with more or less
use of Creole features. This is the pattern apparent in Group A correlations. Back-
ground showed no effect, however, on production of post-vocalic [r], palatalized
stops, retroflex word endings or most consonant clusters.
When a more refined distinction of parent’s occupation is compared with the
patterns that distinguish the (locally) educated and women the following picture
emerges. In Table 3.47 I analyse separately the ‘cleaner’, ‘artisan’, ‘teacher’ and
‘doctor’ categories (* indicates statistically significant).
To a great extent all speakers, regardless of background, tend to use most word
final clusters and post [ɔ] rhoticity here very similarly and approximate to the
pattern of the tertiary+ educated. This suggests that education may normalize
a particular use of many phonological variables. One exception to this is the
production of past clusters and [st] clusters (before a vowel), most consistently
produced by speakers from the “teacher” background. The same observation can
be made for use of [tj] and may well reflect the pattern identified for the LMC in
other speech communities (Labov 1972: 126). Interestingly, the female tendency
100
3.4 Speaker age
Table 3.47: Variables and distribution by parent’s occupation
Feature Women Men Tertiary+ Cleaner Artisan Teacher Doctor

Rhotic [a] * * 26.5% 54.0% 48.0% 31.0%
Rhotic [ɔ] * 62.0% 72.0% 61.0% 76.0%
[tj] * * 24.0% 38.5% 53.7% 23.8%
[ʃɔ n] * 7.0% 19.0% 19.0% 20.0%
[kja] 58.0% 58.5% 51.5% 76.0%
[n’t]##V 67.0% 50.0% 55.5% 37.5%
[n’t]##C 11.5% 18.5% 15.0% 11.5%
[st]##V * 36.0% 43.0% 65.0% 38.0%
[st]##C 13.0% 14.0% 16.0% 9.0%
-ed + V 62.0% 78.0% 90.0% 67.0%
-ed + C * 50.0% 39.0% 53.0% 50.0%
to produce palatalized stops, also more apparent in the mid-strata in the sample,
may add support to the possibility of an emerging prestige form. This feature,
and the analysis of the data from gender, education and background, is discussed
in more detail in the final section of this chapter.
All speakers tend to be rhotic after the back vowel, but the pattern for the
groups after [a] is more diffuse in terms of background of speaker. Speakers
from widely divergent household types (“cleaner” and “doctor”) are more typi-
cally non-rhotic after [a] while being rhotic after [ɔ].
Those from the most high status backgrounds use more [kja] and [dõ(n)] than
others in the sample. In this same group we typically find most use of the PSJ vari-
ants. This suggests a number of things. Firstly, these pronunciations are not of
the same sociolinguistic type as h-dropping or voiceless TH stopping and there-
fore do not necessarily have to be avoided when producing JE. Secondly, it may
be that these are features used across the continuum by most speakers in Jamaica,
and are not indexical of speaking either Creole or English. A third possibility is
that these are Creole features, acceptable in JE and used to show an ability to also
speak Creole. These issues are fully explored elsewhere in this chapter (§3.6).
3.4 Speaker age

JAMPRO is an executive agency of the Jamaican government and conforms to
the government norm that retires employees at 65. The oldest age cohort in my
101
sample is therefore 50–65 years of age. The youngest speaker interviewed was
18 (a single informant), but there were no others under 20. The youngest age
cohort in this study is made up of speakers 20–29 years old. There are a number
of reasons why the social and educational influences on these two age groups
can be described as different.
Those informants 50–65 years old when data was collected, would have grown
up and received most of their formal education before political independence
(1962). This would have been in a social context of British colonial rule, when
access to education was limited for most Jamaicans. According to the data pre-
sented by E. Miller (1989: 218), a high school education was the privilege of a few,
with just 6% of primary school leavers advancing to high school in 1960. Prior to
that, between 1945 and 1950, an average of 2.5% of students advanced out of pri-
mary level education. All 4 of the primary educated informants are in the oldest
age cohort; but the majority of informants in that grouping have at least a high
school education and many a tertiary education (7 of the 16).
Informants in this age cohort would also have been educated at a time when,
according to the Kandel Report on the state of British West Indian Education,
“a secondary education [was organized] to serve the purposes of an external
[British] system of examinations” (Williams 1970: 462), and to impart British cul-
ture. Tertiary education would have been had abroad, as is true for 4 of the 7 in-
formants, or had after the local university was established in 1948 with a largely
expatriate staff. The speakers in this age group are more likely to have been ex-
posed to British norms and education than any others in the JAMPRO sample.
The 20–29 year old group would have had their formative (educational) years
in the 1970s and 1980s. Starting in 1973 high school and university were fully
paid for by the government, with the tuition for both tertiary and secondary
school provided for students. Examinations were also made Caribbean with the
introduction of the CXC. The teacher:pupil ratio in the period following rose
from 1:38 in 1978 to 1:55 in the 1980s. The data in Table 3.48 show the results of
high school English examinations and one of the consequences of what E. Miller
calls this “retrenchment” (1989: 214).
Table 3.48: English language passes in Jamaica 1976–1984
Year Number entered (CXC/GCE) % Passing

1976 7,534 62
1980 6,607 41
1984 11,173 44
102
3.4 Speaker age
Additionally, there has been a perception that standards of English education

have declined even further since then. Some of these are outlined in the discus-
sions in Patrick (1999: 60) and Christie (2003). The following from a letter to the
local newspaper typifies the concerns about the standard of English in public use:
…there is abundant evidence of the rise of Yahoolish, that language spoken

by Jamaicans who believe and insist, yet fail to so demonstrate, that when
the time comes they can produce good English. Whatever the destiny of
written English (…) oral English has an uncertain future. The day may well
come when, at a forum such as the UN, or WTO, the language used by our
representatives will be known as Jamaican.
(Chester Burgess, The Daily Gleaner, August 16 2001)
Like Beckford-Wassink and F. Miller suggest in their studies, as discussed in

the previous section, the above quote also implies, albeit as criticism, the exis-
tence of new or emerging norms of good English that depart from MSE usage,
particularly in a social context where the functional distribution of JC and JE is
less defined than before. Moreover, the greater access to higher education after
independence has also meant a greater number of speakers entering an educa-
tional system which, unlike in the 1940s and 1950s, consciously exposed them to
local norms and models of English.
I wish, therefore, to compare use of the phonological variables in these two
age groups at JAMPRO, and identify any features that distinguish the older from
the younger. For the analysis below, the 4 primary educated informants will be
excluded, so that both groups will be made up entirely of those with at least a
secondary education. The older age cohort therefore has 12 and the younger has
29 informants.
3.4.1 Group A variables and age

For a number of the Group A variables there was no statistical difference found
between the younger and the older informants. Speakers in my sample raised and
educated in colonial Jamaica do not use more or fewer Creole forms than those
raised and educated when access to education was improved or, as has been sug-
gested by some, standards lowered. The data also show that older speakers, at
least for these variables, did not produce significantly more of the JE variants
either, even for those items that are stigmatized like voiceless TH stopping, h-
dropping or the low vowel [a] in not words. It cannot therefore be argued in
relation to these variables here that these phonological norms of (formal) lan-
guage are necessarily changing.
103
The exceptions to this pattern are in the use of the diphthongs.

Table 3.49: Pre-consonantal mid-vowels by age
uo o ie e
20–29 49 (7.5%) 596 (92.5%) 139 (20.0%) 566 (80.0%)
50–65 23 (8.0%) 266 (92.0%) 25 (8.0%) 271 (92.0%)
(p > .80, χ2 = .039) (p < .001, χ2 = 19.23)
Young speakers at JAMPRO are much more likely to produce [ie] pre-conson-
antally than older informants, even while use of the back vowel variable remains
similar for both groups, showing little dipthongization. When the data for the
front diphthong are disaggregated across the age groups at JAMPRO, the patterns
in Figure 3.1 emerged for men and women/secondary+ and tertiary+.
Male Secondary+
35.7
31
Male Tertiary+
30 Female Secondary+
30.7
Female Tertiary+
25 26.1
20 19 17.7
14.5
10 8 7.3 9.3
7.1
1.9
0
50–59 40–49 30–39 20–29
Figure 3.1: Front diphthong use by age, gender and level of education
Younger women generally and younger university graduates use noticeably

more [ie] than their counterparts from older age groups. Young less educated
men, in contrast, use more [e] than both older men of a similar education level
and others in their age cohort in this sample. Indeed, among 20–29 year old in-
formants, there is no statistical difference between general male and female use
(p > .50, χ2 = .24). But in the same age cohort, the tertiary educated are much
more likely to diphthongize than the secondary educated (p < .001, χ2 = 11.48),
104
3.4 Speaker age
whether male or female. The data therefore suggest that use of the front dipthong
may be emerging as a more acceptable feature of JE, as reflected in the pattern
for young women and the more highly educated. This pattern is not apparent in
the speech of young, less educated men who seem to produce more consistently
the [e] that the data in previous sections would suggest is more prestigious. One
possible interpretation of the data here is that young men at JAMPRO are more
careful in their production of the prestige monophthong precisely because when
they use the Creole variant it is more likely to have negative associations than
if used by women. Young men in this context here are the group least presumed
to be educated and least favoured in terms of how they present as prospective
employees of JAMPRO.
Among the older age cohorts note that the data follow a more predictable
pattern. For the 30–39 year olds, men use more diphthongs than women and
the secondary educated use more diphthongs than university graduates. This
would be expected, given the results so far for gender and level of education
that correlated fewer Creole forms with women and the highly educated. And
for the 50–65 age group, university graduates and secondary educated women
showed similar use of the variants, with the latter approximating to the prestige
monophthong variant much more than their male counterparts.
Table 3.50: Pre-rhotic mid-vowels by age

[ɔ] [uo ] [o] [ie r] [er]
20–29 179 (80.0%) 31 (14.0%) 13 (6.0%) 173 (49.8%) 175 (50.2%)
50–65 78 (68.0%) 26 (23.0%) 11 (9.0%) 64 (45.0%) 77 (55.0%)
(p < .05, χ2 = 6.53) (p > .30, χ2 = .73)
When the pre-rhotic data are also considered, it is clear that the use of the [ie]
diphthong is, as Alleyne noted some two decades ago (1980: 41), quite widespread,
particularly in younger speakers; and in these same speakers the back diphthong
is less produced in all environments. So in this sample of JE speech, front diph-
thong use tends to be higher in those educated in the 1970s and 1980s than those
educated in the 1940s and 1950s, whereas back diphthong use is much less, par-
ticularly pre-rhotically where the [ɔ] is favoured.
105
3.4.2 Group B variables and age

There was no statistical difference between older and younger speakers for either
butter type words or for the use of palatalized stops in culture type words. Where
there seems to be some difference in the age cohorts is the -tion ending, though
the significance is above the critical level (p < .10, χ2 = 6.5). What difference there
is suggests that younger speakers are more likely to have a schwa in education
type words, while older speakers tend to produce the [ʃɔn] variant. As with the
other variables along the Creole/English dimension, age does not correlate here
with more, or less, use of the Creole form.
Table 3.51: education type words by age
[ʃan] [ʃʌn] [ʃɔn] [ʃǝn]

20–29 52 (22.0%) 118 (50.5%) 35 (15.0%) 29 (12.5%)
50–65 25 (21.5%) 59 (51.0%) 26 (22.0%) 6 (5.0%)
Interestingly, while use of schwa did not index higher education, [ʃɔn] did, and
has been reported elsewhere as more likely in speakers with longer exposure to
schooling (Shields-Brodber 1996: 4). In addition [ʃɔn] is more a feature of the
age cohort educated during the colonial period and presumably more exposed to
British norms. A more detailed picture of the use of the two variants is presented
in Table 3.52.
Table 3.52: education type words by age and education
50–65 20–29
Tertiary+ Secondary+ Tertiary+ Secondary+
educa[ʃɔn] 14 11 25 10
educa[ʃǝn] 4 2 6 23
Older speakers, regardless of level of education, tend to use the variants sim-
ilarly – here more of the [ʃɔn] variant. It is among young speakers that [ʃɔn]
correlates with higher education and therefore a longer stay in the school sys-
tem. The highly educated young are therefore using the prestige norm of older
speakers, and it would therefore be difficult to argue for [ʃɔn] as an emerging
variant in JE. Schwa use is highest in the young secondary educated speaker and,
106
3.4 Speaker age
given the results for other sociolinguistic correlations, seems to be more periph-
eral in the speech community.
Table 3.53: Rhoticity by age

[partɪ] [pa:tɪ] [fɔrtI] [fɔ:tI]
20–29 54 (58.0%) 39 (42.0%) 77 (85.5%) 13 (14.5%)
50–65 13 (41.0%) 19 (59.0%) 38 (63.0%) 22 (37.0%)
(p < .10, χ2 = 2.91) (p < .01, χ2 = 9.92)
Younger speakers are more rhotic than older speakers, and in particular after
the [ɔ] vowel.
Figure 3.2 compares all age cohorts with education in the production of forty
type words.
Male Secondary+
100 100
93 Male Tertiary+
83 81 Female Secondary+
80 73 Female Tertiary+
67
60 65
50 51 54.5
50
40 36
25
20
0
50–59 40–49 30–39 20–29
Figure 3.2: Rhoticity after [ɔ] by age, gender and level of education.
With the exception of the 4 primary educated informants, all speakers in this
sample seem to be over time normalising rhoticity after the back vowel [ɔ], a
pattern which seemed to have been led by less educated females and more ed-
ucated male speakers. If we argue that the production of 50–65 year old female
graduates, who cannot be characterized as either strictly rhotic or non-rhotic,
reflected the instability of the sociolinguistic variable in terms of prestige, then
there has been a conscious adoption of a rhotic prestige norm.
107
When the patterns for the highest educated groups are analysed in terms of
gender, we see a divergence between men and women as it relates to rhoticity
generally. Young educated men are less rhotic after [a], while following the rhotic
pattern of others in the sample after the [ɔ] vowel.
100 rhoticity [a] F

100
87.5 93 rhoticity [ɔ] F
rhoticity [a] M
80 75
71.4 rhoticity [ɔ] M
67 65
60 62.5
50 51
54.5
50 40
40
22.2
20 25
0
50–59 40–49 30–39 20–29
Figure 3.3: Post-vocalic rhoticity after [a] and [ɔ] by age and gender.
There is evidence, in the hypercorrect use of rhoticity, the distancing from non-
rhotic Creole forms and the female pattern of rhoticity, to suggest that rhoticity
is a normalised feature of SJE. The data for young men, who are less rhotic af-
ter [a], suggest two things however. These young men are, as is found in other
speech communities, producing more non-standard forms of the variable. More-
over, as with the results for earlier correlations, rhoticity after [a] is much more
likely to be the site of sociolinguistic differentiation than after [ɔ]. Arguably,
the sociolinguistic significance of having [ɔ] in the idiolect, production of which
may be more obvious in a rhotic phonetic environment, is one constraint on
variation. Alternatively, the [ɔ] variant is much more likely to have developed
focussed norms of use, in this case a rhotic norm. Masculinity may well be pro-
jected by “sounding tough” through the use of non-standard forms, to the extent
that those forms are not also projecting something which carries greater stigma
in the speech community, such as a lack of education or not being able to “speak
properly” in a context of white-collar employment.
The results for consonant cluster use suggest that younger and older speakers
behave similarly. Generally, both age cohorts have higher rates of cluster produc-
tion before vowels; and higher rates of -nt clusters than -st clusters. The rates of
108
3.5 Hypercorrection in the JAMPRO sample
production of clusters with morphological function are also very similar for both
age cohorts in this sample. In all cases, there was no statistically significant dif-
ference between the age groups.

Hypercorrection in speech has been defined in two ways, both relating to an
overproduction of particular forms in certain groups of speakers or in individ-
uals. Labov (1972) described a pattern of variant use in the lower middle class
New Yorker, such that “…the lower-middle-class speakers go beyond the highest-
status groups in their tendency to use the forms considered correct and appro-
priate for formal styles” (126). In effect, their overproduction was in relation to
the linguistic behaviour in other (higher status) classes of speakers and reflected
a more consistent production of the prestige or standard variants of a variable
such as [θ] or post-vocalic rhoticity. This type of quantitative hypercorrection
suggests more than merely which variants speakers in the community consider
prestigious; it also reveals that these speakers hold an idea that certain (desirable)
social characteristics are indexed by use of such forms, though their patterns of
use betray some insecurity about how they do reflect place in the social structure
(Silverstein 2000: 138). Quantitative hypercorrection is also “a synchronic indica-
tor of linguistic change in progress” (Labov 1968: 245), in particular the conscious
adoption and spread of a perceived prestige form as the standard, noted in the
pattern of rhoticity in the JAMPRO sample.
The other type of hypercorrection involves analogical error (Preston 1989: 117).
Speakers aim to produce what they have identified as the prestige reflex of a
form they have in their vernacular and in the process “replace” too many in
the conversion. Trudgill (1986: 66) gives the example of [bʌʧr] butcher, where
speakers of Northern dialects of English have over-applied a rule that converts
Northern [ʊ] (as in [lʊv] love) to [ʌ] for prestigious RP; and Bobda (2001: 277)
cites the example of the form [plaktɪs] practice in an educated Kenyan speaker
who is conscious of the stigma attached to [l ~ r] variation. It is this latter type of
qualitative hypercorrection (Janda & Auger 1992) and aspects of its occurrence
in speakers of JE at JAMPRO that I wish to discuss in this section.
I should first point out that the features listed below are not the only instances
of qualitative hypercorrection in Jamaica. For example, some Jamaican speakers
produce such items as:
1. [stanʤarin] tangerine or [sprɪkl] prickle. JC does not allow [st] clusters
syllable initially, so that the JC cognates of English ‘stone’ or ‘stop’ become
109
[tuon] or [tap]. In ‘tangerine’ therefore, by analogy, speakers insert the

“missing” onset [s].
2. [a:rva] harbour, where [v] is used for cognate English items with either [b]
or [v].
3. [amaʊn] among or [laʊnz] lungs, as [ʌŋ] typically occurs in JC in English

cognates like down or town, the “correct” ending [aʊn] is substituted.
4. [flɪtaz] fritters, [talabrɛd] thoroughbred, showing the [l ~ r] variation found

in, for example, West African varieties of English (Holm 1988: 135). Alleyne
(1980: 62) points out that this type of variation is somewhat archaic in
Jamaica.
These forms did not occur in my sample and are typically identified with more
basilectal varieties of Jamaican speech (Cassidy 1961: 40–47). Possible instances
of hypercorrection in my sample occurred with the following variables:
1. Word initial glottal fricative /h/. In my sample of speakers, eleven infor-

mants produced the hypercorrect pattern, as in [hoke] ‘OK’ (F70) or [hebl]
able F74), but typically with single attestations.
2. The low back stressed vowel /ɔ/. Forms in my sample like [rɪlɔks] relax
(M47), [sɔlǝrɪ] salary (F53) and [bɔ:trum] bathroom (F96) produced by
seven of my informants would suggest the kind of sensitivity to the fea-
ture discussed above and a belief that [ɔ] is the “correct” reflex of [a].
3. The interdental fricatives /θ/ and /ð/. In this data word initial voiced TH
stopping seemed to be a general feature of JE, but the same informants
tended to avoid the voiceless alveolar variant. I have heard hypercorrect
use of the voiceless fricative in items in some Jamaican speech. It was pro-
duced once in the JAMPRO sample (F76 [θʌg] tug). Hypercorrect use of [ð]
did not occur in my sample.
4. The alveopalatal affricate. Two of my informants (F6, F10), on more than

one occasion, produced the form [djunjʌ] junior, a hypercorrect use of the
palatalized stop; additionally, 6 other informants typically produced [dʌs]
for the word just, a possible avoidance of the affricate. It is, however, specif-
ically the hypercorrect use of the palatalized stop that I wish to focus on
in this section.
110
5. Post-vocalic rhoticity. Hypercorrect rhoticity did occur in one informant

(F16), who produced the forms [ɔrθɔrɪtɪ] authority and [masarʤ] massage.
In light of this, and the findings in this and earlier studies, for some JE
speakers “correct” English is rhotic at least when word internally before
a consonant. No pronunciations such as [ʤǝmekʌr] Jamaica, or [kjubʌr]
Cuba occurred in my sample.
6. The mid-front vowel [e]. According to Christie (2003: 19),

[a]nother example of hypercorrection is the pronunciation by some
teachers and radio personalities in particular, of words such as fear
and here with the same vowel sound as in fair and hair, in an un-
conscious effort to avoid Creole pronunciations such as occurs in fies
‘face’.
In my own sample, this pronunciation also occurred, and was found in the
speech of all but 4 of my sample of 82 speakers. In addition, one informant (F57)
produced the item [lez] liaise. This relatively new item in the JE lexicon is bisyl-
labic. This informant might have first converted this to a monosyllabic [liez] and
then “replaced” the diphthong with its monopthong reflex [e].
Table 3.54 presents the number and types of informants who did produce the
above hypercorrect forms.
Table 3.54: Qualitative hypercorrection in the JAMPRO sample
[h] [ɔ] [θ] [dj] [r] [e]

11 7 1 2 1 78 Number of informants (of 82)
2 0 0 1 1 32 Number with tertiary education (of 32)
3 2 0 2 1 28 Number with tertiary educated parents (of 28)
9 6 1 2 1 64 Number of women (of 67)
2 1 0 0 0 14 Number of men (of 15)
A number of comments can be made about the data presented above. The gen-
eral pattern in this sample is for each informant to have one or two hypercorrect
forms and to produce one or two instances of it in their texts. The data on [e] is
clearly different in type from the others, as it seems to be a normalized feature of
these JE speakers, used in variation with [ie] by nearly all informants (Allsopp
1996: xlvi suggests this is common in most varieties of Caribbean English). And
unlike hypercorrect [h] or [ɔ], for example, [e] in beer type words is typical of
111
the speech of the highly educated and of those from likely JE speaking house-
holds. In any case, to describe [fe:r] fear as hypercorrect is to assume an external
(metropolitan) model of English for the Jamaican speech community and to stip-
ulate that a vowel distinction between words such as fear and fair is necessarily
standard.
By inspection, hypercorrect forms occur more frequently in female speakers,
though there is no statistical difference in the gender distribution of [h] inser-
tion. I am unable to make any meaningful comment on this given the dispropor-
tionate number of women interviewed. However, Patrick identifies speaky-spoky,
hypercorrect use of [h] and [ɔ], as perceived to be more associated with female
speakers (1997: 45). If women are more aware of the prestige pattern in speech
communities, and are more likely to promote what becomes the prestige form,
then it is perhaps not unexpected to find more hypercorrection in their speech.
The incidence of [h] insertion and hypercorrect [ɔ] is much higher than for
the other features discussed. In light of Patrick’s analysis, it appears that the
distribution of these two hypercorrect items in the Jamaican population is also
much more general than the other five. The historical record (see §3.2) would
suggest that they have been aspects of the Jamaican speech community from the
earliest days. Indeed, it is possible that their production is only a function of JC-to-
JE conversion diachronically, and they have become aspects of one variety of JE
selected by speakers from the speech community. It is hypercorrect use of these
two items that have emerged as the register that indexes the locally pretentious
speaker.
Three of the informants produced more than one type of hypercorrect form.
Informant F6 used hypercorrect [h], [ɔ] and [dj] (as in the examples [hon] own,
[kɔ:r] car and [djunjʌ] junior). She is young (under 30), secondary educated, and
comes from a household headed by parents with relatively little education who
in Jamaica are called “higglers”, roadside or market traders. The other feature of
note in her speech is a form like [juʤǝlɪ] usually, with the affricate more typical
of Creole varieties and stigmatized in JE. Informant F74 produced both of the
hypercorrect forms described in speaky-spoky, as in the examples [hebl] able and
[mɔ:kIt] market. She is over 50 years old, with a primary education and very likely
to be a vernacular Creole speaker, given her general sociolinguistic profile. For
example, she typically used Creole forms like [dis] just and an almost invariant
low vowel /a/ in not type words in her text. The other informant, F16, produced
hypercorrect rhoticity. She also tends to produce [ʃɔn] in education type words.
Additionally, she does not TH stop, nor does she have ”problems” with [h] or [ɔ].
She holds a Masters degree and comes from a household with educated parents
(her father is a bursar, her mother an accounts clerk).
112
The distribution of hypercorrect forms in the sample suggests therefore that:
• Feature 7 (described as “hypercorrect” [e]) should not be included in this

discussion which takes an endonormative approach to what is to be termed
standard or good English in Jamaica.
• Hypercorrect [h] and [ɔ] are each most common in the sample and, ar-
guably, across the continuum. Indeed, when used together, they are com-
mon enough to be considered a stereotype of the comically “elevated”
speaker in Jamaica.
• Hypercorrect [h], [ɔ], and possibly [θ], are less likely to be found here
among highly educated JE speakers. These are infrequently occurring
items, at least in this sample, and therefore such associations must be im-
pressionistic and very tentative. However, most of the informants who did
produce these forms were not tertiary educated or from households with
educated parents.
• The hypercorrect forms that I find in female speech include more features
and may suggest a greater sensitivity to a wider set of prestige variants
than held by men. The exaggerated forms in this sample are also more an
aspect of female speech. This, of course, would not be unexpected. Greater
approximation to and awareness of prestige/standard forms is a well-doc-
umented aspect of women in many speech communities.
• The hypercorrect pattern on [dj] may be more a function of correcting a

stigmatized form than the targeting of any prestige form for conversion.
This makes this last feature somewhat different from the other instances of
hypercorrection in which there is a focussed prestige target. In this sample, as
discussed in §2.3 of this study, a number of speakers seem to avoid producing
affricates. Possible substitutes are stops [wɪt] which, fricatives [ʒɛnǝrǝl] general
and the palatalized stop mentioned in this section. Certainly the affricate is the
preferred variant of the highly educated in the particular lexical items in ques-
tion; but in this social context, affricates also occur in Creole and can be stig-
matized in JE, as in [trɛʤa] treasure. I argue that this explains the extent of the
variation found in my sample for this feature.
113
3.6 Constructing the acrolect, Standard Jamaican English

The hypercorrection I describe can be taken as one indication of what is perceived
to be standard. In addition, so can the prescriptions of the school curriculum and
the variants selected by the most educated and those from backgrounds that pre-
dict more access to education. The cumulative data suggest therefore the follow-
ing is salient for speaking SJE according to this analysis of certain phonological
variables:
Consonants
a) the word initial glottal fricative /h/ in words where /h/ has a phone-
mic contrast with its absence.
b) the voiceless interdental fricative in phonemic contrast with /t/.
c) the word initial velar stop [kh ] before the long low central vowel [a:].
d) post [ɔ] rhoticity, specifically before [+ coronal] consonants in words
like forty.
e) the word final phonological stop cluster [nt] before a following vowel.
f) the word final morphophonemic cluster when a past tense marker.
Vowels
g) the low back stressed vowel /ɔ/, in words like not and possible.
h) the mid tense vowel [o] as it occurs pre-consonantally, in items like
goat.
i) the vowel [ɔ] in poor type words.
For all other phonological features there was some interidiolectal variation,
certainly correlated with gender and age, or the suggestion of developing norms
of use in a context of such variation.
When all the sociological data are considered for these 82 speakers, it is clear
that social differentiation is typically located in features along the Creole/English
dimension. Group A variables were included in this study precisely because they
have been identified in the literature as having clear English and Creole variants
of variables. In general the sociolinguistic correlations done for those features
reveal that factors such as higher education, a more affluent background and
being female are associated with a greater use of JE variants. Moreover, the re-
sults for some Group B correlations, specifically the low incidence of [a] in butter
114
and education type words, are also an aspect of avoiding the Creole variant and
favouring the JE [ʌ]. Indeed, the rhoticity in much of the sample may, in part, be
due to the coexistence with a non-rhotic JC.
However, the pattern of variation for many of the variables studied here is
more interesting, and less predictable, than merely the avoidance of Creole forms.
Crucially, for many of the pairs of variables described here, the variant distribu-
tion in Table 3.55 can be abstracted from the patterns that tend to occur in actual
spoken JE.
Table 3.55: Asymmetrical pattern of variation on certain phonological
variables
Typically used variant(s)

Variable 1 Variable 2
Interdental fricative [d ~ ð] [θ]
Mid-vowels [ie ~ e] [o]
[ier ~ er] [ɔr]
Palatalized velar stops [kja ~ kh a] [kh a:]
Rhoticity [ar ~ a:] [ɔr]
Speakers in this sample of formal JE, as used by generally well-educated in-

formants, seem to assign asymmetrical sociolinguistic importance to variants of
related variables. One set of variants in their speech, those in the first column, is
free to vary in the production of JE. The other set of variants is not, and speakers
all seem to demonstrate some consensus in use of one – as shown in the second
column. In that respect, the architecture of sociolinguistic variation in this sam-
ple of Jamaicans, distinguishes what I will call load-bearing or constructional
structures, i.e. those necessary for producing JE, from others that speakers show
no imperative to either produce or avoid. I use the term load-bearing here be-
cause in constructing an edifice, certain elements are necessary for and essential
to supporting the structure; others serve to give the structure its character. I pro-
pose that the function of these load-bearing variants is to identify the variety as
JE, and that without their presence in sufficient frequency, the speaker will not
be interpreted by others as producing “good” English.
An illustration of this asymmetrical pattern of production can be seen in Ta-
ble 3.56. With the exception of the pre-consonantal mid vowels /e/ and /o/ which
vary with diphthongs, for all social groups identified in this study there is a clear
contrast between the rate of production of load-bearing variants (highlighted)
115
Table 3.56: Asymmetrical pattern of frequencies for certain phonolog-

ical variables
Education Gender Background Age

Feature Secondary Tertiary Male Female +Affluent −Affluent 20–29 50–65
[θ] ‘thing’ 82.0% 93% 69.5% 89% 94.0% 79.0% 86.0% 82%
[ð] ‘then’ 44.0% 51% 34.0% 48% 40.0% 52.0% 40.0% 47%
[o] ‘boat’ 91.0% 94% 89.0% 93% 92.0% 89.0% 92.5% 92%
[e] ‘face’ 84.0% 89% 80.5% 87% 89.0% 85.0% 80.0% 92%
[ɔr] ‘poor’ 77.0% 70% 57.0% 77% 81.5% 72.0% 80.0% 68%
[er] ‘beer’ 49.0% 60% 46.0% 56% 63.0% 47.0% 50.0% 55%
[kh a:] ‘calf’ 72.0% 91% 76.5% 81% 91.0% 71.5% 73.0% 73%
[kh a] ‘cap’ 39.0% 48% 41.0% 44% 43.0% 42.0% 46.0% 44%
[ɔr] ‘forty’ 73.0% 61% 62.0% 67% 65.0% 69.0% 80.0% 68%
[ar] ‘party’ 39.5% 53% 28.0% 48% 43.0% 44.0% 50.0% 55%
and non load-bearing ones. A further illustration of this difference is to be found

in the productions of ‘can’t’ or ‘don’t’ in this sample. Speakers produced forms
such as [kjã:], [kh ã:], [kh a:nt] or [duõ], [dõ], [dont]. The first in each set is the
one that seldom occurred in the data. For the other two in each set, it is the pres-
ence of the variant [kh ] before the long vowel or [o] that identifies the form as JE,
notwithstanding the presence of the JC nasal vowel or the cluster simplification
evident in the data presented in this chapter. Moreover, I suggest that a speaker
who carefully uses forms like [kh at] cat and [ðat] that all the time, but produces
[kja:t] cart or [tɪn] thin too often will not necessarily be seen as using JE.
In the data collected here, the Creole variants of the voiceless interdental frica-
tive, the mid back vowel before [r] and the velar stop before a long [a:] for ex-
ample are virtually non-existent in the educated speaker’s formal language use.
Indeed, [uo] is seldom used by anyone in the sample, regardless of level of educa-
tion. In contrast, those same speakers use both the Creole and JE variants of the
voiced interdental, the mid front vowel and /ka/. As such, some features in the
Jamaican language continuum starkly distinguish JC use and JE use – keeping
the varieties discrete in a linguistic context where there is no sharp discontinuity
between a functionally distinct JC and JE.
The logic of this proposition would predict a few things. Firstly, qualitative
hypercorrection would be more likely to occur with load-bearing variables than
with variables that are not load-bearing. And indeed, the previous section point-
ed to the production of hypercorrect [θ] but not [ð] for example. Secondly, be-
cause variant use is more focussed and more normalised for these load-bearing
variables, social differentiation is going to be more starkly signalled by use of
116
these variables, indexing, for example, by their absence the characteristics stereo-
typical of the monolingual JC speaker – little education, membership in the lower
class and the like. The mid-vowel /o/ is the exception in this sample, as all groups
in this study consistently avoid use of the diphthong. As such, and thirdly, group
affiliation among JE speakers is therefore typically not going to be signalled by
use of these load-bearing variables because they function to show primarily the
variety one is speaking. Acts of identity with a group, like gender or age, will
more typically occur with the variables in column one – as was seen with male
rhoticity after [a] and younger speaker’s greater use of [ie].
Devonish & Harry (2004: 271) theorize that the relationship between JC and
JE phonology, for most Jamaicans, represents differential convergence, i.e. a type
of linguistic convergence that facilitates speakers shifting between the two vari-
eties while at the same time maintaining the distinctness necessary for the com-
plementary socio-functional distribution of the two varieties. The results here
suggest that the mechanism for this involves the selection of one variable of a
pair of related variables to attach stigma/prestige, while ignoring the other. The
polar lects are kept distinct for speakers through the salient variables, such as
[θ], [ɔ], [h], [kh a:], use of which signals JE and formality for example. This may
well be a pattern of variation that distinguishes models of diglossia involving dis-
crete varieties from models of diglossia involving varieties on a continuum like
that in Jamaica. The distinction discussed by Gair (cited in Paolillo 1997: 272–
273), between literary Sinhala and formal spoken Sinhala for example, points to
differences between a written and a spoken functionally (H) variety. Educated
Jamaicans who write IAE, as with educated Belizeans (Escure 1997: 67–68), pro-
duce a formal spoken variety of English, SJE, that displays distinct socio-phonetic
characteristics from varieties of English in metropolitan speech communities.
If one examines data from, for example, New York (Labov 1966: 253;1972: 100–
104) or Louisiana (Dubois & Horvath 1998: 254) for the interdental fricative, this
asymmetrical attention to variants does not seem to occur. Certainly, the dis-
cussion in Green (2002: 119) does show that in African American English (AAE)
word initial TH stopping occurs principally with the voiced interdental [ð]. AAE
speakers do not really produce forms like ‘ting’ thing or ‘tree’ three, unless they
are speakers of Gullah, but ‘dese’ these and ‘dem’ them are fairly commonplace.
However, the data from all these communities suggest that, typically, style shift
or the use of more formal speech is signalled by an across the board reduction in
TH stopping, not by reducing TH stopping on one variable and ignoring its coun-
terpart. Certainly, in careful speech both voiced and voiceless TH stop variants
seem to be minimized, as these studies do not suggest differences in patterning
for the two related variables (Ervin-Tripp 2001: 51; Labov 2001: 94). Indeed the
117
two are often analysed as one variable in the published data. And Watt (2000:
86–90) shows that in Tyneside English, variants of the mid-vowels /e/ and /o/ are
used with very similar frequencies to each other. Arguably, the speech communi-
ties of New York, Louisiana and Newcastle operate within a different ideological
framework, one in which there is one language – English – and “best speaker-
hood” (Silverstein 1996: 286) is reflected in use of the Standard of that language.
Moreover, members of those communities believe there is one language, albeit
with dialectal variation that can mark particular sets of speakers.
In Jamaica, there are two related language varieties for speakers, Patwa and
English, and best speakerhood is reflected in the bidialectal speaker. Monolingual
JC speakers are viewed negatively, as are monolingual English speakers. Both are
judged to be deficient in this context, the former “backward” and the latter “not
a real Jamaican”. These related but functionally distinct varieties are managed
through structuring variation in such a way that some variables serve to identify
the variety the speaker is using, given the linguistic overlap particularly at the
phonolexical level.
Generally, in my sample, the idea of what is JC seems to be focussed around
certain features – voiceless TH stopping, [kja:], h-drop, the back diphthong. In
large part it is not using these features that defines the acrolect, so that those with
higher education, from more affluent backgrounds, and women’s language are
characterized by very low frequencies or absence of these variants. Additionally,
JE is also having a greater frequency of other variants – most importantly [ɔ],
being rhotic and with t/d presence before following vowels.
A number of variants are in fact used all the time here by speakers, notwith-
standing differences in social group – the voiceless interdental fricative, [kh a:],
[nt] and past tense clusters before a following vowel. In addition, the highly ed-
ucated and/or those from backgrounds with educated parents tend to produce
/h/ and mid-vowel monophthongs. Interestingly, the data from younger speak-
ers suggest that fewer categorically use [h] or [e], but more of them are rhotic,
particularly after the back vowel.
Except for one informant in this sample, all speakers vary [ɔ] to some extent
with [a]; but with the exception of the 4 with primary education, all informants
produce low frequencies of the [a]. I suggest that rhoticity after the back vowel
and the palatalization of the velar stop also serve to display possession of this
vowel, in a context where not being “able” to distinguish [a] from [ɔ] is “one of
the shibboleths of the speech community” (Devonish & Harry 2004: 272). Rhotic
articulation may distinguish the [ɔ] vowel in an item like [fa:ti] ~ [f ɔ:ti] forty
more clearly; and palatalization of the velar only before [a], again reinforces a
distinction between [kat] ~ [kɔt] which may be either cat or cot in some speakers.
118
Moreover, JE variants like [ʃɔn] in education type words are also a function of
avoiding [a].
The chapter to come explores the sociology of JAMPRO as an organization
and its allocation of staff to certain positions. In doing so I wish to explore how
these phonological features are distributed in the speech of JAMPRO staff and, in
particular, the speech of staff who are successful in the agency and get promoted
into positions that reflect the public face of an agency of the Jamaican state.
119
4 Sociolinguistic Variation in JAMPRO
4.1 JAMPRO: One site of promoting a SJE ideology
In this chapter I present information on the 82 speakers, not as a sample of edu-
cated Jamaicans, but as members of a social collective,1 and specifically as mem-
bers of an agency of the Jamaican state, JAMPRO. Following on the discussion
in Chapter 1, in particular on the ideology of language in Jamaica and the norma-
tive pressure exerted by institutions in their expectation and practice, I selected
one government agency and the speech of its employees for analysis. The type of
agency, the requirements for employment, and the patterns that lead to success
for staff, can certainly shed light on the phonology of “good” Jamaican English.
As was mentioned at the start of Chapter 2, all advertisements for employment
at agencies like JAMPRO insist on “a good command of the English language” or
“excellent oral and written skills”. But the study of staff selection and promotion
can also reveal the type of employee that such an agency perceives and selects
as best able to realize its own image and the image(s) that it constructs for and
about Jamaica. I have selected an agency whose function it is to market Jamaica
to both local and overseas clients, with some of its staff specifically selected as
representatives of the agency through the positions they hold at JAMPRO.
Scollon’s comments (1997) are, therefore, a good starting point for this discus-
sion:
If we think of public discourse in this sense as being outputs of bureaucratic

structures, then a crucial point of interest is the boundary across which
the discourse is generated (…) in a business it might be an advertising de-
partment or a public relations department, in the government it might be a
public affairs office. In any case, one might always expect to find some for-
mally constituted institutional structure along with designated members of
the institution who serve to both generate and legitimate the crossing of
the organizational boundary into public discourse (45).
1 Sealey & Carter (2001) use the term to describe a grouping whose membership is indicated by
an awareness of, and a kind of commitment to, the conventions that constitute the group in
the first place (4).
Agencies of the state like JAMPRO are, in part, engaged in creating an idea of
Jamaica and the Jamaican for a client and therefore a consumer of that construct.
What the agency says or implies about a Jamaican must be reflected in the type
of employee that it selects to symbolize its public face and to interact with its
clients. In its role as one of the agencies that legitimate Jamaica to others over-
seas, JAMPRO puts into the arena of public discourse certain ideas, one of which
concerns language. Its own language use, as reflected by the language behaviour
of its staff, therefore becomes important. This language behaviour provides us
with actual examples of how the agency’s language ideologies become realized
in its allocation and selection of staff for certain positions.
A number of Jamaican government agencies provide information about the
language situation in Jamaica, JAMPRO being one of them. I wish to discuss the
description of the local sociolinguistic situation provided by these agencies of
the Jamaican state, and therefore the nature of the official public discourse about
language in Jamaica. Firstly, this discourse serves as an indicator of the linguis-
tic expectations for staff held by these arms of the state. But such a discussion
also brings into focus the ideas held about language in Jamaica and the ways
in which they (may) have changed over time. In the previous chapters, some of
the comments on the language situation found in the main local newspaper were
mentioned. They represent one aspect of public discourse, which Christie (1998a)
discusses in her study of this aspect of the language debate in Jamaica. I wish to
explore the institutionalised aspect of public discourse, and the image of the Ja-
maica(n) that can be abstracted from published official documents. The agencies
to be discussed here are a) the Jamaica Tourist Board (JTB), b) the Jamaica Infor-
mation Service (JIS) and c) JAMPRO.
4.1.1 Jamaican institutional discourses on language

The marketing of Jamaica for tourism is believed to have started in 1851 (Taylor
1971: 62), when a pamphlet, ascribed to someone named Anderson, was published
urging Americans to come to the island for health reasons, particularly in winter.
Tourism then, as now, was geared to the needs of a largely U.S. clientele, as a great
deal of initial travel to Jamaica was prompted by the United Fruit Company’s ties
to Port Antonio through the banana industry. Titchfield Hill was one of the first
hotels, a “lost garden of Eden – the incomparable combination of American com-
fort, English cleanliness and Italian climate” (1903 pamphlet, unnamed author,
cited in Taylor 1971: 99).
The Jamaica Tourist Association was formed in 1910, and, as Taylor remarks,
scant mention was made of Jamaicans themselves, as they were viewed as more
122
of a liability than an asset in the marketing of the island. Even then there was
concern expressed about letting visitors see all aspects of the society. The only
mention of Jamaicans cited by Taylor is in a 1914 pamphlet: “the hospitality of
the people is proverbial … demonstrated by the smiling faces and happy laughter
of the natives” (4). These are, of course, the production of an association formed
during the time of a colonial government. A comparison with post-independence
publications, when the agency morphed into the JTB, projects a similar image.
Jamaicans are described as: “gentle people named Ivy or Maud or Malcolm who
will cook, tend, mend … who will “Mister Peter, please” you all day long” (JTB
1968, cited in Taylor 1993: 174). Taylor analyses this as marketing for a U.S. sen-
sibility, a clientele more accustomed to and comfortable with the idea of “Negro
servants”, and persuading them that Jamaica was safe in spite of its majority black
population. Language is not mentioned at all, perhaps because interaction was
not expected, though the implication here is that English is at least understood.
More recent advertisements do discuss language. A typical example, directed
specifically at North Americans again, is contained in a JTB pamphlet (1992) and
entitled Say It Again, Mon, “mon” being an attempt to represent the Jamaican
vowel that is more backed when compared to its General American counterpart.
I have numbered particular items for comment.
Listen to two Jamaicans talk and you’ll hear a (1) musical mix of English,
patois (a combination of English, African and Welsh (sic)) and island words
and rhythms. You might think it’s (2) impossible to make sense of it all
without being a native. Not so! While it might be tough to wholeheartedly
jump into a rocket fast discussion, (3) it’s easy to learn a few phrases that
might come in handy at the local rum shack or in the marketplace. Here is
a mini-glossary to get you started.
Cool runnings (4) No bother, no fuss, it’s okay. Also used when parting to
mean “goodbye”.
Gimme a chups Kiss me.
Irie (EYE-re) (5) Everything is cool.
Irie dawta A (6a) sexy, good-looking girl. “Wat a irie dawta!”.
Jah God as in “Praise Jah.” From the (7) biblical name for Jehovah; this
phrase is most often used by Rastafarians.
Kiss me neck Not to be taken literally, this is an expression of surprise, in-
credulity or defiance. (8) “Kiss me neck, I can’t believe you got married
last night and didn’t tell anyone you were going to do it”.
123
Soon cum Meaning someone or something will arrive anywhere (9) be-
tween the next five minutes and the next five hours.
Tek yuh time Take your time; a favourite (10) in the craft market where
you’re being invited to browse. Nobody’s going to rush you here!
Ting This is a tricky one. First and foremost its simply (11) slang for “thing.”
However, if someone says, (6b) “Mek we do a ting” to a woman, it’s
considered a proposition. If it is said to a man, (12) the phrase means
“Lets make an arrangement, do business, or have a drink.” To make
matters more (13) confusing, Ting also happens to be a popular grape-
fruit soda.
Walk good A farewell meaning, “Take care, stay safe, hope all goes well with
you”.2
Comments made about the language situation to be found readily acknowl-

edge Creole (Patois), suggesting it is a mix of English and other ingredients, and
do so in ways that apparently justify adjectives like “musical”, “colourful” and
“rhythmic”. In addition, this mix of language varieties is confused and confus-
ing (2, 3 and 13). Similar terms are used to discuss regional patterns of English,
in comparison with the standard varieties of English. These imply “lack of clar-
ity” and “the charm of the humor in confusion” (Silverstein 1996: 292–293). Note,
however, that in spite of the idiosyncrasies and slang items, the presentation sug-
gests that it is clear that Jamaicans speak English; though the article offers the
tourist the local phrases, the representation of actual Jamaican speech is English
(12 & 8). The tourist to Jamaica, for the most part from the U.S., needs only to
learn a few simple phrases in order to communicate with the Jamaican he/she
is likely to meet. In addition, these phrases are most useful in the rum bar or
marketplace (3), which clearly locates the contexts in which Creole is likely to
be heard and to be appropriate.
What is also of interest are the phrases themselves that were selected for the
article. Four of them project an image of a relaxed, happy-go-lucky people (com-
pare the 1914 pamphlet) with little concern for schedules or time, as in 4, 5 and
9, or serious business, as in 10 and 12. The visitor is also subtly informed that the
women are young and attractive (6a and b) and people, however strange look-
ing, are non-threatening and Christian (7 and 10). The latter piece of propaganda,
the term used here in its denotative sense, is particularly important, as “visitor
harassment” is said to be a problem in the resort areas on the island. More specif-
ically, the arguably negative notion of “defiance” in the Jamaican is reduced to
2 Reprinted from Bridal Guides Magazine.
124
personal concerns that have nothing to do with the tourist or the hardships of
living in a relatively poor country. In summary, Jamaica is said to have two lan-
guage varieties, but the Jamaican is represented as English speaking (with some
regional elements). The people are carefree and laid-back. The Jamaican woman
is presented only in a sexualized way, and the wider socio-cultural context is
Judaeo-Christian and therefore not unfamiliar.
This pamphlet, produced by the JTB, is a reproduction of a collection of articles
on Jamaica that appeared in a US magazine. In that respect, it is an American
image of Jamaica and Jamaicans that the JTB is using to sell the destination to
Americans. Other JTB documents, particularly more recent publications, speak
of the Jamaican language situation using the same terms: “We speak English,
with a few embellishments” (JTB 2001). In that respect, according to the JTB, it
is suggested that Jamaicans speak a variety of English not very different from
other Englishes except for a few localisms.
A second example is taken from the Jamaica Information Service (JIS), the offi-
cial information agency of the government. Interestingly, no explicit reference is
made to English. The reader is invited to infer from the language in the electronic
document that English is the official language of the State. Instead, the following
is said:
The language “patois” is an important part of who we are, giving the peo-
ple a peculiar accent so much so that even in countries outside the region,
we are easily identified. The Jamaican sound is so loved that even persons
who do not sound like us are often quite comfortable being called Jamaican.
While there may be variations in the patois accent across the island, there
are words and phrases that have gone beyond the boundaries of our little
island (Jamaica Information Service, accessed 2002 at www.jis.gov.jm).
This naming of Patois as “language” is clearly unrelated to the actual descrip-

tion provided. For Patois is here considered to be only English with an accent
and some lexical/idiomatic regionalisms. Moreover, this “sound” of a Jamaican
is charming enough to be imitated by others. Since, for the JIS, the difference
between Patois and other Englishes is largely a matter of phonology, this may
account for no direct reference to English as the language of Jamaica, just as there
would be no need to make such a clarification in Australia or the UK.
The sources of these examples represent what Scollon calls the products of for-
mally constituted institutional structures and therefore the language ideologies
of the groups that occupy such positions. These are organizations that are staffed
125
by “middle class” Jamaicans, in terms of education levels, income/occupation, res-

idence and the like. They naturally have particular ideologies which, according
to Austin (1983), are based on
the view (…) that the middle class gained and deserved their access to higher
education and the professions because their socialization and Christian
ways have made them superior to the uneducated, indeed uncivilized, work-
ing class (236).
She cites as an example the “middle class” attitude to traditionally working

class practices like Pentecostalism and Rastafarianism. The former is seen as
“degrading” (233) while the latter was coopted by the middle class and is now
approved for use to sell Jamaica. As if in confirmation of this, one member of
JAMPRO senior management said in his interview that
…if the person don’t (sic) portray the middle class features … am…could be
kept back. For example I know of people who are members of the conser-
vative pentecostal and evangelical type religions who have been kept back
(M20).
JAMPRO was legally incorporated in April 1990 under an arrangement that

merged three previously existing government bodies.3 These three bodies were
created in the first half of the 1980’s “precisely in recognition of the fact that in-
vestors need to be on a fast track to speed up the investment process” (JAMPRO
1997a), but had a forerunner in the pre-Independence Jamaica Industrial Devel-
opment Corporation of the late 1950’s. The three functioned in overlapping, but
different, areas of trade/investment promotion. The first, JNIP, was charged with
identifying potential investors and selling Jamaica to them as a safe and prof-
itable location for their business, i.e. its purpose was as a national marketing
agency. The second, JNEC, targeted both local exporters and foreign investors,
and was created to smooth the way through the maze of the Jamaican trade and
tariff bureaucracy. In that respect, it was there to represent the Jamaican gov-
ernment to business and it promised “enough clout in the public sector…that
the investor is facilitated as quickly and easily as possible” (op.cit.). The third
company, the JIDC, was mandated to oversee the industrial development policy
of the government, and therefore functioned as go-between for buyers of land,
3 Thethree are JIDC (Jamaica Industrial Development Corporation), JNIP (Jamaica National
Investment Promotions) and JNEC (Jamaica National Export Company).
126
equipment and infrastructure and the holders of capital. In 1990 government de-
cided that one umbrella corporation would increase efficiency and inter-agency
communication and created JAMPRO.
It is important therefore to emphasise that JAMPRO is an arm of the Jamaican
state, crucial in representing Jamaica to other individuals and governments as
“the investment capital of the Caribbean” (JAMPRO 1996). It is involved in the
serious enterprise of economic growth and development – without which very
little else of government policy can be realized.
JAMPRO, to further its purpose, produces a number of publications designed
to provide its clientele with information about a) the wider Jamaican social con-
text and b) JAMPRO itself, including the type of employee it selects and that the
investor is likely to encounter when dealing with the company. JAMPRO is one
such “formally constituted institutional structure” that is a state agency, within
which functions its own public relations department. By examining the informa-
tion put out by this department, we can get a sense of the image of Jamaica and
of Jamaicans that JAMPRO constructs for its clientele. In constructing this image
for its clientele, JAMPRO also generates and legitimizes a particular idea of the
Jamaican society and language situation. It is this ideology that I would like to
explore in detail, and how it becomes realized in the allocation of particular staff
members in the agency to certain functions.
Firstly, the ideologies of language and society that JAMPRO documents reveal
are a projection of the ideas of self held by one class of Jamaicans, vis-à-vis others
who occupy the same sociolinguistic space. These ideas are promoted against the
background of and in relation to their own ideas of who the client is. To illustrate,
there are a number of social and linguistic practices that could be described in
JAMPRO publications, but only some are selected. The selection reveals not only
what this class perceives itself and the ‘marketable’ Jamaica to be, but also what
they believe will inspire confidence in the client. To do this, certain assumptions
must be made about themselves and the client, what they have in common – and
can therefore go unexplained – and what is distinctive. To illustrate, no JAMPRO
document tells the investor that Jamaicans live in houses, as this would be un-
necessary as the class that composes the information would view this as normal
for both itself and the client. However, language in Jamaica has to be accounted
for, and in ways that we can use to locate what this class perceives as normative.
Secondly, employees of the company, and in particular those who have to in-
teract with clients, will have to reinforce this image or the client will lose con-
fidence in the information provided. JAMPRO cannot project an identity of the
country and itself that is dissonant with its own behaviour, reflected in the be-
haviour of its staff. Clients to the company who are encouraged to place their
127
confidence, and ultimately their money, in this described Jamaica, do so, in part,
because of the expectations created by the agency. Its staff, therefore, must be
in a position to confirm this image of Jamaica. As I was told by two members of
senior management:
It’s not overtly stated as being important, but I think it is a factor. I think
that when you consider the sort of image that [the agency] wishes to project,
certainly the thinking is, I feel, [that] people from certain backgrounds can
give or protect that image, rightly or wrongly (M101).
A job like this requires a certain type of personality as well as the qualifica-
tions you try to find a combination so its not all qualifications … important
consideration of someone here having to go abroad to represent the orga-
nization to me that is important and you’ll look for the possibility of this
with somebody (F55).
The published JAMPRO information and the personnel that the agency em-
ploys steer the investor in a particular social direction when advising on where
and how to interact with other Jamaicans generally. An analysis of this advice,
including some historical discussion of the type of output, can reveal what as-
pects of Jamaica are considered by agency staff to best conform to the norm that
is created in their own publications.
4.1.2 The JIDC/JAMPRO: Their construct of the Jamaican language

situation
The JIDC was established in 1952. I was able to find three specific references to
the local language situation and the Jamaican people, which I will quote in full
(my emphases).
Jamaica’s Alert, Energetic People – The Jamaican is intelligent, self-respect-

ing, and has a keen technical as well as artistic sense. Jamaicans speak
English with a soft, gentle inflection partially reflecting the British ac-
cent. The US, Canadian or British manufacturer will not face the problem
of a foreign-language-speaking work force which he would encounter in
many other Caribbean, Latin American areas. The advantage of a common
language also facilitates the training of Jamaicans for responsible posts in
manufacturing (JIDC 1957: 11–12).
128
The selling points of Jamaica are that it belongs to the English language com-
munity, in common with the US, Canada and Britain, unlike many other Carib-
bean territories. Moreover, all Jamaicans speak an English that is phonologically
somewhat like British English, though “British” is not further defined. It is not
unreasonable to interpret this as RP, the variety most likely to come to mind for
a Canadian or American reader.
A later publication says much the same thing:
Jamaica is a land of soft-spoken people (…) Jamaicans speak English. The

US, Canadian or British manufacturer arriving in Jamaica does not face a
language difficulty in training and supervision. The advantage of a common
language is that it speeds up training and production (JIDC 1961: 2–8).
The fact that Jamaicans also speak something else, Creole/Patois, is not ad-
dressed. There are a number of possible explanations for this. Firstly, additional
information in these pamphlets gives an indication of where and with whom
the investor was expected to interact and the strata in the society that the JIDC
wished to own. For example, under the heading Cultural Activities, the investor
is told of “lectures on music and art sponsored by the British Council and the
Institute of Jamaica” (13), despite the local popularity of mento bands and the
numbers of venues at which sound systems played (Barrow & Dalton 1997: 7–13).
Additionally, the Social Clubs recommended by the documents were exclusive
and/or member’s only establishments such as Jamaica Club, Liguanea Club and
the Garrison Officer’s club. Secondly, the idea of Patois as a language was not,
at the time, an accepted aspect of public discourse. Indeed, Le Page (1988) asserts
that it was the discussion in academia, by David DeCamp, Beryl Bailey and him-
self, that gave some, however limited, legitimacy to the idea of something called
“Jamaican Creole” as a language in its own right. Interestingly, both quotations
are careful to suggest the gentleness of the people, perhaps to reassure the in-
vestor’s anxieties about the social context, in a similar way to Taylor’s reading
of the JTB information.
Current JAMPRO documents/pamphlets are noticeably different in some
ways, at times presenting the language situation in more measured terms:
The official language of Jamaica is English. However, the majority of the

population speak Jamaican creole called patois, which is a mixture of En-
glish and derivatives of various West African languages. The Jamaican Cre-
ole has been studied by many scholars. There is a Dictionary of Jamaican
English published by Cambridge University Press.
129
There is another variety spoken in Jamaica, legitimised by being in a dictionary.

However, the magazine then immediately goes on to reassure the reader, in two
other places that: “Kingston [the capital] is the largest English speaking city a)
south of Miami and b) in the Caribbean.” (JAMPRO 1997b: 2–3). This insistence
that Jamaica is English speaking dominates the information available to those
interested in more serious ventures – investment, manufacturing or business.
The official language is English. What we call patois, an English-derived

dialect, is widely spoken, though seldom in official or business circles. Patois
tends to prove challenging to foreigners, though a number of publications
ranging from visitor guides to scholarly dissertations are available on the
hotly debated language. Local radio and television dramatic productions
also have high patois content, as do some radio call-in programmes from
time to time, especially when tempers rise! (JAMPRO 1997a).
English-speaking, well-educated and competitively priced labour force

(JAMPRO 1993).
Jamaica’s English speaking population of approximately 2.5 million repre-

sents a diverse mix of ethnic origins (JAMPRO 1994: 15).
English-speaking, non-unionised and experienced professionals (which in-

clude accredited technicians) as [film] crew (JAMPRO 1993: 5).
Here the investor is informed that Creole is widely used but only in specific cul-
tural and discourse contexts. Creole is also not the language of serious business,
suggesting the kind of distinction the BBC made between programmes hosted
by regionally accented and RP accented broadcasters. Moreover, the fact that the
language is “hotly debated” suggests the kind of local controversy about Creole
that is not described for English. The investor is further reassured that the pool
of labour from which comes his/her workforce speaks English and gets a British-
style education:
English is universally spoken. Education is based on the British system…

English is the official language of Jamaica, and this makes communication
with off-shore investors and understanding new processes easier… The lan-
guage of the workplace being English, instruction at all levels is given in
English, even where the workers speak patois and not Standard English
(JAMPRO 1992).
130
Two important things are being said here. Firstly, English is universally spo-
ken and, it is suggested later in the passage, understood. Even if the investor
hears workers speaking patois they should not take this to mean that these work-
ers cannot speak or understand English. Secondly, a distinction is being made
between English, Patois and Standard English. What the investor will find, and
what is being implied, is that the worker’s speech, Patois, is not much more than
a regional non-standard variety of English, for what workers do not speak is
Standard English, the language of official and business circles.
In summary, the JAMPRO documents give the following sociolinguistic infor-
mation:
• All Jamaicans speak English (“universally”).
• Some Jamaicans also speak Patois (“widely spoken”).
• Patois is non-standard English, largely distinguished by some items, possi-

bly of West African origin.
• Patois is used by workers, but they also speak and understand English.
• Patois is used in specific (dramatic) contexts and when tempers flare.
• Patois is controversial.
• Official and business circles use (Standard) English.
• Standard English is only a “problem” for workers.
Generally, we can deduce from these documents that Creole enters the pub-
lic discourse, as generated and structured by these government agencies, as the
other: the non-normal, non-universal aspect of the Jamaican socio-cultural land-
scape. We can further deduce that Patois, being peripheral to the official and
business milieu in which JAMPRO operates, is not going to be heard by the in-
vestor in his interactions with the agency.
I found only one specific text reference to the staff at JAMPRO. The magazine
in which it is contained promises “highly trained professionals [who] provide
friendly, experienced assistance” (JAMPRO 1996: 10). If we relate this information
to the above discussion, then we would expect JAMPRO staff members, who do
operate in official and business circles, at least to speak (Standard) English and to
be very well educated. The speech that these employees actually produce must be
useful in identifying what JAMPRO management considers to be protective of its
131
image, and conforming to the norms they have described above for Jamaica. This
is an agency in which the investor will encounter highly trained and (Standard)
English speaking professionals.
4.2 Frontline staff – The public face of JAMPRO

There are certain employees at JAMPRO designated “frontline staff”. This label
is applied to those employees who actually interact with the public, both local
and foreign, in the course of their duties. In certain departments at JAMPRO,
certain members of staff are required either to deal directly with the agency’s
clients or to represent the agency in overseas departments or trade missions. It
is this subset of my informants – those in the frontline – who are the focus of
this section.
There are a number of reasons why the speech patterns of frontline staff are
crucial to the issues raised in this book, particularly in relation to the legitimizing
of norms of JE usage. Firstly, as has already been discussed in previous chapters,
selection of staff at JAMPRO is heavily dependent on performance during the
job interview. And this is especially so for professional and (senior) management
positions. Frontline staff positions reflect an even more subjective set of selection
criteria on the part of the agency, as these are employees who are the public
face of JAMPRO. As explicitly stated by senior management, those in frontline
positions are selected because they are perceived to be the type of person who can
represent both the agency and the agency’s idea of a well educated, competent
and professional Jamaican.
Secondly, in selecting some employees for frontline positions and rejecting
others, JAMPRO management is communicating the type of Jamaican they think
best fits the desired image, one which may well be informed by the speaker’s use
of English. JAMPRO, as shown at the beginning of this chapter, operates in pre-
cisely the socio-cultural milieu that it advises is reserved for (Standard) English.
Frontline positions indicate a positive judgement made by some Jamaicans, i.e.
the senior management who assign staff to jobs, on the way other Jamaicans
present to them in interviews. This forms, in large part, the basis on which cer-
tain members of staff are selected to interact directly with clients (both local and
foreign) and to represent “Jamaica” and the upper end of the construct of the lo-
cal language situation described in JAMPRO publications. Their promotion into
the frontline provides us with some means of indexing the reactions of others
to the English they produce, for they have been selected from a pool of speak-
ers who themselves had been chosen from a pool of educated Jamaicans with
132
English language qualifications. In that respect, the analysis of the speech of

frontline staff allows me to add an evaluative dimension, albeit an implicit one,
to this study. Within the organization, at least, the kind of behaviour that comes
to be perceived as suitable and successful for many, in terms of advancement and
employment prospects, can be taken to be reflected in frontline staff.
Thirdly, informants themselves held perceptions about the kind of person who
is in the frontline. I would like to examine these perceptions and compare them
to relevant data on the sociology of the agency. For if staff perceptions about
frontline staff are reflected in the findings, then it gives added credence to those
perceptions less amenable to examination – most importantly the ones about
language and speaking “good” English.
In the course of interviews, a number of the informants made comments on
their perceptions of the type of employee who gets selected for frontline posi-
tions. An asterisk indicates that the speaker is in senior management.
yes I think so, am there’s a particular look which you must have, which I
think, you know, I think fair…fairer skin people tend to, especially in terms
of overseas officers…it’s as if they want to portray a certain look of a Ja-
maican (F13).
there are some reasons for it in that you are talking about a promotional
organization and sometimes you will buy into that…somebody with a par-
ticular social background, somebody who has been exposed early has the
ability to present themselves and to cut a dash…but I think it happens a little
bit too much in that some are excluded where they shouldn’t get (inaudible)
as sophisticated they’ll step over her (F64).
I remember once a skin thing came up and it was more in relation to the
film division, there was a time when the perception was that only fair skin
people can deal with movie stars coming from abroad and the colouration
there tended to seem to reflect that (F87).
…but for example if they want persons to go on say overseas assignment…if

you’re going to Korea or one of those you know eastern countries right they
prefer male to go they specifically ask for male (F71).
Yeah I think sometimes…deciding they probably notice the way you act and
the way you speak if you can really represent well at a standard (F73).
133
I think it does I think it does because one has to have a certain amount
of command a certain amount of respect from the people that you have to
interact with and it [a certain background] does help it opens a lot of doors
put it that way…and that’s [education] important because it it equips you
to have a good command of the English language…the subject matter and
it sort of helps you to be able to help the client better (F7).
race especially pigmentation is important, the brownings, I feel so but you

know that fits within the, you know, more with the profile you see some of
these things can be justified on the basis that we are a promotional organi-
zation so the investors are likely to be impress (sic) by a certain type…it’s
just that we want to present the right image and the image would include
a certain kind of, of diction and good looks too…if that person is not obvi-
ously from the middle class and if the person’s manner and, and dress don’t
portray the middle class features [they] could be kept back…(M20*, Public
Relations).
[language] is very important it’s very important in that in JAMPRO you

have to interface a lot with your external clients your ability to deliver is
very important (F32*).
It’s not overtly stated as being important, but I think it is a factor. I think
that when you consider the sort of image that [the agency] wishes to project,
certainly the thinking is, I feel, [that] people from certain backgrounds can
give or protect that image, rightly or wrongly (M101*)
at the stage of the interview I ask is this person somebody I could send away
next week to go and talk to an investor or to go and sell Jamaica. (F55*).
This is a place where you interact a lot with the public, with international
agencies…if you have a, well, you’re well educated and everything your
social class tends to be middle class (F56*)
Generally, these comments speak to a perception among staff, that those in

the agency who interact with the public, especially foreign clients, are selected
on the basis of a) certain social backgrounds, i.e. from the middle class, b) light
skin colour c) “good looks” and d) language. The one mention of gender is specif-
ically related to assignment to some Asian countries where men are specifically
requested by the host country. That suggests an issue outside of JAMPRO’s con-
trol, one which they attempt to accommodate.
134
The first set of data presented discusses those social factors which proved to
have a statistically significant association with having frontline status. The data
is presented in percentages, but the statistical significance was calculated on the
raw data.
Table 4.1: Level of education and selection to frontline duties at
JAMPRO
Education (%)
Position Primary Secondary Secondary+ Tertiary Tertiary+ no data
+Frontline 1.5 3 36.5 27 32
−Frontline 8 21.5 46 11 11 2.5
no data 3 1
By inspection, it is clear that selection for a frontline position becomes more

likely if one is better educated (p < .001, χ2 = 10.91 when analysed as secondary+
and tertiary+). Here, to accept the null hypothesis (H0 ), we would expect to
find more primary and secondary school leavers in the cadre of frontline staff
members. What we do have is a greater proportion of university graduates in
these positions (almost 60%). The one primary educated informant is M40, the
driver/chauffeur for the agency. Staff perceptions that education is important
for frontline positions are confirmed by this data. JAMPRO does cream off its
better-educated employee for interaction with the public.
Table 4.2: Parent’s occupation and selection to frontline duties at
JAMPRO
Parent’s Occupation (%)

Position Cleaner/Artisan Teacher/Doctor no data
+Frontline 46 44 10
−Frontline 65 24 11
no data 3 1
Tests of association suggest that there is some relationship between family

background and selection for a frontline position (p < .05, χ2 = 4.09), when cate-
gories are collapsed to cleaner/artisan and teacher/doctor. Specifically, those who
come from the lowest income/occupation bracket are under-represented in the
135
ranks of frontline staff; 76% of the teacher/doctor group are in frontline positions.
This may be what senior management refers to when it speaks of a preference
for people of “certain backgrounds”, and the implication is that the requisite pre-
sentation for a JAMPRO representative is least likely to be found in the employee
who came from a low income/education household (indeed, only 8% of frontline
staff). Crucially, informants from the more affluent backgrounds are better edu-
cated, most likely to have tertiary level qualifications (§3.3). It is therefore diffi-
cult to delink level of education from background and household. “Background”
is of course a loose term that attempts to capture both linguistic and cultural
practice (see informant F64’s comment on “sophistication”). However, frontline
staff at JAMPRO typically come from households of relative advantage, probably
because they are, in this sample, also highly educated.
Table 4.3: Type of transport to work and selection to frontline duties
at JAMPRO
Transport (%)
Bus Lift Own Car
+Frontline 21 17 62
no data 2 2
How one gets around in Jamaica is a fairly reliable measure of personal income
and, to some extent, type of social contacts maintained. The public transport sys-
tem, certainly at the time of data collection, was chaotic, crowded and unreliable.
Jamaicans, therefore, used the public bus system only if there was no other op-
tion, such as a lift to work. The categories above can be used to reflect both access
to income as well as interaction with other people of a certain income. Informants
who took the bus to work were not only low income earners, but they operated
in networks with similarly low income earners. Those informants who got a lift
to work might not themselves have a car, but they knew people who did. This
suggests a somewhat different socio-economic situation, one with at least some
personal contacts to people of a higher income. Most frontline staff, however,
were able to buy and maintain their own car, and most of these would not have
had access to JAMPRO’s company car salary package. This data supports the
stated perceptions about the preference for “middle class” frontline staff – they
are better educated, from more affluent backgrounds and tend to have better in-
comes than non-frontline staff (p < .001, χ2 = 13.2).
136
Table 4.4: Age of informant and selection to frontline duties at JAMPRO
Age (%)
Position 20–29 30–39 40–49 50–65
+Frontline 22 38 21 19
−Frontline 57 16 8 19
no data 3 1
Grint (1991: 259, citing research by Jenkins 1985) points out that those involved
in exclusionary recruitment, which is what frontline hiring and promotion is
about, tend to have a hierarchy of criteria for acceptance – “the primary criteria
involve appearance, manner (…) and maturity. Secondary criteria relate to ‘gut
feeling’ employment history (…) age, speech style”.
But while middle-aged employees are most typical of those in the frontline
(p < .001, χ2 = 14.01), this does not necessarily mean that JAMPRO selects em-
ployees with most company experience.
Table 4.5: Years with the agency and selection to frontline duties at
JAMPRO
Years in JAMPRO (%)

10+ –10 New
+Frontline 33 38 29
no data 2 2
There is no statistically significant correlation between years in the agency and

frontline status (p > .10, χ2 = 4.49), though there is a tendency to use more expe-
rienced workers. Importantly, new recruits were hired into frontline positions,
suggesting that having JAMPRO experience is neither necessary nor sufficient
for frontline employment. Eighteen (18) members of staff were hired into posts
that entail interacting with the public, notwithstanding their lack of agency ex-
perience. Something in their interviews and qualifications prompted recruiters/
management to do this. Clearly, socialization to JAMPRO linguistic or corporate
norms is not a factor. What may be at issue is socialization to educated linguistic
norms, as evidenced by longer stay in the formal education system. On closer
examination, these 18 had the following profile:
137
Education post secondary 9 (50%), tertiary 3 (17%), graduate 6 (33%);
Parent’s background cleaner/artisan 6 (33%), teacher/doctor 12 (67%);
Transport to work bus 6 (33%), lift 2 (11%), own car 10 (56%);
Age 8 under 30 (44%) and 10 over 30 (56%);
As such, at least half of the new staff who entered the frontline ranks were
university graduates, from relative affluence, had their own car and were over
30. Most of them were hired into professional or managerial positions.
Initially informants were asked to name the racial group to which they belong,
an effort to collect self-ascriptive data. For a number of reasons this proved to be
problematic. Some informants were reluctant to discuss the issue at all – typical
responses were I don’t know or you put what you think. This is not unusual, as
other studies of race in Jamaica have also recorded a similar discomfort with the
subject. As Alexander found in his survey:
This ambivalence is expressed in the touchiness with which the subject is

discussed. Informants all framed the matter the same way: race is a subject
people do not discuss freely and openly; it remains understood (1977: 427).
Additionally, those that did answer did so in terms that would have been dif-
ficult for further sociolinguistic analysis. The first 10 informants gave 5 different
answers, such as Maroon, dark, mixed, Jamaican and Negro. Apart from an in-
ability to correlate such a diversity of answers, it is not at all apparent that even
when informants used the same term to describe themselves that they necessarily
refer to the same thing. It is entirely possible that two informants who describe
themselves as “black” may have differing ideas of what [use of] that label means.
I found (1988), for example, that one informant who described himself as “white”
also included Chinese, Syrians and Lebanese people in the group as the label
referred as much to socio-economic status as it did to race.
In order to have some control over the number and semantic range of labels,
I assigned informants to racial groups. I am aware, however, that by doing so I
am imposing my own perceptions of race on the sample, and that this can yield
misleading results. Labov’s work (1972: 298) in New York City and his analysis of
Italian-American vowel production can be used to illustrate the point. Along sim-
ilar lines, Horvath & Sankoff (1987) found that refinements in method revealed a
somewhat different picture of ethnic variation in Australian English than previ-
ously described.
138
Informants in the JAMPRO sample were grouped according to my perceptions

of skin colour as there is some historical and contemporary justification for be-
lieving that race in Jamaica is essentially defined phenotypically: by colour, hair
type and the like (see Lewis 1968: 20; Nettleford 1972: 24;, 374; Witter & Beckford
1980 for typical discussions). The labels black and brown are used primarily to
refer to dark brown and light brown skin colour respectively.
Table 4.6: Skin colour and selection to frontline duties at JAMPRO (%)
Black Brown no data

+Frontline 44.5 41.25 14.25
no data 1 3
The data in Table 4.6 shows that a smaller percentage of the informants who
I view as “brown”, i.e. have lighter skin, are in non-frontline positions (p < .02,
χ2 = 6.52). Put in another way, there are 32 brown informants for whom I have
data on their position, 26 of them are in the frontline, and therefore more than
four times the number behind the scenes at JAMPRO. In contrast, there are 52
black informants, with 28 in the frontline and 24 behind the scenes. Brown people
therefore are over-represented in the frontline of the agency.
It is important to point out that staff with lighter skin colour are not better
educated; their selection cannot necessarily be explained as a preference for the
highly educated in frontline positions (p > .50, χ2 = .31). However, it is entirely
possible that such a perception exists as part of the cultural traffic (Alvesson
1993: 80) that flows in from the attitudes held in the wider society. In an ear-
lier matched-guise study of 100 educated, affluent Jamaicans (A. Irvine 1994:
61), I found that the guise of JE judged to be that of a black person was also
ranked lower on the intelligence/competence dimension. And while not statis-
tically significant, there seems to be some association between skin colour and
background/household (p > .10, χ2 = 1.68). Brown members of staff are, for exam-
ple, much less likely to come from households with parents who are “cleaners”.
Arguably, brown members of staff are also less likely to come from monolin-
gual JC households and consequently it is possible that an assumption is made
that their use of JE is not as dependent on formal education; moreover, they are
more typically from the backgrounds that JAMPRO seems to favour for frontline
staff.
139
Table 4.7: Gender and selection to frontline duties at JAMPRO
Gender (%)
Female Male
+Frontline 78 22
−Frontline 78 22
no data 4
The only specific reference to gender and position in the company suggests
that, for certain Asian markets, men are more likely to be selected as overseas
officers and at the request of the host country’s agencies. The data show men
and women having virtually the same chances of being selected for the frontline,
since gender does not correlate with position in the agency (p > .95, χ2 = .004).
The social factors that show some degree of association with membership in
the ranks of frontline staff are social class – as indexed by parent’s occupation and
transport – age, level of education attained and skin colour. The perceptions of
staff and the statements of senior management about the type of employee who
tends to be in the frontline did mention background, class, colour and education.
I argue then that since there seems to be some validity to and empirical support
for these perceptions, those that speak specifically to linguistic criteria are also
to be taken as important. Frontline staff are those employees whose English is
believed to be suitable for a representative of JAMPRO, displaying the attributes
of diction, talk and English “at a standard”.
4.2.1 Group A variables in frontline staff

Along the Creole/English dimension we would expect frontline staff to use fewer
Creole forms than those not so selected. The results for a number of Group A
variables are in line with this expectation. Frontline members of staff are much
less likely to drop [h] (p < .001, χ2 = 38.54); they also produce fewer low central
vowels in not words and more of the JE [ɔ] (p < .05, χ2 = 50.4). For other Group A
variables the results are somewhat less predictable, as some Creole features are
used more by frontline staff or in a similar way by both classes of employee.
Voiceless TH stopping, which I argue is the more indexical variable in this pair,
is produced less by frontline staff, in line with h-dropping and the low vowel
(p < .05, χ2 = 4.43). The hypercorrect use of these three features in JE speakers
would suggest a sensitivity to their use and their importance to any idea of good
English. But voiced TH stopping is more apparent in the same select speakers
140
Table 4.8: Word initial TH stopping in frontline staff
Frontline 1472 (52%) 1373 (48%) 63 (12%) 478 (88%)
Non-frontline 818 (37%) 1369 (63%) 54 (17%) 269 (83%)
(p < .001, χ2 = 102.44), which suggests that less attention is paid to its production
in a context of spoken JE (although I would predict that in a word list test the
results would be very different). Arguably, the more speakers are cued to the
variants that matter – here the [θ] – there is less of a requirement to pay attention
to the ones that do not. Certainly, voiced TH stopping does not seem to preclude
promotion to the frontline.
A similar pattern is evident in words beginning with a velar stop.
Table 4.9: Word initial velar stops in frontline staff
kja kh a kja: kh a:
Frontline 77 (56%) 61 (44%) 9 (13%) 60 (87%)
Non-frontline 42 (59%) 29 (41%) 10 (37%) 17 (63%)
(p > .50, χ2 = .19) (p < .01, χ2 = 6.97)
Frontline staff produce much fewer instances of [kja:] than do those in the
background; but use of [kja] does not distinguish the two groups, and it is freely
varied, as has been the case in nearly all correlations with this feature in this
study. Three possible conclusions can be drawn from this. The first is that some
Creole features are perfectly acceptable in spoken JE and have been normalized
as such (as are [kja] or [d ~ ð]). The second is that [kja] and [d] are not necessarily
perceived to be Creole features. The third is that [kja] and [d] are peripheral
to the issue of speaking JE. As such, their use by a speaker who consistently
produces say [θ] is of little import to members of the speech community when
making judgements about “good” JE. When speaking with another Jamaican, as
was the case in their interviews both for selection to frontline duties and for this
study, informants’ use suggest [kja] is an aspect of Jamaican speech attested to
in spoken formal JE.
Table 4.10 shows that the back diphthong is seldom used by any speakers ei-
ther pre-consonantally or before [r]. This would confirm the theory, presented
141
Table 4.10: Mid vowels in frontline staff
uo o ie e
Frontline 87 (8%) 960 (92%) 141 (12%) 994 (88%)
Non-frontline 42 (7%) 537 (93%) 114 (18%) 522 (82%)
(p > .30, χ2 = .56) (p < .01, χ2 = 10.04)
[ɔ] [uo ] [o] [ie r] [er]
Frontline 303 (73.0%) 77 (18.0%) 37 (9.0%) 316 (42.5%) 426 (57.5%)
Non-frontline 127 (72.5%) 35 (20.0%) 13 (7.5%) 149 (52.0%) 138 (48.0%)
(p > .80, χ2 = .42) (p < .01, χ2 = 7.29)
at the end of the previous chapter, that back diphthong use is the more linguis-
tically indexical variable of the pair – a necessary element for being perceived
as speaking good JE. The front diphthong, however, is less likely in the speech
of frontline staff, as would be expected from previous results that also support
the view that [ie] use is more socially indexical in JE than [uo]. Front diphthong
use is more apparent in the young, a group under-represented in the ranks of
frontline staff.
4.2.2 Group B variables in frontline staff

The statistical results for butter type words point to the difference here being who
uses an [a] as opposed to an [ʌ] ending, along the Creole/English dimension
characteristic of Group A variables (p < .001, χ2 = 85.9). As we would expect,
frontline staff use less of the Creole variant. A similar analysis can be made for
that aspect of Table 4.11, but the [ʃɔn ~ʃǝn] variation is much more interesting.
Table 4.11: Articulation of final – tion in frontline staff (p < .001, χ2 =

61.8)

Frontline 69 (14%) 286 (57%) 103 (21%) 41 (8%)
Non-frontline 80 (38%) 91 (43%) 17 (8%) 24 (11%)
Frontline staff are more likely to use the [ʃɔn] that is associated with higher
education and not the schwa typical of MSE. Clearly here speakers are being
142
selected for interaction with the public not only on their avoidance of stigmatized
items, but also on their production of this JE feature, given the peripheral status
of schwa in this speech community. This feature was also correlated with level
of education, and was more typical of women. Within this sample, [ʃɔn] is a
variant in the speech of most, if not all, groups that can be used to locate standard/
prestige forms.
Table 4.12: Articulation of culture type words in frontline staff (p > .10,
χ2 = 2.04)
dj ʤ tj ʧ
Frontline 25 (76%) 8 (24%) 49 (32%) 102 (68%)
Non-frontline 9 (75%) 3 (25%) 24 (44%) 31 (56%)
Use of either the voiceless palatalized stop or the affricate in culture type words
does not significantly distinguish frontline from non-frontline staff. Variation
in this feature was correlated with gender and level of education – the highly
educated favouring the affricate and women more likely to use the palatalized
stop variant. As frontline positions seem not to be allocated on the basis of gender,
the weak tendency for frontline staff to use the affricate more frequently can be
explained by their higher levels of education.
Table 4.13: Rhoticity in frontline staff

Frontline 69 (41.5%) 97 (58.5%) 134 (60%) 89 (40%)
Non-frontline 39 (46.0%) 46 (54.0%) 67 (79%) 18 (21%)
(p > .50, χ2 = .42) (p < .01, χ2 = 9.53)
While there is no statistically significant difference in post [a] rhoticity in

the two groups of employees, rhoticization of forty type words, though clearly
the norm in this total sample, is less a feature of frontline staff’s speech. Recall
that rhoticity after [ɔ] was less a feature of older speakers in this sample (see
Table 3.53 for example). Moreover, this feature was seen to be normalised in JE
among younger speakers and women. Again, persons in frontline positions tend
not to be younger members of staff. The data in total suggests an idea of a rhotic
143
JE, given both the quantitative and qualitative hypercorrection identified. But it
also suggests that frequency of rhotic productions differentiate varying groups
in the sample, with younger and educated female speakers generally the most
rhotic of the groups in both phonetic environments and younger educated males
being less rhotic after [a].
Frontline staff are more consistent in their use of all phonological clusters here
than non-frontline staff. In particular, while [nt] clusters before a vowel tended
to be generally used in the JAMPRO sample, frontline staff also produce more
[st] clusters and clusters before a following consonant (p = .05, χ2 = 3.8). Use
of clusters with morphological content does not distinguish frontline and non-
frontline staff.
4.3 Discussion
The patterns for frontline staff, the selected voice(s) of JAMPRO, suggest that
certain aspects of language use are normalised as good English.
Significant numbers of speakers never h-drop, use voiceless TH stopping or
palatal velar stops before long vowels. And, to a lesser extent, this also applies to
diphthongs, though it appears that [ie] is more and more an aspect of younger
speakers who do pass the “interview” test. In its expectation and practice, this
agency is communicating to its employees that these features of Jamaican speech
[h], [θ] and [kh a:] are necessary linguistic aspects of those that do well; at the
same time features like [d ~ ð] or [kja] do not preclude advancement or employ-
ment. Indeed they are also features of the successful employee. In this way, by
appointing some employees and excluding others, JAMPRO is engaged in con-
structing and legitimizing an ideology of what are the spoken norms of Standard
Jamaican English.
Most speakers who are frontline staff produce affricates in culture type words,
bimorphemic clusters and [nt], at least before a following vowel segment. But the
data here also suggests that saying [dõ wʌrɪ] don’t worry or [la:s je:r] last year is
not unacceptable in a speaker who has been selected by the agency to represent
it to the public. Of course, a discussion in terms of presence/absence of features
does not reflect the complexity of the situation. While it does show which fea-
tures are generally widely used or avoided, it cannot suggest the importance of,
say, using [ɔ] rather than [a] in JE, whether in stressed or unstressed syllables, or
the asymmetrical salience of one variable in a pair of related linguistic variables
in the sample.
144
4.3 Discussion
Table 4.14: Group A variables and distribution in frontline staff
Variant
/h/ 51 0 24 (47%)
Initial[ð] 51 1 (2%) 0
Initial [θ] 51 1 (2%) 28 (55%)
[ɔ] not 51 0 1 (2%)
[o] boat 51 0 19 (37%)
[e] face 51 1 (2%) 20 (39%)
[ɔr] poor 50 2 (4%) 17 (34%)
[ber] beer 51 1 (2%) 2 (4%)
[kh a:] 33 2 (6%) 26 (79%)
Table 4.15: Group B variables and distribution in frontline staff
Variant
-erbutter 51 20 (39%) 0
[ʃɔn]-tion 51 19 (37%) 0
[tj]culture 41 17 (41%) 12 (29.0%)
[r]party 49 19 (39%) 9 (18.0%)
[r]forty 48 5 (10%) 12 (25.0%)
-nt## C 50 6 (12%) 8 (16.0%)
-st## V 45 8 (18%) 7 (15.5%)
-st## C 50 25 (50%) 2 (4.0%)
n’t## C 48 17 (35%) 0
-edbeforeC 43 10 (23%) 6 (14.0%)
145
Frontline staff are selected because they are perceived to be staff who can “rep-
resent well at a standard”, and their behaviour provides a concrete model of spo-
ken JE, which at least this institution considers representative of what it referred
to as Standard English. Unlike the education system, JAMPRO is not engaged in
providing a model for imitation. Rather, the language use of frontline staff here,
in this formal interview context, is a sampling of speech from Jamaicans who are
considered by other Jamaicans to be speakers of good English.
The individual profiles below are examples of some frontline speakers who
used the fewest Creole forms in their interviews.
It would be difficult to argue that any one of the 3 unmarked speakers (M34,
F16, F11) is more acrolectal than the other, or that one could be placed on a
“higher” level on the continuum than some other one of these informants. M34,
for example, is unique in my sample. He was categorical in his use of [ɔ] in not
words, he seldom produced diphthongs, in fact 2 instances in 31 tokens, articu-
lated [st] clusters (before vowels) and so on. Is he to be placed on a “higher” level
on the continuum than F16 for example, given M34’s use of [kja] or [tj]/[dj] or
[dõ]? If one takes MSE as point of reference for Jamaican linguistic norms, then
the answer must be no. In F16’s interview she is categorical in her use of morpho-
logical clusters and in line with other MSE speakers in her use of phonological
clusters. But she also h-drops more and uses diphthongs more often in most en-
vironments. Her use of variants, if one takes an endonormative approach, may
well be evaluated by others as less acrolectal than his because of the particular
features that are found in her speech. But she is a highly educated female and
from a household with educated parents; he is a less educated man. Their lan-
guage use is likely to be filtered through the social perceptions that others in the
society hold of members of those groups. Kulick (1998) suggests that
…language ideologies seem never to be solely about language – they are

always about entangled clusters of phenomena, and they encompass and
are bound up with aspects of culture like gender, and expression, and being
“civilized” (100).
H-drop in the driver M40’s speech, for example, is not therefore sociolinguis-
tically the same as it is in, say, speaker F16. And it would be an interesting ex-
tension of this study to see if the occasional h-drop in someone like F16 is even
“heard” by the listener. M40 is a speaker who typically uses Creole phonology.
The exceptions are the back diphthong [uo] and h-drop. These features were both
infrequent in my total sample, and I have classified them both as two of the load-
bearing variables in Jamaica. M40 also uses [tj] in culture words and past tense
146
4.3 Discussion
Table 4.16: Group A Variables and distribution in 4 frontline speakers.
%
Feature M34 F16 F11 M40
h-drop 0 3 0 0
Voiced TH stop 30 16 19 100
Voiceless TH stop 0 0 11 100
[a] notwords 0 9 7 67
[uo] boatwords 6 8 6 0
[ie] facewords 0 0 0 100
[u or] poorwords 0 44.5 0 100
[bier] beerwords 33 25 28 0
[kja] catwords 75 0 67 no data
[kja:] cartwords 0 0 0 100
Table 4.17: Group B Variables and distribution in 4 frontline speakers.
%
Feature M34 F16 F11 M40
[a]butterwords 0 0 0 86
[ʃan]-tionwords 7 0 0 100
[ʃɔn]-tionwords 21 40 82 0
[ʃǝn]-tionwords 7 20 0 0
[tj]culturewords 100 no data 100 100
[r]partywords 20 0 100 0
[r]fortywords 33 no data 100 no data
-nt## before V 33 67 100 no data
-nt## before C 33 33 0 50
-st## before V 100 67 100 no data
-st## before C 33 40 0 no data
-n’t## before V no data 100 no data no data
-n’t## before C 14 33 0 0
-ed before V 100 100 100 100
-ed before C 50 100 33 no data
147
inflections. In that respect, there are aspects of his speech that index “better En-
glish” and this may suggest why he has been selected for a frontline position.4
The data from frontline staff, the selected public face of JAMPRO, is taken from
speakers who are there because they have been judged suitable to represent the
construct of the language situation described by the agency. Patois and Standard
English are the varieties described in these publications, the latter, I would sug-
gest, reflected in the formal usage of the majority of professionals appointed to
the frontline ranks. And the features present or more frequent in this majority
represent what is good spoken SJE, the acrolect in this speech community. In its
expectation for staff and its practice of promotion, JAMPRO is legitimizing norms
of speech, necessarily mediated by the ideas of those who control it about who
the speaker is and what place they are expected/believed to occupy in the society.
That the norm is essentially reflected in educated female speech at JAMPRO is
also a function of the social context and the place women occupy in that agency.
4 In the ideology of the Jamaican middle-class (Austin 1983: 236) a position like driver/chauffeur
cannot be held by someone at the professional level in the agency or a woman and therefore
must be occupied by an employee like M40.
148
5 Conclusion
This study set out to answer a set of questions about English in Jamaica, and
to discuss what is called the acrolect or acrolectal in Jamaica. One central point
that has been made in this study is that the construct called the Jamaican lin-
guistic continuum, of which the acrolect is an integral part, has to be defined
and analysed in relation to local norms of language use and not an external MSE.
Approaches that use MSE as benchmark for what is standard in the Jamaican
speech community, I argue, can be problematic both in analysis of data and in
some of the conclusions that follow from that analysis. For example:
• Forms are typologized as basilectal in Jamaica even though they are at-
tested to in the speech of all Jamaicans generally, irrespective of social
context of use;
• Forms are labelled (upper) mesolectal even when found in the formal
speech of educated Jamaicans or appear to be accepted as standard locally;
• Conclusions are arrived at that suggest that no one in the speech commu-
nity speaks Standard English, even as members of the speech community
hold to an idea that there is such a thing and that there are such speakers;
• The presence of non-MSE forms in the speech of educated Jamaicans is

described as an emerging or new trend even though such variation was
identified some two centuries ago;
• Discussions about decreolization and language change are carried out on

the assumption of a vaguely defined “local standard” as one target, even
though in practice that local standard is typically treated as either RP or
General American.
This exploration of the acrolect takes an endonormative approach and seeks to

explore the phonology of JE, as used by 82 educated speakers using language in
a formal context of interaction. From the data collected, I define what is spoken
SJE or the Jamaican acrolect. Further, the speech of select groups of informants
5 Conclusion
was singled out for analysis, most notably frontline JAMPRO staff, in order to
present an actual example of good spoken JE as judged by other Jamaicans. In so
doing I wished to address the following questions:
• What patterns of use are to be found in this sample of educated Jamaican

speakers when in formal interaction?
• How do the various groups identified differ in their use of these phonolog-
ical variables?
• What is the relationship between acrolectal phonology and other varieties

in the speech community?
• Does JAMPRO select speakers of a particular sociolinguistic type for high

status positions?
• Is JAMPRO, in its practice and expectation, legitimizing certain speech pat-

terns?
• What, based on all the above, constitutes an endonormatively determined

set of phonological features for SJE?
5.1 The architecture of phonological variation in the

Jamaican acrolect
Two types of phonological variables were created for the purposes of analysis.
One type was a group of variables previously described in the literature as having
Creole and English variants (Groups A); the other was a less studied group with
variant options that are not solely describable as Creole vs English and which
suggest another aspect of JE variation (Group B). The general patterns of use
of these variables is shown in Tables 5.1 and 5.2, with reference to particular
sociolinguistic correlations.
As shown in Table 5.1, there are some Group A variables that are statistically
significant with most of the social categories used to stratify this population.
Most of them, for examples [h] in hot, [ð ~ d] in them, [θ ~ t] in thing, [ɔ ~ a]
in pot, [ɔr ~ uor] in poor and [e ~ ie] in face, are aspects of education, parent’s
background and high job status in the same speakers means a high degree of
similarity in the way these groups pattern in their use of Group A variants. The
Creole variants of the first four are also the features that women tended to pro-
duce less of. We can identify the standard and prestige variants of these as [h],
150
5.1 The architecture of phonological variation in the Jamaican acrolect
Table 5.1: Overview of Group A data indicating statistically significant

associations
Group A Education Background Gender Age Frontline Significant

[h ~ ∅] * * * * 4
[ð ~ d] * * * * 4
[θ ~ t] * * * * 4
[ɔ ~ a] * * * * 4
[o ~ uo]
[e ~ ie] * * * * * 5
[ɔr ~ uor] * * * * 4
[er ~ ier] * * * * 4
[kh a ~ kja]
[kh a: ~ kja:] * * * 3
Significant 8 8 7 2 7
Table 5.2: Overview of the Group B data including statistically signifi-

cant associations
Group B Education Background Gender Age Frontline Significant

-tion * * 2
culture wds. * 1
party wds. * * 2
forty wds. * * * 3
-ed before V * 1
-ed before C * 1
-nt before V * 1
-nt before C * * * 3
-st before V * * 2
-st before C * 1
Significant 6 1 3 2 5
151
5 Conclusion
[θ], [ð], [ɔ], [ɔr] and [e], the first four also identified as part of the prescriptive
SJE of the school curriculum. Further, the patterns of use in the sample indicate
that [kh a:] calf and [o] boat are also the standard and prestige forms in JE.
Indeed, in the sample of speakers who received at least a secondary level ed-
ucation, there was very consistent use of the JE variants of these variables and
very little use of either [kja:] or [uo].
Further support for this is to be had in the hypercorrect productions of some
informants for some of these variables. This hypercorrection does suggest that
these variants, for example /h/:[h], /ɔ/:[ɔ] and /θ/:[θ], can be said to characterize
and define acrolectal speech.
Among Group B variables, those that also presented variation among JE op-
tions, the sociolinguistic variation that occurs is more diffuse than with Group A
variables. Notably, only level of education and frontline status distinguish speak-
ers’ use of more than a few features. The association of education, parent’s back-
ground and status in the agency was manifest in Group A variation, where JC and
JE variants were typically the speakers’ options for use. But with these Group B
variables, the patterns of the highly educated were not necessarily a predictor of
the patterns of speakers from more affluent backgrounds or high status positions
at JAMPRO.
The variants [ʃɔn] – tion, and [tj]/[dj] culture / soldier reinforce the point that
MSE forms cannot be used to identify acrolectal forms. In both instances speakers
here do not necessarily use the MSE variant (schwa and the affricate) but are
varying with options that reflect the coexistence with Creole and what speakers
believe to be Creole forms. These are features that are produced in response to
an idea of what is stereotyped Creole (affricates), distancing speakers even when
the Creole form is like the non-local standard that the literature has assumed
the acrolect to be. As such, the acrolect is not “more or less well-defined and
discrete” (Winford 1997: 241), at least phonologically. Forms that are perceived
to be Creole, such as [ʃan] – tion or [ʤ] treasure, affect the productions of some
speakers and as a consequence the kind of variation found in the acrolect.
However, the distribution of the phonological variants in the data presented
here is more complex than merely a low frequency of Creole forms or high fre-
quency of prestige forms in the speech of formal JE speaking informants. And it
is this distribution that turns out to be the important feature of acrolectal phono-
logical variation.
There is one category of phonological variables that appears to be salient for
producing SJE. Speakers, when producing these variables, show a remarkable
conformity in variant use.
These are:
152
a) the word initial glottal fricative [h] in words like hand;
b) the voiceless interdental fricative in words like thin;
c) the word initial velar stop [kh ] before the long low central vowel [a:] in
words like calf ;
d) [ɔ] in words such as cotton;
e) post [ɔ] rhoticity, specifically before [+ coronal] consonants in words like

forty;
f) [ɔr] in words such as poor;
g) [o] in words such as boat;
h) the word final phonological stop cluster [nt] before a following vowel, as
in hint at.;
i) the word final morphophonemic cluster when a past tense marker, as in

looked.
This type of variable I label a load-bearing phonological variable in the acro-

lect, for they seem to function primarily to define the variety the speaker is using.
Without these variants being produced in significant quantities, the speaker will
not be interpreted in the Jamaican speech community as someone speaking En-
glish. Moreover, it appears that it is use of these load-bearing variants, and not
English variants per se, that defines someone as speaking SJE.
As an example, high status speakers typically produce voiceless interdental
fricatives, so much so that qualitatively hypercorrect use of the variant occurred.
A different analysis, however, would have to be made of these same speakers’
use of [ð ~ d]. The voiced TH stop variant occurs in their speech in high frequen-
cies, and occurs more in the speech of frontline staff than those not selected to
interact with the public. The kind of asymmetrical patterning I find for this pair
of variables does not seem to occur in spoken MSE.
This same kind of asymmetrical sociolinguistic pattern is found elsewhere in
the data, and is evident in the mid-vowels, the velar stops and post vocalic rhotic-
ity. [e ~ ie] is sensitive to social factors here and more frequently varied, but [o ~
uo] is not; [kja] is attested in all speakers, but [kja:] is seldom produced by any-
body; and while speakers vary rhotic productions after [a], they generally are
rhotic after [ɔ].
153
5 Conclusion
It is important to point out, also, that this is the pattern of production in careful
speech. This asymmetrical attention to variants produced of related variables is
then fundamentally different to the patterns I can identify in the literature on
metropolitan English varieties, and is not a function of style shift in speakers.
Informants here, when producing speech in circumstances that would suggest
care and attention to forms used, focus on one variable of the pair and show
some consensus on which variant to select. At the same time, use of the other
variable in the pair is typically less focussed and is characterized by much greater
levels of variation.
This suggests that use of the variants [ð] in them, [e] in face, [kh ] in cat or the
affricate [ʧ] in culture, does not necessarily define SJE. This makes these phono-
logical features a different type from their load-bearing counterparts, because
their production does not seem to be either necessary for or an indicator of speak-
ing SJE.
The data also shows that speakers, in performing acts of identity or signalling
group affiliation, are more likely to manipulate use of these non load-bearing JE
variants for such purposes. I argue that this is because their presence or absence
is not already assigned particular social meaning in the speech community, i.e.
an inability to speak English, and therefore the speaker can manipulate these
phonological variables in contexts of interaction that require SJE.
Younger informants used the diphthong [ie] in face more frequently; in the
same speakers there was very little use of the back diphthong [uo] in boat. When
compared with the oldest speakers, it appeared that front diphthong use was in-
creasing. However, Alleyne’s observations about the more widespread use of
[ie] in Jamaica were made over two decades ago, and therefore I remain cau-
tious about saying that its more frequent occurrence in younger speakers re-
flects a changing norm. It is possible therefore, that front diphthong use here
distinguished younger speakers from older ones, and therefore is an aspect of
signalling group affiliation.
Variation in rhoticity, specifically after [a] in words such as party, also cor-
related with age as well as gender. Young men were more typically non-rhotic
after [a], distinguishing them from older speakers and women’s patterns gener-
ally. Males in this sample also more typically produced an affricate in culture type
words, as distinct from the female pattern which showed higher frequencies of
[tj] use.
I argue that because the variables discussed above are not crucial to defining
the variety being used, whether JC or JE, the type of variation that occurs is going
to be less focussed and therefore more a function of signalling group affiliations
154
like gender, or age. Variation on load-bearing variables will be constrained by

their function of either indexing “speaking Creole” or “speaking English” or the
identities of “Creole speaker” or “English speaker”. But this is not the only type
of social differentiation that is marked sociolinguistically.
The question then becomes, how does one project membership in other types
of social groups, particularly to an unfamiliar hearer? One cannot manipulate
variants to which social meaning is already widely assigned unless one is inter-
acting with intimates. When interacting with intimates, who are familiar with
one’s background and who generally have information about one’s social posi-
tion, use of JC features is typical, indeed it is expected. However, with strangers,
variation in one’s use of [h] or [uo] in JE is likely to indicate something undesir-
able, such as “backwardness” or being “low class” or “uneducated”, particularly
in a (work) context that favours the English speaking, middle class persona. As
such, the pattern of variation on load-bearing and non load-bearing variables is
different, more uniform on the former than on the latter.
I wish to therefore to reformulate my initial classification of variables as Group
A or B, and instead distinguish load-bearing variables from the other phonologi-
cal variables used in this study. The absence or low frequencies of the appropriate
variant of a load-bearing variable will be negatively evaluated in social contexts
where use of SJE is expected.
I further propose that this asymmetrical attention to variants is possibly a func-
tion of the coexistence of two varieties for speakers, English and Creole, in an
ideological context where use of both is the ideal and not just use of “the stan-
dard”. Models of language situations like Jamaica point to the structural relation-
ship among varieties, i.e. the linguistic continuum, or the functional relationship
among varieties, i.e. diglossia. The variation that occurs here in this JAMPRO
data, suggests that these speakers do not merely shift from the L variety JC to
the H variety SJE when switching from informal to formal situations. Rather,
these are speakers who seem to signal a shift in both variety and function by ma-
nipulating some variables, those load-bearing variables which are indexical of
variety and which therefore signal use of the H code. At the same time, the abil-
ity to use JC, the L code, is signalled by the presence of other non load-bearing
variables because the ability to use both is ideal in the Jamaican social context.
I can abstract four types of variables from my data, reflecting the general pat-
terns of their distribution in the speech of this JAMPRO sample.
Load-bearing and non load-bearing variables can be further distinguished in
terms of whether or not they function as indexing membership in or affiliation
with particular social groups.
155
5 Conclusion
Variables
Load-Bearing Socially Indexical Non Load-Bearing
Type 1 Type 2 Type 3 Type 4
/o/:[o]C /o/:[ɔr] /e/:[e ~ ie]
/h/:[h]
-ed -nt:[nt]V -st:[s ~ st] -n’t:[n ~ nt]
/ɔ/:[ɔ]
/θ/:[θ] /ð/:[ð ~d]
/k/:[kh ]a: /k/:[kh ~ kj]a
[ɔ]+rhotic [a]+/−rhotic
Not used in Jamaican Creole Used in Jamaican Creole
Figure 5.1: Load-bearing and non load-bearing variables in JE
• Type 1 load-bearing variables typically are produced by speakers with very

similar patterns of use. All speakers tended to produce -ed and [o] in goat
and, in fact these features were seldom statistically significant when sub-
ject to the test of association.
• Type 4 non load-bearing variables are also produced with very similar pat-
terns of use. All speakers tended to vary the Type 4 variants [kj ~ kh ] in
cat or [nt ~ n ~ õ] in don’t regardless of how the sample was stratified and
analysed.
• Types 1 and 4 are not usually sensitive to sample stratification. What dif-
ferentiates Type 1 from Type 4 is that the generally used variant of a Type
1 load-bearing variable is not used in JC, while the variant used in a Type
4 non load-bearing variable does occur in JC as well.
• In contrast, Types 2 and 3 variables are typically sensitive to sample strat-
ification. However, they (are used to) index different things in the speech
community.
• Type 2 variables, which are load-bearing, typically can be correlated with
social factors like level of education and social status and therefore also
index an ability to speak SJE.
• Type 3 variables tend to correlate with factors like gender or age without
necessarily also signalling an (in)ability to use SJE.
156
5.2 An endonormative definition of SJE phonological features
Dyer & Beckford-Wassink’s (2001) matched-guise test results, in which a

speaker who clearly used Creole forms was judged to be a likely speaker of JE,
can be explained in terms of her use of load-bearing and non load-bearing vari-
ables1 . The speaker judged as “uses English”, even though she “produced 16 of 19
forms with basilectal variant” (26) in the test, did produce – ed turned and [kh ]
(lexical item unclear).
As there is nothing to suggest that the phonology of a system would have
patterns of variation peculiar to it alone, which are not also evident in other areas
of structure, it is very likely that use of morphosyntactic forms in JE is structured
in a similar way to the phonological variation abstracted from the data presented
here.
5.2 An endonormative definition of SJE phonological

features
The data from this JAMPRO sample reveals that education, social background
and gender are the three social factors that have a significant association with
speakers productions, a not unexpected finding. Crucially, being female, univer-
sity educated and from a relatively affluent background to a great extent reflects
the type of employee JAMPRO hires; senior management explicitly says so. The
JAMPRO data as a whole shows that typically the features that tend to occur
in female speech also tend to distinguish the select employees at JAMPRO, i.e.
the frontline cohort. Female speech also patterns the variant use that charac-
terizes those with more access to (longer) schooling. This suggests either that
the phonology of formal acrolectal varieties is essentially going to be a female/
educated norm, or that women at JAMPRO approximate more to the norms con-
sidered to be acrolectal in the speech community. It is perhaps one reason why
Senior Management at JAMPRO articulates a preference for female employees
when selecting from interviewed candidates.
Generally, there are very few, if any, phonological features that are not shared
with other varieties along the Jamaican linguistic continuum. What distinguishes
acrolectal speech is the frequency of use of these features and what characterizes
it is a low production of stigmatized Creole items. This suggests that the acrolect,
and here I mean the spoken SJE, will be described as either having [h], [θ] and the
like or as not having the Creole reflex of these features. It cannot be described as
being typified by an absence of [kja] or [ð] them or even mid-vowel diphthongs,
1 This data is discussed in Chapter 1, Table 1.1.
157
5 Conclusion
although it is very probable that if asked, speakers will describe an idealized SJE
that is devoid of these features.
The acrolect, as actual spoken SJE, cannot therefore be located by identifying
MSE forms; nor can the features along the linguistic continuum be characterized
with reference to MSE without leading to paradoxical conclusions or misinter-
pretations of data. Indeed, an idealized acrolect as “local standard” can also be
problematic. This can be illustrated by looking again at the example of Meade’s
work (2001), cited at the beginning of this study. In his study MSE phonology
was used to determine what he labels acrolectal phonology. But his data on use
of dental fricatives and their variants makes no distinction between voiced and
voiceless forms. It is entirely possible, therefore, that his sample of Jamaican chil-
dren with educated caregivers are being exposed to [ð ~ d]. The data I have would
suggest that it is the pattern on the voiceless fricative, or other such load-bearing
variables, that would need to be examined to determine a time frame for acqui-
sition of JE phonology.
The speaker’s notion that two varieties exist in Jamaica, JC and JE, is therefore
important. The form English takes in the Jamaican social context, particularly its
pronunciation, is shaped in part by the idea speakers have of what Creole is. In an
informal survey of 42 students at the University of the West Indies, when asked
how they know someone is speaking good English, most responded by saying
“because they are not” - dropping aitches, saying ‘mi’, ‘did’ a and so on, listing a
number of stereotypical Creole practices. For them, English is, in part, negatively
defined in relation to what Creole is believed to be and is assessed in terms of
the speaker’s avoidance of those characteristic items. Wolfram & Schilling-Estes
(1998: 12) comment on similar definitions of the standard in the US. But there are
other items that are present in the acrolect that have been typically associated
with Creole that speakers use as frequently as, if not more than, the variants
which also appear in MSE.
Patrick (1999: 9) observes, of his urban, mesolectal sample, that the speakers
he interviewed all seem to have
a rich and nuanced ability to vary their speech (…) but a very limited set
of metalinguistic labels for it, essentially binary oppositions of which the
bluntest instrument is the English/patois distinction.
He does not take the strong position of DeCamp, who described this belief
as a persistent myth (1971: 350), but argues that this distinction, though psycho-
logically powerful, should not form the basis of linguistic description. But the
importance of the speaker’s distinction should not be underestimated either. It
158
5.2 An endonormative definition of SJE phonological features
is central to the way variation patterns asymmetrically on a number of phonolog-

ical variables. Moreover, issues like language acquisition and language change in
the Jamaican context are going to be affected by what speakers think is Creole
and English, to the extent that they are trying to target the latter. For them, target-
ing English is not reproducing foreign norms.2 Many speakers here for example,
notably women and the secondary educated, are avoiding affricates, perceived
to be Creole, and producing [tj] in culture type words.
JAMPRO was specifically selected because of its role as a one of the Jamaican
state’s promotion agencies, conducting its business in a context that it explicitly
states is reserved for Standard English. JAMPRO is one of many agencies that
select receptionists, broadcasters, managers and the like and reject or pass oth-
ers over. In its practice of selecting some staff for frontline positions, Jamaicans
who represent the agency in this milieu, JAMPRO is implicitly favouring some
patterns of speech over others and determining which features are acceptable or
unacceptable as spoken SJE.
The data here demonstrate that a number of social and linguistic factors in-
form their selection. Frontline staff are typically highly educated, they come from
households more likely to be JC/JE speaking and they are perceived by others in
their work environment to be selected because they can represent JAMPRO well,
socially and in the way they speak English. Arguably, frontline staff reflect the
agency’s idea of “best speakerhood” (Silverstein 1996: 286) – at least of the upper
end of the bidialectal/diglossic construct JAMPRO says exists in Jamaica. Their
speech shows the pattern of variation uncovered in this study – little if any voice-
less TH stopping, but freely varied [d ~ ð]; a virtual absence of [kja:], but frequent
use of [kja]; little use of diphthongal variants for the back mid-vowel and vari-
ation between [e ~ ie], especially pre-rhotically. In addition, frontline staff are
more likely to produce [ɔ] – whether in stressed syllables or in words ending –
tion – and seldom h-drop.
I suggest that this frontline data presents an actual example of what “good En-
glish” is in Jamaica. Speakers who seek certain kinds of jobs will perhaps target
features like [ʃɔn], or even [kja]; certainly they will identify [h] and [ɔ] as as-
pects of speech that the more successful use, not because they belong to British
or American varieties of English but because in Jamaica these are features of
speakers presented as having the right diction. Speakers who seek certain kinds
of jobs will carefully avoid voiceless TH stopping, [uo], [kja:] because these are
“Creole”, not because British or American speakers do not use them. And they
2 An early discussion of this preference for and use of local norms is found in Eersel’s study
(1971: 320) of educated Surinamese use of Dutch.
159
5 Conclusion
will perhaps freely vary word final [st ~ s] in a word like cast, [d ~ ð] or [e ~ ie]
because they do not necessarily index “bad English”.
It is important therefore to separate discussion of the (historical) linguistic
processes that may have accounted for most of the forms I have described above,
from a speaker-driven endonormative discussion of the Jamaican acrolect. Un-
doubtedly, hypercorrection, distancing from Creole or Creole influence, may ex-
plain the presence of most of the features in the production of informants. But
the idea speaker hold of SJE, the acrolect, as it occurs now will come from their
perceptions of who is successful and how they speak and who is presented as
having a good command of the English language in the Jamaican social context.
It cannot be assumed, however implicitly, to be a foreign model of English be-
cause English is not a foreign language in Jamaica. Mühleisen’s (2002) criticism
that “CELCs [Caribbean English-lexicon creoles] are (…) never defined in isola-
tion but always ’in comparison with’ the high prestige language English” (74)
is made more pointed if that English is itself situated outside of the speaker’s
community.
It is not clear how we can proceed to discuss issues of decreolization, or of lan-
guage change generally, with assumptions about the acrolect, or for that matter
the basilect, that are not based on the norms of speakers in their speech com-
munities. Jamaican English is, after all, a national variety (or set of varieties) in
its own right. The speakers in this speech community operate in a social context
where their two languages, Jamaican English and Jamaican Creole, have coex-
isted for centuries. The two are perceived as distinct, but connected elements
of being Jamaican. The data presented here demonstrates both – the speaker’s
differentiation and integration of JE and JC in the Jamaican diasystem.
160
Appendix A: Prepared script for all
interviews of JAMPRO
informants
104 informants were interviewed using this script. 82 of these interviews were
recorded and transcribed phonetically for later sociolinguistic analysis.
1. Was the informant told of the purpose of the interview? 13 (yes)
A.1 Personal data collected

1. Sex: 82 (female), 22 (male)
2. Age: 35 (20–29), 33 (30–39), 16 (40–49), 20 (50–65)
3. Residence: 65 (Portmore), 4 (Franklin Town), 21 (Liguanea), 9 (Red Hills),

4 (Stony Hills), 1 (no data)
4. Race: 53 (black), 35 (brown), 1 (Chinese), 1 (Indian), 14 (no data noted)
5. Years at JAMPRO: 30 (10+), 38 (5+), 28 (–5), 8 (new)
6. Level of Education: 4 (primary), 10 (secondary), 43 (post-secondary), 22

(tertiary), 24 (graduate)
7. Salary Scale: 7 (ancillary), 31 (secretarial), 42 (professional), 10 (director), 14

(senior management)
8. Spouse’s Occupation: 1 (“cleaner”), 7 (“artisan”), 23 (“teacher”), 11 (“doctor”),

2 (business), 60 (none)
9. Parent’s Occupation: 17 (“cleaner”), 39 (“artisan”), 30 (“teacher”), 8 (“doc-

tor”), 6 (business) 4 (no data)
10. Floor: 32 (1), 17 (2), 14 (3), 28 (4), 10 (5), 3 (overseas posting)

A Prepared script for all interviews of JAMPRO informants
11. Job Description: 24 (frontline - local clients), 39 (frontline - local & foreign
clients), 41 (no clients)
12. Employment History: 26 (never promoted), 37 (one promotion), 30 (multi-
ple), 11 (no data)
13. Transport to Work: 30 (bus), 50 (own car), 24 (lift)
14. Organizational Section: 24 (JNIP), 15 (JNEC), 3 (JIDC), 1 (JNIC), 61 (none)
A.2 Data on patterns of workplace interaction

1. Which members of staff do you lunch with?
2. Which members of staff live in your residential community?
3. Are any members of staff related to you?
4. Which members of staff do you consider friends?
5. Which members of staff do you see outside of work?
6. Are you active on the Staff Association (JSA, hereafter)?
7. Do you hold any (elected) position on the JSA?
A.3 Data on working at JAMPRO

1. What type of person does well and moves ahead in JAMPRO?
2. Why do you think staffers feel as they do about the JSA?
3. Who is responsible for your promotion?
4. Do you see yourself here in 5 or 10 years? 36 (yes), 50 (no), 18 (don’t know)
5. How important do you think these are in determining who does well here:
6. gender (36.5%) b) colour (16%) c) class (40%) d) education (87.5%) e) other
(15%) (These percentages are calculated out of a possible 104 mentions of a
particular social factor.)
7. Which levels in JAMPRO do you think have authority and constitute top
management?
162
Appendix B: Profiles of JAMPRO
informants
Table B.1: Profiles of male JAMPRO informants
Males
Informant # Education Parent’s Age Status
5 tertiary tradesman 38 5 (frontline)
8 tertiary teacher 38 4 (frontline)
9 tertiary builder no data 6 (frontline)
14 tertiary carpenter 51 7 (frontline)
15 graduate farmer 42 7
18 primary farmer 52 1
20 graduate tailor 35 6 (frontline)
27 secondary farmer 33 1 (frontline)
29 graduate teacher 55 5 (frontline)
34 secondary+ civil servant 29 5 (frontline)
40 primary mason 51 1 (frontline)
47 secondary supervisor 58 1 (frontline)
49 graduate farmer 27 5
54 graduate messenger 32 5
86 graduate salesman 44 6
88 secondary+ shoemaker 42 5 (frontline)
89 tertiary shopkeeper 45 6 (frontline)
92 tertiary farmer 27 4
94 secondary+ helper (maid) 26 4
100 secondary+ dressmaker 28 5 (frontline)
101 graduate teacher 60+ 7 (frontline)
102 graduate no data 49 7 (frontline)
B Profiles of JAMPRO informants
Table B.2: Profiles of female JAMPRO informants
Females
2 secondary+ no data 29 3 (frontline)
4 secondary higgler 58 1
6 secondary+ nurse 25+ 3 (frontline)
7 graduate no data 40+ 5 (frontline)
10 tertiary contractor 35+ 5 (frontline)
11 tertiary farmer 31 4 (frontline)
12 tertiary engineer 45 6 (frontline)
13 tertiary business man 40+ 6 (frontline)
16 graduate bursar 33 5 (frontline)
17 tertiary overseer 40 2 (frontline)
19 secondary self-employed 35+ 3 (frontline)
21 secondary+ dressmaker 30 3
22 tertiary civil servant 30 4 (frontline)
23 secondary+ farmer 30 4 (frontline)
25 secondary+ cleaner 27 3 (frontline)
28 tertiary pub. health insp. 38 6 (frontline)
30 secondary+ mason 24 4 (frontline)
31 secondary+ factory worker 25 3
32 secondary+ civil servant 50+ 7 (frontline)
33 graduate teacher 50+ 6 (frontline)
35 tertiary business man 37 6 (frontline)
36 graduate financial contr. 40+ 6 (frontline)
37 secondary+ carpenter 33 3 (frontline)
38 secondary+ clerk 30+ 4 (frontline)
39 secondary+ supervisor 27 3 (frontline)
41 secondary+ farmer 28 3
42 tertiary civil servant 39 6 (frontline)
43 secondary+ business man 27 3
44 secondary+ clerk 25 3
45 secondary bus conductor 22 3
164
46 tertiary teacher 32 5
48 secondary+ tobacconist 63 3
50 secondary+ nurse 40 4 (frontline)
51 graduate civil servant 52 6 (frontline)
53 secondary farmer 36+ 1
55 graduate cashier 42 7 (frontline)
56 graduate insurance agent 31 6 (frontline)
57 secondary+ prison warder 52 5 (frontline)
58 tertiary pub. health insp. 40 5
59 secondary+ butcher 60 5 (frontline)
60 secondary+ construction 26 2
61 secondary+ farmer 21 no data
62 tertiary plumber 37 5 (frontline)
63 secondary electrician 23 3
64 tertiary pastor 33 6 (frontline)
65 tertiary dressmaker 43 3 (frontline)
66 secondary+ secretary 37 5
67 secondary+ cook 30 3 (frontline)
68 tertiary meteorologist 28 5 (frontline)
70 secondary+ policeman 22 3 (frontline)
71 secondary+ business man 26 3
72 secondary draftsman 23 5
73 secondary+ carpenter 29 3
74 primary labourer 50 1
75 graduate engineer 33 6 (frontline)
76 secondary+ plumber 28 2 (frontline)
77 no data tailor 57 5
78 secondary+ tiler 26 2 (frontline)
79 tertiary teacher 25 4
80 secondary+ supervisor 32 4
82 secondary+ policeman 24 3
83 graduate manager 26 6
84 secondary+ no data 23 3
85 secondary business man 27 3
165
B Profiles of JAMPRO informants

87 graduate farmer 41 7 (frontline)
90 secondary+ matron 26 4 (frontline)
91 secondary+ farmer 27 2
95 secondary+ self-employed 54 2
96 primary cleaner 51 1
97 secondary+ teacher 23 2
98 secondary+ clerk 52 4 (frontline)
99 secondary+ doctor 31 4
103 tertiary policeman 43 3 (frontline)
105 graduate doctor 30 4 (frontline)
166
Appendix C: Parent’s Occupation
The following is a detailed list of parent’s occupations given by informants in
this study. The data has been grouped according to the five headings used in the
body of the study.
Table C.1: List of parent’s occupations
“Cleaner” “Artisan” “Teacher” “Doctor” “Business”

small farmer carpenter teacher engineer contractor
labourer plumber policeman physician shopkeeper
cleaner butcher draftsman manager pastor
messenger tailor superintendent accountant overseer
driver mason nurse financial controller
cook electrician meteorologist
tobacconist clerk
dressmaker supervisor
shoemaker public health inspector
tiler warder
References
Akers, Glen. 1977. Phonological variation in the Jamaican continuum. Harvard
University dissertation.
Akers, Glen. 1981. Phonological variation in the Jamaican continuum. Ann Arbor:
Karoma Publishers.
Alexander, J. 1977. The culture of race in middle-class Kingston, Jamaica. Ameri-
can Ethnologist 4(3). 413–435.
Alleyne, Mervyn C. 1980. Comparative Afro-American. Ann Arbor: Karoma Pub-
lishers.
Allsopp, Richard. 1996. Dictionary of Caribbean English usage. Oxford: Oxford
University Press.
Alvesson, Mats. 1993. Cultural perspectives on organizations. Cambridge: Cam-
bridge University Press.
Austin, Diane. 1983. Culture and ideology in the English speaking Caribbean: A
view from Jamaica. American Ethnologist 10(2). 223–240.
Bailey, Beryl. 1966. Jamaican Creole syntax. Cambridge: Cambridge University
Press.
Bailey, Beryl. 1971. Jamaican Creole: Can dialect boundaries be defined? In Dell
Hymes (ed.), Pidginization and creolization of languages, 341–348. Cambridge:
Cambridge University Press.
Barrow, Steve & Peter Dalton. 1997. Reggae, the Rough Guide. London: Rough
Guides Ltd.
Baugh, Albert & Thomas Cable. 1993. A history of the English language. London:
Routledge.
Beal, Joan. 2002. English pronunciation in the eighteenth century. Oxford: Oxford
University Press.
Beckford‑Wassink, Alicia. 1999a. A sociophonetic analysis of Jamaican vowels. Uni-
versity of Michigan dissertation.
Beckford‑Wassink, Alicia. 1999b. Historic low prestige and seeds of change: Atti-
tudes toward Jamaican Creole. Language in Society 28. 57–92.
Beckford‑Wassink, Alicia. 2001. Theme and variation in Jamaican vowels. Lan-
guage Variation and Change 13. 135–159.
References
Beckles, Hilary. 2000. A “riotous and unruly lot”: Irish indentured servants and
freemen in the English West Indies, 1644–1713. In Verene Shepherd & Hilary
Beckles (eds.), Caribbean slavery in the Atlantic world, 226–238. Kingston: Ian
Randle.
Bell, Allan. 1984. Language style as audience design. Language in Society 2. 145–
204.
Bell, Allan. 2001. Back in style: Reworking audience design. In Penny Eckert &
John Rickford (eds.), Style and sociolinguistic variation, 139–169. Cambridge:
Bickerton, Derek. 1973. On the nature of a Creole continuum. Language 49(3).
640–669.
Bickerton, Derek. 1975. Dynamics of a Creole system. Cambridge: Cambridge Uni-
versity Press.
Bobda, Augustin Simo. 2001. East and Southern African English accents. World
Englishes 20(3). 269–284.
Bourdieu, Pierre. 1984. Distinction. London: Routledge.
Bourdieu, Pierre. 1991. Language and symbolic power. Cambridge: Polity Press.
Brereton, Bridget. 1995. Text, testimony and gender: An examination of some
texts by women on the English-speaking Caribbean, from the 1770’s to the
1920’s. In Bridget Brereton Verene Shepherd & Barbara Bailey (eds.), Engen-
dering history, 63–93. Kingston: Ian Randle Publishers.
Brodber, Kathryn. 1989. Standard English in Jamaica: A case of competing norms.
English World-Wide 10(1). 41–53.
Bryan, Patrick. 1996. The black middle class in 19th century Jamaica. In Hilary
Beckles & Verene Shepherd (eds.), Caribbean freedom, 284–296. Princeton:
Markus Wiener Publishers.
Bryan, Patrick. 2000. The Jamaican people, 1880–1902: Race, class and social control.
Kingston: The University of the West Indies Press.
Busby, Margaret. 1992. Daughters of Africa. New York: Ballantine Books.
Cameron, Deborah. 2000. Styling the worker: Gender and the commodification
of language in the globalized service economy. Journal of Sociolinguistics 4(3).
323–347.
Campbell, Carl. 1996. Social and economic obstacles to the development of pop-
ular education in post-emancipation Jamaica, 1834–1865. In Hilary Beckles
& Verene Shepherd (eds.), Caribbean freedom, 262–268. Princeton: Markus
Wiener Publishers.
Carmichael, A.C. 1833. Domestic manners and social condition of the White,
Coloured, and Negro population of the West Indies. London: Whittaker, Treacher.
Also published in New York by Negro Universities Press, 1969.
170
Carpenter, Joseph E. 1868. A handbook of poetry. London: Sampson Low, Son, &
Marston.
Cassidy, Frederic. 1961. Jamaica talk. London: Macmillan.
Cassidy, Frederic & Robert Le Page. 1967. Dictionary of Jamaican English. Cam-
bridge: Cambridge University Press.
Chaudenson, Robert. 2001. Creolization of language and culture. London: Rout-
ledge.
Cheshire, Jenny. 1999. Spoken Standard English. In Tony Bex & Richard Watts
(eds.), Standard English: The widening debate, 129–147. London: Routledge.
Christie, Pauline. 1989. Questions of standards and intraregional differences in
Caribbean examinations. In Ofelia Garcia & Ricardo Otheguy (eds.), English
across cultures: Cultures across English. 243–262. Amsterdam: Benjamins.
Christie, Pauline. 1998a. History and language. UWILING, Working Papers in Lin-
guistics 3. 74–90.
Christie, Pauline. 1998b. Trends in Jamaican English: Increasing deviance or
emerging standards? UWILING, Working Papers in Linguistics 3. 19–34.
Christie, Pauline. 2003. Language in Jamaica. Kingston: Arawak Publications.
Coye, Dale. 1998. Orthoepic piracy: Spelling pronunciations and Standard En-
glish. American Speech 73(2). 178–196.
Craig, Dennis. 1982. Toward a description of Caribbean English. In Braj Kachru
(ed.), The other tongue: English across cultures, 198–209. Urbana: University of
Illinois Press.
Crompton, Rosemary. 1993. Class and stratification: An introduction to current de-
bates. Cambridge: Polity Press.
DeCamp, David. 1961. Social and geographic factors in Jamaican dialects. In
Robert Le Page (ed.), Creole language studies 2. 61–84. London: Macmillan.
DeCamp, David. 1971. Toward a generative analysis of a post-Creole speech com-
munity. In Dell Hymes (ed.), Pidginization and creolization of languages, 340–
376. Cambridge: Cambridge University Press.
Deuchar, Margaret. 1988. A pragmatic account of women in their speech com-
munities. In Jennifer Coates & Deborah Cameron (eds.), Women in their speech
communities, 27–32. London: Longman.
Devonish, Hubert. 1978. The selection and codification of a widely understood and
publicly usable language variety in Guyana, to be used as a vehicle for national
development. University of York dissertation.
Devonish, Hubert. 1992. On the existence of autonomous language varieties in
‘Creole continuum situations’. In Pauline Christie, Barbara Lalla, Velma Pollard
& Lawrence Carrington (eds.), Studies in Caribbean language II. 178–196. St.
Augustine: Society for Caribbean Linguistics.
171
References
Devonish, Hubert. 2002. Talking rhythm, stressing tone. Kingston: Arawak Publi-
cations.
Devonish, Hubert. 2003. Language advocacy and ‘conquest’ diglossia in the ‘an-
glophone’ Caribbean. In Christian Mair (ed.), The politics of English as a world
language, 157–177. Amsterdam: Rodopi B. V.
Devonish, Hubert & Otelemate Harry. 2004. Jamaican Creole and Jamaican En-
glish phonology. In Edgar Schneider (ed.), Varieties of English, volume 1, the
Americas and the Caribbean, 450–480. Berlin: Mouton de Gruyter.
Devonish, Hubert & Walter Seiler. 1991. A reanalysis of the phonological system
of Jamaican Creole. Society for Caribbean Linguistics Occasional Paper 24.
Dubois, Sylvie & Barbara Horvath. 1998. Let’s tink about dat: Interdental frica-
tives in Cajun English. Language Variation and Change 10(3). 245–261.
Dyer, Judy & Alicia Beckford-Wassink. 2001. Taakin braad and talking broad:
Changing indexicality of phonological variants in two language contact situa-
tions. Paper presented at the University of the West Indies (Mona), Kingston,
May 7, 2002.
Eersel, Christian. 1971. Prestige in choice of language and linguistic form. In Dell
Hymes (ed.), Pidginization and creolization of languages, 317–322. Cambridge:
Ervin-Tripp, Susan. 2001. Variety, style shifting, and ideology. In Penny Eckert
& John Rickford (eds.), Style and sociolinguistic variation, 44–56. Cambridge:
Escure, Genevieve. 1997. Creole and dialect continua (Creole Language Library 18).
Amsterdam: John Benjamins Publishing Company.
Escure, Genevieve. 2000. Neva: From Caribbean Creoles to meso-acrolects. In,
vol. 2000, 141–145. Papers presented at the biennial meeting for the Society for
Caribbean Linguistics 2000, Kingston, University of the West Indies (Mona),
August 16–19.
Fisher, John H. 2001. British and American, continuity and divergence. In John
Algeo (ed.), The Cambridge history of the English language, vol. VI, English in
North America, 59–85. Cambridge: Cambridge University Press.
Freeborn, Dennis. 1998. From Old English to Standard English. London: Macmil-
lan.
Gbedemah, G. 1995. Fun with phonics: Skill-building. 7th edition. Kingston: Car-
long Publishers.
Giegerich, Heinz. 1992. English phonology. Cambridge: Cambridge University
Press.
Giles, Howard & Peter Powesland. 1975. Speech style and social evaluation. New
York: Harcourt Brace.
172
Gimson, A. C. 1980. An introduction to the pronunciation of English. London: Ed-
ward Arnold.
Gordon, Derek. 1986. Class, status and social mobility in Jamaica. Kingston: ISER.
Görlach, Manfred. 1987. Colonial lag? The alleged conservative character of
American English and other ‘colonial’ varieties. English World-Wide 8(1). 41–
60.
Graff, David, William Labov & Wendell Harris. 1986. Testing listener’s reactions
to phonological markers of ethnic identity: A new method for sociolinguistic
research. In David Sankoff (ed.), Diversity and diachrony, 45–58. Amsterdam:
John Benjamins Publishing Company.
Green, Lisa. 2002. African American English. Cambridge: Cambridge University
Press.
Grint, Keith. 1991. The sociology of work. Cambridge: Polity Press.
Gupta, Anthea. 2001. Realism and imagination in the teaching of English. World
Englishes 20(3). 365–381.
Guy, Gregory. 1980. Variation in the group and the individual: The case of final
stop deletion. In William Labov (ed.), Locating language in time and space, 1–36.
New York: Academic Press.
Hancock, Ian. 1994. Componentiality and the Creole matrix: The Southwest En-
glish contribution. In Michael Montgomery (ed.), The crucible of Carolina: Es-
says in the development of Gullah language and culture, 95–114. Athens: Uni-
versity of Georgia Press.
Hernández-Campoy, Juan & José Jiménez-Cano. 2003. Broadcasting standardis-
ation: An analysis of the linguistic normalisation process in Murcian Spanish.
Journal of Sociolinguistics 7(3). 321–347.
Hewitt, Roger. 1986. White talk Black talk. Inter-racial friendship and communica-
tion amongst adolescents. Cambridge: Cambridge University Press.
Ho, Mian-Lian & John Platt. 1993. Dynamics of a contact continuum: Singaporean
English. Oxford: Oxford University Press.
Holm, John. 1988. Pidgins and creoles: Volume 1, theory and structure. Cambridge:
Holmes, Janet. 1997. Setting new standards: Sound changes and gender in New
Zealand English. English World‑Wide 18(1). 107–142.
Horvath, Barbara & David Sankoff. 1987. Delimiting the Sydney speech commu-
nity. Language in Society 16. 179–204.
Hudson, Richard. 1980. Sociolinguistics. Cambridge: Cambridge University Press.
Hymes, Dell. 1974. Foundations in sociolinguistics: An ethnographic approach.
Philadelphia: University of Pennsylvania Press.
173
References
Irvine, Alison. 1988. A study of the linguistic markers of social differentiation in

two jamaican communities. M. Phil. dissertation, University of York.
Irvine, Alison. 1994. Dialect variation in Jamaican English: A study of the phonol-
ogy of social group marking. English World Wide 15(1). 55–78.
Irvine, Judith & Susan Gal. 2000. Language ideology and linguistic differentiation.
In Paul Kroskrity (ed.), Regimes of language, 35–83. New Mexico: School of
American Research Press.
Jamaica Tourist Board. 2001. Jamaica: Visitor guide.
JAMPRO. 1992. Invest Jamaica.
JAMPRO. 1993. Choose Jamaica.
JAMPRO. 1994. Jamaica: The investment capital of the Caribbean.
JAMPRO. 1996. Jamaica: The investment capital of the Caribbean.
JAMPRO. 1997a. Business traveller’s guide to Kingston. Volume 1. 1–10.
JAMPRO. 1997b. Travel agent reference guide.
Janda, Richard & Julie Auger. 1992. Quantitative evidence, qualitative hypercor-
rection, sociolinguistic variables – and French speakers’ ’eadhaches with en-
glish /h/. Language and Communication 12. 195–236.
Jenkins, R. 1985. Black workers in the labour market: The price of recession. In
Ruth Finnegan Bryan Roberts & Duncan Gallie (eds.), New approaches to eco-
nomic life, 169–183. Manchester: Manchester University Press.
JIDC. 1957. New opportunities for industrial development.
JIDC. 1961. Opportunities for industrial investment.
Judd, Karen. 1998. The name of the people: Populist ideology and expatriate
power in Belize. In Howard Johnson & Karl Watson (eds.), The White minority
in the Caribbean, 133–158. Kingston: Ian Randle.
Kachru, Braj. 1982. Models for non-native Englishes. In Braj Kachru (ed.), The
other tongue: English across cultures, 31–57. Urbana: University of Illinois Press.
Kirwin, William. 2001. Newfoundland English. In John Algeo (ed.), The Cambridge
history of the English language, vol. VI, English in North America, 441–455. Cam-
Krapp, George Philip. 1925. The English language in America: Volume 11. New York:
The Century Company.
Kroskrity, Paul. 2000. Regimenting languages: Language ideological perspectives.
In Paul Kroskrity (ed.), Regimes of language, 1–34. New Mexico: School of
American Research Press.
Kulick, Don. 1998. Age, gender, language shift and the politics of revelation in
a Papua New Guinea village. In Bambi Schieffelin, Kathryn Woolard & Paul
Kroskrity (eds.), Language ideologies, 87–102. Oxford: Oxford University Press.
174
Labov, William. 1966. The social stratification of English in New York City. Wash-
ington: Center for Applied Linguistics.
Labov, William. 1968. The reflection of social processes in linguistic structures.
In Joshua Fishman (ed.), Readings in the sociology of language, 240–251. The
Hague: Mouton.
Labov, William. 1972. Sociolinguistic patterns. Philadelphia: University of Penn-
sylvania Press.
Labov, William. 1980. The social origins of sound change. In William Labov (ed.),
Locating language in time and space, 251–265. New York: Academic Press.
Labov, William. 2001. The anatomy of style-shifting. In Penny Eckert & John Rick-
ford (eds.), Style and sociolinguistic variation, 85–108. Cambridge: Cambridge
University Press.
LaCharité, Darlene. 1996. The segmental status of [st]: Evidence from basilec-
tal and mesolectal Jamaican Creole. In Päivi Koskinen (ed.), Proceedings of the
1995 annual conference of the Canadian Linguistic Association. Toronto: Toronto
Working Papers in Linguistics.
Lalla, Barbara & Jean D’Costa. 1989. Voices in exile. Jamaican texts of the 18th and
19th centuries. Tuscaloosa: The University of Alabama Press.
Lalla, Barbara & Jean D’Costa. 1990. Language in exile: Three hundred years of
Jamaican Creole. Tuscaloosa: The University of Alabama Press.
Le Page, Robert. 1960. Jamaican Creole. In Robert Le Page (ed.), Creole language
studies 1. Cambridge: Cambridge University Press.
Le Page, Robert. 1968. Problems to be faced in the use of English as the medium
of education in four West Indian territories. In Joshua Fishman (ed.), Language
problems of developing nations, 431–442. London: John Wiley & Sons.
Le Page, Robert. 1980. Projection, focussing, diffusion, or steps towards a soci-
olinguistic theory of language. York Papers in Linguistics 9.
Le Page, Robert. 1988. Some premises concerning the standardization of lan-
guages, with special reference to Caribbean Creole English. International Jour-
nal of the Sociology of Language 71. 25–36.
Le Page, Robert & Andrée Tabouret-Keller. 1985. Acts of identity. Cambridge:
Lehmann, Winfred. 1973. Historical linguistics. New York: Holt, Rinehart & Win-
ston.
Lewis, Gordon. 1968. The growth of the modern West Indies. New York: Monthly
Review Press.
Lippi-Green, Rosina. 1997. English with an accent. London: Routledge.
Long, Edward. 1774. The history of Jamaica. London: T. Lowndes.
175
References
Marshall, Hayden. 1983. Unpublished BA Caribbean study. University of the West

Indies (Mona).
McArthur, Tom. 1992. The BBC and spoken English. Tom McArthur (ed.). Oxford:
Oxford University Press.
McArthur, Tom. 1998. The English languages. Cambridge: Cambridge University
Press.
Meade, Rocky R. 2001. Acquisition of Jamaican phonology. LOT: Netherlands
Graduate School of Linguistics.
Miller, Errol. 1989. Educational development in independent Jamaica. In Rex Net-
tleford (ed.), Jamaica in independence, 205–227. Kingston: Heinemann Publish-
ers.
Miller, Errol. 1991. Men at risk. Kingston: Jamaica Publishing House Ltd.
Miller, Faye. 1987. Acrolectal Jamaican speech: A descriptive analysis of its phonol-
ogy and morpho-syntax. University of the West Indies (Mona) M.Phil. thesis.
Milroy, James. 1999. The consequences of standardisation in descriptive linguis-
tics. In Tony Bex & Richard Watts (eds.), Standard English: The widening debate,
16–39. London: Routledge.
Milroy, James & Lesley Milroy. 1985. Authority in language. London: Routledge.
Milroy, Lesley. 1980. Language and social networks. Oxford: Basil Blackwell.
Milroy, Lesley. 2002. Introduction: Mobility, contact and language change –
working with contemporary speech communities. Journal of Sociolinguistics
6,1 1. 3–15.
Milroy, Lesley & Matthew Gordon. 2003. Sociolinguistics: Method and interpreta-
tion. Oxford: Blackwell Publishing.
Ministry of Education and Culture, Jamaica. 1999. Revised primary curriculum.
Government of Jamaica/Inter-American Development Bank, Primary Educa-
tion Improvement Programme.
Moag, Rodney. 1982. The life cycle of non-native Englishes: A case study. In Braj
Kachru (ed.), The other tongue: English across cultures, 270–288. Urbana: Uni-
versity of Illinois Press.
Mufwene, Salikoko. 1996. The founder principle in Creole genesis. Diachronica
13. 83–134.
Mufwene, Salikoko. 2001. The ecology of language evolution. Cambridge: Cam-
bridge University Press.
Mugglestone, Lynda. 1995. ‘Talking proper’: The rise of accent as social symbol.
Oxford: Clarendon Press.
Mühleisen, Susanne. 2002. Creole discourse (Creole Language Library 24). Ams-
terdam: John Benjamins Publishing Company.
176
Nair-Venugopal, Shanta. 2001. The sociolinguistics of choice in Malaysian busi-
ness settings. International Journal of the Sociology of Language 152. 21–52.
Nettleford, Rex. 1970. Mirror, mirror. Jamaica: William Collins & Sangster’s Ltd.
Nettleford, Rex. 1972. Identity, race and protest in Jamaica. New York: William
Morrow & Company.
Neu, Helene. 1980. Ranking of constraints on /t,d/ deletion in American English:
A statistical analysis. In William Labov (ed.), Locating language in time and
space, 37–54. New York: Academic Press.
Nichols, Patricia. 1983. Linguistic options and choices for Black women in the ru-
ral South. In Barrie Thorne, Cheris Kramarae & Nancy Henley (eds.), Language,
gender and society. Rowley, MA: Newbury House.
Paolillo, John. 1997. Sinhala diglossia: Discrete or continuous variation? Language
in Society 26. 269–296.
Patrick, Peter. 1997. Style and register in Jamaican Patwa. In Edgar Schneider (ed.),
Englishes around the world 2. 41–55. Amsterdam: John Benjamins Publishing
Company.
Patrick, Peter. 1999. Urban Jamaican Creole. Philadelphia: John Benjamins Pub-
lishing Company.
Patrick, Peter. 2000. Social status and mobility in urban Jamaican Patwa. In Pro-
ceedings of the 13th biennial conference of the Society for Caribbean Linguistics.
Kingston, Jamaica: University of the West Indies.
Patrick, Peter. 2002. Caribbean Creoles and the speech community. Society for
Caribbean Linguistics Occasional Papers 30.
Patterson, Orlando. 1973. The sociology of slavery. London: Associated University
Press.
Pegge, Samuel. 1814. Anecdotes of the English language: Chiefly regarding the local
dialect of London and its environs. London: Nichols, son, & Bentley.
Planning Institute of Jamaica (PIOJ). 2000. Jamaica: Human development report
2000. Kingston: PIOJ.
Pollard, Velma. 1998. The lexicon of dread talk in standard Jamaican English. In
Velma Pollard Pauline Christie Barbara Lalla & Lawrence Carrington (eds.),
Studies in Caribbean language II. 178–196. St. Augustine: Society for Caribbean
Linguistics.
Preston, Dennis. 1989. Sociolinguistics and second language acquisition. Oxford:
Blackwell.
Pyles, Thomas & John Algeo. 1993. The origins and development of the English
language. Fort Worth: Harcourt Brace Jovanovich.
Rickford, John. 1987. Dimensions of a Creole continuum. Stanford: Stanford Uni-
versity Press.
177
References
Rickford, John & Faye McNair-Knox. 1994. Addressee- and topic- influenced
styleshift: A quantitative sociolinguistic study. In Douglas Biber & Edward
Finegan (eds.), Sociolinguistic perspectives on register, 235–276. Oxford: Oxford
University Press.
Roberts, Peter. 1988. West Indians & their language. Cambridge: Cambridge Uni-
versity Press.
Romaine, Suzanne. 2001. Contact with other languages. In John Algeo (ed.), The
Cambridge history of the English language, vol. VI, English in North America,
59–85. Cambridge: Cambridge University Press.
Sahgal, Anju & Rama Kant Agnihotri. 1985. Syntax: The common bond. Accept-
ability of syntactic deviances in Indian English. English World-Wide 6(1). 117–
129.
Santa Ana, Otto. 1991. Phonetic simplification processes in the English of the Barrio:
A cross-generational sociolinguistic study of the Chicanos of Los Angeles. Univer-
sity of Pennsylvania dissertation.
Schneider, Edgar. 1998. Negation patterns and the cline of Creoleness in English-
oriented varieties in the Caribbean. In Velma Pollard Pauline Christie Barbara
Lalla & Lawrence Carrington (eds.), Studies in Caribbean language II. 204–227.
St. Augustine: Society for Caribbean Linguistics.
Scollon, Ron. 1997. Handbills, tissues and condoms: A site of engagement for the
construction of identity in public discourse. Journal of Sociolinguistics 1(1). 39–
61.
Sealey, Alison & Bob Carter. 2001. Social categories and sociolinguistics: Apply-
ing a realist approach. International Journal of the Sociology of Language 152.
1–19.
Sharwood-Smith, Michael. 1999. British shibboleths. In Eddie Ronowicz & Colin
Yallop (eds.), English: One language, different cultures, 46–82. London: Cassell.
Sherlock, Philip & Hazel Bennett. 1998. The story of the Jamaican people. Kingston:
Ian Randle Publishers.
Shields-Brodber, Kathryn. 1996. Old skeleton, new skin: The relationship between
open syllable structure and consonant clusters in Standard Jamaican English.
In Pauline Christie (ed.), Caribbean language issues, old and new, 4–11. Kingston:
Caribbean University Press.
Shields-Brodber, Kathryn. 1997. Requiem for English in an “English-speaking”
community: The case of Jamaica. In Edgar Schneider (ed.), Englishes around
the world 2, 57–67. Amsterdam: John Benjamins Publishing Company.
Shields-Brodber, Kathryn. 1998. Hens can crow too: The female voice of authority
on air in Jamaica. In Pauline Christie, Barbara Lalla, Velma Pollard & Lawrence
178
Carrington (eds.), Studies in Caribbean language II, 187–203. St. Augustine: So-
ciety for Caribbean Linguistics.
Shields, Kathryn. 1984. The significance of word-final t/d consonant clusters for
syllable structuring in Standard Jamaican English. Paper presented at the 5th
biennial meeting for the Society for Caribbean Linguistics, 1984, Kingston, Uni-
versity of the West Indies (Mona), August 29–September 1.
Shields, Kathryn. 1987. Language variation in the classroom dissertation. D. Phil.
Dissertation, University of the West Indies (Mona).
Silverstein, Michael. 1996. Monoglot “standard” in America: Standardization and
metaphors of linguistic hegemony. In Donald Brenneis & Ronald Macauley
(eds.), The matrix of language: Contemporary linguistic anthropology, 284–306.
Boulder: Westview Press.
Silverstein, Michael. 1998. Contemporary transformations of local linguistic com-
munities. Annual Reviews of Anthropology 27. 401–426.
Silverstein, Michael. 2000. Whorfianism and the linguistic imagination of nation-
ality. In Paul Kroskrity (ed.), Regimes of language, 85–138. New Mexico: School
of American Research Press.
STATIN. 1991. Population census. Vol. 1. Kingston: Statistical Institute of Jamaica.
Stewart, Michele. 2002. The emergence of basilectal varieties in kingston.
Manuscript, Department of Language, Linguistics & Philosophy, University
of the West Indies (Mona).
Stewart, William A. 1965. Urban Negro speech: Sociolinguistic factors affecting
English teaching. In Roger Shuy, Alva Davis & Robert Hogan (eds.), Social di-
alects and language learning: Proceedings of the Bloomington, Indiana conference
1964, 10–19. National Council of Teachers of English.
Stone, Carl. 1980. Democracy and clientelism. New Jersey: Transactions Inc.
Strevens, Peter. 1982. The localized forms of English. In Braj Kachru (ed.), The
other tongue: English across cultures, 23–30. Urbana: University of Illinois
Press.
Taylor, Frank. 1971. The foundation of the Jamaica tourist industry (up to 1914).
University of the West Indies (Mona) MA thesis.
Taylor, Frank. 1993. To hell with paradise: A history of the Jamaica tourist industry.
Pittsburgh: University of Pittsburgh Press.
Thakerar, Jitendra & Howard Giles. 1981. They are – so they speak: Noncontent
speech stereotypes. Language and Communication 1. 251–256.
Thakerar, Jitendra, Howard Giles & Jenny Cheshire. 1982. Psychological and lin-
guistic parameters of speech accommodation theory. In Colin Fraser & Klaus
Scherer (eds.), Advances in the social psychology of language, 205–255. Cam-
179
References
Thaxter, Kenneth. 1977. Language variation in a Jamaican training college. Uni-

versity of the West Indies (Mona) MA thesis.
Trudgill, Peter. 1972. Sex, covert prestige and linguistic change in the urban
British English of Norwich. Language in Society 1. 179–195.
Trudgill, Peter. 1978. Sociolinguistic patterns in British English. London: Edward
Arnold.
Trudgill, Peter. 1986. Dialects in contact. Oxford: Blackwell.
Trudgill, Peter. 1999. Standard English: What it isn’t. In Tony Bex & Richard Watts
(eds.), Standard English: The widening debate, 117–128. London: Routledge.
Trudgill, Peter. 2002. Sociolinguistic variation and change. Washington DC:
Georgetown University Press.
Trudgill, Peter & Jean Hannah. 1994. International English. London: Edward
Arnold.
Walker, John. 1781. A rhetorical grammar. London.
Walters, Keith. 1996. Gender, identity and the political economy of language use:
Anglophone wives in Tunisia. Language in Society 25. 515–555.
Watt, Dominic. 2000. Phonetic parallels between the close-mid vowels of Tyne-
side English: Are they internally or externally motivated? Language Variation
and Change 12(1). 69–101.
Wells, John C. 1973. Jamaican pronunciation in London. Oxford: Basil Blackwell.
Wells, John C. 1982a. Accents of English 1: An introduction. Cambridge: Cambridge
University Press.
Wells, John C. 1982b. Accents of English 2: The British Isles. Cambridge: Cambridge
University Press.
Wells, John C. 1982c. Accents of English 3: Beyond the British Isles. Cambridge:
Williams, Eric. 1970. From Columbus to Castro. The history of the Caribbean 1492–
1969. New York: Vintage Books.
Wilmot, Swithin. 2002. From bondage to political office: Blacks and vestry poli-
tics in two Jamaican parishes, Kingston and St. David, 1831–1865. In Kathleen
Monteith & Glen Richards (eds.), Jamaica in slavery and freedom: History, her-
itage and culture, 307–323. Kingston: University of the West Indies Press.
Winford, Donald. 1985. The concept of “diglossia” in Caribbean Creole situations.
Language in Society 14. 345–356.
Winford, Donald. 1991. The Caribbean. In Jenny Cheshire (ed.), English around
the world, 565–584. Cambridge: Cambridge University Press.
Winford, Donald. 1997. Re-examining Caribbean English Creole continua. World
Englishes 16 2. 233–279.
180
Witter, Michael & George Beckford. 1980. Small garden, bitter weed. Morant Bay:
Maroon Publishing House.
Wolfram, Walt & Natalie Schilling-Estes. 1998. American English. Oxford: Basil
Blackwell.
Wolfson, Nessa. 1976. Speech events and natural speech: Some implications for
sociolinguistic methodology. Language in Society 5. 189–209.
Wright, Philip (ed.). 2002. Lady Nugent’s journal of her residence in Jamaica from
1801 to 1805. Kingston: The University of the West Indies Press.
Yallop, Colin. 1999. English around the world. In Eddie Ronowicz & Colin Yallop
(eds.), English: One language, different cultures, 26–45. London: Cassell.
Young, Colville. 1973. Belize creole: A study of the creolized English spoken in the
city of Belize, in its cultural and social setting. University of York D.Phil. disser-
tation.
181
Name index
Agnihotri, Rama Kant, 30 Carpenter, Joseph E., 48

Akers, Glen, 3–5, 34, 35, 39, 41, 43, 45, Carter, Bob, 121
52, 55 Cassidy, Frederic, 34, 35, 37, 39, 43,
Algeo, John, 39 45, 46, 110
Alleyne, Mervyn C., 1, 34, 43, 110 Chaudenson, Robert, 1, 2
Allsopp, Richard, 2, 13, 39, 60, 111 Cheshire, Jenny, 5, 15, 20
Alvesson, Mats, 139 Christie, Pauline, 21, 43, 48, 103, 111,
Auger, Julie, 109 122
Austin, Diane, 126, 148 Coye, Dale, 52
Craig, Dennis, 21
Bailey, Beryl, 1, 3–5 Crompton, Rosemary, 92
Barrow, Steve, 129
Baugh, Albert, 40 D’Costa, Jean, 7, 9, 10, 38, 42
Beal, Joan, 37, 41, 42, 51, 53 Dalton, Peter, 129
Beckford, George, 139 DeCamp, David, 1, 3
Beckford‑Wassink, Alicia, 13, 17, 22, Deuchar, Margaret, 80
27, 32, 34, 41, 43, 46, 52, 53, Devonish, Hubert, 1, 4, 34, 35, 41, 43,
81, 84, 96, 157 56, 60, 117, 118
Beckles, Hilary, 38 Dubois, Sylvie, 117
Bell, Allan, 26, 27 Dyer, Judy, 17, 22, 27, 46, 157
Bennett, Hazel, 38
Bickerton, Derek, 1, 3, 6 Ervin-Tripp, Susan, 117
Bobda, Augustin Simo, 109 Escure, Genevieve, 3, 5–8, 117
Bourdieu, Pierre, 13, 21
Fisher, John H, 11, 51
Brereton, Bridget, 10
Freeborn, Dennis, 41, 43, 51
Brodber, Kathryn, 47, 49, 97
Bryan, Patrick, 10, 11, 93 Gal, Susan, 4
Busby, Margaret, 10 Gbedemah, G., 19
Giegerich, Heinz, 48
Cable, Thomas, 40
Giles, Howard, 22, 86
Campbell, Carl, 51
Gimson, A. C., 40
Carmichael, A.C., 10
Gordon, Derek, 78
Name index
Gordon, Matthew, 26, 27 McArthur, Tom, 2, 14, 20

Graff, David, 27 McNair-Knox, Faye, 26
Green, Lisa, 117 Meade, Rocky R., 5, 34, 35, 39, 41, 43,
Gupta, Anthea, 15, 31 55
Guy, Gregory, 54 Miller, Errol, 12, 63, 78, 102
Miller, Faye, 13, 39, 45, 53, 81, 96
Hancock, Ian, 51 Milroy, James, 4, 15, 97
Hannah, Jean, 40, 48 Milroy, Lesley, 15, 26, 27, 31, 80, 97
Harry, Otelemate, 4, 117, 118 Moag, Rodney, 11
Hewitt, Roger, 31 Mufwene, Salikoko, 1, 4, 11, 38, 74
Ho, Mian-Lian, 15 Mugglestone, Lynda, 31, 37, 41, 45,
Holm, John, 110 47, 56, 59, 96
Holmes, Janet, 86
Horvath, Barbara, 117, 138 Nair-Venugopal, Shanta, 31
Hudson, Richard, 30 Nettleford, Rex, 93, 139
Hymes, Dell, 14 Neu, Helene, 54, 58, 90
Nichols, Patricia, 80
Irvine, Alison, 17, 45, 53, 75, 81, 139
Irvine, Judith, 4 Paolillo, John, 117
Patrick, Peter, 4, 6, 36, 41, 45, 54, 55,
Janda, Richard, 109 90, 97, 103, 158
Jenkins, R, 137 Patterson, Orlando, 11
Judd, Karen, 9 Pegge, Samuel, 56
Platt, John, 15
Kachru, Braj, 11, 13
Pollard, Velma, 13
Kirwin, William, 38
Powesland, Peter, 86
Krapp, George Philip, 38
Preston, Dennis, 109
Kroskrity, Paul, 4
Pyles, Thomas, 39
Kulick, Don, 146
Rickford, John, 12, 14, 15, 26, 31
Labov, William, 14, 31, 33, 54, 56, 58,
Roberts, Peter, 37
70, 80, 100, 109, 117
Romaine, Suzanne, 9
Lalla, Barbara, 7, 9, 10, 38, 42
Le Page, Robert, 1, 2, 12, 14, 19, 34, 35, Sahgal, Anju, 30
37, 39, 43, 45, 46, 56, 81, 87, Sankoff, David, 138
129 Santa Ana, Otto, 54
Lehmann, Winfred, 51 Schilling-Estes, Natalie, 158
Lewis, Gordon, 139 Schneider, Edgar, 5
Lippi-Green, Rosina, 13, 20, 30 Sealey, Alison, 121
Long, Edward, 10 Seiler, Walter, 34, 35, 41, 43
184
Name index
Sharwood-Smith, Michael, 20
Sherlock, Philip, 38
Shields, Kathryn, 18, 51, 55
Shields-Brodber, Kathryn, 21, 48, 86,
106
Silverstein, Michael, 20, 21, 31, 75,
109, 118, 124, 159
Stewart, Michele, 10
Stewart, William A., 2
Stone, Carl, 93
Strevens, Peter, 11
Taylor, Frank, 122, 123

Thakerar, Jitendra, 22, 63
Thaxter, Kenneth, 13, 31, 39
Trudgill, Peter, 15, 31, 40, 48, 51, 63,
80, 86, 109
Walker, John, 45
Walters, Keith, 80
Watt, Dominic, 118
Wells, John C., 4, 33–35, 37, 39, 41, 43,
45, 47, 52, 53
Williams, Eric, 102
Wilmot, Swithin, 12
Winford, Donald, 1, 4, 12, 17, 58, 152
Witter, Michael, 139
Wolfram, Walt, 158
Wolfson, Nessa, 14, 26
Wright, Philip, 42
Yallop, Colin, 60
Young, Colville, 9, 26, 39
185
Subject index
accent, 15, 22, 31, 34, 47, 60, 125, 128 Belizean, 8, 39
acrolect, 2–9, 11–14, 16–18, 24, 29, butter type words, 88
32–34, 39, 41, 52, 54, 5511 , 60,
63, 81, 84, 118, 148, 149, 152, Canadian, 5, 34, 38, 128, 129
153, 157, 158, 160 careful speech, 43, 44, 48, 57, 117
adoptive, 47–49 cluster simplification, 47, 56, 57, 92,
affricate, 49–52, 59, 73, 74, 747 , 89, 99, 116
110, 112, 113, 143, 152, 154 see also word-final clusters
age, 16, 70, 101–106, 109, 114, 117, 137, colonial period, 12, 40, 106
140, 154–156 see also frontline staff,
see also age cohorts see also senior manage-
age cohorts, 105–109 ment
AJE, 13, 16, 39, 45 Creole, 1–7, 10, 103 , 104 , 16–19, 21–
alveolar stop, 39, 41, 49, 52 23, 237 , 24, 31, 38, 39, 41–
alveopalatal affricate, 33, 110 43, 45–47, 49–51, 519 , 52, 55,
see also culture type word 56, 59, 68–74, 747 , 75, 768 ,
American English, 20, 5310 , 117 78, 82, 84–86, 88, 95, 96, 98,
architecture, 115 100, 101, 103, 105, 106, 108,
111–116, 124, 129–131, 140–
back diphthong, 68, 105, 118, 141, 146, 142, 146, 150, 152, 155, 157–
154 160
see also front diphthong
Barbados, 1 data collection, 24, 79, 136
basilect, 2–6, 12, 18, 34, 41, 52, 5511 , decreolization, 1, 7, 12, 15, 149, 160
84, 160 diglossia, 117, 155
see also linguistic contin- diphthong use, 105, 142
uum education level, 76, 88, 104
basilectalization, 1
see also education type
BBC, 8, 20, 130
endonormative approach, 32, 113,
see also mid vowel 146, 149
Belize, 1, 7–9, 115 , 21, 268
Subject index
female speech, 16, 45, 84, 85, 88, 89, implicational scale, 5, 52
99, 113, 148, 157 implicational scales, 13
see also gender interdental fricative, 5, 39, 74, 82, 117
formal speech, 13, 26, 27, 33, 40, 45,
Jamaican continuum, 5, 18, 33, 34, 52,
50, 58, 70, 81, 8710 , 97, 117
96
see also careful speech
Jamaican Creole, 18, 35, 38, 41, 51, 519 ,
front diphthong, 47, 104, 105, 142, 154
52, 60, 768 , 129, 156, 160
see also front vowel, Jamaican social context, 14, 24, 127,
see also mid vowel 155, 158, 160
front vowel, 44, 84, 111, 116 Jamaican speech community, 24, 42,
frontline staff, 67, 132, 133, 135, 136, 112, 149, 153
139, 140, 142–144, 146, 148, JAMPRO, 23, 24, 26–34, 38, 39, 42,
153, 159 44, 47–50, 53, 58, 64, 65, 654 ,
665 , 69, 78, 79, 86, 89, 90,
gender, 11, 16, 26, 43, 54, 60, 69, 78,
99, 101–105, 109, 110, 119, 121,
79, 799 , 80, 82, 84, 87–90, 92,
122, 126–129, 131, 132, 134–
95–99, 101, 105, 108, 112, 114,
137, 139, 140, 144, 146, 148,
117, 134, 140, 143, 146, 154–
150, 152, 155, 157, 159, 161,
157, 162
162
General American, 9, 20, 48, 123, 149
JE, 24, 33–36, 42, 44–51, 519 , 52–54,
glottal fricative, 32, 35, 110, 114, 153
5511 , 56, 58, 59, 61, 69–73,
see also h-drop
75, 82, 88–90, 92, 94–99, 101,
good English, 50, 75, 80, 103, 113, 140,
103, 105, 106, 109–119, 132,
144, 146, 158, 159
139–144, 146, 149, 150, 152,
see also local standard 154, 155, 157–160
group affiliation, 117, 154 JIDC, 126, 1263 , 128, 129, 162
Guyana, 1, 115 , 15, 9712 JNEC, 126, 1263 , 162
Guyanese, 12, 14, 15, 31 JNIP, 126, 1263 , 162
h-drop, 37, 38, 66, 69, 70, 95, 118, 144, JSA, 162
146, 147, 159 JTB, 122, 123, 125, 129
see also hypercorrection Kingston, 10, 12, 16, 17, 19, 22, 24, 36,
hypercorrection, 38, 39, 41, 477 , 58, 38, 78, 84, 130
74, 109–114, 152, 160
see also qualitative hyper- Lady Nugent, 9, 42
correction language acquisition, 3, 34, 159
language contact, 1, 3, 9, 115 , 39
IAE, 13, 21, 36, 42, 117 linguistic continuum, 1, 149, 155, 157,
idiolect, 108 158
188
Subject index
literary standard, 15, 31, 34 see also implicational scales

load-bearing, 115 prestige form, 85, 90, 101, 109, 112, 113
load-bearing variable, 155, 156 see also acrolect
local standard, 3, 4, 18, 34, 81, 149, 152, prestige norm, 52, 106, 107
158 see also local standard
long vowel, 45, 46, 58, 85, 116 prestige variety, 6, 74, 80, 81
low back vowel, 68
low central vowel, 32, 41, 45, 53, 114, qualitative hypercorrection, 109, 116,
153 144
see also low back vowel quantitative hypercorrection, 70, 75,
low vowel, 41, 59, 69, 95, 103, 112, 140 109
see also prestige variety
Malaysia, 31
mesolect, 1, 34, 55, 5511 race, 631 , 134, 138, 139
metropolitan norm, 59, 73 see also skin colour
see also MSE retroflexion, 12, 48, 86, 88
Ministry of Education, 18, 63 rhoticity, 17, 32, 33, 47, 48, 52, 53,
morphophonemic cluster, 114, 153 5310 , 59, 74, 75, 78, 89, 92,
see also phonological clus- 98, 100, 107–109, 111, 112, 114,
ter 115, 117, 118, 143, 153, 154
MSE, 4–10, 13, 16–19, 21, 216 , 24, 43, see also post-vocalic rhotic-
45, 49–51, 59, 72, 73, 81, 103, ity
142, 146, 149, 152, 153, 158 RP, 10, 12, 15, 18, 20, 31, 47, 487 , 51,
109, 129, 130, 149
native speaker, 8, 9 see also British Standard
see also non-native English English
NBC, 20
Niger-Congo, 1 senior management, 29, 30, 66, 79,
non-native English, 8, 11 126, 128, 132, 133, 136, 140,
157, 161
occupation, 92 Sinhala, 117
official language, 9, 125, 129, 130 SJE, 6, 13, 14, 18, 19, 21, 23, 24, 26,
orthography, 37, 47, 50 27, 30, 34, 55, 56, 60, 61, 63,
64, 97, 108, 114, 117, 148–150,
palatal glide, 45, 46, 9611 , 97 152–160
parent’s background, 53, 150, 152 skin colour, 631 , 134, 139, 140
patois, 123, 125, 129–131, 158 socio-economic class, 13, 16, 81
phonological cluster, 54, 89
see also some other term
PJS, 34, 68, 74, 85, 86, 88
also of interest
189
Subject index
speaky-spoky, 36, 41
spelling pronunciation, 47, 48, 50, 52
spirantization, 50
Standard English, 2, 4, 7, 8, 13–15, 21,
24, 268 , 28, 29, 34, 80, 81,
130, 131, 146, 148, 149, 159
standard variety, 8, 476 , 63, 80, 131
style shift, 80, 117, 154
tense marking, 32, 54, 55, 58, 90

see also interdental frica-
tive
Trinidad, 1, 51, 58, 9712
Trinidadian, 32
velar stop, 32, 35, 46, 74, 114, 116, 118,

141, 153
vernacular, 3, 7, 9, 10, 17, 24, 33, 47,
74, 97, 109, 112
voiced interdental fricative, 27
voiceless interdental fricative, 32,
114, 116, 118, 153
see also hypercorrection,
see also th stopping
vowel inventory, 34
word-final clusters, 54
190
Did you like this
book?
This book was brought to you for
free
Please help us in providing free access

to linguistic research worldwide. Visit
http://www.langsci-press.org/donate to
provide financial support or register as
language
a community proofreader or typesetter science
press
at http://www.langsci-press.org/register.
The acrolect in Jamaica
An ability to speak Jamaican Standard English is the stated requirement for any manage-
rial or frontline position in corporate Jamaica. This research looks at the phonological
variation that occurs in the formal speech of this type of employee, and focuses on the
specific cohort chosen to represent Jamaica in interactions with local and international
clients. The variation that does emerge, shows both the presence of some features tradi-
tionally characterized as Creole and a clear avoidance of other features found in basilectal
and mesolectal Jamaican. Some phonological items are prerequisites for “good English”
– variables that define the user as someone who speaks English – even if other Creole
variants are present. The ideologies of language and language use that Jamaican speak-
ers hold about “good English” clearly reflect the centuries-old coexistence of English and
Creole, and suggest local norms must be our starting point for discussing the acrolect.
ISBN 978-3-96110-114-6
9 783961 101146

The Acrolect in Jamaica: G. Alison Irvine-Sobers

Uploaded by

Copyright:

Available Formats

The Acrolect in Jamaica: G. Alison Irvine-Sobers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Acrolect in Jamaica: G. Alison Irvine-Sobers

Uploaded by

Copyright:

Available Formats

The acrolect in

Chief Editor: John R. Rickford

1. Irvine-Sobers, G. Alison. The acrolect in Jamaica: The architecture of phonological

This title can be downloaded at:

Cover and concept of design: Ulrike Harbort

Language Science Press

Storage and cataloguing done by FU Berlin

1 Ideologies of the acrolect and English in Jamaica 1

2 Phonological variation in the Jamaican acrolect 29

3 Sociolinguistic variation in Jamaican English 63

3.1.2 Group B variables and education . . . . . . . . . . . . . 71

4 Sociolinguistic Variation in JAMPRO 121

Appendix A: Prepared script for all interviews of JAMPRO informants 161

Appendix B: Profiles of JAMPRO informants 163

Appendix C: Parent’s Occupation 167

… a continuous spectrum of speech varieties whose extremes are mutually

1 Chaudenson (2001) describes a similar process of basilectalization for French-lexicon Creoles.

It is to this sociolinguistic discussion that W. A. Stewart (1965: 15–16) added

I will refer to this topmost dialect in the local sociolinguistic hierarchy as

• the “topmost dialect in the local sociolinguistic hierarchy” (W. A. Stewart

• the stereotype of the educated standard, focussed by institutions of educa-

• a “high, prestigious language or variety of a language” in any speech com-

Acrolect here is essentially a community of speakers’ idea of the speech pat-

• the educated counterpart of the lexifier in the Creole speech community

• the English of educated nationals of the Caribbean used in formal social

Central to these views of the acrolect is the speaker, as it is a particular set

• the local standard, similar to other standard Englishes except in “a few

Finally, the acrolect is seen as the outcome of Creole speakers’ acquisition of

• a non-native version, derived from spontaneous or guided learning, of the

2. as a consequence of language acquisition, as it is suggested that the acro-

3. sociolinguistically, in relation to its speakers and the situational contexts in

4. ideologically, an idea of English that is legitimised by institutions that con-

1.2 Sense 1: The acrolect as defined by structure

tual production of acrolectal speakers. His pool of informants consisted of 10

… variables are here counted in their presence or absence so that maximum

Speaky-spoky is associated with lower mesolectal and basilectal speakers

1.3 Sense 2: The acrolect as the outcome of language

A non-native version of the standard language, which is acquired through

She goes on to add that the label “English” is:

extremely confusing to Belizeans, primarily because English is an external

The above quotes suggest the following:

b) Acrolect is the formal, educated Belizean variety of English (EBE hereafter),

…acrolects develop independently because the officially prescribed standard

the existence, from the eighteenth century, of extensive variation (among

Moreover, as Michele M. Stewart (2002: 14) notes,

The acrolect then, as a macrosociolinguistic phenomenon, cannot be learned

For high schools, a “limited number of elementary school children [were]

English in terms of MSE will result in paradoxical conclusions. Some of these

1.4 Sense 3: The acrolect as defined by its speakers

And of course, education as a defining characteristic of “standard” speech is

The speakers have a wider range of language usage in phonology than

probably because morphology and syntax decreolize faster than phonology

What Thaxter explains as an aspect of the rate of decreolization, can be ex-

The English of the classroom is pronunciation neutral, based as it is interna-

The high frequency of occurrences of palatal realizations (44.3%), a 30 per-

Beckford-Wassink 2001: 31) in the speech of the upwardly mobile, speaks to a

though some phonological features are shared by mainstream RP and Ja-

1.5 Sense 4: The acrolect as an idea of Jamaican English

Le Page mentions, in particular, the emergence of the Caribbean Examinations