Singapore English - Structure, Variation, and Usage PDF

more information - www.cambridge.
org/9781107027305
Singapore English
In recent years the study of English and its global varieties has grown
rapidly as a field of study. The English language in Singapore, famous
for its vernacular known as ‘Singlish’, is of particular interest to linguists
because it takes accent, dialect, and lexical features from a wide range
of languages including Malay, Mandarin, Hokkien, and Tamil, as well
as being influenced by the Englishes of Britain, Australia, and America.
This book gives a comprehensive overview of English in Singapore by
setting it within a historical context and drawing on recent developments
in the field of indexicality, World Englishes, and corpus research. Through
application of the indexicality framework, Jakob Leimgruber offers readers
a new way of thinking about and analysing the unique syntactic, semantic,
and phonological structure of Singapore English. This book is ideal for
researchers and advanced students interested in Singapore and its languages.
jakob r. e. leimgruber is Lecturer in the English Department at the

University of Freiburg, Germany.
studies in english language
General editor
Merja Kytö (Uppsala University)
Editorial Board
Bas Aarts (University College London), John Algeo (University of Georgia),
Susan Fitzmaurice (University of Sheffield), Christian Mair (University of
Freiburg), Charles F. Meyer (University of Massachusetts)
The aim of this series is to provide a framework for original studies of English,
both present-day and past. All books are based securely on empirical research,
and represent theoretical and descriptive contributions to our knowledge of
national and international varieties of English, both written and spoken. The
series covers a broad range of topics and approaches, including syntax,
phonology, grammar, vocabulary, discourse, pragmatics and sociolinguistics,
and is aimed at an international readership.
Already published in this series:

Geoffrey Leech, Marianne Hundt, Christian Mair and Nicholas Smith: Change
in Contemporary English: A Grammatical Study
Jonathan Culpeper and Merja Kytö: Early Modern English Dialogues: Spoken
Interaction as Writing
Daniel Schreier, Peter Trudgill, Edgar Schneider and Jeffrey Williams: The
Lesser-Known Varieties of English: An Introduction
Hilde Hasselgård: Adjunct Adverbials in English
Raymond Hickey: Eighteenth-Century English: Ideology and Change
Charles Boberg: The English Language in Canada: Status, History and
Comparative Analysis
Thomas Hoffmann: Preposition Placement in English: A Usage-Based Approach
Claudia Claridge: Hyperbole in English: A Corpus-Based Study of Exaggeration
Päivi Pahta and Andreas H. Jucker (eds.): Communicating Early English
Manuscripts
Irma Taavitsainen and Päivi Pahta (eds.): Medical Writing in Early Modern
English
Colette Moore: Quoting Speech in Early English
David Denison, Ricardo Bermúdez-Otero, Chris McCully and Emma Moore
(eds.): Analysing Older English
Jim Feist: Premodifiers in English: Their Structure and Significance
Steven Jones, M. Lynne Murphy, Carita Paradis and Caroline Willners:
Antonyms in English: Construals, Constructions and Canonicity
Christiane Meierkord: Interactions across Englishes: Linguistic Choices in Local
and International Contact Situations
Haruko Momma: From Philology to English Studies: Language and Culture in the
Nineteenth Century
Raymond Hickey (ed.): Standards of English: Codified Varieties around the World
Benedikt Szmrecsanyi: Grammatical Variation in British English Dialects: A
Study in Corpus-Based Dialectometry
Daniel Schreier and Marianne Hundt (eds.): English as a Contact Language
Bas Aarts, Joanne Close, Geoffrey Leech and Sean Wallis (eds.): The Verb
Phrase in English: Investigating Recent Language Change with Corpora
Martin Hilpert: Constructional Change in English: Developments in Allomorphy,
Word Formation, and Syntax
Jakob R. E. Leimgruber: Singapore English: Structure, Variation, and Usage
Earlier titles not listed are also available

Singapore English
Structure, Variation, and Usage
jakob r. e. leimgruber
University of Freiburg
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town,
Singapore, São Paulo, Delhi, Mexico City
Cambridge University Press
The Edinburgh Building, Cambridge cb2 8ru, UK
Published in the United States of America by Cambridge University Press,
New York
www.cambridge.org
Information on this title: www.cambridge.org/9781107027305

C Jakob R. E. Leimgruber 2013
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2013
Printed and bound in the United Kingdom by the MPG Books Group
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data

Leimgruber, Jakob R. E. (Jakob Robert Eugen), 1980–
Singapore English : structure, variation, and usage / Jakob R. E.
Leimgruber, University of Freiburg.
pages cm. – (Studies in English language)
Includes bibliographical references and index.
isbn 978-1-107-02730-5 (hardback)
1. English language – Singapore. 2. English language – Variation – Singapore.
3. English language – Social aspects – Singapore. 4. Language and culture –
Singapore. 5. Singapore – Languages. I. Title.
pe3502.s5l46 2013
427 .95957 – dc23 2012049291
isbn 978-1-107-02730-5 Hardback
Cambridge University Press has no responsibility for the persistence or

accuracy of URLs for external or third-party internet websites referred to
in this publication, and does not guarantee that any content on such
websites is, or will remain, accurate or appropriate.
For Marie, Lukas, and David
Contents
List of figures page xi

List of tables xii
Acknowledgements xiii
List of abbreviations, glosses, and symbols xv
1 Introduction: Singapore and its Englishes 1

1.1 Historical and present-day background 1
1.1.1 Historical background 1
1.1.2 Linguistic diversity 6
1.1.3 The place of English 8
1.2 The Englishes of Singapore and their models 13
1.2.1 Genesis 15
1.2.2 Variation 16
1.3 Data and methods 21
1.4 Conclusion 25
2 Variation in Singapore English: old and new models 26
2.1 Early models and diglossia 26
2.1.1 The continuum hypothesis 26
2.1.2 Polyglossia 35
2.1.3 Diglossia 38
2.2 Contemporary models and indexicality 46
2.2.1 Cultural orientation model 46
2.2.2 Indexicality 52
2.3 Conclusion 62
3 Description: phonology and lexicon 64
3.1 Phonology 64
3.1.1 Vowels 64
3.1.2 Consonants 65
3.2 Lexicon 67
3.2.1 Etymology 67
3.2.2 Borrowing and code-switching 68
3.3 Conclusion 70
ix
x Contents
4 Description: grammar 71
4.1 Syntax and morphology 71
4.1.1 The verb 71
4.1.2 got 77
4.1.3 Aspect marking 80
4.2 Semantics and pragmatics 82
4.2.1 Discourse particles 82
4.2.2 Lexical semantics 96
4.3 Conclusion 99
5 Indexicality: a model for Singapore? 100
5.1 Indexicality as an analytic tool 100
5.2 Eckert’s indexical field 103
5.3 Indexicality in Singapore 108
5.4 Conclusions 112
6 Conclusion: the variety as a structural unit 114
6.1 Codes, varieties, and code-switching 114
6.2 English in Singapore and Southeast Asia 117
6.2.1 Language policies and Singapore English 117
6.2.2 English in Southeast Asia 119
6.3 ‘Varieties’ of English 123
6.4 Conclusion 126
Appendix A: Chinese romanisation 127
A.1 Mandarin: Pı̄nyı̄n 127
A.2 Cantonese: Jyut6ping3 128
A.3 Hokkien: PĲeh-ōe-jı̄ 129
Appendix B: Informants 130
Appendix C: Chinese languages 133
References 134
Author index 146
Subject index 149
Figures
1.1 Location of Singapore within Asia page 2

1.2 Map of Singapore 5
1.3 Kachru’s concentric circles 13
1.4 Post-creole continuum according to Platt (1975) 17
1.5 ‘Expanding triangles of English expression’ (Pakir 1991) 19
1.6 Modified triangles (Poedjosoedarmo 1995) 20
1.7 Post-secondary enrolment in 2005 22
2.1 Post-creole continuum according to Platt (1975) 31
2.2 Indexical field of /t/ release (Eckert 2008) 55
4.1 Occurrences of (-ed) and (3sg) in ii.M.gr 73
4.2 H and L variants of auxiliaries and inversion 74
4.3 Verbal variables’ H and L proportions 76
4.4 Three variables in ii.M.gr 77
4.5 Lexifier filter 81
4.6 Discourse particles and discourse markers 83
4.7 Gupta’s scale of assertiveness 85
5.1 Indexical field of SgE 106
5.2 Indexing process with an example 107
6.1 Home language shift, 1980–2010 120
6.2 Dialect continuum 125
C.1 Phylogenetic tree of Sino-Tibetan 133
xi
Tables
1.1 Language most frequently spoken at home page 7

2.1 Idiolectal distribution of Jamaican English features 29
2.2 Speakers from Table 2.1 rearranged by variants used 29
2.3 Speaker ranking from Table 2.2 with features rearranged by variants
used 30
2.4 Combining diglossia and multilingualism 35
2.5 ‘A model for polyglossia with multilingualism’ (Platt 1977) 36
2.6 Verbal repertoire of an English-educated Chinese of Singapore
(Platt 1977) 37
2.7 Use of diglossic varieties for selected situations (adapted from
Ferguson 1959) 41
2.8 Features of diglossic SSE and CSE (adapted from Gupta 1994: 10–13) 42
2.9 Ferguson (1959) applied to SgE 44
2.10 Features of the cultural orientation model (Alsagoff 2007) 50
2.11 Selected variables and the stances they index 59
3.1 CSE vowel system 65
3.2 CSE consonant inventory 66
4.1 Aspect in Chinese, CSE, and English according to Bao (2005) 80
4.2 Terminological definitions 84
4.3 Definitions of CSE particles 92
4.4 A selection of SgE initialisms 97
5.1 Features of the cultural orientation model (Alsagoff 2007) 105
5.2 Overview of existing variationist models for SgE 109
6.1 Population figures of ASEAN members states 123
B.1 Number of recordings of each type, by school and ethnic group 130
B.2 Informants and their demographic and socio-economic background 131
xii
Acknowledgements
This book builds on a D.Phil. thesis (Leimgruber 2009b) submitted to the

Faculty of English Language and Literature at the University of Oxford in
2009, with corrections, updates, and expansion. Both the thesis and this book
would have been impossible without the help of many people, to whom I
would like to express my heartfelt thanks.
Special thanks go to my D.Phil. supervisors Suzanne Romaine and Deb-
orah Cameron and to my college advisor Lynda Mugglestone for their guid-
ance, and to two anonymous reviewers who suggested helpful modifications
to the original book proposal. I benefited from the help of and discussions
with John Coleman, Ben Rampton, Elinor Keane, Greg Kochanski, Francis
Marriott, Bao Zhiming, Peter Tan, Norhaida Aman, and Lubna Alsagoff.
Thanks also to my examiners Anthea Fraser Gupta and David Cram, whose
feedback on the thesis were of immense value, and to Pamela Macdonald,
Eddie Williams, Robert and Caroline Rehder, as well as both my parents,
who proofread all or parts of the thesis or this book.
The fieldwork in Singapore was made possible thanks to the Singapore
Ministry of Education, which granted me access to their Junior Colleges.
In addition, I would like to reiterate my gratitude to Samuel Wee, Emily
J. L. Low, Eden Liew, Connie Lim, Virginia Cheng, Mark Lo, Chua Mui
Ling Dorothy, Meena M. Kaur, and Wan Wai Sum for their invaluable
cooperation in allowing their institutions to take part in the fieldwork, re-
cruiting my informants, organising the interview venues, and providing an
exciting insight into the Singapore education system. I extend my thanks
to the thirty-six informants who took part in this study, and who willingly
sacrificed their time for my benefit. Without them, this book would not have
been possible.
A word of thanks also goes to my fellow postgraduate students in the
Oxford linguistics department for the stimulating discussions and exchanges
in the course of those three years in Oxford. Thank you also to Chandraselven
Bavani and Lavanya Sankaran for help with Tamil, and to Suriel Mofu for
help with Malay transcriptions. Further thanks go to Helen Barton and her
team at Cambridge University Press, for their help with seeing this book
through to its current shape.
xiii
xiv Acknowledgements
Financial support was given throughout my studies by the Pembroke

Dean of Graduates Fund, which covered significant proportions of travel
expenses. The first half of the fieldwork in 2006 was made possible by the
Maxwell & Meyerstein Special Research Committee Grant, awarded by the
English Faculty. In my third year I also benefited from the Pembroke College
Browning Senior Scholarship. My thanks go to all these awarding bodies.
Finally, and most importantly, my gratitude goes to my parents, for making
my time at Oxford possible and for always believing in me. I also thank Marie,
my wife, who patiently supported me during often trying times, and to Lukas,
our son, for his patience while this book was being finalised. Marie’s family
also deserves mention here, not least for their wonderful hospitality in the
course of my numerous trips to Singapore.
Abbreviations, glosses, and symbols
Abbreviations
C consonant or Chinese
CSE Colloquial Singapore English
dia dialogue interview
f female
gr group recording
H high (diglossia)
i junior college
ii polytechnic
iii vocational school
I Indian
ind individual interview
ISE International Singapore English
ITE Institute of Technical Education
JC Junior College
L low (diglossia)
LSE Local Singapore English
m male
M Malay
MOE Ministry of Education
rm radio-microphone recording
RP Received Pronunciation
SgE Singapore English
SSE Standard Singapore English
StBE Standard British English
StdE Standard English
V vowel
Glosses
1 first person
2 second person
3 third person
xv
xvi Abbreviations, glosses, and symbols
clf classifier
pl plural
q question particle
sg singular
Symbols
* ungrammatical
? questionable grammaticality
(***) indecipherable passage
(*. . .*) tentative transcription
1 Introduction: Singapore and
its Englishes
1.1 Historical and present-day background

1.1.1 Historical background
Present-day Singapore is an island-state of approximately 710 km2 located
at the southern tip of the Malay Peninsula. Early written records name it
Pu Luo Zhong (蒲羅中, from Malay Pulau Ujong ‘island at the end [of the
Malay Peninsula]’, Savage and Yeoh 2005; third century, Turnbull 1996),
Temasek (from Javanese Tumasik ‘sea town’; fourteenth century, Prapañca
1995: 14.2, l. 3), and finally Singapura (from Sanskrit si_
mha ‘lion’ and puram
‘city’; sixteenth century, Brown 1983). This latter name became established
and the island changed hands repeatedly, belonging in turn to the Śrivijaya
thalassocracy, to the Javanese Majapahit Empire, to the Thai Kingdom, and
to the Malacca Sultanate, when it was destroyed by the invading Portuguese
in 1613 (Brown 1983: 41, Turnbull 1996: 4).
In the early nineteenth century, the Johor Empire founded a village on
the site of Singapura. Aboriginal Malays (Orang Kallang and others) lived
scattered over the island, but had limited interaction with the ruling classes.
In early 1819, Singapore had around a thousand inhabitants, among whom
were some thirty Chinese (Turnbull 1996: 5).
On 28 January 1819, Sir Stamford Raffles, investigating possible locations
for an East India Company station, anchored near Singapore. A treaty with
the Sultan of Johor was signed on 6 February, leasing the island to the
Company: modern Singapore was born. After disputes with the neighbouring
Dutch were resolved, Singapore became a permanent British settlement
under the direct administration of Calcutta (Turnbull 1996). Before the
Anglo-Dutch Treaty, colonial administrators in Singapore appealed to the
existing colonies Malacca and Penang1 for settlers and traders, promising
a tariff-free port. A great many Chinese arrived early on, and by 1821, of
the 5,000 inhabitants, 3,000 were Malays and over 1,000 Chinese (Turnbull
1996: 13). Malays flocked in from the entire archipelago. Indians were mostly
soldiers in the British Indian Army, but there were a few businessmen too,
1
Previously established British dependencies on the west coast of the Malay Peninsula.
1
1000 km
600 mi
Figure 1.1. The location of Singapore within Asia. The city-state is located at the southern tip of the Malay Peninsula; it is
sandwiched between Malaysia to the north and Indonesia to the south and southwest. The island is just 1◦ 22 north of the equator.
Source (Asia): http://d-maps.com/carte.php?lib=asia map&num car=13293&lang=en
Source (Singapore): http://d-maps.com/carte.php?lib=singapore map&num car=15621&lang=en
some from Penang and Malacca (1996: 14). The first official census of 1824
reported 11,000 inhabitants, with the Malays still in the majority – three
years later they were overtaken by the Chinese.
Between 1827 and 1836 the population almost doubled (Turnbull 1996: 36)
and in 1860 it stood at 81,000. The Chinese, representing 65 per cent in 1867,
came predominantly from the southern provinces of Guangdong and Fujian.
The Hokkien (Fùjiān) were the largest group and dominated commerce,
followed by the closely related Teochew (Cháozhōu). Cantonese and Hakka
(Kèjiā) came mostly as labourers and craftsmen. More Indians arrived in this
period, mainly from southern India, but also from Punjab, and at this time
they formed the second largest group (Turnbull 1996: 39–40). The Malays
came third, and the Europeans formed a tiny minority, with fewer than 300
British, nearly all men who, however, held key positions in the civil service
and business.
Singapore became a Crown Colony in April 1867, and in 1911 numbered
over 185,000 inhabitants, of whom almost three-quarters were Chinese.
Immigration continued, Europeans increased their numbers slightly; the
Indians were the only ethnic group to see their number decline: most had
used Singapore as a transitory port to seek employment in the neighbouring
Malay States (Turnbull 1996), which had come under British rule with the
Anglo-Dutch Treaty of 1824.
Under colonial rule, Singapore saw its education system develop slowly.
At first, the authorities were only interested in providing teaching in Malay,
which was seen as the only viable means of communication in the region –
Chinese was divided into too many dialects,2 and the Indians spoke many
different languages. Leading positions in politics were reserved for Euro-
peans anyway, and the few Chinese who became Justices of the Peace or
members of the Legislative Council were wealthy and spoke English well,
and consequently sent their sons to Britain for their studies. A number of
privately run schools offered classes in Hokkien, Cantonese, and Tamil, but
they had no government backing. English-medium instruction was largely
the responsibility of Christian missionaries. Only the Raffles Institution pro-
vided secondary education in English; but the few Malay and Chinese classes
it offered had to be closed down due to lack of funds and interest in 1894
(Wijeysingha 1963: 97). It was only after World War I, which left Singapore
2
On the question about the dialect/language status of the various Chinese varieties, see
DeFrancis (1986: 53–67), Ramsey (1987: 16–17, 28–9), and Mair (1991), among others.
Although speakers normally refer to them as dialects, they are often mutually unintelligible
(Cheng 1996), and some (such as Cantonese) have their own, more or less standardised
writing system. Ultimately, such naming issues are of a social, rather than linguistic, nature
(Trudgill 1995). Therefore, and in accordance both with linguists specialising in the field
(Bao 2001; Tan 2003, among others) and with vernacular Singaporean usage, varieties of
Chinese are herein called dialects.
4 Introduction: Singapore and its Englishes
unaffected except for a quickly subdued mutiny of its only regiment,3 that
education became more important, and by 1939, of the 550,000 inhabitants,
72,000 children were enrolled in school, of whom 38,000 were in Chinese,
27,000 in English, 6,000 in Malay and 1,000 in Tamil schools (Doraisamy
1969: 38).
The Japanese invasion on 15 February 1942 resulted in three years of
occupation and the death of some thirty thousand. When Commonwealth
troops retook Singapore and Malaya in 1945, it was a devastated city and
although British rule was welcomed back, confidence in the colonial mas-
ters had been shattered. In April 1946, civil administration being restored,
Singapore was again a Crown Colony, and Peninsular Malaysia became the
Malay Union. The latter was modified into the Malay Federation in 1948.
At that stage, Singapore’s population of 941,000 consisted of 78 per cent
Chinese, 12 per cent Malays and 7 per cent Indians (Turnbull 1996: 229,
234), a situation that roughly prevails today. The Singapore Improvement
Trust was established – the future Housing Development Board – to better
the situation for the thousands that were left in precarious housing condi-
tions after the war. Education was stepped up, with the government finally
acknowledging the demand for English-medium primary education. The
University of Malaya was created in 1949, and a Teacher Training College
was also founded. In 1954, the English-medium primary school intake was
higher than the Chinese-medium one.
When the first elections were held in 1955, Lee Kuan Yew’s People’s
Action Party (PAP) formed a coalition government which had to tackle a
number of strikes and internal political and racial turmoil. After almost four
years of agitation and negotiation, self-government was effective from 1959,
with the PAP winning forty-three of Parliament’s fifty-one seats (Turnbull
1996: 263).
In a referendum held in September 1962, Singapore voters agreed to
the government’s proposal for a merger with the Federation of Malaya.
This was a short-lived experiment, which ended on 9 August 1965, when
Singapore was ejected from the Federation and became independent again.
Initial distress caused by the shock of being left alone, without an economic
hinterland and natural resources, was overcome by successful policies, which
soon attracted foreign investors. Singapore flourished, and the Malaysian
episode was soon put behind.
The education system was developed, strongly emphasising pragmatic dis-
ciplines rather than the humanities. Racial relations were an issue, especially
after serious riots in the sixties. Various schemes were introduced, among
them an annual Racial Harmony Day which aims to promote mutual respect
and understanding, and a quota system for public housing. The government
3
The Indian Army 5th Light Infantry, consisting solely of Punjabi Muslims, ‘were bitter that
the British were fighting against Muslim Turkey’ (Turnbull 1996: 126).
Figure 1.2. A map of Singapore, showing major highways, Mass Rapid Transit network, and residential areas, as well as the
three schools where data were collected for this study (marked by , , and ). The Central Business District () is just east of
Chinatown. The highest population density is found in the CBD and adjoining areas, although settlements (Housing Development
Board ‘New Towns’) exist all over the island. A nature reserve and water catchment area is located in the geographical centre of
the country. Labels and refer to the discussion on pp. 10–11.
introduced a massive public housing construction scheme, which now houses

over 80 per cent of the population. Tenants own their flats which they buy at
subsidised rates. Living conditions further improved with strong economic
growth (averaging 8 per cent yearly between 1960 and 1999 (Singstat 1999)).
This situation prevails, and thanks to efficacious policies, the Asian Crisis of
the late nineties had only a minor impact on Singapore’s growth. The same
seems to hold true for the crises of the late 2000s: in 2011, GDP per capita
was at SGD 62,008 (GBP 31,581), which was second only to Japan in East
Asia, and the growth projected for 2012 was 4.41 per cent (IMF 2011).
1.1.2 Linguistic diversity

When the constitution was written in 1963, the thorny issue of linguistic
diversity was addressed. ‘Malay, Mandarin, Tamil and English [are] the
four official languages of Singapore’ (Constitution: §153A). Technically,
they all enjoy the same constitutional status, §53 also stating that they may
all be used for debates in Parliament. Simultaneous translation during sit-
tings is provided, but English as the working language of the government is
used overwhelmingly.4 All government websites are in English without any
translation – except for some important information leaflets.5 Unlike other
countries with more than one official language, Singapore’s legislation is
entirely in English. The other three official languages are known as ‘mother
tongues’. Every ethnic group has one assigned to them, and they are taught
it as a second language at school – since 1987 English has been the only
medium of instruction for all groups. Thus all Chinese are taught Mandarin,
all Malays Malay and all Indians Tamil as a second language6 – in English.
The aboriginal language, if Singapore can be said to have one, is Malay.
Widely spoken in the area, Malay is the sole official language of Malaysia, and
its dialectal and mutually intelligible sister Indonesian is Indonesia’s official
tongue.7 Given the regional predominance of Malay (over thirty million
speakers in the two neighbouring countries (Lewis 2009)), it may come as a
surprise that in the 2010 Census, only 17.7 per cent of Singapore’s population
declared literacy in it (Wong 2011).
4
Between 11 February 2006 and 20 September 2007, there were 641 questions asked in
Parliament. Of these, 34 (5.3 per cent) were in Malay, 81 (12.6 per cent) in Mandarin, and
just a single one (0.2 per cent) in Tamil (Parliament of Singapore 2007).
5
For instance the information booklets on measures to counteract infectious diseases (avian
flu, SARS, dengue fever), which are made available in the four official languages (MOH
2005b,a). Generally, information pertaining to far-reaching government schemes is available
in all four languages as well.
6
Recently, this has been somewhat relaxed, with parents allowed to enrol their children for
‘mother tongue’ classes other than their assigned ethnic language.
7
The names given to the different varieties of Malay eloquently reflect their sociopolitical,
rather than linguistic, difference: Bahasa Malaysia ‘Malaysian language’ in Malaysia, Bahasa
Indonesia ‘Indonesian language’ in Indonesia, and Bahasa Melayu ‘Malay language’ or simply
Melayu in Singapore and Brunei.
Table 1.1. Language most frequently spoken at home. Data for 1980 from
Foley (1998: 221) citing Lau (1993: 6), for 1990 and 2000 from Leow
(2001), and for 2010 from Wong (2011)
1980 1990 2000 2010
English 11.6 % 18.8 % 23.0 % 32.3 %

Mandarin 10.2 % 23.7 % 35.0 % 35.6 %
Chinese dialects 59.5 % 39.6 % 23.8 % 14.3 %
Malay 13.9 % 14.3 % 14.1 % 12.2 %
Tamil 3.1 % 2.9 % 3.2 % 3.3 %
Others 1.7 % 0.8 % 0.9 % 1.1 %
Singapore’s Constitution gives Malay a special place: the Malays are re-
garded as ‘the indigenous people of Singapore’, and, therefore, their language
deserves particular support and encouragement (Constitution: §152). Fur-
thermore, Malay is officially designated as the ‘national language’ (§153A),
which is different from the official language status that it also holds. In prac-
tice, this status as a national language means that Malay is the language of
the national anthem and of drill commands in the army and other marching
bodies. It is also part of the national coat of arms, which bears the motto Ma-
julah Singapura or ‘onwards Singapore’ – and of the President’s residence,
which is called the Istana, ‘the palace’.
Socially, Tamil plays a minor role: the fourth official language is used
along with the others in warning signs on buses, in the information booklets
mentioned above (MOH 2005b,a) as well as in announcements on MRT8
platforms to ‘stay behind the yellow line’. It is the main home language of
only 3.26 per cent of the population (Wong 2011) and its fate seems sealed:
since 1911, these figures have been dropping (Bao 2001: 281) and nothing
seems to indicate a change in this trend. One of the problems for Tamil is that
the Indian community is fragmented. It encompasses linguistic and ethnic
groups as diverse as Tamils, Telugus, Malayalees (Dravidian), Punjabis,
Bengalis, and Sinhalese (Indo-Aryan). Together, they amount to 9.23 per
cent of the population (Wong 2011); but their linguistic background makes
Tamil a non-native language for a high proportion of them.
The Chinese, who represent over three-quarters of the republic’s popula-
tion, are also a fragmented population, although less so. Most are of southern
Chinese extraction, the majority originally hailing from Fujian Province
(Hokkien, 福建) and Guangdong. Linguistically, the most sizeable dialect
group is Hokkien (49 per cent of the 2010 dialect speakers in Table 1.1),
which is mutually intelligible with Teochew (Cháozhōu, 潮州), the third
largest group (19 per cent). These are both varieties of the Southern Min
(閩南) dialect group. Other varieties include Cantonese, a sizeable community
8
Mass Rapid Transit, Singapore’s underground train system.
(25 per cent) which forms a distinct dialect group (Yuè, 粵), as well as Hakka
(Kèjiā, 客家) (Bao 2001: 282). Ancestral Mandarin speakers, however, have
never been represented in any significant number (but see next paragraph).
This has changed in recent decades, with the increasing immigration of Chi-
nese professionals, but northern Chinese of the Mandarin dialect group still
make up only a negligible proportion of the community.
Malay and Tamil, besides their status as official languages and as edu-
cational ‘mother tongues’, have received little encouragement from the gov-
ernment. The same cannot be said of Mandarin: launched in 1979, the
‘Speak Mandarin Campaign’ aims to promote the use of Mandarin among
the ethnic Chinese. First directed at speakers of Chinese dialects, it later ex-
panded to address English-educated Chinese too and encouraged them to use
Mandarin in their everyday transactions (SMC 2009). Critics have argued
that in an attempt to unite the Chinese and to preserve their cultural identity,
the campaign has actually severed cross-generational communication, with
young Singaporeans having increasing difficulties communicating in dialect
with their grandparents. Proponents of the shift point to the enhanced cross-
dialectal communication, and not less importantly, to the improved position
in commercial dealings with the emerging superpower China.
The campaign has been a success. The majority of Chinese have embraced
Mandarin, and in the 2010 Census, 35.6 per cent of Singaporeans indicated it
was their dominant home language, outnumbering the other dialects which
stood at 14.3 per cent (Table 1.1). Besides the emphasis on Mandarin in
education (some élite schools offering bilingual programmes), many parents
are generally supportive of the movement and speak Mandarin to their chil-
dren. Mandarin has also replaced Hokkien as the lingua franca for everyday
intra-Chinese transactions. This means that Mandarin, despite not being
the language of immigrant Chinese, has become a native (or nativised) va-
riety in Singapore. This shift within Chinese does not, however, entail a
shift towards Mandarin culture (of course dialect groups, being much more
than simple linguistic entities, each exhibit their own traditions and culture).
Typically, the traditional9 is now associated with ‘the dialects’, as opposed to
Mandarin, a new variety, associated with modernity and openness (commu-
nication). Young Chinese tend to overlook differences between dialects, with
which they are no longer familiar, using instead Mandarin and English, and
a whole body of traditions is in danger of being forgotten – a fact sometimes
lamented in public (Chiang 2009; Abu Baker 2009).
1.1.3 The place of English

Ever since Raffles’ arrival, English has had a privileged place in Singapore.
As the language of the colonial masters, it was remote from the populace,
9
‘Traditions’ in this setting refers to a wide array of cultural practices, including religious
customs, associated with weddings, funerals, etc., as well as more everyday concerns such as
superstitions based on dialectal homophony or age-old legends.
who used Bazaar Malay (a pidgin form of Malay) as a lingua franca, and
their own varieties within their community (Hokkien within the diverse
Chinese). Provision for education was poor, and it was only slowly that
schools, mainly in Malay and English, were opened by the government (Erb
2003: 20). After the Japanese Occupation, however, enrolment in English-
medium schools increased every year, reaching 50.4 per cent in 1962 (Platt
1975: 366). In 1987, English was made the only medium of education, with the
three mother tongues effectively taught as second languages. This move was
a logical consequence of the fact that enrolment in Malay-medium schools
had diminished substantially over time, and the last Tamil-medium school
had closed three years earlier due to a lack of pupils (Gupta 1994: 145–6).
Recent calls for a return to providing Chinese-medium education have so
far met with little action from the government (Gupta 1994: 147), although
various schemes for more advanced mother tongue training are seeing some
success: the so-called ‘Special Assistance Plan’ schools are élite institutions
where some subjects are taught in the mother tongue; they currently only
exist for Mandarin (MOE 2008). Statistics on pupils’ home language show a
similar trend, too: among Chinese students starting their primary education,
the proportion coming from ‘English-speaking homes’ was 20 per cent in
1988, 40 per cent in 1998, and 54 per cent in 2007 (MOE 2008).
English, therefore, can be considered a main language in Singapore. As
the language of politics, of the courts, and of education, its status is such
that non-proficient speakers are significantly disadvantaged. It does not come
as a surprise, therefore, that 32 per cent claim to use it as their dominant
home language (Table 1.1). And this cannot be a result of interethnic mar-
riage alone, which stands at 20.2 per cent of all marriages (Singstat 2010).
There must therefore be parents who speak the same mother tongue, but
who decide to use English with their children or between themselves. And
this is not surprising, seeing that all the years they spent in schools meant
daily conversation in English. Furthermore, many parents see English as
an important language, and rightly so: as the medium of education and a
central subject in schools, it is crucial to pupils’ performance. This presents
a further motivation to use English at home, at least with the children. One
example of the position of English with young speakers is given by one
of my informants (i.M.3.m),10 whose hobby is poetry writing on a blog.
As a Malay, he has had Malay as a second language since starting com-
pulsory schooling. Nonetheless, he prefers to write in English, for obvious
reasons:
10
The data used here are referenced according to the following coding scheme: ii.M.gr, where
the first element refers to the school (i = Junior College, ii = polytechnic, iii = vocational
training college), the second to the ethnicity of the participants (C = Chinese, M = Malay,
I = Indian), and the third to the setting of the recording (dia = dialogue, gr = group,
rm = radio-microphone). Individual interviews as well as informants themselves are coded
as e.g. i.C.1.m, where the first two identifiers are as above, the third is a running identifier,
and the last indicates the informant’s sex (m = male, f = female).
I don’t think in Malay . . . My thoughts are mostly in English, so when I

write [in Malay], when I want to associate words with my ideas, they tend
to come out English first, then I translate into Malay. And so it’s hard
to get the smooth flow kind of writing in Malay. So I write my poems in
English.
The predominance of English over the other local languages can also be
seen in less obvious domains. The Singapore Post, for instance, expects ad-
dresses to be written in English (more appropriately, in the Latin script – see
below). A similar point can be made about the translation of place-names. All
languages have contributed something to Singapore’s toponymy: the abo-
riginal nature of Malay is apparent in its references to geographical features
(bukit ‘hill’, sungei ‘river’, telok ‘bay’, pulau ‘island’, etc.), the many Chinese
immigrants named their settlements in their dialects (Ang Mo Kio ‘rambu-
tan bridge’ (Hokkien); Yew Tee ‘oil pond’ (Teochew)), and the British used
‘imperial’ names (Queenstown, Canberra Road, Victoria Crescent, Dover
Road, Commonwealth Avenue, etc.). Few Indian toponyms remain, Dhoby
Ghaut11 being a famous exception (Savage and Yeoh 2005: passim).
Translations or transliterations of place-names also show an interest-
ing pattern. Every street in the Singapore Street Directory (Singapore
Land Authority 2003) has been transliterated into Chinese. Translations of
English place-names are usually quite consistent, with an accurate semantic
rendering: Redhill Road is 紅山路, each character bearing the exact semantic
content of the three English morphemes, and they are even in the same order.
This is obviously to the detriment of phonemic resemblance, the Chinese
being Hóngshān Lù. In contrast, Jalan Mata Ayer is a road in northern Sin-
gapore with a Malay name (jalan ‘road’; mata ayer ‘spring’, lit. ‘eye water’):
its Chinese transliteration is rělán mǎdá yàyı̀, which may well display some
phonemic similarity, but which bears no semantic content relevant to the
original Malay word – rělán, which is consistently used for Singapore’s jalan,
does not refer to a road, path, or anything of the sort, other than by its pho-
netic closeness to the Malay original,12 and mǎdá yàyı̀ is equally nonsensical,
spring being quán in Mandarin.
Perhaps the most striking example, however, is that of Redhill Road
(Hóngshān Lù, marked by on the map on p. 5), which lies not far off Jalan
Bukit Merah ( on the same map). This latter road’s Malay name means
Redhill Road (jalan ‘road’, bukit ‘hill’, merah ‘red’)! It is translated into
Chinese as Rělán Hóngshān. It would appear, then, that if in the ‘English’
(or rather, Latin-spelling) road name, there is a distinction between redhill
11
Dhoby means ‘launderer’ and a dhoby ghaut is the location where washing would take place
(Savage and Yeoh 2005). Today it is the name of a major MRT interchange and its immediate
surroundings.
12
Pinyin r represents the phoneme /õ/ (sometimes [ü], see Appendix A), which is, arguably,
>
similar to the Malay affricate /dZ/, spelt j.
and bukit merah, but in Chinese, both are translated as the same, then their
equal semantics is highlighted in Chinese, but left opaque in ‘English’. The
Malay bukit merah thus becomes little more than a loanword in an otherwise
SSE construction,13 its meaning restricted to those with some knowledge of
Malay.
The hierarchy of languages in Singapore is, therefore, quite clear. The
language of international business, English, comes first, closely followed by
Chinese, the ‘mother tongue’ of the majority and the language of economic
potential. The national language Malay comes third: it enjoys a united speech
community14 but few real advantages from its exceptional legal status. Tamil
fares worst, as it has a small base of speakers and does not benefit from
government incentives as much as the other three.
Language planning in Singapore has been described by several authors
(Bokhorst-Heng 1998; Dixon 2005; Rappa and Wee 2006; Alsagoff 2007;
Wee 2010a), and is evident at several levels. Most importantly, the education
system is usually the first place to feel the effects of new or adjusted policies.
The change in 1987, for example, to a system where English was to be the
only medium of instruction, had a significant impact – although it was only
the official implementation of a shift to English education that parents’ choice
had all but completed (see above). Beside the crucial issue of the language
of instruction, the mother tongue policy bears significant importance to
pupils and parents alike. The mother tongue policy is closely linked to the
ethnic policies of the government, which classify the population into ‘racial’
categories: there are four top-level categories (‘Chinese’, ‘Malay’, ‘Indian’,
and ‘Other’), and one’s membership in one of these is determined by one’s
father’s ‘race’. The classification is then refined, such that in census data
one can find several subgroups (the Indians have eleven, such as ‘Tamil’,
‘Malayalee’, ‘Punjabi’, ‘Hindi’, ‘Gujarati’, etc.). The term race is commonly
used in everyday discourse, void of the negative connotations it typically has
in Europe. A recent change (Parliament of Singapore 2010) has relaxed the
patrilineal nature of these categories, with allowances for ‘double-barrelled
13
It should be noted that while jalan is indeed Malay, it is found, in Singapore, as a ‘classname’
(Pullum 2007) combined with a multitude of languages: Jalan Kayu (Malay), Jalan Bukit
Ho Swee (Malay and Hokkien), Jalan Seaview (English), etc.
14
There are two main ways in which this is true: 82.7 per cent of Malays speak Malay as their
main home language, whereas only 47.7 per cent of Chinese use Mandarin and 36.7 per
cent of Indians Tamil. More strikingly, 98.7 per cent of Malays are Muslims: this is in stark
contrast to the fragmented Indian and Chinese communities with 59.0 per cent Hindus and
43.0 per cent Buddhists, respectively (Wong 2011).
Furthermore, and this is a somewhat delicate issue, a close analysis of the occupational
categories listed in Singstat (2000) reveals that while Chinese, Indians and the ‘others’
group all have more than a quarter of their workers in the two highest-ranking categories
(‘senior officials and managers’ and ‘professionals’), only 7 per cent of the Malay group do.
Conversely, in the three lowest-ranking classes (manual labour and production), the Malays
represent 38 per cent of the workforce, whereas the other three groups all feature with less
than a quarter of theirs.
race options’ (Kor 2010), as well as the option for a child with a ‘Caucasian’
and a ‘Chinese’ parent to be called ‘Eurasian’, a category previously reserved
for children of Eurasian fathers (Wee 2010b).
The initial idea was for each of the three major ethnic groups (Chinese,
Malay, Indians) to be given one of the three official languages, to work as a
mother tongue: Mandarin for the Chinese, Malay for the Malays, and Tamil
for the Indians. The declared aim of the policy was to enable a cultural
grounding in the language of the respective ethnicity’s ancestral language
(notwithstanding the fact that few of Singapore’s Chinese ancestors spoke
Mandarin); in this view, the mother tongue was the vehicle of traditions
and cultural values, whereas English was the useful language of international
trade and regional competitiveness. This separation, which voids English
of its perceived ‘Western’ cultural baggage, has the logical consequence that
English is unsuitable for mother tongue status (Alsagoff 2007). This aspect of
Singaporean language policy and its linguistic effects on Singapore English
will be considered further in the following chapter.
In the education system, the bilingual (English + mother tongue) system
has been in place since 1966 (Pakir 2001). What has changed since then is
the greater flexibility with regard to the choice of the mother tongue: while
Indian pupils of non-Tamil ethnicity have long had the possibility to choose a
mother tongue other than Tamil, ‘such as Bengali, Gujarati, Hindi, Punjabi,
or Urdu’ (MOE 2010b), the traditional link between ethnicity and mother
tongue seems to have (at least legally) now disappeared, since registration
forms for Primary 1 now ask not just for the child’s race, but also for the
mother tongue, listed as either Malay, Chinese, Tamil, or non-Tamil Indian
language, in that order (MOE 2010a).
The language policies in Singapore are under constant review and scrutiny,
being adapted by the planners as the perceived linguistic needs change, and
being subject to the population’s critical appraisal of the requirements. Thus
the compulsory element of mother tongue education has often been criticised,
in particular by parents concerned for their children’s academic success. For
instance, a recent mention (Davie 2010) by the minister for education of the
problems inherent in the weighting of the mother tongue component of the
Primary School Leaving Examination (PSLE), currently at the same 25 per
cent as English, mathematics, and science, has led to various reactions both
online and in letters to the editor, both for and against a lowering of that
weighting. While the concern was primarily about students who excel in the
other three subjects but not in the mother tongue, and who, as a result, end
up disadvantaged in their PSLE results, there were also concerns ranging
from socio-economic and national cohesion to considerations of international
competitiveness (in dealings e.g. with China) and sheer love for the mother
tongues (Chang and Hussain 2010). The debate was promptly resolved by
acknowledging the issue and promising reforms in the teaching methods, but
also by maintaining the original weighting (Goh 2010).
Expanding circle
Brazil
Outer circle d
ea
spr
of
tion
ec
Inner circle dir Russia
India
UK, USA
Nigeria
Canada, Australia
China
Malaysia
Singapore
Germany
Indonesia
Figure 1.3. Kachru’s concentric circles of World Englishes (Kachru

1982, 1985)
1.2 The Englishes of Singapore and their models

It is useful, at this stage, to situate this book within the larger tradition of
research into World Englishes. The extensive body of literature available
on varieties of English worldwide, in particular those beyond the traditional
English-speaking countries in the British Isles, North America, and Aus-
tralasia, has largely been built on the foundations laid by Braj B. Kachru
(1982, 1985, 1992, inter alia). One of his more widely cited contributions
remains the formulation of a model of World Englishes in the form of three
concentric circles (see Figure 1.3), which describe the spread, uses, and
statuses of English around the globe.
In this model, the ‘inner circle’ comprises those countries and places
where English is traditionally spoken as a native language by a large or
significant number of people. This includes the United Kingdom and Ire-
land, the United States of America, Canada, Australia, New Zealand, South
Africa (despite native English speakers being a minority there), and several
Caribbean countries and territories. The number of speakers in this circle has
been estimated at 320 to 380 million (Kachru 1985); Crystal (2006) puts this
number around 400 million. The ‘outer circle’ comprises countries and terri-
tories where English is a ‘second language’: these are typically former British
colonies, where the English language has been retained after decolonisation
for several purposes, in politics, public administration, and/or education.
Countries in the outer circle include Singapore, Malaysia, the Philippines,
India, Bangladesh, Pakistan, Nigeria, Kenya, Ghana, Zimbabwe, and several
others. Here the number of speakers is less easily computed, because English
is never the only language present (hence the label ‘second language’), and
because of the difficulty in obtaining reliable information on the whole pop-
ulation’s English language skills. Estimates range from 150 to 300 million
(Kachru 1985) to 400 million (Crystal 2006). Such numerical computation
is even more difficult in the case of the ‘expanding circle’, which comprises
those countries where English is, essentially, taught as a foreign language.
These are countries that recognise the important role of English in the current
world order, but who have no history of British or American colonisation.
They include China, Japan, Russia, Indonesia, Brazil, Germany, France,
Morocco, and almost every other country on the planet, since it is becom-
ing increasingly difficult to find any country where teaching of English as a
foreign language is not available at all. As a result, the number of speakers
of English within the expanding circle is constantly increasing, and has been
estimated by Kachru (1985) to lie between 500 and 1,000 million.
There is also an evolutionary component to the model, which shows
how the English language has spread geographically from its original core
(the British Isles) to new locations around the globe. Certainly the former
settler colonies found in the inner circle also were among the first to adopt
a form of English as their main language in most or all settings, followed
later by communities in the outer circle colonies, which also happen to
be exploitation colonies, thus receiving fewer inner circle settlers during
colonisation. Countries in the expanding circle, though perhaps exposed to
English before countries in the outer circle (e.g. in Continental Europe) have
not really experienced the current levels of English language teaching before
the establishment of outer circle varieties. Notwithstanding this diachronic
element of the model, it is arguably less powerful than other attempts at
formalising the history and development of postcolonial varieties, such as
Schneider (2007).
This concentric circles model has been subjected to other criticism, largely
for failing to take into account the more complex sociolinguistic realities at
work in the individual settings within each circle (Bruthiaux 2003). Re-
gardless, the model, as formulated by Kachru, was primarily intended to
emphasise the global spread of English, to question ‘traditional notions of
codification, standardisation, models and methods’, and to challenge the in-
ner circle’s ‘prerogative to control its standardisation’ (Kachru 1985: 29–30).
This valorisation of outer and expanding circle varieties, which enables re-
searchers to focus on individual forms of English around the globe without
constantly having to contrast them against British or American English,

has been referred to as the ‘Kachruvian paradigm’ (Canagarajah 1999: 180).
Many of the descriptions of Singapore English referenced herein would be
situated within this Kachruvian approach. They certainly avoid statements
like the ‘frequent sub-standard forms’ found in Tongue (1974: 111), which
implicitly rob the variety of any legitimacy.
The approach taken in this book will, therefore, be grounded in this belief
that outer circle varieties (of which Singapore English is one, in Kachru’s
typology) are to be described in their own right, not simply as weaker versions
of an exonormative inner circle standard, but as fully fledged systems of
their own. Where it is useful, comparisons will be drawn with Standard
British English, and the pronunciation chapter will make use of Received
Pronunciation as a convenient reference accent. In the larger context of
the final chapters, and in the discussion of internal variation, other, post-
Kachruvian models will be applied in order to address the sociolinguistic
shortcomings of the paradigm.
1.2.1 Genesis
English arrived in Singapore with the British East India Company in 1819.
Singapore, at this stage, was little more than a tiny Malay trading settle-
ment (see p. 1). The modern city, therefore, is in fact English-founded. The
colonial administration was conducted in English, but little was undertaken
to increase English proficiency amongst the population. In fact, the mostly
British civil servants were taught Malay throughout much of the early set-
tlement days (Turnbull 1996: 84). It was only after World War II, and more
importantly after independence, that instruction in English became available
to an increasing number of Singaporeans (see p. 4).
Ho and Platt (1993) see the education system as the main contributor to
the emergence of Singapore English (SgE).15 In its early stages, English-
medium education relied on teachers brought in from around the Empire: a
few British and Irish, but also an important number of Indians and Ceylonese
(Ho and Platt 1993: 6), who are still well represented in the profession. This
might account for some of the Indian English features found in Singapore
English, to which I shall turn later. The standard used in English-medium
education was obviously British English, which is reflected in, for example,
SgE’s non-rhoticity.
15
Platt and Weber prefer the term Singaporean English rather than Singapore English, which
is ‘sometimes used in a pejorative manner to suggest a “substandard” variety of English’
(1980: 1). The inflected form is seen as more appropriate since it follows the style of native
varieties (Australian English, British English, etc.). With due respect to these considerations,
I shall continue to use ‘Singapore English’, as it seems to be the accepted term both in the
field of World Englishes and in general usage.
Further and more fundamental influences on the education medium were

the languages of those learning it: early on, few Malays were sent to English
schools, with the result that few of its structural features were carried over
into the emerging variety. Those lexical items that were transferred had
their origin in Bazaar Malay, the pidgin which was used for interethnic com-
munication in the colony, and which itself had strongly Chinese-influenced
structures (Ho and Platt 1993: 8), or indeed in Baba Malay (Pakir 1986). This
latter variety emerged among the first Chinese settlers in Malacca, proba-
bly where they intermarried with the local Malays. Their Hokkien-based,
Malay-lexifier pidgin creolised to become the language of the descendants
of these mixed marriages (Lim 1975: 8, cited in Ho and Platt 1993: 9, see
also Lim 2010). These Babas or Peranakans were among the first to embrace
English as a home language, and very often also educated their children in
English (Gupta 1994: 41, 1998: 108–13; Lim 2010), which brought them into
wealthy and influential positions (former Prime Minister Lee Kuan Yew is
a Peranakan).
The Indian influence, apart from that of the teachers mentioned above,
was limited. Lexical loans are rare and limited to culture-specific terms, and
substratal grammatical influence was nearly non-existent. A major reason
for this is simply the number of speakers. Chinese and Malays greatly out-
numbered Indians, and, therefore, the cross-ethnic variety that emerged was
more highly Chinese-based.
1.2.2 Variation
Since the first book published on Singapore English (Tongue 1974), it has
been apparent that the speech community under investigation is not a ho-
mogeneous one. Initial descriptions of variation were of a prescriptive nature
(the colloquial form of SgE being decried as ‘frequent sub-standard forms’
in Tongue 1974: 111), but subsequent researchers took a more enlightened
approach. The seminal article by Platt (1975) on the nature of this variation
introduced the important concept of the continuum that is still being used
today. More recent researchers (Gupta 1989, 2001) see Singapore’s speech
community as diglossic, whereas others regard variation as reflecting lan-
guage learning proficiency (Pakir 1991). Yet others take speaker agency into
account, citing cultural orientation as a major influence on variation (Alsagoff
2007). This section introduces these approaches, complemented by an alter-
native framework based on work in the field of indexicality (Silverstein 2003;
Eckert 2008; Leimgruber 2009b). All will be investigated more thoroughly
in the next chapter.
The continuum
In describing SgE, John T. Platt used the term creoloid, which he coined
for varieties that match several criteria, including their use as a ‘lingua
Singapore speech SE speech con-

community tinuum
F
Speaker 1
SF
Co
ll
F SF
Speaker 2 SF
Coll
F
Speaker 3
Coll Basilect ‘Singlish’
Pidgin English
F – formal
SF – semi-formal
Coll – colloquial
Figure 1.4. ‘Relation between socio-economic factors and the usage of

sub-varieties of [SgE] available to a speaker’ (Platt 1975: 369)
franca in inter-ethnic group communication’ (Platt 1975: 372). The concept,

however, has fallen out of use, particularly for its grounding in a view of
contact linguistics that presupposes a pidgin to creole development (cf. its
other criteria that say that a creoloid is creole-like in linguistic form without
having sprung from a pidgin (Platt 1975: 372)). In Platt’s view, a ‘multilingual
society where an attempt has been made to tolerate or even further the
existence of several national languages side by side’ (1975: 372–3) would
provide the ideal setting for such a variety to emerge – in short, a place like
Singapore.
In his analysis of variation within the variety, Platt sees SgE as consisting
of a number of lects, positioned on a continuum ranging from a basilect
to an acrolect. As in Jamaica (DeCamp 1971), different lects may be used
by the speaker to interact in different social environments. The speaker’s
‘position on the continuum, coupled with socio-economic and educational
factors, [determines] the number and types of sub-varieties which are at
his disposal’ (Platt 1975: 369) (see Figure 1.4). The ‘higher’ the sociolect a
speaker achieves, the wider his range of available lects: while not everyone
masters Standard Singapore English (SSE, the acrolect), all SgE speakers are
assumed to be able to use Colloquial Singapore English (CSE or ‘Singlish’,
the basilect) when deemed appropriate. The example Platt gives is that of
university students who converse in SSE with their lecturers, but use CSE
to address a waitress, the latter only having CSE at her disposal.
This model accounts well for variation within SgE: SSE speakers do not
indeed stick to their acrolect when conversing about mundane matters, and
the idea of the speaker sliding up and down the sociolectal scale according to
the situational setting is particularly well-suited to explain the large variation
found within SgE. Its shortcoming is that not all speakers of SgE are, in fact,
proficient in the basilect. This may be a recent development, but reports of
English-educated Singaporeans struggling to understand it (Hussain 2006)
point to the weaknesses of this model.
Diglossia
Anthea Fraser Gupta applies ‘Ferguson’s use of diglossia’ (1994: 7) to Singa-
pore: Charles Ferguson (1959) uses the term diglossic for a speech community
in which there is a superposed (H) variety, which is learned through formal
education rather than acquired natively. This H variety is used in written
and formal contexts and is in complementary distribution with the everyday
L variety, which is the normal code for communication in the community. In
Gupta’s (2001) model, Standard Singapore English (or SSE) is H, and Collo-
quial Singapore English (CSE, or ‘Singlish’) is L. While she does not assume
that everyone speaks H, she accounts for variation as a function of switching
between H and L, the speaker being aware of doing so and exploiting it for
functional purposes.
This explanation is attractive, firstly because it shows variation as a matter
of personal choice (Low and Brown 2005), rather than as a function of a
speaker’s educational level (cf. Platt 1975), and secondly, this is how Singa-
poreans typically perceive SgE themselves: ‘Singlish’ versus ‘Good English’.
Equally, it stresses the fact that SgE is acquired natively, rather than via the
school system, as used to be the case. However, explaining variation as the
result of mixing of two codes is problematic: diglossic speech communities
do not normally code-switch between H and L intrasententially, the two
codes being distributed functionally. Ferguson’s definition, on which Gupta
(1994) bases her analysis, states clearly that H ‘is not used by any section of
the community for ordinary conversation’16 (Ferguson 1959: 435).
Alternative models
Pakir (1991) proposes a different model to describe the English of ‘English-
knowing bilinguals’. She sees SgE as varying in two dimensions: formality
and proficiency. In Figure 1.5, Standard (Singapore) English is at the top,
CSE at the bottom. The triangles reflect the fact that speakers who are more
proficient in English have a wider array of styles at their disposal. Speakers
with the largest triangles will be able to shift ‘downwards’ and use CSE for
various purposes. Speakers with fewer years of education (Pakir’s diagnostic
16
This is certainly true of speech communities such as those of Czech and Swiss German, for
instance.
SSE
Formal Advanced
Cli
ty
ali
ne
rm
o
fp
f fo
rof
o
Careful Adept
ne
icie
Cli
ncy
Consultative Intermediate
Casual Basic
Intimate CSE Rudimentary
Figure 1.5. ‘Expanding triangles of English expression’ (Pakir

1991: 174)
indicator of proficiency) will have access to a smaller range of stylistic varia-

tion. In other words, ability to style-shift is again tied to education. This is in
fact not very much different from Platt (1975), except that she calls his ‘lectal
continuum’ a ‘cline of proficiency’ – education remains the main diagnostic
feature.
Poedjosoedarmo (1995) modifies Pakir’s (1991) triangles model, and uses
the labels acrolect, mesolect and basilect. However, as shown in Figure 1.6,
she places the triangles in such a way that their bases do not match, which pre-
vents acrolectal speakers having the full range of the mesolect. Her basilect,
which she puts on the same level as a ‘pidgin form’ used by speakers of other
languages, directly connects, with its summit, to the mesolectal triangle’s
base. Low and Brown (2005) recognise that this is problematic in that it
suggests that neither acrolectal nor mesolectal speakers ever use the basilect,
which, however, they seem to do, if only for humorous purposes.
New models
Alsagoff (2007) reviews the models presented, and finds faults firstly in Platt’s
continuum because of its dated view of the variety as largely non-native, and
secondly in Gupta’s diglossia for its unsatisfactory dealing with features of
H in otherwise L discourse or vice versa. She recasts the variation in SgE in
a new dimension, using a ‘cultural orientation model’: this model renames
Singlish as ‘Local Singapore English’ (LSE) and the standard ‘International
Singapore English’ (ISE), the two being extremes at a continuum of cultural
orientation marked by localism on the one hand and globalism on the other.
The orientations themselves come in several guises, such as sociocultural vs
RP Pronunciation (media & American Pronunciation (media an-

some professionals in formal nouncers on game shows and some
daily speech) sports & music shows)
ACROLECT
Syntactically & lexically standard

with local pronunciation features
Cline of formality Cline of

proficiency
Stylistic range for

educated speakers
Colloquial style for many educated speak-

ers. Primary code in English for secretaries,
some taxi drivers, etc. Distinctive syntactic
& lexical as well as pronunciation features
MESOLECT
Pidgin form: uneducated speakers of other lan-

BASILECT
guages use this for interethnic communication
Figure 1.6. Modified triangles (Poedjosoedarmo 1995)
economic capital, informality vs formality, and community membership vs

educational attainment, to name but three pairs. Speakers then combine
features of LSE and ISE in any interaction, depending on which orientation
they choose, and to what extent they decide to do so: a politician may,
for instance, deliver a speech largely in ISE, underlining his suitability for
the job by displaying educational attainment, but with the occasional, well-
placed element of LSE, in order to display community membership – both
important stances in the given context. In short, the cultural orientation
model goes beyond diglossia as far as speaker agency is concerned. The
model will be explained in more detail in Chapter 2.
A further extension of Alsagoff ’s model can be seen in variationist models
based on indexicality (Silverstein 2003; Eckert 2008). This approach con-
siders every linguistic variable to index (to point to, to mark) one or more
social meanings, understood by the speaker and addressee consciously or
unconsciously. The range of a particular variable’s social meanings can be
captured by an ‘indexical field’ (Eckert 2008), which, in Eckert’s example of
/t/ release in American English, includes stances such as ‘angry’, ‘formal’,
‘careful’, ‘emphatic’, and so on.
Indexical analysis takes into account several layers of reference – ‘orders

of indexicality’ in Silverstein’s (2003) terms. By way of example, the sen-
tence in (1) has a layer of ‘pure’ semantic meaning, a first-order indexical
(pragmatic) meaning, a second-order indexical (metapragmatic or sociolin-
guistic) meaning and, possibly, a ‘higher-order’ (conventionally recognised)
indexicality.
(1) Kot.ikirtuku need three days ah. (ii.I.gr)
¯
The referential, semantic meaning is a question whether one would need
three days to eat the amount of food discussed (colloquial Tamil kot.ikirtuku
‘to eat a lot’); at the first indexical level, it is a rebuke to the preceding¯ turn
suggesting a three-day trip to Thailand in order to sample the local fare. A
second level of indexicality operates at the level of the discourse particle ah,
which is commonly ascribed to Singlish, and, as a result, places the utterance
within that subvariety, and, therefore, in an informal style. Finally, the Tamil
word at the beginning of the utterance may be an instance of a higher-order
indexicality, firstly, in that a shared background is needed in order to simply
understand the switch, but also because of the effect the switch has on
the conversation: the topic of food (particularly the concept of eating large
quantities thereof) may be removed from the relatively formal atmosphere of
the setting in which this recording was made. A switch away from the school
language English (in particular into the L variety of an official language)
may therefore be recognised by the participants as indexing a particularly
anti-prescriptive stance.
1.3 Data and methods

The methodology used here has a lot in common with previous studies of SgE,
primarily in its reliance on the education system as a source of informants
(see e.g. Pakir 1991; Chew 1995), where others relied on corpora (Deterding
2003b; Bao 2005; Bao and Hong 2007) or on observation in informants’ homes
(Kwan-Terry 1991; Gupta 1994; Bokhorst-Heng et al. 2007). Similarly, the
use of sociolinguistic interviews in the Singapore context is not new (Platt and
Weber 1980; Ho and Platt 1993; Bokhorst-Heng et al. 2007). The internal
structure of the interviews, however, is, even though they use, of course,
a lot of the methods pioneered in early sociolinguistic works (Labov 1966;
Trudgill 1974).
The informants in this study come from three post-secondary institu-
tions. This choice was motivated by the educational landscape of the Re-
public, which makes a three-tiered post-secondary selection simple. After
secondary school, 83.8 per cent of students carry on with some sort of training.
They have three options: the top 26.1 per cent (MOE 2006: xii) proceed to
junior college (JCs), a pre-university institution that awards the Singapore–
Cambridge ‘A’ Levels after two years of study. The second option is one of
Junior college
Polytechnic
Vocational training
Figure 1.7. Post-secondary enrolment in 2005 (data from MOE

2006: xii)
the five polytechnics: they are chosen by 34.9 per cent, and offer three-year
diploma courses, with the possibility of pursuing higher degrees afterwards.
The third option is the Institute of Technical Education (ITE), the voca-
tional training institute: it awards a certificate after a two-year course, and
is subdivided into three colleges; 22.9 per cent of secondary-school leavers
take this option.17 This distribution of students can be further subdivided
according to the proportion of post-secondary students per type of school.
Thus junior colleges account for 31.1 per cent, polytechnics for 41.6 per cent,
and the ITE for 27.3 per cent (see Figure 1.7). This near-equal distribution
(with a standard deviation of 7.4) just about justifies the use of equally sized
samples from these three categories.
Entry requirements to these institutions offer some information on the
competence of their students. Admission to any post-secondary institution
is decided during the yearly Joint Admissions Exercise, during which every
student’s ‘O’ Level results are computed into different types of aggregates.
Junior college requires an aggregate taking into account English and five other
subjects relevant (abbreviated to L1R5, MOE 1995) to the course chosen (e.g.
students wanting to take the chemistry option in junior college will have their
‘O’ Level chemistry results in the aggregate). Polytechnics use an aggregate
based on English, two subjects relevant to the course chosen (e.g. geography
for business-related courses), and the two subjects where the best marks
were obtained (ELR2B2) at ‘O’ Level. The Institute of Technical Education
simply requires passed ‘O’ Levels. This system gives an indication of the
kind of student competence one might expect in each of these schools, and,
17
An important advantage of the Singapore system is that it is highly permeable: students
almost always have the possibility to further their studies, if they fulfil certain requirements.
Thus an ITE graduate may proceed to polytechnic, and a polytechnic student to university,
if the necessary grades are achieved.
by extension, of English language proficiency: clearly the academic standard

is highest for junior college and lowest for vocational school.
Within each school, an equal number of informants were selected from the
three majority ethnic groups: four Chinese, four Malays, and four Indians.
All were in their first post-secondary year, and of an average age of 17.5
years. English being the medium of education and the language generally
associated with school, students of this age range, who have had nine or ten
years of schooling in the language, and who will probably have more, are at
a stage where their use of the variety peaks, and adolescents are generally
accepted to be influential actors in language change (Rampton 1995; Kerswill
1996; Aitchison 2001). Furthermore, I hoped that at this age, there would be
fewer inhibitions about speaking openly, even to a stranger.
Recordings were made with a Sony® Hi-MD ‘Walkman’ MZ-NH700
mini-disc recorder, using a Sony® ECM-MS907 ‘Electret Condenser’ stereo
microphone. For the radio-microphone recordings, a K&K® AT-668 ‘Super
Professional’ wireless microphone was used. The resulting recordings were
of a high quality, although certain radio-microphone recordings suffered
from various electromagnetic interferences, such as mobile phone signals.
The data on mini-disc were then transferred onto a computer using the
Sony® ‘SonicStage’ software (version 4.0.00.05080). From there, data were
exported and converted into Microsoft waveform audio format (.wav) for
easy accessibility. With one file per recording session, this resulted in seventy-
two .wav files, totalling over 16 12 hours (16:32:12) of recording time and
9.78 GB of disk space.
The recordings themselves were of four types, deliberately chosen to
create different levels of formality. It was assumed that individual and
dialogue recordings would involve more formal English than group and
radio-microphone recordings, due in part to the heavy involvement of the
researcher in the first two. A closer look at each situational setting is given
here:
r Individual interview: each informant was singly interviewed for 15 min-

utes by me. We both sat in a (ostensibly) sound-proof room, facing each
other. The setting was deliberately formal, with the informant being
asked open-ended questions. While it was made clear to them before-
hand that they were expected to be doing most of the talking, in some
cases this proved problematic: one young woman (iii.I.2.f), for instance,
had problems overcoming her shyness and more often than not, resorted
to smiles and single-word replies to evade talking too much. Of course
this could also be explained by my inability to make her feel at ease.
Questions asked were of a general nature, and it was hoped that they
would elicit as many of the variables under investigation as possible.
Priority was given to topics that it was hoped would be informal, such as
holidays, which, fortunately, were always within a few weeks before or
after the recording dates. Questions about the students’ activities in their
free time, their hobbies, and part-time job experience were also used, as
well as the question about their career plans. More school-related topics
were co-curricular activities,18 relationships within the class, and course
content.
r Dialogue interview: Once all four members of the group had been inter-
viewed individually, informants paired up (they decided with whom), and
were interviewed. The format was that of a debate, with the two sides de-
fending opposing views. The topic was, again, set by the researcher such
that it was hoped that it would not be overly formal. Holidays seemed,
again, a good choice. The initial task was, therefore, for one of the infor-
mants to defend the point of view that it is best to stay within Singapore
during the holidays, while the other was to defend the opposing view that
going abroad19 was the better choice.
r Group recordings: Here all four informants from the same ethnic group
were asked into the recording room, and given a task. I would then leave
them to their discussion, physically leaving the room, and returning fif-
teen minutes later. In line with the topics addressed in earlier interviews,
the task also focussed essentially on holidays. Informants were asked to
formulate a plan for the following holiday, given a budget of SGD 300
per person, involving a trip to a place where none of them had ever been.
They were told to plan travel, accommodation, and activities.
The only physical clue to the unusual nature of the situation was the
presence of the recording device and the microphone, to which some
allusions were made (ii.C.gr), and maybe the situation of finding oneself
in a room where one would not normally encounter one’s classmates.
r Radio-microphone recordings: The last fifteen minutes were recorded with
the use of a radio-microphone, in a location that was meant to be less for-
mal: usually a cafeteria or a school canteen. While informality was indeed
achieved in this way, background noise and interference (particularly
from mobile phones) were a major factor in diminishing the quality of
the recordings. In this setting, informants were not given any particular
topic or task, they were simply asked to talk among themselves.
The rationale behind this structure of four different types of recordings
was to proceed from a formal setting (individual interview), through a less
formal one (dialogue interview) and a more informal one (group recording), to
an informal setting (radio-microphone recording). It was thought, and later
confirmed, that informants would feel initial unease at the situation, with
18
In the Singapore context, co-curricular activities, or CCAs, are extra-curricular activities such
as participation in school-based sports teams, marching bodies, or music groups. While they
do not form part of the curriculum as such, they are compulsory and organised by the school.
19
Overseas was the preferred term in all contexts, with 101 occurrences (78.3 per cent)
compared to 28 (21.7 per cent) for abroad.
1.4 Conclusion 25
doubt about what was expected of them, giving rise to more careful speech:
it was therefore only natural to begin with the more formal setting, working
my way through to the less formal ones. Once they had grown accustomed
to me, more spontaneous speech would occur – particularly once I left them
alone to talk during the last two kinds of recordings.
1.4 Conclusion
This introductory chapter has given background information that will be
of importance with reference to the issues central to this book. The histor-
ical approach taken in the first half of the chapter sets the variety under
investigation into sociohistorical perspective; this is important for chiefly
sociolinguistic research. The diachronic account of SgE and its defining fea-
tures provide a background against which the more detailed investigation
of the following chapters will be more easily discerned. Finally, the variety
of models proposed for the definition of SgE already point to the main en-
deavour of this book: the refinement of these attempts at modelling variation
more accurately.
2 Variation in Singapore English: old and
new models
2.1 Early models and diglossia

2.1.1 The continuum hypothesis
In his seminal 1975 article, John T. Platt introduces two important concepts
into the way in which SgE is viewed. Firstly, he applies the concept of a
post-creole continuum, championed by DeCamp (1971), to the SgE speech
community. His is a much more dynamic model than Ferguson and Stewart’s
formulae. Secondly, he introduces the term creoloid to describe a particular
type of contact variety. I will first examine the concept of the post-creole
continuum, before explaining the reasons for Platt’s choice of the term
creoloid.
The Jamaican continuum

Much of Platt’s inspiration for the post-creole continuum analysis comes
from DeCamp’s work on Jamaican English, which considers ‘post-creoles’,
i.e. creoles that ‘gradually merge with the corresponding standard language
[that is, their lexifier]’ (1971: 349). This is not always the case; similar sit-
uations may result in quite different outcomes: Haitian Creole French, for
instance, has continued co-existing with Standard French in a diglossic sit-
uation without any sign of merger (Ferguson 1959; DeCamp 1971: 351).
DeCamp (1971: 350) defines a post-creole situation as one where he says
there is ‘no sharp cleavage between creole and standard’. Rather, ‘there is
a linguistic continuum, a continuous spectrum of speech varieties ranging
from . . . “bush talk” or “broken language” . . . to the educated standard’ (De-
Camp 1971: 350). This continuum, made up of its extremes, the basilect and
the acrolect, and the intervening mesolects, presents a seamless succession of
subvarieties: it is in fact so fine-grained that it cannot be subdivided into dis-
crete lects. Speakers, and this is crucial, use a span of the continuum, rather
than a single lect. The ‘breadth of [this] span depend[s] on the breadth of
[the speaker’s] social contacts’ (DeCamp 1971: 350).
In defining DeCamp’s continuum, Rickford (1987) characterises it
by (i) non-discreteness and (ii) unidimensionality. The first of these,
26
non-discreteness – the impossibility of subdividing the continuum into iden-

tifiable lects – is, by its very nature, in direct conflict with the diglossia ap-
proach also proposed for some creoles, such as Haitian (Ferguson 1959).1
Patrick (1999) gives a good review of the major question this raises, namely
that of the nature of the grammar or grammars of the acrolect and the basilect
in such a continuum: it can either be two discrete ones, distinct from each
other, or two grammars linked by variation. Rickford (1987: 15) argues that
there are two grammars that are ‘polar varieties between which there is con-
tinuous variation’, whereas Winford (1993: 8–11) sees the two grammars as
‘two systems’ with ‘radically different (and competing) grammars’ where
there is ‘code-switching’. Winford seems to have difficulty accepting the
idea of a continuum: his view is largely reminiscent of diglossia, and in no
way reflects the non-discrete character observed in continua. According to
Patrick (1999: 9), the concept of creole speech communities as consisting
of ‘separate, self-contained linguistic systems’ is a ‘structuralist idea’, and
according to him, therefore, ‘weak’. The kind of code-switching advocated
by Winford is by necessity unsystematic ‘since no grammar apparently con-
tains rules governing proper alternation with a separate grammar’ (Patrick
1999: 9). He thus refutes the generative approach taken by DeCamp (1971),
arguing that only if the ‘basilect or acrolect prove to be homogeneous and
invariant, structuralist or generative approaches are useful’ (Patrick 1999: 9–
10) – variationist linguistics, according to Patrick, is the key to heterogeneous
speech of this kind.
The second of Rickford’s characteristics is unidimensionality: the variation
in the continuum ‘can be ordered in terms of a single dimension’ (Rickford
1987: 22). This is very much what DeCamp argues for when he rejects ‘the
pigeonhole technique[s]’ (1971: 355) of sociolinguistics: in his data, he finds
variation along the lines of ‘age, sex, occupation, ethnic group’ relatively
insignificant (1971: 357). When such variation happens, he brings it down
to his continuum of features: the example is that of children and older
speakers, who ‘tend more towards the creole end of the continuum than
do young adults’ (DeCamp 1971: 357). Geographical variation is the only
exception which DeCamp allows, but he is quick to confine this to the
lexicon, and says it does ‘not seriously interfere with the analysis of the
linear continuum’ (1971: 357). Unidimensionality, however, is criticised by
Le Page and Tabouret-Keller (1985), and also by Patrick, who questions this
theory by suggesting that it ‘seems unlikely to do justice to the complexity
of Caribbean social life’ (Patrick 1999: 15; a statement reflecting his insider
knowledge of the Jamaican speech community).
1
DeCamp (1971: 368) notes that many speakers of Jamaican Creole ‘persist in the myth that
there are only two varieties: the patois and the standard’, i.e. essentially a diglossic situation.
He refutes this by pointing to the varying types of ‘patois’ and ‘standard’ used by different
speakers.
28 Variation in Singapore English: old and new models
The conditions listed by DeCamp for the emergence of a post-creole

continuum are twofold: firstly, the dominant official language must be ‘the
standard corresponding to the creole’ (1971: 351), i.e. the dominant language
must be the lexifier. This condition would seem an obvious one. It is clear
that an unrelated standard language could not act as a catalyst towards a
continuum of subvarieties leading from the creole to the standard. Consider
the case of Sranan Tongo, an English-lexifier creole spoken in Surinam,
where the official language has been Dutch for the last three hundred years
(Sebba 1997: 146–55). The presence of a related lexicon in the official lan-
guage would certainly have helped such a continuum to emerge there; as it
is, no such continuum exists in Surinam. Secondly, the social stratification
so typical of early creole speech communities must have ‘partially (not com-
pletely) broken down’ (DeCamp 1971: 351, emphasis in the original): there
needs to be the possibility of at least some upward social mobility in order
for intermediate varieties to emerge. According to these criteria, therefore,
communities where post-creole continua are most likely to emerge are ones
that allow a fair amount of social (upward) mobility, and, crucially, where
the lexifier is in continued contact with the creole, acting as the standard
(possibly the official) language of the polity. Unsurprisingly, Jamaica fits
these conditions, as does Singapore (Platt 1975).
It should be noted that this approach presupposes an initial stage where
only the lexifier and the creole were in existence, and the mesolects came
about later, when speakers of the creole started having access to the standard
by such means as education, and access to occupations that would previ-
ously have been the sole preserve of speakers of the superstrate. These are
assumptions that have been challenged (Alleyne 1980; Rickford 1987), and
the current view seems to be that where a continuum exists, it grew nat-
urally together with the creole itself. In other words, rather than a simple
two-way segregation of lexifier- and creole-speakers, there were a number
of intermediaries between the two, who would have had varying exposure to
the lexifier (Alleyne 1980: 184). The distinction between domestic and field
slaves, for instance, would have helped with the emergence of a hierarchy of
lects within the substrate population.
I now turn to the actual nature of DeCamp’s continuum. Table 2.1 shows
six features used to grade the continuum. Each one (A–F) has two antithetical
categories, a standard ([+A], [+B], . . .) and a creole ([−A], [−B], . . .) variant.
A, B, and E are lexical variables (where the standard has child, the creole
has pikni), C and D are phonological (the standard has a contrast between
thick and tick, the creole does not), and F is morphosyntactic (the standard
negates the past auxiliary verb have morphologically, the creole does so
syntactically). While the acrolect would have the standard variant for each
of these, the basilect would have the creole variant for all of them. It is the
intervening lects, which combine standard and creole variants, that make up
Table 2.1. Continuum ‘of seven speakers, each of which differs from the
other six by one or more of six features’ (DeCamp 1971: 355, table 1)
Features Speakers
+A child −A pikni 1. +A +B +C −D +E +F
+B eat −B nyam 2. −A +B −C −D +E +F
+C /T ∼ t/ −C /t/ 3. −A +B −C −D −E −F
+D /D ∼ d/ −D /d/ 4. −A −B −C −D −E −F
+E granny −E nana 5. +A +B +C +D +E +F
+F didn’t −F no ben 6. +A +B −C −D +E +F
7. −A +B −C −D +E −F
Table 2.2. Speakers from Table 2.1 rearranged

by variants used
Ranking + −
5. +A +B +C +D +E +F 6 0
1. +A +B +C −D +E +F 5 1
6. +A +B −C −D +E +F 4 2
2. −A +B −C −D +E +F 3 3
7. −A +B −C −D +E −F 2 4
3. −A +B −C −D −E −F 1 5
4. −A −B −C −D −E −F 0 6
the continuum linking the two. An illustration of these can be seen in the
column ‘Speakers’ in Table 2.1.
Speaker 5 is using only standard variants, and speaker 4 only creole vari-
ants. All other speakers use a combination of standard and creole variants.
These data can be handled in two different ways. First, speakers can be
ranked according to their use of one or the other variants, effectively re-
sulting in a continuum of sociolects (illustrated in Table 2.2): the speaker
with six acrolectal variants (5) appears at one end, the speaker with five next,
and so forth down to speaker 4, who has no acrolectal variants. The features
themselves can also be ranked in terms of their co-occurrence: for instance,
it appears that feature D has [+D] only once, namely when all other features
are also present, while [−F] only occurs where all of A, C and D also have
the [−] variant. This can be replicated for all features, giving rise to the
implicational ranking given in Table 2.3.
This procedure enables an implicational ranking of linguistic features:
when a speaker makes a distinction between there and dare, he/she will also
make the distinction between thought and taught, and use child, eat, granny
and didn’t rather than pikni, nyam, nana and no ben, i.e. only standard variants.
Table 2.3. Speaker ranking from Table 2.2 with

features rearranged by variants used
Ranking + −
5. +B +E +F +A +C +D 6 0
1. +B +E +F +A +C −D 5 1
6. +B +E +F +A −C −D 4 2
2. +B +E +F −A −C −D 3 3
7. +B +E −F −A −C −D 2 4
3. +B −E −F −A −C −D 1 5
4. −B −E −F −A −C −D 0 6
If on the other hand pikni and didn’t are used by the same speaker, the data in
Table 2.3 predict that the speaker also has eat and granny, but no distinction
between interdental fricatives and alveolar stops. We are thus presented with
a true scale of subvarieties, co-defining one another by the realisation of a
particular variable: once the list of variables is established, and the variables
are ranked ([±A], [±B], . . . [±n]), one can formulate a rule ‘whereby the
presence of any index feature implied the presence of all other index features
of lower number’ (DeCamp 1971: 353).
The Singaporean continuum

This model, developed for Jamaica in the 1970s, has been applied to several
other speech communities by subsequent authors (see, inter alia, Bickerton
(1973) for Guyana, Chaudenson and Carayol (1979) for La Réunion, and
Rickford (1977) for African American Vernacular English). The model is also
‘on the whole’ accepted for Singapore by Platt (1975: 366) four years after
its inception by DeCamp. Unlike DeCamp’s data, however, which suggest
that speakers are ‘normally distributed’ (1971: 358) along the continuum,
with a complete absence of the possibility of identifying subvarieties, Platt’s
data do seem to suggest a lesser degree of non-discreteness, with a certain
amount of what he calls ‘bunching – particularly at the [basilectal] end of the
continuum’ (Platt 1975: 366). The lack of an explanation of this ‘bunching’
does however not stop him from postulating the existence of an identifiable
basilect (‘Singlish’).
Grammatical variables used by Platt include noun plural marking, copula-
deletion, past tense marking, and reduplication (Platt 1975: 370, also in Ho
and Platt 1993), but here the emphasis is on phonological variables, which
are ‘the most distinctive qualities of [SgE]’ (1975: 370). While these variables
too come in the binary form introduced above for DeCamp’s features (e.g.
for copula deletion: (be) = [∅], [be] and for word-final consonant deletion:
(C#) = [∅#], [C#], i.e. one CSE, the other SSE), they do not display the
implicational hierarchy that was illustrated in Table 2.3. It is therefore not
Singapore speech SE speech con-

community tinuum
F
Speaker 1
SF
Co
ll
F SF
Speaker 2 SF
Coll
F
Speaker 3
Coll Basilect ‘Singlish’
Pidgin English
F – formal
SF – semi-formal
Coll – colloquial
Figure 2.1. ‘Relation between socio-economic factors and the usage of

sub-varieties of [SgE] available to a speaker’ (Platt 1975: 369)
possible, as in Jamaica according to DeCamp (1971), to predict the behaviour

of a given variable by that of another.2 A large sample of the speaker’s speech
is required in order to establish the occurrence rates of CSE and SSE variants,
and by that figure place the speaker’s variety on the continuum.
This is perhaps less of a concern for Platt, since his model is mostly
concerned with speakers’ abilities to use a certain span of the continuum for
functional and stylistic purposes. The model, illustrated in Figure 2.1, shows
an increase in the number of subvarieties available to a speaker the further
up the speaker is on the socio-economic scale. As a working hypothesis, the
three speakers in Figure 2.1 can be taken to be representative of three broad
social classes: upper, middle, and lower. All three have in common their
colloquial subvariety: the basilect, the lower end of the speech continuum.
The lower-class speaker has, in addition to his/her colloquial subvariety, a
formal subvariety, which is, however, only slightly less basilectal than his/her
colloquial one – in fact both fall within the span of the continuum defined by
Platt as ‘Singlish’. The span or range of subvarieties available to this speaker
is, therefore, quite narrow. The middle-class speaker, on the other hand, has
an already much wider range of subvarieties at his/her disposal: in addition
2
Except for the variable (be), which can be analysed as subvarieties according to environment
(e.g. pre-adjectival, pre-locative, post-pronominal, clause-final, and ten others (Ho and Platt
1993: 31–2, 40–52)).
to the colloquial form (‘Singlish’), there is an intermediate ‘semi-formal’ one,

and a ‘formal’ subvariety that corresponds to his/her position on the socio-
economic scale. This speaker’s span would appear to be more than three
times as wide as that of the lower-class one. Finally, the upper-class speaker
commands a range of subvarieties spanning almost the complete continuum:
a formal subvariety (presumably SSE), a colloquial one (a form of ‘Singlish’ of
a nature comparable to the lower-class speaker’s formal subvariety), and two
semi-formal intervening lects, which are distributed roughly equidistantly
between the formal and colloquial extremes.
The model seems to suggest that from a continuum of a technically infi-
nite number of subvarieties, speakers at different positions on the socio-
economic scale have at their disposal a fixed number of lects: certainly
Figure 2.1 shows the speaker I have called ‘upper-class’ as speaking an
acrolect, a basilect, and two mesolects, whereas the lower-class one speaks
two basilects (a formulation which is in itself unfortunate). It is not immedi-
ately obvious how the width of the span influences the number of subvarieties
commanded: in the case of a discrete hierarchy of subvarieties, this would not
pose any problems (the speaker commands an individual high,3 an individual
low, and all intervening lects), but if there is a non-discrete continuum (as
proposed in DeCamp 1971 and taken on board by Platt 1975: 366), then it
is hard to see how discrete subvarieties are selected from that continuum to
form part of the speaker’s repertoire.
One could of course interpret the subvarieties illustrated in Figure 2.1 as
abstractly standing for a range of subvarieties, resulting in a speaker being
able to ‘slide’ up and down the span delimited by his/her individual high and
low. Platt’s formulation, however, leaves little doubt that this was not his in-
tent: there is reference to ‘the number and type of sub-varieties’ (suggesting
distinct, identifiable subvarieties), and to ‘drop[ping] comfortably . . . into
the basilect’ (1975: 369) (suggesting discreteness of the continuum). A sim-
ilar point is made in Platt and Weber (1980: 109), where four subvarieties
(acrolect, upper and lower mesolect, and basilect) are identified by means
of educational achievement. This is unlike the Jamaican continuum, where
there is no lectal ‘grouping’ and which is defined as non-discrete (DeCamp
1971: 350, 358; Rickford 1987: 15). A related point where Platt’s continuum
differs from the Jamaican one is in the nature of the basilect: the ‘bunching’
(1975: 366) at the lower end of the continuum can be seen in Figure 2.1,
where several subvarieties group to form what Platt labels ‘Singlish’. This he
considers a subvariety in its own right, which ‘warrants particular attention’
(1975: 370) due to its resemblance to creoles (see below). Again, the illustra-
tion given seems to suggest variation within this range, despite it being called
the basilect – which would suggest a uniform variety.
3
The terms high and low are herein taken to mean ‘SSE-oriented’ and ‘CSE-oriented’ respec-
tively, and do not carry any value judgements.
The second dimension4 of the model in Figure 2.1 – its left-hand, socio-
economic scale – is defined in terms of social class and education: it is
‘the speaker’s social status and educational background’ (Platt 1975: 368)
that define his/her position on the social scale. Throughout Platt’s article,
however, as well as in Ho and Platt (1993) and Platt (1977), the emphasis
is very much on the level of education achieved by his informants, ranging
from primary to tertiary (i.e. the highest educational qualification obtained).
This is then taken as the sole social variable used to define their position
on the scale labelled ‘Singapore speech community’. This position, Platt
found, was closely correlated with ‘the highest point [they] could reach on
the continuum’ (Platt 1975: 368). In Figure 2.1, the two scales (social and
linguistic) are linked by arrows, standing for the various subvarieties at the
disposal of the speaker. The scales are arranged in such a way that the
speakers’ highest lect on the continuum and their position on the social scale
are at the same horizontal level,5 underlining the importance of the highest
lect in determining a speaker’s repertoire. An individual’s lowest lect, as
part of the model, is the basilect: notice the arrow labelled ‘colloquial’,
which in every speaker points to a subvariety located within the span of
the continuum designated as ‘basilect (“Singlish”)’ – the ‘bunching’ Platt
refers to (1975: 366) – with the implication that every speaker of the speech
community is proficient in the basilect. It is unclear how true this assumption
still is nowadays. Certain members of the speech community’s upper social
classes have expressed intelligibility concerns when confronted with the
basilect: Hussain (2006), for instance, reports on an ‘upper middle-class,
English-speaking’ Singaporean who was first exposed to CSE during his
National Service, and ‘often . . . could not understand his platoon mates’.
4
This departs from the unidimensional model proposed by DeCamp, who argues that social
variables have a relatively insignificant correlation with his continuum (1971: 357, see also
p. 27 above).
5
This is an important point of abstraction in the model: while it is straightforward to have
these two points coincide for any individual speaker (illustrated in (i) below), a system with
two speakers may already necessitate a readjusting of the two continua’s scales (cf. (ii)). As
soon as more than two speakers are involved, however, this readjustment may no longer
work since it would also have to take into account the relative social and linguistic positions
of subsequent speakers – at which point the system might collapse, as in (iii) below. Platt
does not address this issue, which must mean Figure 2.1 stands for a much more abstract
model than what might have been.
(i) 100 100 (ii) 100 82.5 (iii) 100 82.5
S1 S1
90 80 90 80
S1
90 80 80 77.5 80 77.5
S2 S2
70 75 70 75
80 60 60 72.5 60 72.5
S3
70
67.5
While this may be a recent phenomenon – and certainly not a widespread

one – the fact that the said Singaporean has since set out to create his own
online dictionary (Lee 2004) further weakens objection, since it is primarily a
lexical issue,6 with minor phonological examples: grammar is not considered
at all. It would therefore seem that every member of the speech community
does indeed have at least passive knowledge of the basilect (barring certain
lexical items, particularly highly socially marked ones).
Diachrony
Platt’s article also took a diachronic approach, trying to explain the apparent
likeness of Singlish to creoles. He considers Singlish only creole-like because
it did not develop from a pre-existing pidgin, and labels this type of language
a ‘creoloid’. By way of definition, Platt (1975: 372) provides a list of features
that such a ‘creoloid’ must have:
1. It has similar structural variables to post-creoles based on the same
‘standard’ language.
2. It did not develop from a pidgin but by some other process.
3. It developed from the transference of features into the ‘standard’ lan-
guage from the languages of several (sometimes unrelated) ethnic groups.
4. The superordinate language is usually only one of the official languages.
5. It is used as one of several ‘native’ languages by the speech community.
6. It is usually also used as the lingua franca in interethnic group communi-
cation within the speech community where it is one of the subvarieties.
Thirty years later, this approach was described by Ansaldo (2004: 129) as
‘obsolete’ since, on the one hand, the concept of the creoloid is ‘rather con-
troversial’ and incompletely defined, and on the other hand, because a pidgin
ancestry is no longer seen as necessary for creoles. Furthermore, the validity
of the category creoloid is questioned, since, much like creole, it does not con-
stitute a structurally definable class, but ‘is purely sociohistorical and political
in nature’ (2004: 129–30). While Ansaldo recognises that there is no clear ev-
idence for a Pidgin English in Singapore, ascribing this to the precolonial
presence of Bazaar Malay as a vital lingua franca throughout the archipelago,
with the result that even Europeans were fluent in this variety (Ansaldo
2004: 141–2, a point previously made by Gupta 1994: 38), he also agrees
with Gupta (1994, 1998) in stressing the importance of two ethnically mixed
groups: the Babas and the Eurasians. The former, also called Peranakans
(see Section 1.2.1, p. 16), are descendants of Chinese (typically Hokkien)
men and Malay women, and the latter of European (typically Portuguese)
men and Asian (mostly Indian and Malay) women (2004: 142; see also
6
Furthermore, many of the lexical items in Lee’s dictionary are (predictably) from the military
register, and would, therefore, only be known actively to that segment of the population
subjected to it (roughly, men older than 16.5 (age of enlistment) and younger than 60
(approximate age of the first cohort), crucially excluding all but a minority of women).
Table 2.4. Proposal of a sociolinguistic typology

linking diglossia and bilingualism (adapted from
Fishman 1972: 75)
Diglossia Bilingualism
(1) + +
(2) − +
(3) + −
(4) − −
Gupta 1994: 41, Ansaldo and Matthews 1999, and Rappa 2000). While the
Babas traditionally use a contact variety of Malay known as Baba Malay (Pakir
1986), and the Eurasians a Portuguese-based creole called Kristang (Baxter
1988), these groups, originally hailing from pre-existing Straits Settlements
such as Malacca (Gupta 1994: 41; Ansaldo 2004: 142), were among the first
to use English for purposes of social advancement, enrolling their children in
English-medium schools, where they were available (Gupta 1994: 37). It is,
therefore, in these groups that one must look for an ancestor to modern SgE:
the Peranakans and the Eurasians had their own, typologically converged En-
glish, long before the emergence of widespread English-medium education
and the ready availability of Standard English (Ansaldo 2004: 143–4).
To sum up, this subsection outlining Platt’s take on SgE discussed two
major points: the model used to explain the variation between the Standard
English used as a target language in education, and the basilect ‘Singlish’.
The model takes into account intervening varieties on a continuum linking
the two extremes, much in the way that has been suggested by DeCamp
(1971) for Jamaican English. It also offers an explanation of speakers’ usage
of these subvarieties, linking their social status with the number of lects at
their disposal for stylistic use. The second point discussed is that of CSE’s
typological status as a contact language; Platt introduces the concept of
the creoloid – in essence, a variety that has creole-like features but lacks a
historical pidgin stage. The relevance of this term is questionable, due to its
many assumptions which have been shown to be outdated.
2.1.2 Polyglossia
Two years after his article describing SgE as a continuum, Platt (1977) pro-
posed a model for polyglossia, based on Singapore English, with substantial
reference to Malaysian English. He was referring to Fishman’s (1972: 75)
proposal of a typology that linked diglossia and bilingualism, which included
one category where both diglossia and bilingualism exist (see Table 2.4).
Considering cases where this situation applies, Platt argues that there are
Table 2.5. ‘A model for polyglossia with

multilingualism’ (Platt 1977: 367)
Order of status
Speech varieties H1 –Hn M1 –Mn L1 –Ln L−
SV1
SV2
SV3
..
.
SVn
speech communities that ‘are not cases of mere bilingualism and that a divi-
sion into H and L only, or even one H and several Ls may not be sufficient’
(1977: 362, emphasis in the original). Such multilingual speech communi-
ties, as his examples of Singapore and Malaysia show, exhibit numerous
‘subcodes within each of the separate codes’ and an important ‘interaction
between the bi-(and often multi-)lingualism [sic] of the individual . . . with
particular spheres of social activity and particular social attitudes’ (Platt
1977: 362). He calls this situation one of polyglossia, where several speech
varieties are distributed functionally across the community (in a compara-
ble way to H and L in diglossia). This can be illustrated in the form of a
table listing the speech varieties involved on one axis, and their respective
polyglossic function on the other, as in Table 2.5.
Listed in the leftmost column are the speech varieties in use within the
community: in the Singaporean case, illustrated in Table 2.6, there are eight
(one of which, ‘Other Chinese dialects’, comprises several7 ). These varieties
can then be assigned to any of the statuses listed at the top of the table: Platt
allows for several levels of H(igh), M(edium), and L(ow). L− is a Low variety
which is used as such by all speakers and has no associated H. DH stands for
‘Dummy H’ – a variety that is given a high rating because it is recognised
by some segments of the population as carrying prestige (in Singapore the
government, the media, etc., see Table 2.6), but that is not widely used in
the community (again, in the case of even educated Chinese Singaporeans,
Malay and especially Tamil, it is hardly used at all).
The ranking of these varieties along the H–M–L scale is based on two
major criteria: domains of use and speaker attitudes. Domains include ‘fam-
ily, friendship, employment, religion, education, government, media’ (Platt
1977: 368), and can be subdivided into subdomains: in Singapore, the family
domain typically involves generational subdomains, so that in the case of a
Chinese family, grandparents would use the native Chinese dialect, parents
7
Teochew, Hakka, etc. See pp. 3 and 7.
Table 2.6. Verbal repertoire of an English-educated Chinese of

Singapore (Platt 1977: 375)
H1 H2 DH1 DH2 M L1 Ln L−
Formal S[g]E ×
Mandarin ×
Standard Malay ×
Tamil ×
Colloquial S[g]E ×
Hokkien/Cantonese ×
Other Chinese ×
Bazaar Malay ×
the dialect and CSE, and children CSE (Platt 1977: 369). This example of
Platt’s would need some updating to match the current situation, where
Mandarin has become an important addition to the repertoire. Nonetheless,
it highlights the complexity of the situation and the relation between the
choice of speech variety and the level of publicity of the domain. Platt ar-
gues that varieties used most often in the public domain are highest on the
scale, and those associated with private, informal domains would be among
L. Varieties used in ‘semi-formal/semi-public domains’ (Platt 1977: 370) are
M.
Several types of attitudinal factors were considered. Departing from the
more mainstream model of accommodation theory (Giles et al. 1973), Platt
(1977: 371) confines himself to the speaker’s choice of a variety based on
his/her previous experience of the addressee’s repertoire, and to the speaker’s
elicitation of said repertoire. Another attitudinal factor is the speaker’s ‘poly-
glossic reasons’ (1977: 371), i.e. attitudes towards the individual varieties and
their appropriateness to the domain at hand. These attitudes are: (a) emotive
(like–dislike), (b) linguistic (vocabulary range, etc.), (c) pragmatic (usefulness
in communicating, gaining better treatment, financial gain, etc.), (d) custom
or policy (established practice, government policies), and (e) appropriateness
(Platt 1977: 371–2).
Taking these two criteria, domains and attitudes, into account, Platt was
able to classify the speech varieties of Singapore along the H–M–L scale
illustrated in Table 2.6. In this example, the speaker, a Chinese Singaporean,
has SSE (or ‘Formal S[g]E’) as his/her H1 , and Mandarin as the H2 : SSE
is the variety used in education, in politics, in the workplace and in dealings
with the authorities, while Mandarin, a co-official language, also features
prominently in the curriculum and may present advantages at the workplace.
Standard Malay and Tamil, as noted above, are DH1 and DH2 respectively:
while the speaker, ethnically Chinese, is unlikely to use either of them very
much if at all, they are nonetheless considered H because of their co-official
status. Malay is DH1 because of its additional status as a national language – in
fact, the speaker, if male, will have had at least minor exposure to Malay
during his National Service (see p. 7). CSE is an M variety: it is lower than
SSE and standard varieties of the official languages, but higher than other
varieties, such as the officially discouraged, non-Mandarin Chinese varieties,
and it is used in a variety of semi-public domains, particularly in interethnic
communication. If the speaker is proficient in Hokkien and/or Cantonese,
then Hokkien/Cantonese will be L1 : L because of their non-official status and
because they are the target of official disapproval in favour of Mandarin, they
are nonetheless higher within the Ls because of their currency as the variety
of a dominant ethnic group (see pp. 3 and 7). Any other Chinese variety in
the speaker’s repertoire would be L1−n . Bazaar Malay, less used nowadays
than it was at the time of Platt’s work, would have been an L−, considered
as carrying the lowest prestige and restricted to the domain of transactions
(e.g. at the market, whence ‘Bazaar’, from Malay pasar, ‘market’).
An interesting aspect of this model is that it lists the repertoire of a given
speaker but couches it in the speech community’s language ecology. It is
also more complete than other attempts (Platt 1975; Gupta 1994, 2001) in
that it takes a holistic view of the speech varieties involved: it is not just the
diglossia between SSE and CSE that is considered, it also accounts for other,
unrelated varieties. After all, as is evident from the list of domains given
above, English (in its H and L forms) is but one component of the speech
community’s resources.
Notwithstanding this latter point, I now turn to analyses of the variation
observed within a subset of the varieties enumerated in Table 2.6. The
following will look at the interaction between the subvarieties of English
used by the speech community. Diglossia between CSE and SSE has already
been mentioned, e.g. in Table 2.6. What follows, then, is a review of how the
interaction between these two has been analysed.
2.1.3 Diglossia
Another approach is taken by Anthea Fraser Gupta, who proposes a binary
system where Standard Singapore English (SSE) and Colloquial Singa-
pore English (CSE, ‘Singlish’) are in a diglossic relationship (Gupta 1989,
1994, 2001). Diglossia is a term first used by Rhoı̈dis (1885), who used
it to mean a ‘type of collective bilingualism, among other possible types’
(Mackey 1993: xv)8 – writing about the Greek situation, he says ‘on souf-
fre aujourd’hui de diglossie’9 (Rhoı̈dis 1885, quoted in Psichari 1885: 211,
cited in Fernández 1993: 309). In the same year, Psichari (1885: 267), citing
Rhoı̈dis, uses the term to qualify the relationship between two contemporary
8
The word used by Rhoı̈dis, διγλωσσία, is Greek for bilingualism (Mackey 1993: xv), a point
which may have contributed to his definition.
9
‘it nowadays suffers from diglossia’.
varieties of Greek (Mackey 1993: xv). Several years later, Marçais (1930),
describing the language situation in the Arab world, extends the term to
cover cases of bidialectalism. It is Ferguson (1959) who was the first to use
it in English-speaking academia. His definition has been very influential and
many subsequent ones draw on his (Mackey 1993: xv).
Fergusonian diglossia
Gupta describes the Singaporean case as one of ‘diglossic English . . . in
terms of Ferguson’s use of diglossia’ (Gupta 1994: 7). The original definition
of Ferguson’s diglossia, which I shall explore next, is given here in full:
Diglossia is a relatively stable language situation in which, in addition
to the primary dialects of the language (which may include a standard
or regional standards), there is a very divergent, highly codified (of-
ten grammatically more complex) superposed variety, the vehicle of
a large and respected body of written literature, either of an earlier
period or in another speech community, which is learned largely by
formal education and is used for most written and formal spoken pur-
poses but is not used by any section of the community for ordinary
conversation.10 (Ferguson 1959: 336)
Ferguson’s definition is a very precise one, and is often called ‘classic’ or
‘narrow’ diglossia (Fasold 1984: 53; Wardhaugh 2002: 93; Winford 2003:
112–13). The emphasis is clearly on H, which is described in structural,
cultural and functional terms, as opposed to L, which is the ‘primary’ code,
and, therefore, something of a default – it is H that is restricted in use,
and limited to ‘written and formal spoken purposes’ (Ferguson 1959: 336).
Furthermore, the definition is limited to situations where H and L are ‘vari-
eties of the same language’, with the ‘analogous situation where two distinct
(related or unrelated) languages are used side by side’ being expressly ex-
empted from it (1959: 325).11 Diglossia characterises a ‘language situation’
(1959: 336), i.e. a property of the speech community, typically exempli-
fied by those of Haiti (H: French, L: Haitian Creole), German Switzerland
(H: Standard German,12 L: Swiss German dialects), and Egypt (H: Clas-
sical Arabic, L: Egyptian Arabic). Important components of Ferguson’s
10
The first of these, the vernacular, is called the ‘low’ variety, or L, and the latter, ‘high’ or H.
11
This is a point revisited by Fishman (1967), who argues that there is nothing preventing
unrelated languages from being in a diglossic relationship. In fact, he stresses that in the
case of related languages, these need to be ‘sufficiently different from one another that,
without schooling [H] cannot be understood’ by L speakers (attributed to Fishman 1980: 4
in Wardhaugh 2002: 94). This has been referred to as ‘broad’ diglossia (Fasold 1984: 53,
Winford 2003: 112–13, see below), two flagship examples being those of Alsace (H: French,
L: German) and Paraguay (H: Spanish, L: Guaranı́).
12
Switzerland’s Standard German is however endonormative, in that it differs from Standard
High German in several ways (much in the way that SSE does from StBE): examples in
phonology (initial devoicing), the lexicon (wischen for fegen ‘to sweep’), and spelling (ss
for ß) abound (see e.g. Rash 1998: 150–79).
definition include the stability of the situation (unlike in other language

contact situations, diglossia typically does not give rise to language shift
(1959: 327, 340)), and more importantly, the functional distribution of H
and L: they are in complementary distribution, so that where one is used, the
other is not, ‘with the two sets overlapping only very slightly’ (1959: 328).
Ferguson (1959: 329) gives a list of situations with their associated variety,
reproduced in Table 2.7. Practically all conceivable situational settings call
for a clearly defined variety, and a speaker using the wrong one ‘is an object of
ridicule’ (Ferguson 1959: 329). The ones in Table 2.7 are fairly predictable,
with anything remotely formal (religion, education, media) being assigned
H, and the less formal ones (conversation, soap operas), L. Worth pointing
out is the use of H in all writing (except restricted uses in folk literature,
but including ‘personal letters’) – the absence of a standardised orthography
for L being a prime motive (together with any ‘descriptive and norma-
tive studies[, which are] either non-existent or relatively recent and slight’
(Ferguson 1959: 332)). L is, however, the default language for conversation,
so that after a lecture in H, or reading an article in H, one can then ‘proceed
to discuss the contents in L’, even ‘with the speaker [or author] himself ’
(1959: 329). This points to a final and important defining factor of the Fer-
gusonian diglossia: its ‘democratic’ distribution of H–L usage. As illustrated
in the Greek13 name for its L variety (δημοτική, ‘[language] of the people’),
there is ‘no segment of the speech community’ that ‘regularly uses H as a
medium of ordinary conversation’ (1959: 336–7).
Ferguson lists other attributes that characterise diglossic communities:
prestige (H is always regarded ‘as superior to L’, or at the very least ‘better
able to express important thoughts’ (1959: 329–30)), a literary heritage in H
(shared with another speech community or another era), aspects of language
acquisition (L is always acquired first, H learned formally), H standardisation
(as opposed to L, which is described, if at all, ‘first and chiefly by scholars
from outside the speech community’ (1959: 332, emphasis in the original)),14
stability of the language situation over time, significant grammatical differ-
ences in the two varieties (with H typically less regular and more complex
than L), the presence of lexical doublets (which instantly mark an utterance
as either H or L), and divergent phonologies (the L phonology being the
basic one and H diverging from it to form a subsystem (1959: 335)). All these
are important features of diglossic communities; the most important role,
however, is played by the functional distribution seen in Table 2.7 – the
13
The Greek diglossia (H: καθαρεύουσα, L: δημοτική), which was used by Ferguson (1959)
as one of his examples, has now broken down due to a variety of factors widely reported in
the literature (Trudgill 1995: 103–4, Frangoudaki 1992, inter alia): the new official language
(the erstwhile L), however, shows substantial admixture from καθαρεύουσα (Kazazis 1993).
14
Ferguson does allow for situations where L becomes a local standard: his example is that of
Cairo Arabic, which serves as an Egyptian Standard Arabic (1959: 332).
Table 2.7. Use of diglossic varieties for selected situations

(adapted from Ferguson 1959: 329)
Situation H L
Sermon in church or mosque ×

Instructions to servants, waiters, workmen, clerks ×
Personal letter ×
Speech in parliament, political speech ×
University lecture ×
Conversation with family, friends, colleagues ×
News broadcast ×
Soap opera ×
Newspapers, caption on picture ×
Caption on political cartoon ×
Poetry ×
Folk literature ×
restriction of use for each variety to particular situational settings is certainly

the prime factor defining this particular language situation.
Diglossia in Singapore
Motivated by Platt’s (1975) outdated treatment of SgE as a non-native vari-
ety, Gupta (1994, 2001) sees the Singapore speech community as diglossic
‘in terms of Ferguson’s use of diglossia’ (Gupta 1994: 7). She posits the ex-
istence of an H variety, which is called Standard Singapore English (SSE):
as outlined in Section 1.2.2, SSE is mostly indistinguishable from other
standard Englishes around the world (see e.g. Trudgill and Hannah 1994),
although it bears a closer resemblance to StBE than to other exonormative
standards (Gupta 1986), as its genesis would predict.15 The only noteworthy
points of difference between SSE and StBE lie within the lexicon, with some
loanwords from substrate languages, as well as in some aspects of seman-
tics.16 This H variety is used in ‘formal circumstances, in education, and
in all writing’ (Gupta 1994: 7), except in restricted contexts (such as online
writing, see Gupta 2006b). The L variety, Colloquial Singapore English
or ‘Singlish’, is radically different from H, and of the form described in
Section 1.2.2 above. Unlike H, it is used in everyday conversation and gener-
ally in situations deemed informal, as well as in some writing (e.g. in instant
text messages and informal computer-mediated communication). The differ-
ence between the two varieties is such that members of the speech community
are aware of them, and can exploit them for stylistic purposes.
15
This is one of the reasons why Gupta uses Standard English (StdE) for H most of the time,
whereas I use SSE here. Since the former encompasses the latter, this is not deemed a major
issue.
16
E.g. the use of slippers for the footwear called sandals in StBE, or SSE bath for StBE shower.
Table 2.8. Features of diglossic SSE and CSE (adapted from Gupta 1994: 10–13)
SSE (H) CSE (L)
Inversion in interrogatives (except with be and Use of pragmatic particles: ‘This place
can): ‘Where do you live?’ very noisy leh.’
Presence of verbal inflexions: ‘He lives here.’ Verb groups without subjects
(PRO-drop): ‘∅ go where?’
Noun inflexions (genitive and plural): ‘My Conditional clauses without subordinating
brother’s flats are big.’ conjunction: ‘∅ you go there, sure can see.’
Certain complex verb groups: ‘I have been V-ing as finite verb and verbless
there several times.’ complements (copula deletion): ‘This one
∅ very cheap.’
Gupta (1994: 10–13) uses a number of features to differentiate between the

two varieties, four for H and four for L (given here in Table 2.8). While they
are not an exhaustive list of features, they are ‘analytically straightforward’
and salient (1994: 9), covering morphology, syntax, and lexis. Many of these
are ‘required in [SSE] but optional in [CSE]’ (Gupta 1994: 9, emphasis in
the original). For instance, nominal and verbal morphology can be variably
realised in CSE; this optionality is of course absent in SSE. Similarly, in-
version can occur in conjunction with CSE features. The only SSE feature
that escapes this is the last one: complex verb groups (modal auxiliaries other
than can, and uses of have and do) ‘are associated with [SSE] only’ (Gupta
1994: 13). The reverse does not, however, hold true: the CSE features in
Table 2.8 are restricted to the L variety, and using them instantly marks the
utterance as ‘Singlish’: the simple presence of a particle within an otherwise
H sentence turns it into an L utterance. One could easily conclude from this
that CSE is simply anything that is not SSE, which is clearly fallacious:17 the
features provided, while unidirectional (in the sense that SSE features, with
the above caveat, can occur in CSE but not vice versa), are nonetheless called
for to identify a particular SgE utterance as H or L – they form a working
basis on which such decisions can be made, even if these features ‘are not the
only differences between the two varieties’ (Gupta 1994: 9, emphasis in the
original).
Gupta concedes that there might not be a ‘hard division between H and
L’ but rather ‘degrees of aim at H and L’ (1994: 8). In an earlier paper
discussing written SgE, she also notes that ‘speakers [are] using a wide
span on the scale of standardness in speech’ (1986: 77), which seems to run
counter to the basic assumptions of Fergusonian diglossia. Gupta nonetheless
feels that it is possible, with the above features, to draw a line between one
and the other, since ‘in practice [they] tend to constellate’ and result in
17
Consider the wealth of nonstandard Englishes outside of Singapore, which are by definition
not CSE.
identifiable varieties (1994: 8), i.e. to members of the speech community,

it is always clear whether H or L is being used. As Gupta notes, Pakir
(1991) felt able to clearly mark strings of her discourse data as H or L, based
on Gupta’s features. Willemyns (1987: 37) reports a similar situation from
Belgium where, despite the presence of a continuum of varieties between the
archetypal H and L, speakers ‘experience a clear gap’ between what he calls
[−dialect] (H, Standard Dutch) and [+dialect] (L, Westflemish).
To account for this heterogeneous component of SgE’s diglossic varieties,
Gupta (2006b: 22) calls the Singaporean diglossia ‘leaky’, a term introduced
by Fasold to qualify diglossic situations where ‘one variety “leaks” into
the functions formerly reserved for the other’ (Fasold 1984: 41).18 This
leakage clearly brings the two varieties closer, and code-switching becomes
more frequent. While it is certainly the case that in Gupta’s data (1994,
2006b) there is substantial switching between H and L, sometimes within
the same sentence, it is difficult to see how a situation such as this can
felicitously still be called ‘diglossia’. It might be that Fasold’s definition
of ‘broad diglossia’ is more appropriate here; in this much less restrictive
use of the term, it is ‘highly valued segments’ of the continuum that are
reserved ‘for situations perceived as more formal and guarded’, and ‘less
highly valued segments . . . for situations perceived as more informal and
intimate’ (Fasold 1984: 53). As in Fasold’s words, such a redefinition of
diglossia, which eliminates the binary nature of the Fergusonian diglossia19
(viz. the plural reference to ‘segments’), which also specifies that H and L
can be ‘of any degree of linguistic relatedness, . . . from stylistic differences
to separate languages’ (Fasold 1984: 53), and which, crucially, removes the
strong, almost deterministic functional/situational element of Ferguson’s
definition (see Table 2.7 above) and replaces it with an H–L distinction
on grounds of perceived levels of formality, almost removes the necessity
for a separate term – the only reason for keeping it being historic and that
‘it would be a shame to give it up’ (Fasold 1984: 53). While it is certainly
the case that the wealth of research that has sprung forth from Ferguson’s
article has diluted the concept somewhat, particularly with regard to the
more restrictive elements of the definition (relatedness of H and L, language
18
Leaky diglossia is at the origin of the situation described as ‘bilingualism without diglossia’
by Fishman (1972: 75) – see situation (2) in Table 2.4 above. This would seem to spell
impending changes for the Singaporean case, since bilingualism without diglossia has two
possible outcomes: a new, mixed (H & L) variety, or a shift towards H at the expense of L
(Fasold 1984: 41).
19
The binary nature of diglossia had previously been questioned by the application of the term
to situations in Khalapur (Abdulaziz Mkilifi 1978) and Tanzania (Gumperz 1964), which
were termed ‘double nested diglossia’ and ‘double overlapping diglossia’ respectively: in
the first case, H and L are different languages, each with their respective ‘nested’ H and L,
resulting in four interacting varieties, whereas in the second case, there are three varieties
(hence ‘triglossia’ in Gumperz’ terminology), one of which, spoken by the whole community
(Swahili in Gumperz’ example), functions as an H for lower socio-economic and educational
groups (the local vernaculars being L), and as an L for the higher classes (where English is
H) (see Fasold 1984: 44–50 for an overview of these).
Table 2.9. Ferguson (1959: 336) applied to SgE
Fergusonian criteria Presence in SgE
(a) Stability of the situation ±

(b) L and H are dialects of the same language +
(c) Linguistic distance +
(d) Codification of H +
(e) H more complex +
(f) Literature in H +
(g) H used in another speech community +
(h) H learned formally ±
(i) H used in writing +
(j) H used for formal speech +
(k) H not used for ordinary conversation −
acquisition), it nonetheless remains a convenient benchmark from which a

language situation can be described, if anything, as more or less divergent.
In the case of SgE, it might be useful to test Gupta’s (1994, 2001) claim
of Fergusonian diglossia by comparing the Singaporean situation with the
original definition itself (Ferguson 1959: 336; see p. 39 above). Table 2.9 takes
each assertion from the definition and shows its applicability to SgE. Stability
(criterion (a)) has been marked as [±] since despite the posited prevalence of
the situation nowadays, a diglossic situation cannot have existed in the early
stages of contact: the number of English speakers was very small indeed in the
nineteenth century (Ho and Platt 1993; Gupta 1994, 1998; Ansaldo 2004),
and an English diglossia could only have emerged with the appearance of CSE
as a distinct variety (see Section 1.1.3). Diglossia among English speakers in
Singapore can, therefore, only be described as a relatively new phenomenon,
hardly a century old. Criterion (b) refers to the ‘primary dialects of the
language’ in Ferguson’s (1959: 336) definition. Clearly this raises a number
of issues, and perhaps the reformulation in Table 2.9 is not the best, but
it seems intuitive that both H and L are dialects of English, if one takes
Trudgill’s (1999) definition of the term: their differences include phonology
and grammar, but these are not great enough to render the two mutually
unintelligible – they are, therefore, (b) [+dialects] but also (c) [+divergent].
That H is codified (d) warrants no further explanation, and neither do
criteria (f) and (g). H is also the default variety for writing (i), despite the
recent move towards vernacularisation with examples of written L (Gupta
1986, 2006b). Criterion (j) is a very basic one, and it is certainly the case
that in Singapore, as in any diglossic situation, H is the variety of choice
for formal speech.20 Complexity (e), however, is a much less clear-cut term.
While, for instance, the rules governing the syntax and semantics of the
20
It is, however, possible to have discussions on formal topics in L (as also e.g. in Swiss
German, see Trudgill 1995: 182–3).
discourse particles are by no means simple, certain features of H, as previously

established (see e.g. Section 1.2.2 or Table 2.8), such as plural marking and
tense inflexions, are obligatory in H but optional in L. Inversion rules and
variable question tags are also arguably complex elements in H that have
no counterpart in L (for the former) or an invariant one (for the latter). A
reduction of the level of complexity in grammatical structure is, of course, also
an expected outcome of contact languages (see Sebba 1997; Winford 2003),
whence the decision on [+] for (e). The formal nature of H acquisition (h),
i.e. via the education system, is marked here as [±] for the reason that it
might well be the case for a majority of the speech community, but it is
not true that there are no exceptions to this rule. This is strongly related to
criterion (k), the use of H for ordinary conversation. While this is not the case
in traditional diglossia, and at the origin of the equations ‘L = acquired, H =
learned’, it seems that in some sections of the speech community, H is used
informally – this, at least, would explain certain speakers’ difficulties with L
(see Hussain 2006)21 – a usage which would enable its acquisition as a first
language. It would not be surprising at all to see such a development: SSE
is a highly valued asset in terms of educational performance, and educated
Singaporean parents have been known to try and provide their children with
‘useful’ linguistic exposure (consider for instance the success of the Speak
Mandarin Campaign, which saw parents forsaking ‘dialects’ in interaction
with their offspring (see e.g. Bokhorst-Heng 1998; Simpson 2007: 382–4)).
Furthermore, in terms of domains of use, at least two of the situations listed
in Table 2.7 need to be reevaluated for SgE: after some English soap operas
in Singapore were aired in CSE, which sparked wide public discussion and
resulted in a ban on ‘ungrammatical English’ in the media, all are now in
SSE for reasons of language policy (Simpson 2007: 387). Conversations with
colleagues are not invariably L either, as the topic involved can skew usage
towards H (Koh, p.c. June 2008). (h) is, therefore, marked as [±], because of
the heterogeneity of the speech community: both H and L can be acquired
natively; (k), however, is given [−], because it would imply that H is not
used ‘by any section of the community for ordinary conversation’ (Ferguson
1959: 336), which is not the case in this categorical form.
The features of Table 2.9, therefore, suggest that Singapore does not
present a clear-cut example of Fergusonian diglossia. On an individual per-
formance level, however, as well as in terms of language attitudes, there
is a strong sense in which diglossia cannot be wholly dismissed: in terms
of speakers’ perception of things, it has already been established above
that speakers ‘are conscious of these differences’ (Gupta 1994: 8, see also
Willemyns 1987: 37), and it is certainly the case that in public discourse,
21
It is safe to assume that this represents a socio-economically advantaged minority, which
may still have some passive knowledge of L – it is hard to imagine a situation where an active
member of the speech community would be completely sheltered from contact with L.
‘Singlish’ and ‘Good English’ are often-used terms. Importantly, DeCamp

also notes that ‘on the level of performance’, the ‘verbal behaviour of the
individual may turn out to be discrete, even though the composite behaviour
of the community . . . is continuous’ (1971: 368). The attempt to distinguish
between an individual and a society diglossia, as it were, is laudable in that
it allows for more speaker agency. It does not, however, solve the problem
entirely; the whole notion of diglossia needing to be rethought, at least when
applied to the Singaporean case.
2.2 Contemporary models and indexicality

2.2.1 Cultural orientation model
Lubna Alsagoff (2007, 2010) sees the need for a new approach to SgE varia-
tion, arguing for ‘a more culturally grounded perspective’ (2007: 25), which
essentially treats variation as a result of cultural forces pulling SgE in two
opposing directions: globally and locally. This approach is situated within
a recent take on variation, which the term ‘sociolinguistics of globalisation’
(Blommaert 2010) succinctly summarises. Alsagoff herself grounds her con-
tribution in work by Pennycook (1994, 2007), among others, who also took
a global, cultural view of variation in global English. As far as Singapore is
concerned, the emphasis on culture and identity is something of a departure
from the previous models, which focus primarily, as established above, on
educational achievement (as a measure of language proficiency). For instance,
the implication of Platt’s model, Alsagoff argues, is that ‘Singlish is not borne
out of choice but of a lack of education’, thus branding it as ‘undesirable’
(2007: 27) – certainly the continuum suggests a dichotomy between the ex-
onormative ‘Good English’, StBE, and the substrate-influenced, uneducated
‘creoloid’.22 Her major target, however, is the diglossic model, since it does
not account for those members of the speech community who do not speak
English natively – there is still a sizeable group that ‘continue[s] to use English
as an “other” language, often as a lingua franca’ (Alsagoff 2007: 30–1). This
exclusion of ‘learners’ of English from the diglossia model effectively pre-
cludes proficiency-based variation. Alsagoff thus settles on Pakir’s (1991)
triangles as the least flawed model, since it includes both proficiency and
formality as its variables. The only problem is its incorporation of diglossia.
Alsagoff struggles with Gupta’s data, which suggest that L features occur
in domains where H is predicted. She observes that ‘such pervasive inclusion
of [CSE] features in almost all speech appears to be endemic . . . even in formal
contexts’ (2007: 32). Since diglossia does not have the explanatory power
needed to characterise the kind of variation observed in SgE, Alsagoff turns
22
Alsagoff (2007: 28) acknowledges Platt’s (1975) efforts in providing a purely descriptive
analysis of CSE, highlighting his use of ‘non-standard’ rather than the previously used
‘sub-standard’ (e.g. Tongue 1974: 111).
to a wider perspective of the situation, considering the concept of ‘culture’

and the way in which English contributes to two distinct ends. Culture, here,
refers to ‘a macro-culture’ related to ‘notions of citizenry, and nationalist
identity based on a collective disposition and history’ (2007: 34). This culture
is externalised in the two main roles of English in Singapore: English is,
firstly, a global language, the language of international trade and commerce,
that of finance and technology. Secondly, English is the lingua franca used for
interethnic communication between the various ethnic groups in Singapore.
The first of these roles is intrinsically external; the second is unambiguously
internal to the speech community as such. Alsagoff’s argument is that the
variation in SgE is a direct result of the opposing forces of these two roles,
the first of which is directed towards globalisation, and the second towards
localisation; variation is a result of this ‘conflict between “being global” and
“being local”’ (2007: 34).
As far as the ‘global’ target of SgE is concerned, the discussion is always
formulated in pragmatic and primarily economic terms. Its necessity is ad-
vocated by political leaders on grounds of ‘trade, science and technology’,
which renders English ‘essential as Singapore becomes a leading financial
and banking centre’ (Minister of State Prof. Jayakumar, The Straits Times,
19 August 1982, cited in Bokhorst-Heng 1998: 290). This concern for com-
petitiveness on the international market is probably further strengthened
by the potentially small economic weight of the island-state: the character-
isation of Singapore as a ‘little red dot’23 by Indonesian President Habibie
(Wall Street Journal Asia, 4 August 1998) left a deep impression and ‘entered
the psyche of every Singaporean’ (MITA 2003). The widespread aware-
ness of vulnerability (economic and otherwise) has its share in the pursuit
of economic growth and of an internationally (rather than regionally) ori-
ented public image (Alsagoff 2007: 35). In order to achieve this, English
is needed, but an English that is exonormative, i.e. the acrolect of Platt
(1975), the SSE of Pakir (1991) and the Standard English of Gupta (1994,
2001) – in short, a global English. This Alsagoff (2007: 35) calls ‘Inter-
national Singapore English (ISE)’, the global brand of English as used in
Singapore.
This ISE is part of a discourse on the global nature not only of the economy
that Singapore thrives on, but also of its citizens’ achievements: while the first
of these was and is crucial in bringing foreign investment into the country,
the latter has recently become more important with significant numbers of
Singaporean entrepreneurs ‘going global’, thereby ‘bringing Singapore to the
world’ (Alsagoff 2007: 35), e.g. by creating global brands, and thus effectively
acting as ambassadors of Singapore in the global arena. These interactions
23
Referring to the cartographic symbol for capital cities (e.g. ), which in the Singaporean
case, at a given scale, would cover the entire island due to its small surface area of c. 710 km2
(see map, Figure 1.2).
naturally take place in ISE, and are actively encouraged. However, and
this is an important point, ISE is always referred to in economic terms,
and is stripped of any cultural and ethnic associations. Culturally, language
planners had to further ISE without encouraging ‘corrupted Western values’
(Alsagoff 2007: 36; see also Alsagoff and Ho 1998): English is a global, not a
Western24 language. It is the ‘mother tongues’ that have the prerogative of
being vehicles of local25 culture(s). Ethnically, too, English is seen as ‘neutral’
with respect to the ‘mother tongues’. In the government’s ideal view, every
Singaporean is of a given ethnicity (Chinese, Malay, Indian), speaks ISE for
global and economic purposes, and uses their ‘mother tongue’ (Mandarin,
Malay, Tamil) for cultural and local identity.
This is of course far from reality. The lingua franca that is English
has evolved to become a vehicle of intercultural communication and co-
hesion, and as the only language common to every Singaporean (at least for
the younger generation), it takes on cultural meaning in a local, national
sense. Alsagoff (2007: 36–7) explains this by the contact-induced change that
saw the emergence of CSE: contact of English with the several cultures
(Chinese, Malay, Indian) is at the origin of the ‘enculturation’ of the new
variety. She continues by saying that Singlish ‘is used by people, so it be-
comes part of the cultural history of Singapore’ and ‘shapes the culture
which it serves’ (2007: 37). While seemingly circular (culture shapes lan-
guage which shapes culture, a point acknowledged by Alsagoff (2007: 37)),
the argument essentially underlines the interaction between the two, and
their mutual reinforcement. Using Singlish has become ‘a means of ex-
pressing local identities’ (2007: 37); it is this localism that prompts her to
rename Singlish ‘Local Singapore English (LSE)’26 – a variety which is
used to express local and national identity, reflecting a uniquely Singaporean
culture.
The two varieties serve two opposing cultural orientations: the first is
global, outward-looking, and requires ISE, whereas the second is local,
inward-looking, and requires LSE. In the first case, ISE is devoid of cultural
elements, stresses similarities with the rest of the world, and emphasises
24
The term Western culture, in the Singaporean context, is often taken to mean British or
American culture, rather than any other ‘Western’ one. Many Singaporeans, for instance,
find it hard to understand that there are ‘Westerners’ who do not speak English (Gupta,
p.c. 28 November 2007).
25
Or rather ancestral cultures, i.e. one for each recognised ethnic group, and not the local,
Singaporean culture per se.
26
It may be appropriate at this stage to summarise the various appellations used to describe
the two varieties. Based on the diglossic model, H has been called ‘Standard English’,
‘Standard Singapore English (SSE)’, ‘acrolect’, ‘International Singapore English (ISE)’,
and, by non-linguists, ‘Good English’. L, on the other hand, has been referred to as
‘Singlish’, ‘Colloquial Singapore English (CSE)’, ‘basilect’, and ‘Local Singapore English
(LSE)’. ‘SCE’ is sometimes used for CSE in the literature, but this has been standardised
here.
standardisation – Alsagoff (2007: 37) uses Bakhtin’s (1981) concept of cen-

tripetal force to describe ISE, which enables Singapore and Singaporeans,
as it were, to interact with the outside world. LSE, on the other hand, is
highly associated with Singaporean culture, thus setting it apart from the
rest of the world, and emphasises differentiation from exonormative stan-
dards: Bakhtin’s (1981) centrifugal force is appropriate to describe LSE’s
function of uniting Singaporeans among themselves while at the same time
setting them apart from the outside world (Alsagoff 2007: 37). For Alsagoff,
variation within SgE is, therefore, the negotiation between these two orien-
tations (globalism and localism), which results ‘in English being “pulled” in
two different directions’ (2007: 38) – it is this interaction between two macro-
cultural perspectives (embodied linguistically by LSE at the local end and
ISE at the global end of the spectrum) that accounts for the various types of
English found in Singapore. Uses of either, or shifts towards either of these,
indicates a shift towards the respective cultural orientation; ‘the degree or
extent’ of linguistic shift ‘can be seen as a measure of the strength of the orien-
tation’ (Alsagoff 2007: 38). Proficiency-based variation is also accounted for,
again grounded in cultural terms: ‘a speaker who is more proficient in English
is also likely to have a wider macro-cultural repertoire’ (Alsagoff 2007: 39) –
there is, therefore, a correlation between use of ISE and English proficiency,
which is, again, ‘equated with educational attainment’ (2007: 39). Alsagoff
calls this approach the ‘cultural orientation model’, sensibly abbreviated as
COM.
The two orientations within COM are characterised by the features re-
produced in Table 2.10 (Alsagoff 2007: 39, table 1). These dichotomous
features are vaguely reminiscent of diglossic H–L in that they assign certain
situational settings to one or the other variety. Thus, for instance, in situ-
ations where authority is important, such as in the ticking off of a pupil in
the classroom, ISE will be used, whereas when camaraderie is stressed, i.e.
friendliness, etc., LSE is the norm. Similarly, dyad (d) is closely related to
this, in that authority is almost inherently associated with distance (albeit
not vice versa), and camaraderie with closeness (again, not vice versa). Dyad
(c), however, while being equally straightforward, harks back to previous
models’ stylistic variation (Platt 1975; Pakir 1991; Poedjosoedarmo 1995),
rather than to the diglossic model: while all features present a certain degree
of continuity, it is probably fair to say that formality is much less binary than,
say, authority.
Dyad (e) is a central part of the model since, as described above, it is
the one that accounts for English proficiency – the ISE used by speakers
reveals their level of educational attainment, as reflected in their English
(ISE being the variety used in education, and LSE the one acquired natively
or otherwise). The choice between ISE or LSE at this level, however, is
again one of orientation, stressing either one’s educational attainment (ISE) or
one’s membership of the community (LSE) – an important fact which will be
Table 2.10. ‘Features of the two orientations in the

cultural orientation model’ (Alsagoff 2007: 39, table 1)
ISE LSE
Globalism Localism
(a) Economic capital Sociocultural capital

(b) Authority Camaraderie
(c) Formality Informality
(d) Distance Closeness
(e) Educational attainment Community membership
discussed in more detail below. Similarly, in (a) the model presents two poles
of orientation towards economic and sociocultural capitals.27 The economic
capital is undeniably embodied in ISE, which is constantly championed as
the medium that enhances Singapore’s economic standing, whereas LSE has
sociocultural capital, namely a connection to local culture and identification
with Singaporean (as opposed to foreign) society. Speakers can decide, on
this scale, which of the two orientations is more appropriate in a given setting
(and to what extent), and opt for the associated subvariety.
This element of choice is central to COM. Alsagoff (2007: 40) notes that
the two poles, ISE and LSE, can be exploited within the same speech, where
for instance authority is necessary, but local features are incorporated ‘in
order to stress membership in the community’. The use of L features in con-
texts where H would be required in a diglossic framework can, therefore, be
reanalysed as the insertion of ‘local’ features into speech that would be char-
acterised, overall, as ‘global’, thus enabling speakers to mark themselves as
insiders to the society – an ability which can have tangible advantages (think,
for example, of a political speech), since it underlines ‘community member-
ship alongside educational attainment’ (Alsagoff 2007: 40). Alsagoff equates
these possibilities with style-shifting, a term that she prefers to code-switching,
since the latter suggests ‘a binary movement between two varieties’ (2007: 40),
as opposed to the more fine-grained shifting implied by a stylistic scale.
Lastly, the community membership feature of LSE allows users to ‘bridge
educational differences’ (Alsagoff 2007: 41): while in Platt’s (1975) model
(and to a lesser extent in Pakir 1991), speakers could use a ‘lower’ lect to
communicate with speakers of a different educational level, COM stresses the
27
Alsagoff (2007) uses Bourdieu’s (1986) definition of capital in its economic, social, and
cultural form: the first refers to ‘richesse matérielle sous la forme d’argent, de biens et de
valeurs mobilières [material riches in the form of money, goods, and transferable securities]’,
while cultural capital consists of ‘savoirs, de compétences, et d’autres acquisitions culturelles
[knowledge, competences, and other cultural acquisitions]’, such as educational qualifica-
tions (Bourdieu 2001: 26–7). Social capital, similarly, is held to be that capital mobilised
within the field (‘champ’ in Bourdieu’s terminology, see 1981: 113–20) of social interaction,
e.g. networks, relationships, etc., but excluding what is termed ‘symbolic capital’, which
includes ‘prestige, réputation, renommée’ (2001: 295).
localism of the variety chosen rather than its association with low proficiency.
This incorporates a diachronic dimension into the variation experienced in
SgE: Singlish, decried as ‘uneducated’, ‘poor’ English, is the result of having
to ‘accommodate a wide range of grammaticality’, thereby acknowledging
the important ‘group of the poorly educated’ (Alsagoff 2007: 41–2).28 This
‘structural inclusivity’ (2007: 42) of LSE accounts for the wide variation
observed in LSE, since it extends not only to proficiency, but also to ethnicity
(with LSE being able to accommodate various ethnic subvarieties; cf. Alsagoff
2007: 41, but also Deterding and Poedjosoedarmo 2000). Much of this can be
subsumed under the label ‘community membership’ (LSE feature of dyad
(e) in Table 2.10), but certainly the other features play a significant role too.
In conclusion, Alsagoff’s (2007) COM analyses variation in SgE as being
the result of opposing cultural orientations. One is globalist, serves eco-
nomic ends, represents authority, formality, and distance, and also signals
educational attainment; it results, linguistically, in ISE. The other is localist,
serves sociocultural ends, represents camaraderie, informality and closeness,
and signals community membership; its outcome is LSE. In so doing, COM
presents a significant improvement over diglossia, in that it does take into
account functional distribution, but at the same time accounts for the ‘leaks’
in Gupta’s (1994) data more satisfactorily than by code-switching. The im-
portance of culture in the model also recasts speakers as acutely aware of
their orientation when choosing between ISE and LSE as their target: users
of SgE are now ‘agents of culture, not merely bearers of culture’; a fact
that has effected the changes of LSE to accommodate this function (Al-
sagoff 2007: 43, 2010: 346). This inclusiveness of LSE increases its value
and endows it with the local cultural capital which it is often stripped of in
policy-making discourse.
As mentioned at the beginning of this subsection, the approach taken
in this model can be included in a wider philosophy of linguistic variation
and language use, termed the ‘sociolinguistics of globalisation’ (Blommaert
2010). Blommaert in particular, but also Pennycook (1994, 2007), argues for
a new thinking on how language is seen to vary globally. One major ele-
ment is the move away from a repertoire of linguistic variables, or features
(such as those often listed in anthologies of World Englishes, for instance),
to a repertoire of linguistic resources (see also Heller (2006) on the com-
modification of such linguistic resources). These resources may come from
languages that are (traditionally) from quite different parts of the world,
brought together largely through the effects of globalisation, mediated or
not by actual speakers. The examples given by Blommaert (2010) include
28
Alsagoff (2007: 42) does, however, point to the fact that LSE is not entirely to be equated
with learner varieties and Pidgin English. She cites Chew (1995: 165) who argues that the
two, while similar, are not the same in that ‘an educated English speaker speaking informally
[is distinguishable] from an uneducated speaker’. These differences, however, have not been
clearly investigated to date.
the various languages used in the multilingual and multicultural inner cities
of Europe, where migration has diversified the linguistic ecology (resulting
in ‘super-diversity’ in some select cases (Vertovec 2007)) and resulted in a
wider set of linguistic resources available to the ‘indigenous’ population, but
also the case of French-language advertising in central Tokyo (2010: 29).
The argument goes that in many such instances, it is not actually features or
bits of language(s) (French, English, Japanese, etc.) that are being used, but
resources that are being tapped into in order to achieve something: to convey
a sense of sophistication, to appeal to an international audience, to project a
certain attitude about the sender or the receiver of the message, and so on.
These resources are said to ‘index’ a particular stance, for instance – this
concept of indexicality is what I am now turning to in more detail.
2.2.2 Indexicality
Background
The models presented in Section 2.1 could be described as working within
what has been termed a ‘first-wave’ approach (Eckert 2005), with variables
taken to be markers of a given, predefined social category. These variables
carry prestige or stigma based on social class, a (subsequently recognised
as not easily definable) category which is regarded as a ‘map of social space’
(Eckert 2005: 3). Certainly the large-scale surveys of e.g. Ho and Platt (1993),
and the previous formulation of models correlating formality with levels of
education (Platt 1975, 1977; Pakir 1991) fall within this type of study. The
concept of style, in these studies, refers to the attention speakers pay to their
speech, and, according to Eckert, is controlled by an orientation towards the
prestige or stigma associated with the variables used (Eckert 2005).
It is particularly the relationship between variables and social categories
that is problematic in this approach. Of course the correlations between,
say, the variable (A:) in Norwich English29 (Trudgill 1974) and the social
classes taken into account are beyond statistical doubt. Trudgill (1974: ch.3,
cited in Chambers 2003: 51) also went to considerable length to ensure a
reasonably accurate picture of his informants’ social class, by computing
a ‘socio-economic index’ that took into account not only the informants’
occupation, but also their housing type, their income, their level of education,
their father’s occupation, and the neighbourhood they lived in, with each
given a score and weighted appropriately. Nevertheless, the socio-economic
index thus generated remains a predefined social category – much like sex or
age. This category is then ascribed certain variables, such as (A:), which has
a variant [A:] for those with a higher index, and a variant [ä:] for those with
a lower index. Speakers use one or the other based on how much attention
29
In Norwich the variable (A:), which occurs in words such as father, cart, after, has a prestige
variant [A:] and local fronted variants [ä:] or [a:] (Trudgill 1974: 97–9).
they pay to their speech; word lists and reading passages usually triggering
a higher rate of the prestige variant, and casual styles usually using more of
the non-prestigious variant.
However, Trudgill recognised that the selection of one or the other of
these variants can be more than simply a result of attention (which would
mean that the prestige variant is always the one desired). He explained some
of this variation as reflecting different kinds of prestige: overt prestige, which
is associated with the variant that is overtly recognised as prestigious (i.e.
the more RP-like), and covert prestige, carried by the local variant. This latter
kind of prestige has local currency, so much so that in the self-reporting tests
administered post hoc, informants quite often over-reported their use of the
local variant. The overt vs covert prestige distinction nevertheless continues
to regard variables largely as markers of social categories, with the choice
of one variant over the other being a result of an orientation to prestige or
stigma – with the latter potentially reinterpreted as being locally prestigious.
A departure from this approach can be seen in what Eckert calls ‘second-
wave’ studies. These are characterised by an ethnographic approach, where
the informants consist of typically small-scale, geographically defined com-
munities (Eckert 2005: 15). Rather than the ‘top-down’ methodology of first-
wave studies, where social categories are decided upon before the study
commences, and are then imposed on the sample, here the process is re-
versed, and it is locally defined categories that are seen to link to large-scale
demographics. An example can be found in Eckert’s own study of two
groups of US high school students, which she called ‘jocks’ and ‘burnouts’
(Eckert 1989). The definition of these two groups was not premeditated,
and arose in the course of the researcher’s interaction with the students. In
fact, these groups were not homogeneous, and did not have a clear delimi-
tation as a ‘clique’ in which membership is explicitly acquired or rejected.
Rather, students could be more or less strongly part of the group, interacting
with other students sharing their taste in music, clothing and, obviously,
linguistic features. The fact that many, but not all, of the ‘jocks’, who had
a culture of identifying more closely with school values, and incidentally of
using the more standard features of General American, came from what can
be called a middle-class background, was methodologically unconnected to
the definition of the group. The group’s behaviour, linguistic and otherwise,
is something that transpired from the grouping itself, rather than from an
extrinsic social category ‘class’.
Second-wave studies also broke with the first-wave idea of what a lin-
guistic variable stands for. In ethnographic studies of the second-wave kind,
variables are taken to index locally defined categories, rather than the large-
scale predefined social categories. For example, in Labov’s (1963) study of
variation in Martha’s Vineyard, the two often-cited variables (ay) and (aw)30
30
Labov’s notation; IPA (ai): [aI] vs [@I ∼ 5I] and (au): [aU] vs [@U ∼ 5U].
take on several meanings. Firstly, a centralised onset has a traditional, dialec-

tological meaning; it can be seen as a feature of the local accent. Lowering
of the onset, as a result of contact with the mainland, has occurred in more
recent generations, and this has given rise to a second, more symbolic mean-
ing. Those identifying more strongly with the island used the raised variants
(some even more so than in the traditional accent), whereas those intent on
leaving for the greener pastures of the mainland used the lowered variant.
The variables, therefore, point to a social meaning that is locally created.
The style thus created represents an ‘act of affiliation’ (Eckert 2005: 15) to
a particular category, which, however, is local in that it is both geograph-
ically and temporally restricted. Unlike in first-wave approaches, speakers
are not seen to be born into a particular social category, complete with
its sociolinguistic variables. Speakers use the variables as a means to be
part of or to distance themselves from the social categories that they find
themselves exposed to. Thus, a speaker can be part of several categories in
the course of his/her life, perhaps even concurrently, like Eckert’s (1989)
‘in-betweeners’.
Building on this new approach, ‘third-wave’ studies are concerned with the
ethnographic analysis of ‘communities of practice’ (Eckert and McConnell-
Ginet 1992; Wenger 1998; Eckert 2000). These communities are charac-
terised not by geographical or social categories like speech communities, but
by a common set of practices, goals, or values. Here the local categories dis-
cussed previously are defined not in terms of higher-order social categories
(first-wave) or as being linked to demographics (second-wave), but as being
built on a set of common stances (Eckert 2005: 30). These stances (activi-
ties, characteristics) are indexed by the use of sociolinguistic variables; these
variables then combine to form a style, which is here seen to be used for the
active construction of the speaker’s persona.
Eckert (2008) explains this by postulating an ‘indexical field’ attached to
every sociolinguistic variable. The field covers the potential range of social
meanings of the variable. One of the examples she gives is that of word-
final /t/ release in American English, illustrated in Figure 2.2. The variants
>
themselves, i.e. the actual realisation of /t/ (e.g. [th ], [∅], [t^], [P], [tP], etc.),
are not the main focus here. It is a list of the social meanings associated
with those variants where the /t/ is audibly released. The field consists
of attributes falling into three types: social types, permanent qualities, and
stances. Eckert is quick to point out that this distinction does not refer to
distinct categories of social meanings, but ‘emphasise[s] the fluidity of such
categories and the relation between [them] in practice’ (Eckert 2008: 469).
The local nature of the system is evident in that the variable can index, in
Eckert’s field of /t/ release in American English, a social type ‘British’. In
a British context, however, the same variable (/t/ release) would index a
different social meaning.
Educated
Articulate Annoyed
Formal Clear Emphatic Angry
Elegant Careful
Polite Effortful Exasperated
Prissy
Figure 2.2. ‘Indexical field of /t/ release. Boxes = social types, bold =
permanent qualities, italic = stances’ (Eckert 2008: 469)
The social meanings covered by /t/ release are quite diverse, and show
the usefulness of an indexical field. The stances, which range from ‘formal’
and ‘polite’ to ‘annoyed’ and ‘angry’, are but momentary expressions of one’s
state of mind; they are occasional in use and useful to mark one’s attitude at
a particular moment in time, in a particular situation. Permanent qualities,
on the other hand, are already part of the speaker’s identity, due to a habitual
use of certain stances. A person who repeatedly takes an exasperated stance
might come to be seen as an exasperated person through ‘stance accretion’
(Rauniomaa 2003, cited in Bucholtz and Hall 2005: 596, cited in Eckert
2008: 469). There is, therefore, considerable fluidity in the way in which
the adjectives in Figure 2.2 can be used. The permanent quality ‘educated’
refers to the social identity a speaker is constructing, in this case by the use of
released /t/, which may, in other speakers, be simply an occasionally taken
stance with the same meaning.
Clearly, this kind of approach to sociolinguistic variation is a major depar-
ture from the first-wave approaches described earlier. The level of detail with
which speakers’ variation in speech is analysed is much more fine-grained,
and enables a more relevant and localised explanation for a particular use of
a given variable. Of course, the conclusions one can draw from this kind of
analysis are not easily reproducible to the population as a whole, but they
do offer explanations for complex phenomena of linguistic variation such as
those observed in SgE.
Indexicality in Singapore English

The reproduction, for the Singaporean case, of an indexical field such as
the one in Figure 2.2 is not the main purpose an indexical analysis of SgE
should have. More importantly, the localised social meaning of sociolinguistic
variables, as well as their use by the participants involved, is of the essence.
Considering the widespread use of co-occurring instances of what has been
called, in diglossic terms, H and L variants, a closer look at their use in
conversational interaction will enable an attempt at a formulation of this
local social meaning. In the examples that follow, straight underlines indicate
features commonly associated with L, wavy underlines indicate features
commonly ascribed to H.
Consider, for instance, example (1), where i.C.1.m introduces, in the first
utterance, the term hor fun.31 In the second sentence, the speaker turns
towards the microphone (as reflected audibly in the actual recording), and
gives a definition of the term, presumably for the benefit of the researcher.
The explication triggers laughter in the other three participants, a reaction
the speaker obviously expected.
(1) We can eat hor fun there, I heard
:::::
that the hor fun ∅ quite famous.
[to microphone] er hor fun me:::
ans rice noodles. (i.C.1.m)
Attention is drawn to the absence of copular be in the first sentence,
where hor fun is first used, and crucially, to the presence of a 3sg inflexion
in the second sentence. This second sentence is addressed, more or less
directly, to the researcher. Apart from the explication of hor fun, which does
in itself highlight a (perceived) difference in local knowledge, the shift from
an L variant of (be) to an H variant of (3sg) at the point where the addressee
changes seems too much of a coincidence. It is conceivable that [−be] indexes
a stance that can be described as ‘local’, likely to be triggered by the use of
hor fun (a local, or at least regional dish). In contrast, the inflexion on means
can be considered as indexing a non-local, if not global stance, which brings
the concept hor fun out of its local meaning into the reach of the outsider’s
understanding.
Example (1) is of course only one of many where L and H variants co-
occur. Interestingly, the two sentences cannot be identified as being entirely
L and H respectively, given the use of an inflected heard between two L
variants. To illustrate another intertwined use of H and L variants, as it
were, consider example (2):
(2) How long we :::
are supposed to talk? (iii.M.1.f )
31
Term of Cantonese origin, 河粉 ho4 fan2 (Mandarin hé fěn). A Chinese dish consisting of
broad, flat rice noodles, fried with prawns, fishcakes, vegetables, etc., and served in gravy
(Lee 2004). The anglicised spelling with non-rhotic r and u for /a/ is the usual one in
Singapore and is the one reported in Lee (2004).
Here we have a non-inverted interrogative – traditionally diagnostic of L –

combined, in the same clause, with a realised copula, [+be]. While the [+be]
variant is not per se diagnostic of either H or L, it is nonetheless the only one
accepted in H. This raises the question of why, given a choice, the speaker
did not opt for a fully L-marked utterance: if anything, the use of the copula
lessens the ‘L-ness’ of the question.
The explanation for the use of the variants in (2) is not as straightforward
as in (1), where the presence of (perceived) local lexical items helps in the
identification of an indexical stance taken by the speaker. There is also,
in (1), a clear shift between the first and the second clause. This is not the
case in (2): the question is a single clause featuring one instance of an L
variant immediately followed by, if not combined with, a non-L feature. The
turn itself is nested between others that use overwhelmingly non-L and H
features, as shown in (3):
(3) 3: So we ::::
have:::::::::
decided. Three hundred bucks per person,
Sarawak:’s the place. God bless Sarawak man. Yeah. (***)
ok. So ∅ everybody agree with Sarawak? Yes! Well done!
1: How long we ::: are supposed to talk?
2: So that’:s all?
3: Yeah, because we agre:: ed on Sarawak, Sabah. (iii.M.gr)
This exchange is predominantly in a variety of SgE that would be de-

scribed as H – the only exception being the question of the first informant
(iii.M.3.m) ‘So everybody agree with Sarawak?’, which lacks do-support.
It seems therefore unusual, from a diglossic point of view, for iii.M.1.f to
switch to a style that includes L features. A possible explanation could be that
iii.M.1.f has a lower proficiency level than her interlocutors – an assumption
not borne out by the rest of the data, however. Likewise, it is possible for the
non-inversion to have been triggered by the lack of do-support in the pre-
ceding turn. Much more likely, however, is the nature of the question itself:
it is, after all, an expression of impatience, signalling the speaker’s annoyance
with the task at hand (the turn is uttered 10 24 into the 15 recording). It is
conceivable, therefore, that the use of an L variant in this utterance indexes
a stance ‘annoyed’ or ‘impatient’. The concurrent use of [+be] might then
be seen as a remnant of the overall H nature of the interaction, where StdE
features predominate.
That the variables investigated here can be used to index a wide array of
stances is further exemplified in (4). Here I have highlighted two instances
of noun plural inflexion (toiletries and beaches), reported by Gupta (1994: 13)
as being another StdE feature. It thus appears that after four H variants,
four consecutive L variants occur, only to be closely followed by another H
variant.
(4) 4: Oh, very very important. Where are we ::::: going:::

to:::
go for
sanitary, our toiletr:::
ies and everything?
1,3: [exasperated sighs and laughs]
2: Oh my god!
4: Aah, yes, nobody :::::::
thought of that!
3: There the beaches :
never got toilet∅ ah, got public toilet.
4: Yes yes got, I think they:::::
have :::
got it. They protect the beach,
a heritage site for them. (i.I.gr)
It seems reasonable to suggest that the L variant of (got) in i.I.4.m’s reply to
i.I.3.m is triggered by the latter’s twofold use of this variant in the preceding
turn. The speaker then repeats his objection, this time with an auxiliary have,
an H variant.
What might have triggered i.I.3.m’s L variants of existential constructions
could well be the nature of the topic discussed. There is a certain light-
heartedness about the issue, as evidenced by the laughs of the participants
once the topic is introduced. A similar orientation can be observed in i.I.4.m’s
‘Nobody thought of that!’, which has a mildly sarcastic overtone, but here,
an H variant is used. The difference between the two is that in the case of
thought the H variant mockingly indexes a stance ‘important’, whereas in the
case of got, the L variant indexes ‘light-heartedness’. The inflected beaches
may simply be a remnant of the preceding turn’s H-fullness, but need not
index H at all, since, as explained earlier, inflexions can also variably occur in
CSE (this applies to both verbal and nominal inflexions, see Gupta 1994: 13).
Thus far, examples (1)–(4) have shown several possible stances that can
be taken, using the grammatical and lexical variables we have focussed on.
Table 2.11 summarises them. Some, like ‘local’ vs ‘global’, are reminiscent
of the dyads proposed by Alsagoff (2007) in her ‘cultural orientation model’
(COM, see the previous section). This is not a coincidence: her model
takes an approach to SgE variation that is very similar to Eckert’s (2008)
indexicality. Alsagoff’s COM is, however, much more concerned with the
idea of a continuum between two varieties (Local and International Singapore
English), whereas an indexical approach would, in the present case, leave the
definition of distinct varieties aside. This point will be considered in more
detail in the concluding chapter.
While grammatical and lexical variables of SgE are the prime elements
in this study, I will here briefly explore one additional feature occasion-
ally observed in the group recordings: code-switching.32 This is an unusual
variable in that it is unsuitable for first-wave studies, but potentially inter-
esting for the kind of ethnographic, small group-based approach this chapter
has taken. Code-switching, however common in everyday interaction in
32
I use a relatively loose definition of code-switching here, with any non-StdE lexical item
classified as a switch.
Table 2.11. Selected variables

and the stances they index
L variant H variant
local global
regional serious
exasperated important
annoyed
light-hearted
Singapore, does not occur widely in the data. This is due to many factors,
chief among them the setting in school and the fact that the informants are
aware of being recorded. In the group recordings, however, the first time that
all four informants were together and without the researcher, some instances
of code-switching did occur – and the same is true for the radio-microphone
recordings. Two instances of Tamil were observed:
(5) a. Kot.ikirtuku need three days ah. (ii.I.3.f )

b. And we¯ mustn’t forget Mr Man.matan. (iii.I.2.f )
¯
In (5a), kot.ikirtuku is a colloquial Tamil verb phrase that translates as ‘to
eat a lot’, and in¯ (5b), man.matan is a noun that roughly translates as ‘good-
looking male person’. The context ¯ makes it clear that the referent, in this
case, is the researcher.
It is reasonable to consider code-switching as a feature of CSE, or certainly
of LSE – code-switching is, more often than not, highly localised in nature.
What, then, are the triggers for these switches? They occur in isolation and
are not surrounded by any other Tamil switch. Perhaps the fact that all
participants are of the same ethnic background could be an explanation,
particularly with the official ‘mother tongue’ policy that links language and
ethnicity. This, however, does not take into account ii.I.gr, where ii.I.1.m is
Punjabi and openly admits to not understanding much Tamil at all. In fact,
the explanation for the switches could well be not much different from those
offered for the previous examples. Consider (5a), for instance, where the
topic at hand is that of food and of the necessity of allowing enough time at
the chosen holiday destination (Thailand) in order to appropriately sample
the local delicacies. A similar instance was observed in (1), where the topic
was food as well, and where the term hor fun (arguably not a switch, but at
the very least a lexical item for a region-specific dish with local relevance)
was accompanied by another CSE or L feature, be-deletion. The same is
true of (5a), where a discourse particle ah is used. The data, therefore, seem
to suggest that the topic of food has a strong influence on the choice of
variants.
Similarly, references to the researcher are likely to be indexed by L vari-

ants. In (5b), the researcher, who is not immediately present, is referred
to with an ostensibly opaque Tamil noun. The reason for this choice, one
can surmise, is an assumption that the referent is unable to understand the
flattering description – a correct assumption, until I enrolled the help of a
native speaker. Note that here, there is no other L variant in the sentence; in
fact, mustn’t forget is an auxiliary-supported structure that is considered H
by Gupta (1994: 13). This could be explained by the fact that the utterance,
if anything, carries positive overtones, with a suggested invitation of the
researcher to the group’s weekend getaway. To put (5b) into perspective,
remember example (2), where instead of a direct reference to the researcher,
there is a mild criticism of the recording process, a sign of exasperation
possibly meant to be heard by the researcher. And again, the speaker uses
an L variant (lack of inversion in an interrogative) followed by an H vari-
ant ([+be]). What this seems to show is a kind of struggle between two of
Alsagoff’s orientations, ‘local’ and ‘global’, in their widest sense. There is a
desire to connect locally with the other participants, discussing the outsider
or subverting the interview setting, in a manner that is local and not immedi-
ately penetrable to the researcher. At the same time, however, the interview
setting itself and its ‘global’ nature – due to the (physical or otherwise) pres-
ence of a foreign interviewer – is given consideration as well. This is achieved
by the simultaneous use of variants that have previously been ascribed to a
given subvariety of SgE (CSE/L/LSE/basilect or SSE/H/ISE/acrolect,
depending on the author and the model).
There are many more examples of this in the data, but I shall conclude
this section by mentioning a different stance that can be indexed, in SgE,
by several variables. Ethnic accents of SgE have been investigated in some
detail (see e.g. Deterding and Poedjosoedarmo 2000; Lim 2000; Tan 2005;
Deterding 2007b), although there has not been a clear account of what exactly
makes one ethnic variety distinct from the other. The differences in vowel
quality investigated by Deterding (2007b) are not large enough to account for
the high success rates observed in ethnic accent identification tests (Deterding
and Poedjosoedarmo 2000). Besides code-switching, which is a (generally)
obvious marker of ethnicity, Deterding identifies intonation as a possible
marker. Thus, when searching for examples of ethnic indexing in the data,
phonological variables are of paramount importance. Example (6),33 taken
from the radio-microphone recording ii.C.rm, and described in Leimgruber
(2009a), is, however, more than simple ethnic self-indexing. Here, a group
of Chinese polytechnic students use (what they perceive to be) variables
indexing an Indian ethnicity – they are basically imitating, if not mocking,
an Indian accent.
33
A note on the transcription used in (6): passages of interest (those highlighted with braces)
are in IPA. Intonation is indicated by means of tone marks. While this is not entirely
orthodox, it gives a useful impression of what the actual recording sounds like.
(6) 1: Do you know want to slap you?

4: ’scuse me?
A

1: Ě
Oh, na: £TIN na:Ě£TIN
Ě
4: Oh, na-.. oĘ£ na:Ę£úIN na:Ę£úIN. maãí £ na;ãíŁŔ£

B C
2, 3: [laugh]
4: diNgíĚ£ na;ãíĘ£ g̊Imíno:ŁŘ£
" "
D
All: [laugh]Ă Ă
Ě Ę£ mi olĚ£ dIsĂ£ kaInãĚ£ OfĂ£ úINĘ£
4: ma £ãó £ ãonĚ£ kam £ æn sEí
" Ę£ b@tsĂ£ plisĚ£. baãĚ£ itsĂ£ onliĘ£ inaf fOĽŐ£ wanĘ£ b@d^Ă£ E
aĄ£. fa;Iv
Ă Ă
2: w5n Ę
£ b@d^£ noĚ£ lak^£
F
?: [singing]
All: [laughing] (ii.C.rm)
The context of this exchange shows the spontaneous nature of the use of
ethnic indexing. Speaker ii.C.1.f makes a good-natured threat to ii.C.4.m,
upon which the latter asks for clarification. ii.C.1.f then shields herself by
dropping the subject (‘nothing, nothing’, A in (6)), but doing so with length-
ened first syllables and an unusual intonation pattern. ii.C.4.m starts to repeat
the utterance, checks himself, and restarts, this time with a much stronger
mock-Indian accent (B) – notice the use of retroflex [ú], where ii.C.1.f pre-
viously had [T]. He then carries on with two nonsense ‘Indian’ syllables (C),
with abundant use of retroflexion and sing-songy intonation. This is met
with laughter by the other two participants, a reaction that the speaker seems
to enjoy: the following turn (D) is a similar nonsense construct, again char-
acterised by several retroflex consonants, length contrast and a wide pitch
range. After another round of laughter by the whole group, ii.C.4.m em-
barks on a longish turn in this mock-Indian accent (E), which includes two
sentences which are (not entirely accurate) quotes from a reasonably famous
Visa credit card advertisement visibly set in India and featuring Richard
Ě Ă
Gere (the term mother [ma £ãó £] refers to ii.C.3.f, who, for unknown reasons,
"
is given this nickname throughout ii.C.gr and ii.C.rm, mostly by ii.C.4.m).
The reference is taken up by the group, with ii.C.2.m continuing the par-
ody ostensibly with another quote (although this one does not appear in
the original), and with an unidentified group member starting to sing the
advertisement music, resulting in general laughter.
Clearly, this kind of indexical usage of sociolinguistic variables is highly
reminiscent of ‘crossing’ (Rampton 1995, 2006), although it does not work
in exactly the same way. Typically, in crossing, the speaker uses a stylised
version of another participant’s variety, and this participant is somehow
involved in the exchange (as addressee or overhearer), or, in the case of
‘stylised Asian English’ (Rampton 1995, cited in Coupland 2007: 139–42),

the speakers themselves are associated with the variety into which they cross
(but which they do not normally use, e.g. Asian British using stylised Asian
English when they normally use the local vernacular, see Rampton 1995).
Example (6), however, features four Chinese participants, with one of them,
ii.C.4.m, starting an imitation of Indian-accented English (marked with ex-
tensive use of retroflexion, unusual vowel length, and varying intonation),
and even imitating a nondescript ‘Indian’ language (C and D above). No par-
ticipant is of Indian background; it is only the ‘nothing, nothing’ by ii.C.1.f
that triggers the ‘crossing’. Central to the exchange is the introduction, in
E, of snippets from the Visa advertisement, immediately recognised by the
participants. Not only does this strengthen ii.C.4.m’s position in the group,
it also explicitly indexes the stylised accent as ‘Indian’, thus removing all
doubts as to its social meaning. It is also an example much less of crossing
than of ‘ritualised interaction’ (Rampton 2009).
From an indexical perspective, then, there is a subset of ethnic stances that
can be indexed linguistically by using several variables: a stance ‘Indian’ is
achieved, in SgE, even by non-Indian speakers, by using retroflex consonants
and a distinctive intonation – regardless of whether these features are actually
used by real Indian SgE speakers – and other features, such as topic and
non-linguistic elements (e.g. singing) can act in a similar way. The extreme
example in (6) is one where the participants immediately recognise the stance
indexed. This need not always be so, but in this case, the connection between
the stylisation and the real-world referent category, as mentioned, was made
explicit.
2.3 Conclusion
This chapter has presented various attempts to explain the variation ob-
served in SgE. The difficulty of the task is apparent in the number of models
proposed: of the six given in this chapter, each was trying to improve on
previously presented ones. Platt’s (1975) was the first to attempt an in-
clusive model to capture the complete speech community’s behaviour; his
continuum model, while generally considered outdated (Ansaldo 2004), has
become an influential starting point for other models. His later work on
polyglossia (1977) was similarly groundbreaking, and has the merit of being
the only model to capture the multilingual element in Singaporean society:
the other models either ignore the presence of other languages (Platt 1975;
Pakir 1991; Poedjosoedarmo 1995), or acknowledge their presence but do
not integrate them into their model (Gupta 1994; Alsagoff 2007). This gives
the polyglossic model a certain edge over the others in recognising an added
level of complexity that often escapes analyses which focus on SgE alone.
However, since it is built essentially on a complex diglossic framework, it
suffers from the same shortcomings associated with diglossia (outlined in
2.3 Conclusion 63
Section 2.1.3). The diglossic model (Gupta 1989, 1994, 2006b) also has its
merits, firstly because it explicitly casts SgE speakers as native speakers of
English, having at their disposal two subvarieties of the language, used in
different situations in accordance with diglossic rules (Ferguson 1959), and
secondly because it reflects the speakers’ perception of the situation. This
tempting application of diglossia, termed ‘leaky’ (Gupta 2006b: 22; Fasold
1984: 41) because of the non-categorical (Chambers 2003: 26–38) distinction
between H and L, has been called into question due to its breaches of several
of Ferguson’s (1959) criteria (see p. 44). I concluded Section 2.1.3 by arguing
that while Singapore may not present a case of classic Fergusonian diglossia,
a two-way distinction between a perceived H and L in a situation that is,
linguistically speaking, a continuum is not unheard of (cf. Willemyns 1987,
but also DeCamp 1971: 368), and might, therefore, have some merit.
More modern approaches include Alsagoff (2007), who proposes a model
that takes cultural orientation as its main element. COM takes a markedly
different approach from previous models by taking a step back from the
more narrowly linguistic analyses to a more macro-level view, which takes
into account issues normally not captured in traditional, first-wave varia-
tionist sociolinguistics. The interaction between two antithetical orientations
(global and local) results, at the linguistic level, in the variation observed in
SgE between the poles ISE (target of the globalist orientation) and LSE (its
localist counterpart). Within this model, cultural orientation also explains
style-shifting, issues of power and solidarity, and negotiates the thorny is-
sue of language proficiency. This presents a very satisfying alternative to
the diglossic framework and its reliance on ‘leaks’ or code-switching, and
to the continuum model and its quasi-deterministic slant regarding socio-
economic factors. A step beyond COM is the indexicality model introduced in
Section 2.2.2: the arguments made by Alsagoff are extended beyond cultural
orientation, such that any utterance is interpreted as potentially indexing a
particular social meaning. These social meanings (called stances, permanent
qualities, and social types by Eckert 2008) are indexed by the use of fea-
tures (sociolinguistic variables) which the speech community recognises as
belonging to either Singlish or Standard English. Speakers may then choose
features from either set when building their utterances, enabling them to
express a host of nuanced social meanings, sometimes (as in COM) conflict-
ing ones. It is this indexical framework that will underpin the description of
Singapore English that follows.
3 Description: phonology and lexicon
When describing a variety such as Singapore English, the first question

that arises is: which Singapore English? The Singlish used by the lesser-
educated hawker is quite unlike the polished Standard English employed
by a minister when addressing parliament, and not just in the choice of
vocabulary: pronunciation and grammar differ, too. And yet both have to
be considered instantiations of Singapore English, as they are, obviously,
English, and used by Singaporeans. This variability is omnipresent, as I
hope to make explicit in the following description.
3.1 Phonology
3.1.1 Vowels
The monophthongs of SgE are generally given as in Table 3.1 (see e.g. Bao
1998 or Deterding 2007a). This inventory shows the absence of phonemic
length distinction (i.e. beat and bit are pronounced identically); this absence
seems generalised in the speech community. There is, however, conditioning
by phonological environment: vowels are ‘relatively short, except in open
syllables’ (Bao 1998: 156). Thus beat and bit are both /bit/, but bee, being
/bi/, is realised as [bi:].
SgE has five diphthongs (face and goat1 are monophthongised to /e/ and
/o/ respectively) which are phonologically identical to RP’s,2 but phoneti-
cally much narrower, the two centring diphthongs particularly so. Deterding
(2007a: 26–7) mentions that the diphthong /u@/ distributes differently in SgE
from how it does in RP. In the words poor, tour, and sure the vowel is usually a
diphthong [u@], whereas in pure and cure it tends to be [jO]. This, Deterding
notes, is ‘the inverse of the common pattern in modern RP’ (2007a: 26).
A merger that is not restricted to Singapore, the dress–trap merger,
collapses quite a number of minimal pairs into a single vowel: bed–bad,
pet–pat, dead–dad, pedal–paddle, and so on. The merger has been shown to
1
‘Lexical sets’, introduced by Wells (1982) to easily refer to the vowel used in that keyword.
2
The exogenous Received Pronunciation (Roach 2000, 2004) is used here for convenient
comparison.
64
3.1 Phonology 65
Table 3.1. Vowel system of Singapore English
Front Central Back
Close i u
Close-mid e @ o
Open-mid E O
Open a
Diphthongs Oi ai au i@ u@
be less pronounced in educated speakers and more careful styles (Suzanna

and Brown 2000), although measurements by Deterding have again shown
it to be present in educated speakers (Deterding 2003a, cited in Deterding
2007a). What seems clear is that the existence of the two vowels is no secret
to speakers, as the following example from Leimgruber (2012b) reveals.
During a light-hearted post-dinner conversation with a group of similarly
aged Singaporeans, in which I was the only ‘Western’ foreigner, a question
arose from one of the group regarding the culinary preferences of, precisely,
ang mohs. Ang moh (see next section) is a SgE term of Hokkien origin (紅毛,
âng mĲo, lit. ‘red hair’), and is either a noun for ‘a Caucasian, a white person’, or
an adjective meaning ‘having the nature or attributes of a Caucasian or white
person’ (Lee 2004). It is always pronounced [aN mo] in regular Singlish. In
this particular instance, however, the speaker pronounced it [EN mo], which
resulted in laughter by all present, including the linguist addressed, much
to the speaker’s embarrassment. What must have happened was that the
speaker, having heard the outsider’s version of the the trap vowel (a less
open [æ]) several times that evening, and assuming that ang moh contained
the ‘short a sound’, perhaps due to its usual spelling, moved the item ang moh
from its original phonological system into another English one, reassigning
it to the trap set, which is, of course, pronounced [E] in most varieties
of Singapore English. The resulting laughter can be interpreted, too, at
several levels, firstly by being directed at the misplaced pronunciation, more
likely, however, as a reaction to the pretentious stance such a pronunciation
would index, but also, crucially, as a recognition of the irony inherent in
that pretension being indexed through an ‘ang moh’ pronunciation (albeit a
mistaken one) of that very same item, and therefore backfiring.
3.1.2 Consonants
In terms of consonants, the SgE repertoire is essentially the same as that
of RP. Wee (2008) gives the inventory in Table 3.2. Differences lie in the
realisation of plosives, which are almost never aspirated in SgE. There is
some evidence for a correlation between sociolect and aspiration, with higher
Table 3.2. CSE consonant inventory (based on Wee 2008: 267)
Post-alveolar
Labiodental
Alveolar
Bilabial
Glottal
Dental
Palatal
Velar
Plosive p b t d k g
> >
Affricate tS dZ
Fricative f v (T D) s z S Z h
Nasal m n N
Approximant w ô j
Lateral l
mesolects showing higher rates of aspiration (Leimgruber 2005: 59); informal

field notes of overheard conversations suggest that this association of aspira-
tion with some sort of exonormative standard is also exploited for indexical
purposes. Final consonant clusters are usually reduced and single final con-
sonants unreleased or glottalised (Deterding 2007a: 17, 19), while a released
(and sometimes aspirated) final plosive is used as an index of a pretentious,
if not pedantic stance, as in (1). SgE is generally non-rhotic, lacking linking
and intrusive /r/ (Trudgill and Hannah 1994: 135). Conversely, Tan and
Gupta’s (1992) research suggests a different picture, with a possible change
in progress: some of their informants used non-prevocalic /r/, and more so
when formality increased. Age correlated very closely with rhoticity, leading
to their conclusion that a ‘change . . . may be in progress’ (Tan and Gupta
1992: 39–40), at the expense of the non-rhotic variant.
(1) Ah, how much would it cos[th ] (.) to travel there? (i.I.4.m)
The lateral approximant /l/ is often velarised throughout ([ë]), sometimes
vocalising to [U]. This can have consequences: by regressive assimilation, it
can monophthongise the /aI/ in aisle, for instance, into [a]. When preceded
by a nasal, /l/ tends to be deleted and /n/ palatalises to [ñ] when the
following segment is /i/ or /j/: thus only /onli/ > [oñi] (Bao 1998: 165) and
mainly /mEnli/ > [mEñi] (Leimgruber 2005: 11). Word-final consonants are
often devoiced, fricatives most consistently. Bao (1998: 154) notes that dental
fricatives are pronounced as labiodentals when they occur word-finally (as
in British th-fronting), but as dental stops prevocally: thus, healthy may
be [hElti] or [hElTi] but health is realised as [hElf] or [hElT] (Bao 1998: 154).
Other positions are not affected by this process: Deterding (2005) reports that
CSE speakers have considerable difficulty understanding southern British
English speakers’ th-fronting (the pronunciation of /T/ and /D/ as [f] and [v]
3.2 Lexicon 67
respectively, regardless of position, as in thought [fO:t], something [s2mfIN],

and youth [ju:f]).
3.2 Lexicon
3.2.1 Etymology
A major feature of colloquial SgE is its stock of substrate-derived lexical
items. Chinese languages have provided the strongest input, but Malay has
also contributed significantly, due to its original status as a lingua franca in
the colonial era (pp. 9 and 15). The Indian languages, however, have only left
limited traces – and words designating cultural artefacts and practices or re-
ligious holidays, such as Deepavali (the Hindu Festival of Lights, of Sanskrit
etymology), cannot be said to be restricted to CSE, or, indeed, SgE. Con-
versely, Malay borrowings designate items of everyday usage: roti ‘bread’,
barang-barang3 ‘belongings, luggage’, makan4 ‘food, to eat’, bodoh ‘stupid’,
etc. The same holds for Hokkien, which contributed ang moh ‘Westerner’
(lit. ‘red hair’), shiok ‘exceptionally good’, kiasu ‘characterized by a grasping
or selfish attitude arising from a fear of missing out on something’,5 jia lat
‘terrible’ (lit. ‘sapping strength’), and many more. Cantonese had a lesser
influence, but provided sap sap sui ‘insignificant’ (lit. ‘water drops’).
Another feature of CSE’s lexicon is the use of English lexical items with a
semantic field different to that of Standard British English.6 These include
send in ‘I’ll send you home’ (to mean ‘I’ll give you a lift home’), follow in
‘Can I follow your car?’ (for ‘Can you give me a lift?’), keep in ‘keep the
glasses, please’ (for ‘put away the glasses, please’) and on and off as verbs in
phrases such as ‘on the light, please’ or ‘off the fan’. The verb to renovate,
in SgE, refers to the action of furnishing, decorating, etc., an empty flat one
has just acquired, and which may be brand new. The British meaning of
this verb is found again in SgE to upgrade. Upgrading most commonly refers
to government-funded renovation schemes in public housing estates, which
can include the adding of lift shafts, and sometimes even additional rooms,
to existing blocks of flats.
A third subset of the SgE lexicon is derived from English words which
have since lost some currency in StBE: spectacles for glasses and alight as used
in everyday conversation. Similarly, to patronise (a shop, etc.) is frequently
3
Reduplication of nouns in Malay marks the plural.
4
‘In Malay “food” is makanan; makan means “to eat”’ (Platt et al. 1983: 15). In SgE, makan is
both verb and noun.
5
Definition from the Oxford English Dictionary (Simpson and Weiner 2000, draft entry March
2007, sense B). Notwithstanding the order of the OED entry, the adjective is more common
than the noun.
6
I will use Standard British English (henceforth StBE) as the original exonormative model
for ease of reference and comparison. Certainly in the early stage of SgE formation, this was
the superstrate variety involved, and until very recently, any non-British Standard English
was seen as ‘mistakes’ (Gupta 1986: 80).
used across the continuum of SgE, whereas in StBE it is much less used, and
then mostly in its condescending sense.
There is variation, in SgE, with regard to the use of these lexical items.
While the English items above are usually found across the continuum,
borrowed words are normally replaced: makan is unlikely to appear in any
SSE7 utterance. This does not apply without exception, however: thus blur
‘ignorant, confused’ is restricted to CSE, whereas ang pow (from Hokkien
âng-pau), which would only be partially rendered by the English New Year’s
gift,8 is perfectly acceptable in the Standard. Other examples, such as ki-
asu, have made it into the OED, and do, therefore, appear in formal speech
and printing: especially kiasu, which is often used as a mild and jocular
self-criticism of a perceived stereotypically Singaporean trait, even in parlia-
mentary speeches.
3.2.2 Borrowing and code-switching

The ‘Singlish words’ described above, most of which have their origin in
Hokkien or Malay, are the typical loanwords that are described in popular
descriptions of the variety, as well as attempts at more serious documentation.
It seems reasonable to assume that a sentence such as (2) is an example of
borrowing rather than code-switching: the utterance is completely Standard
English, with the exception of one word.
(2) I wish that the Government Ministers do not become infected
with the same kiasu syndrome that they themselves have advised
other people against. (Parliamentary debate, 1990)
While it is possible to argue that kiasu is in fact Standard English, being,
as it is, included in the OED, the fact remains that it is a loanword of
Hokkien origin, that has been borrowed into (Singapore) English – even
the OED notes that it is used ‘orig[inally] and chiefly in South-East Asia,
esp. Singapore’. Examples such as (2) are distinct from code-switching,
exemplified, for instance, in (3). Here there are whole clauses in two or more
languages, rather than the odd single borrowing.
(3) a. He’s very nice, ∅ you know why? Because tā yě shı̀ xı̀ng X
leh. Because keoi
:::::
jat1 yap:::::
5 :::::: dou3::::
lei4:::::::::::::::
zoek6 tai2 zoek6 ::::::::
ngo5 ge3
san4 zu3 paai
::::::::::: 4 leh, aa 3 maa
::::::: 4 de X mun
:::::4 .
b. He’s very nice, do you know why? Because his surname is also
X. After he came in, he was looking at our ancestral tablet,
[and saw] grandmother’s name.9
7
CSE = Colloquial Singapore English, SSE = Standard Singapore English. See p. 41.
8
The ang pow (紅包, lit. ‘red packet’) is part of the Chinese New Year tradition. The small red
envelopes contain money and are given to relatives to celebrate the Lunar New Year.
9
Cantonese romanisation in Jyutping, Mandarin romanisation in Pinyin, see Appendix A.
3.2 Lexicon 69
The speaker in (3) begins her turn with an English question, which she
answers herself, starting with the English conjunction because. She then
switches into Mandarin, and ends the sentence with one of the famous
Singlish discourse particles, marking a tentative suggestion (see Section 4.2).
The following utterance again begins with the English conjunction because,
followed by an immediate switch into Cantonese. Cantonese remains the
main code for the rest of the utterance, with the exception of another leh and
a quick switch back to Mandarin (de X).
Among other things, it is interesting to note that the speaker’s surname
(which has been obscured here) is uttered in Mandarin – this is unsurprising
in a speech community where the same Chinese surname often comes in
several forms. For instance, a person may be called Tan in his passport and
official records, and called Chén when speaking Mandarin and Can4 when
speaking Cantonese. When written in Chinese characters, it can be either 陳
(‘traditional’ or ‘full-form’) or 陈 (‘simplified’). This multitude of forms is in
part due to the various policies in place at different times; early registration
under colonial rule often saw the Hokkien pronunciation being chosen as
the basis for the romanisation in official records (thence Tan even for
Cantonese Can4 ).
To come back to (3), there is little by way of explanation as to why the
switches occur where they do – a wealth of research on code-switching
(Muysken 1995; Myers-Scotton 1997, 2000, to cite but three) has posited
numerous linguistic constraints on when and where switches can occur, most
of which were refuted by subsequent findings. What matters here is that the
speaker has at her disposal three codes, which she more or less freely mixes in
regular conversation, provided the addressees also command the three codes.
The expectation that when speaking to someone who does not understand
Cantonese, but only Mandarin and English, the switching will be limited to
these two varieties, was confirmed in observation.
What deserves consideration are the two ‘switches’ involving the particle
leh. This pragmatic particle (see Section 4.2.1), found throughout Singlish
with the function of indicating a tentative suggestion, is postulated by Lim
(2007: 463) as having a Cantonese origin, namely le1 , which ‘indicates “what
about?”’ (2007: 461). This raises the question whether it belongs, in the
context of (3), to Singlish or to Cantonese. The Singlish option seems more
obvious in the first case, as it co-occurs with the English because, but the
Cantonese option seems more obvious in the second case, which comes after
a rather long string of Cantonese.
If one were to consider leh Singlish, one can again ask whether the two
particles are the only instances of Singlish, with the remaining English bits
being Standard. Certainly, there is little by way of nonstandard features
in the English strings (e.g. the lack of do-support in you know why, but
see the overt copula on the first he). A diglossic approach would probably
suggest switches between Standard English and Singlish here, but it is more
enlightening, as suggested in Section 2.2, to consider individual features of

Singlish (lack of do-support, particles) and Standard (copular be) to occur
sporadically during nondescript ‘English’ discourse.
The data presented here show the kind of interlocking of varieties that is
common in everyday Singaporean speech. Whether they are loanwords, as in
(2), or clear examples of code-switching, as in (3) (where features from two
or more languages’ grammars, not just lexicons, mix), this co-occurrence of
elements from originally distinct codes results in a combination that, more
often than not, is considered the default. In other words, it is monolingual
(non-switching) discourse that is marked.
3.3 Conclusion
This chapter attempted a description of the phonology and the lexicon of SgE.
It highlighted, in both cases, the high level of variation within the variety
‘Singapore English’, and, as a result, the difficulty of such a description
using a traditional, variety-based approach. The example of code-switching,
presented in Section 3.2.2, further highlights the high degree of linguistic
heterogeneity in Singapore: descriptions of SgE rarely take the whole range
of codes used into account. The reasons for this are obvious enough: the
additional work required in treating not just English, but also a speaker’s
collection of codes belonging to other languages (a repertoire which can
change from speaker to speaker), is considerable. An approach based on
indexicality, as shown in some of the examples here, has a lot more to tell us
about how Singlish, English, and other ‘varieties’ interact and merge together
in actual spoken interaction.
4 Description: grammar
This chapter is divided into two sections: Section 4.1, ‘Syntax and mor-
phology’, will consider three major topics: features connected with the verb
(inflexions, auxiliaries, copula deletion), existential constructions with got,
and aspect marking. Section 4.2, ‘Semantics and pragmatics’, is concerned
with discourse particles and lexical semantics. The grammatical analysis in
this chapter draws largely on the group recordings (see Section 1.3 for an
explanation of the data collection procedure). Data are referenced according
to the coding scheme described in note 10, p. 9 above.
4.1 Syntax and morphology

4.1.1 The verb
Verbal inflexions
In Gupta’s diglossic analysis of Singapore English (1994: 9–13), there are
two types of ‘features’: those that are diagnostic of L, and those that are
diagnostic of H. One such feature is the verbal inflexion marking the third
person singular, exemplified in (1) below. The use of a zero inflexion in (1a) is
diagnostic of L, and marks the utterance as L. However, the use of -s in (1b),
while required in H, is optional in L (Gupta 1994: 9): therefore, when the
verb form has an H inflexion, it is not unambiguously part of an utterance in
H, since -s as a marker of 3sg is optionally present in L.
(1) a. He want∅ to see how we talk. (ii.M.2.f )
b. Hor fun means rice noodles. (i.C.1.m)
Another inflexional morpheme that occurs variably in SgE is past tense
inflexion, represented here by (-ed) (but covering also irregular past forms).
Again, its absence marks the utterance as L, but its presence does not auto-
matically turn the utterance into H. Examples from the data include irregular
(2a) and regular (2b) past forms, both marked and unmarked. The (-ed) vari-
able includes preterites and past participles.
71
(2) a. (i) My auntie went there, she went shopping and she wants
to go back just to shop. (ii.I.4.f )
(ii) He doesn’t say we cannot include our own money.
(ii.M.3.f )
b. (i) He’s rather experienced. (i.I.4.m)
(ii) You know, he marry a Singaporean. (iii.C.2.f )
Therefore, within a diglossic framework, and paraphrasing Gupta’s
(1994: 10–13) definitions, the effect of verbal inflexion can be formalised
as in (3) below. This formula applies to both (-ed) and (3sg).
(3) a. Absence of inflexion −→ L
−→ H
?
b. Presence of inflexion −
→ H
?
−
→ L
As examples (1) and (2) show, these inflexional features occur relatively
widely in the data. The variable (-ed) had 21 instances of the L variant,
and 77 of the inflected variant. Similarly, (3sg) had 13 instances of [∅],
and 52 of the inflected type. The examples in (4) and (5) illustrate the use
of these variables. It is interesting to note, as for instance in the first two
examples of (4), that both variants can occur in the same utterance. This is
sometimes a result of self-correction, but not always, as the change can go
in both directions. This latter point is of relevance, since it may well be that
the co-occurrence of H and L variants is not coincidental in the context of
the exchange. Section 2.2 has suggested explanations for this; Chapter 5 will
consider this in more detail.
(4) a. That’s for the part of Cambodia that I come, that I came across
on the internet. (i.C.1.m)
b. They never say the date, he didn’t state the duration, right.
(i.I.4.m)
c. That’s what him say to us just now. (ii.M.2.f )
d. My brother one went there and he took the budget airline and
then he come back with the normal airline. (ii.M.3.f )
e. Then we should have buy bikini. (iii.I.gr)
f. Desaru? When you say Desaru? (iii.M.3.m)
(5) a. If he have like a rubber plantation or something. (i.C.4.m)

b. Oh you do ah, ok, i.M.3.m have relatives there. (i.M.4.m)
c. Because he want to see how we all talk, normally. (ii.M.2.f )
d. It cover up everything in the floss. It’s not nice. It look very
cheap. (ii.I.1.m)
e. We can at least spend three days two night in a hotel which
only cost thousand plus. (iii.M.3.m)
H
1
L
0
1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321 341
Figure 4.1. Occurrences of (-ed) and (3sg) in ii.M.gr. The X axis shows
the number of the turn in which the variant is used, and the Y axis the
number of tokens in each turn.
In sum, verbal inflexions are an often-used feature, and show interesting

patterns of use. Particularly in the groups that had a high count of overt
inflexions in their recording, the feature can be analysed in a longitudinal
way. Figure 4.1 shows at which point in the interaction a given variant is
used: each turn is numbered and plotted on the X axis, and for each turn, the
number of H and L variants for (-ed) and (3sg) is given (either none, one, or
a maximum of two in the present case). This gives an idea of the distribution
of these variants, and the picture that emerges shows a true mix of H and
L variants, without an immediately obvious order to the distribution. This
seemingly haphazard use of variants, which will be further highlighted below,
calls into question the existence of two clearly identifiable diglossic varieties.
Auxiliaries
Features considered under this heading include inversion in interrogatives
and the use of modal auxiliaries. According to Gupta (1994: 12–13), inversion
in SgE is a feature of H, except when used with be and can, where it can be
either H or L. Similarly, the modals have and do are used in H only; other
modals (except can) are H too. This can be formalised as follows:
(6) a. Absence of inversion −→ H
−→ L
b. Presence of inversion −−−−→ H
iff ¬be/can
−−−−−→ L
Examples from the data abound, with, again, several cases where the H and
L variants are used in the same turn. In the examples reproduced below, there
is lack of inversion (7f–h) and absence of modals required in H (7a–c,e,i), as
well as missing do-support in negatives (7d). On the other hand, there is also
the presence of auxiliaries (7b,c,i) and of inversion (7e). There are, again,
interesting juxtapositions of H and L usages: consider (7c), which features
self-correction, as well as (7i), where deontic modality is marked (albeit in
H
i.C i.M i.I ii.C ii.M ii.I iii.C iii.M iii.I
Figure 4.2. Auxiliaries and inversion in all group recordings. Each

instance is marked by a tick; the H variants on the lower line and the L
variants on the upper line.
a pro-dropped construction) but the next verb (featuring the subject) lacks
the modal. This latter example is particularly interesting, since it shows the
close interplay between features of H and L: pro-drop, which is an L feature
(Gupta 1994: 10–11), occurs here in an otherwise H verb group, whereas in
the following L verb group, the pronoun is phonologically realised.1 This
poses problems for a diglossic analysis, a point which will be taken up in the
following section.
(7) a. Three cents leh, how to pay? (i.C.1.m)
b. So how about Genting? Never been there before, I’ve never
been there in my life. (i.C.4.m)
c. But budget airline, how much ah they cost, how much do they
cost (i.I.4.m)
d. Why not we chose our holiday back in Singapore? (ii.C.2.m)
e. So the ferry how much is this roughly, some of you been to
Bali actually? (ii.C.4.m)
f. How much it will be? (ii.M.2.f )
g. How long we are supposed to talk? (iii.M.1.f )
h. What the cruise is like? (iii.I.gr)
i. Cannot go Batam, must be a place where we never been before.
(ii.M.3.f )
A graphic representation of the distribution of H and L variants of aux-
iliaries and inversion is given in Figure 4.2. In this graph, the horizontal
dimension represents the internal temporal structure of each recording, with
turns as the unit. All nine group recordings are then concatenated together
for a complete picture. The two horizontal lines represent the two variants
of a combined variable (auxiliaries + inversions), with the H variants as the
base line and the L variant as the upper line. Thus, a picture emerges as to
the distribution of H and L variants.
What transpires from Figure 4.2 is, firstly, that the H variants (use of modal
auxiliaries other than be and can and presence of inversion in interrogatives)
predominate: there are 240 H variants and just 28 instances of L. Secondly,
1
Of course, pro-drop being a feature of L does not mean that its absence is diagnostic of H.
the distribution of L variants is unequal: ii.I.gr has none, i.M.gr and i.I.gr
have one each, whereas ii.M.gr has the most, seven. Thirdly, there is no
clear break in the use of H or L: many L variants occur in the same turn
as H variants, or are surrounded by turns with H variants. There are only
two major gaps in H usage (in ii.M.gr and iii.C.gr), during which L variants
occur intermittently, but this is not the default pattern. Such a distribution
is very unlikely to be indicative of a diglossic situation. As Chapter 5 will
show, an indexical approach has much more explanatory power.
The copula
The copula refers, here, to the verb be in its use as a ‘link or mark of
relationship between one element and another’ (Matthews 2007: 82). Its use
in the Singaporean context has been described at length in Ho and Platt
(1993) and Gupta (1994), to cite but two. It is the be used in the examples
below, and can link the subject with adjectives (8a), nominals (8b), an -ing
inflected verb (8c), and passives (8d).
(8) a. It might be perfect for the mind, but not perfect for the body.
(i.C.1.m)
b. That’s the point. (ii.C.4.m)
c. People will like really notice what you’re wearing. (ii.M.2.f )
d. It’s all being recorded, hello! (ii.I.2.f )
Its absence is often termed copula deletion, and is encountered in several
contact varieties (e.g. African American Vernacular English, see Labov (1969,
1982: 179), but also Singlish, as Ho and Platt (1993) make clear). Examples
of copula deletion from the data include those in (9).
(9) a. I heard that the hor fun ∅ quite famous. (i.C.1.m)
b. First we ∅ thinking of going to Malaysia. (i.M.3.m)
c. My uncle ∅ staying there. (ii.M.1.m)
d. What ∅ your dialect? Punjabi? (ii.I.1.m)
e. That boat ∅ very short one. (iii.C.2.f )
f. These ∅ like houses. (iii.I.gr)
As with the previous variables, the distribution of [−be] and [+be] varies
within each recording, although occurrence rates are lower. Of the 605
instances of the variable (be) in the group recordings, only 35, or 5.8 per
cent, were [−be].2 Again, too, there are (several) instances where [−be] and
[+be] co-occur in the same turn. Consider, for instance, example (10). Here
the speaker uses (be) twice, the first time with [−be] and the second time
with [+be]. Note that the subjects of the two clauses do not have the same
referent; that boat is not the antecedent of it.
2
The inclusion of it’s accounts for 224 of the 570 instances of [+be]. No instances of it+[−be]
were observed, presumably for articulatory phonetic reasons: often, it’s as a whole was
omitted.
94.2 % 89.6 % 79.2 %
BE Auxiliaries Inflexions
L H
Figure 4.3. Proportion of H and L variants in the three verbal variables.
The percentages given are those of H variants.
(10) That boati ∅ very shorti one, but itj ’s very longj . (iii.C.2.f )
Summary
The verbal features described in this section can be summarised as behaving
in the following way. Firstly, there is no sense of an agreement to use
exclusively H or L variants in a given turn or in a given succession of
turns. Secondly, the H variants are, in all cases, in the majority. Thirdly,
the motivation for the use of one or the other variant is, from the diglossic
perspective taken here, unclear. This final point will be addressed in the
following chapter; this summary merely lists the overall behaviour of the
variables.
A broad representation of the distribution of H and L variants in the group
recordings is given in the graphs in Figure 4.3. Here the predominance of the
H variants is illustrated graphically, with an indication as to which variables
are more likely to be used in their L form. Copula deletion is the least favoured
variant, whereas uninflected verbs are more likely (albeit at a modest 21.8 per
cent). These pie charts alone, however, do not explain the internal structure
of the interactions very well. Figure 4.4 is an attempt at a more comprehensive
analysis, where the Malay group from the polytechnic (ii.M.gr) will serve as
an example. In this chart, the horizontal axis is akin to a time axis, with each
unit being one turn (or one line-break in the transcription). In the top half
of the chart are the L variants and in the bottom half the H variants of four
separate variables: auxiliaries (including inversion), be, inflexions, particles,
and existential constructions. All but the particles are binary, with instances
in both H and L.
A few things can be said about this chart, which could be repeated for the
other eight group recordings. Firstly, and this has been noted several times
in this section, H variants co-occur with their L counterparts throughout the
recording. There is no sense of two clearly defined varieties, with their set of
variants used exclusively. In Gupta’s (1994) terms, there is active switching
1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321 341
Aux.
Cop.
L Infl.
Part.
GOT
Aux.
H Cop.
Infl.
GOT
1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321 341
Figure 4.4. Occurrences of H and L variants of the variables inflexions

(‘Infl.’), copula deletion (‘Cop.’), and use of auxiliaries and presence of
inversion (‘Aux.’) in the group recording ii.M.gr. Particles (‘Part.’) and
existential constructions (got) have been added for comparison. The
horizontal axis shows the number of the turn in which the variant is
used (Leimgruber 2012a: 58).
from H to L and vice versa. The ‘leaky’ (Gupta 2006b: 22) quality of the
SgE diglossia is here seen to be extreme. Secondly, while we cannot talk
about two distinct varieties, there do seem to be instances where the use of a
particular type of variant peaks: there are clusters of H variants and clusters
of L variants. Note, for instance, the rich number of L variants in turns
250–330, or the similarly prolific use of H variants in 250–320. The problem
with these is that they occur quasi-simultaneously, which, again, renders an
identification of the passage as either H or L difficult. Thirdly, and perhaps
less interestingly, L variants are vastly outnumbered by H variants. The
particles are the only L variable that is being used extensively. This analysis
unveils the need for a closer look at what actually triggers the use of what we
have called, so far, ‘H’ and ‘L’ variables. This is what Chapter 5 sets out to
do.
4.1.2 got
I briefly discuss here how a single ostensibly English word, in this case got,
has been reassigned a whole range of meanings in CSE. This serves two
purposes: firstly, to show how such a reassignment works, and secondly,
to show the indexical properties of such a lexical item. The example of
got has been largely described by Teo (1995) and Lee et al. (2009); similar
explorations have been undertaken for other such lexical items (e.g. until
(Bao and Wee 1998), one (Bao 2009), and already (Bao 1995)).
Firstly, got is used in CSE much as it is in StBrE, at least as far as the
following meanings are concerned: possessive (11a), passive (11b), to receive /
to obtain (11c), and to become (11d). There is little to add to this, but attention
is drawn to the use of an alternative, uninflected form get in (11c.ii), where
it marks present habituality, and in (11d.ii), where it seems to be marked for

the future.
(11) a. I got two brothers, one sister. (Lee et al. 2009: 295)
b. I got scolded by the teacher. (Lee et al. 2009: 299)
c. (i) I got flowers.
(ii) She get flowers every day. (Lee et al. 2009: 300)
d. (i) The white skirt got dirty.
(ii) Later your dress get dirty. (Lee et al. 2009: 300)
It is the use of got for existential marking that deserves more attention.
As Lee et al. (2009) note, Platt and Weber (1980) already noted this use,
reproduced in (12). It is often also used with a locative element. The element
got is a verb which ‘denote[s] the existence of a theme’ (Lee et al. 2009: 295).
The origin of this variant of got is clearly Chinese, as both Mandarin (13)
and Hokkien (14) have an existential verb that functions in a similar way to
CSE got.
(12) a. Got two pictures on the wall. (Lee et al. 2009: 295)
b. Do we look serious? No, already got this mike in front of us.
(ii.M.rm)
c. In Malay culture group there’s a lot of branches like that, er
dance, then got silat, which is martial arts. (ii.M.4.m)
(13) Yǒu yı̌ ge rén hěn xı̌huan nı́.
have one clf person very like you
‘There is a person who likes you very much.’ (Hu and Pan
2007: 133)
(14) Chit-tah ū chin chē lâng.
here got very many people
‘There are a lot of people here.’ (Teo 1995: 72)
In SgE, existential constructions with the use of got are markedly Singlish,
and correlate with lower levels of formality. Thus individual interviews
and dialogue recordings both showed more than 95 per cent of the
standard there+be construction, whereas in the unsupervised group and
radio-microphone recordings, there was an average 31.6 per cent of
got-usage.
Further, Lee et al. (2009) note that got can take on several meanings related
to realis modality: it can indicate temporal location, as in (15a–b), aspectual
meanings, as in (15d–f), and emphasis, as in (15g).
(15) a. I got go Japan.
‘I have been to Japan before.’ (past)
b. I got go Japan these days.
‘I go to Japan on a regular basis these days.’ (present)
c. I got go Japan next time.

‘I am going to Japan some time in the future.’ (indefinite
future)
d. You got play tennis last time?
‘Did you play tennis regularly?’ (habitual)
e. You got stay in Ang Mo Kio?
‘Have you ever lived in Ang Mo Kio?’ (experiential)
f. You got wash your hands?
‘Did you wash your hands just now?’ (completive)
g. A: You never sweep the floor ah?
‘You didn’t sweep the floor, did you?’
B: I got sweep!
‘I did sweep the floor.’ (emphatic)
(all examples from Lee et al. 2009: 296–8)
Finally, got occurs in the collocation where got, which is an idiomatic
expression typically signalling disagreement or challenging a proposition, as
in (16). Here too, the origin of the construction lies in Hokkien, which has
a similar construction toh-lĲoh ū ‘where got’, that works in exactly the same
way.
(16) A: This dress very red.
‘This dress is very red.’
B: Where got?
‘Is it? I don’t think so.’ (Lee et al. 2009: 299)
That where got can be used sociolinguistically to index a particular stance
is illustrated in (17), where a first paragraph complains, in entirely Standard
English, about the increase in price of some food items. The second paragraph
markedly breaks with Standard English, by introducing a sarcastic retort
‘Where got inflation?’, followed by a CSE sentence featuring copula-deletion
and a final particle (see Section 4.2.1). The sarcasm is that of mimicking
the government’s upbeat commentary on the economic situation, which is
clearly – in the mind of the writer – at odds with the daily experiences of
the average citizen. At the first level of indexicality, therefore, (17) indexes a
sarcastic stance, as well as, at a second order, a Singlish (localist) orientation.
A potential third-level (or ‘higher’-level) index would reference an unspoken
criticism of governmental communicative strategies.
(17) My bowl of bak chor mee3 at the coffee shop nearby recently
upped their prices from $2 to $2.20. Some have even upped it
to $2.50. And don’t even talk about aircon food court prices.
Inflation? Where got inflation? Singapore economy booming
what. (MrBrown.com 2007)
3
A Teochew dish consisting of noodles with minced pork (Lee 2004).
Table 4.1. Aspectual categories of Chinese, CSE, and

English (adapted from Table 38 in Bao 2005: 251)
Chinese [CSE] English
(a) Perfective
i. Completive V le S already V-ed, V-en
ii. Experiential V guo ever V ≈ ever V-en
iii. Emphatic yǒu V got V –
V wán finish V –
(b) Inchoative S le S already –
(c) Inceptive S le S already –
(d) Imperfective
i. Dynamic zài V V-ing V-ing
ii. Stative V zhe. . .(ne) ≈ V-ing ≈ V-ing
iii. Stative V zhe V ≈ V-ing ≈ V-ing
(e) Tentative V-V – –
4.1.3 Aspect marking

Aspect in Singapore English has been widely analysed in scientific articles
(Bao 1995, 2005; Bao and Wee 1998), whereas general textbooks usually give
a less in-depth account (Platt and Weber 1980; Ho and Platt 1993; Alsagoff
and Ho 1998; Deterding 2007a). Bao (2005) is particularly interesting, as it
analyses CSE’s aspect system as being the result of a system transfer from
the substrate Chinese languages present in the contact situation, ‘filtered’
through the lexifier’s surface forms.
Put simply, many aspectual meanings are encoded in CSE by the use of
English adverbs. They include an inchoative marker already, an experiential
marker ever, a completive already, and an emphatic completive finish. These
are exemplified in (18).
(18) a. We don’t know what to say already lah. (ii.M.dia.23, inchoa-
tive)
b. You ever [take photo]? (iii.M.gr, experiential)
c. I kind of grew out of it already leh. (i.C.rm, completive)
d. You stay in Singapore, four culture already, you haven’t
understand finish? (ii.I.dia.34, emphatic completive)
Bao (2005: table 38) provides us with a comparative summary of aspec-
tual classes for CSE, English, and Chinese,4 showing which Chinese as-
pect markers have been transferred and relexified in CSE (reproduced here
as Table 4.1). Thus completive/inchoative/inceptive le becomes already,
experiential guo turns into ever, emphatic yǒu and wǎn are transferred as
got and finish respectively. The imperfective zài V (dynamic) and V-zhe
4
Bao takes the Mandarin system to stand for Chinese in general. See footnote 2 on p. 3 for a
discussion on the uses of the term Chinese in this book.
(i) Ràng wǒ kàn-kan nèi-ben shuˉ.

(ii) Wǒ kàn-guo nèi-ge rén.
relexification
(i) *Let me read-read that book.

(ii) *I see-ever that man.
filter
(i) Let me read that book.

(ii) I ever see that man.
Figure 4.5. Lexifier filter. Example (i) is from Bao (2005: 24).
(stative) correspond to V-ing in CSE. The delimitative (called tentative by

Bao), which in Chinese is rendered by reduplication, is, according to Bao,
not transferred into CSE.
Bao’s analysis of system transfer is tempting. According to this theory,
the whole of the substrate’s grammatical subsystem of aspect is transferred
into the new variety. The lexifier then provides not only the new forms
for the transferred morphemes, but its grammar also acts as a ‘filter’: the
surface grammatical requirements of the lexifier language exert pressure on
this transfer and thereby condition the use and transfer of these ‘new’ mor-
phemes, a process illustrated in Figure 4.5. This results in a ‘morphosyntactic
exponence of the transferred system [which] conforms to the (surface) struc-
tural requirements of the lexical-source language’5 (Bao 2005: 257). Thus
for example ‘stative imperfective’ and ‘tentative’ are not transferred into
CSE, since it would not meet the requirements of English grammar (verbal
reduplication being absent in English). This analysis results in a fairly clear
correspondence between the substratal and the CSE aspect systems.
Problems with Bao’s take on the CSE aspect system exist, however. For
one, there is disagreement on the absence of productive verbal reduplication:
Wee (2004: 105–17) provides plenty of data suggesting otherwise, with verbal
reduplication indicating continuity or attenuation. Examples from my own
data show continuity (19) (a case of triplication) and attenuation (20), too.
(19) A regular day? Erm get up in the morning, come to school,
study study study, head home, wash up, and go to work. Yeah,
that’s my regular day. (ii.I.3.f )
(20) No that means she talks is like an old lady, like that, read read
and talk. (iii.I.rm)
5
Bao further observes that his Lexifier Filter can be seen in other properties of contact
languages, such as word order, which tends to be dictated by the lexifier (2005: 258).
Furthermore, aspect markers are an uneasy variable to work with: their

occurrence rates are typically relatively low, which makes an informed, data-
driven analysis difficult. Some descriptions of the CSE aspect system have
taken an introspective approach (one where the researchers rely on their own
or their informants’ intuitions about the variety). While this has resulted in
worthwhile theories, it is hard to reconcile with the fact that the various vast
corpora on CSE have had little to say on these features.
4.2 Semantics and pragmatics

4.2.1 Discourse particles
So-called discourse particles are used in all languages, and come in various
forms: English has an elaborate system of question tags (isn’t it, can we), hesi-
tation markers (erm, y’know), and many others such as markers of mitigation
(I mean, you see). Stereotypically, Canadian English has a sentence-final eh
(Ferguson and Ferguson 2003: 65–8), and Swiss varieties of French an equiv-
alent ou bien (Poirier 2008). They share a number of features, first among
others that of not being essential to the immediate semantic content of the
utterance: the truth value of the utterance is not influenced by the presence
or absence of these items. They are syntactically optional. More often than
not, these particles are also associated, by the speech community in general,
with the variety in which they are used: like is (or used to be, at least) a shib-
boleth of teenage speech6 (Siegel 2002), ou bien is jocularly commented on by
French speakers as being Swiss and, consequently, provincial, and high rates
of question tag usage (particularly with first-person referent) are typical of
Estuary English (Crystal 1995: 327). This element of public awareness also
plays an important role in the case of particles in SgE.
Terminology and definition

When discussing what I call here discourse particles, a number of different
terms are used in the literature, sometimes even by the same author. These
terms, discourse markers, discourse particles, pragmatic markers, pragmatic parti-
cles, are all used relatively interchangeably (Wouk 1999; Beeching 2002). The
ambiguity is also observed in studies of SgE (Lim 2007: 446): Ho and Platt
(1993), Ler (2006), and Gupta (2006a) use discourse particles, whereas Gupta
(1992) and Low and Brown (2005) use the term pragmatic particles. In all
cases it is clear what is being discussed, but these terminological differences
need to be addressed before proceeding to a more precise definition.
The reason for this variety of terminology is that it does not automati-
cally follow that the interchangeable use of particles and markers, pragmatic
and discourse, enjoys universal approval. Aijmer and Simon-Vandenbergen
6
There are different uses for like; the one referred to here is the quotative in ‘She was like
“Get out of here”’ (Siegel 2002: 37).
pragmatic functions
items fulfilling discourse functions
lexicalised items
discours mar
discourse markers
ers
discourse
particles
particles
Figure 4.6. Discourse particles and discourse markers (adapted from

Fischer 2006b: 7)
(2006: 2–3), for instance, use pragmatic marker as a ‘broad term’, encom-
passing discourse marker which is reserved for coherence relations, but also
covering markers of other dimensions of communication beside discourse,
such as any part of an utterance that ‘does not contribute to [its] propositional,
truth-functional content’ (Aijmer and Simon-Vandenbergen 2006: 2). The
term pragmatic itself conveys the idea of a wider field than simply discourse.
The other terminological point of contention is that of particle versus
marker. Fischer (2006b: 4) says that particle ‘suggests a focus on small, un-
inflected words that are only loosely integrated into the sentence structure,
if at all’. She further establishes that particle has the advantage of distin-
guishing between these items and larger strings, such as ‘phrasal idioms,
that fulfil similar functions’ (2006b: 4). Figure 4.6 is from Fischer (2006b: 7)
and shows the extent to which particles and markers overlap. The diagram
combines two approaches: a functional one (pragmatic vs discourse) and a
formal one (lexicalised vs non-lexicalised items, with particles a subset of the
former). In the words of Fischer, ‘discourse markers may be both lexicalised,
including particles, and non-lexicalised items that fulfil discourse functions’
(2006b: 6).
Discourse particles, in this analysis, are, therefore, lexicalised items that
fulfil discourse functions, and which have the additional functional prop-
erty of being particles, while discourse markers is the term used for any item
fulfilling discourse functions, lexicalised or otherwise (thus including parti-
cles). Although not explicitly stated in Figure 4.6, pragmatic markers, in this
view, would comprise all items fulfilling pragmatic functions (including all
Table 4.2. Definitions of the various terms of pragmatic/discourse markers/particles

(based on Aijmer and Simon-Vandenbergen 2006; Fischer 2006b)
Pragmatic markers Discourse markers
Any item, lexicalised or otherwise, that Any pragmatic marker, lexicalised or otherwise,
fulfils pragmatic functions that fulfils discourse functions
Examples: Discourse makers but also Examples: Discourse particles but also Eng. you
adverbial connectors, reformulations, know, question tags, Ger. nicht wahr, Fr. pour
hesitations, restarts (Fischer 2006b: 5) ainsi dire, tu vois comment
Pragmatic particles Discourse particles
Small, uninflected lexicalised items that Small, uninflected lexicalised items that fulfil
fulfil pragmatic functions discourse functions
Examples: Discourse particles but also Examples: Eng. well, like, Ger. eben, ja,
Eng. like, Fr. enfin (Beeching 2001) Fr. donc, hein, It. cioè
discourse markers), whereas pragmatic particles would be all particles except

those not fulfilling pragmatic functions. Table 4.2 attempts to present these
four terms in a more orderly manner.
While Table 4.2 brings some sort of order into what seemed a terminologi-
cal muddle, it does not eliminate confusion completely: ‘pragmatic functions’
and ‘discourse functions’, for instance, are not explained. However, the four-
way classification in Figure 4.6 takes into account that discourse particles are
a subclass of both pragmatic particles and of discourse markers, and discourse
markers are a subclass of pragmatic markers. In what follows, I shall use the
term discourse particles, which is in line with the practice that the majority of
scholars on the variety (e.g. Ho and Platt 1993; Wee 2004; Gupta 2006a; Ler
2006) have established. It seems to be the term most suited in light of their
discourse function and their syntactic properties, and also follows the defi-
nition given in Table 4.2: they are lexicalised, monosyllabic items fulfilling
discourse functions.
Discourse particles in Singapore English

Discourse particles are a stereotypical feature of Colloquial Singapore En-
glish. The category covers a range of monosyllabic items used to convey the
speakers’ attitude or mood, to establish solidarity, or to emphasise the obvi-
ousness of a statement (Gupta 1992), among other functions. The literature
on discourse particles is extensive, as will become apparent. What follows is
a brief overview of the most common particles.
A useful way of looking at CSE particles is given in Gupta (1992), where
she analyses them as a system. The particles are displayed on a ‘scale
of assertiveness’ (1992: 37), hierarchically progressing from least assertive
mah
+assertive
contradictory
what
meh
geh
leh
assertive
na
lah directive
lor
hor
−assertive
hah tentative
ah
Figure 4.7. Gupta’s scale of assertiveness (1992: 37)
(‘tentative’) to most assertive (‘contradictory’). Her scale is reproduced here

in Figure 4.7: ah, hah, and hor are tentative particles, lor, lah, na, leh, geh, and
meh assertive. Together, they form the group of ‘directive’ particles. Finally,
what and mah are contradictory particles.
This latter group of contradictory particles are used to ‘explicitly contra-
dict something a prior speaker has said’ (Gupta 1992: 37), while the assertive
particles ‘express speakers’ positive commitment to what they are saying’,
and the tentative ones ‘offer a less positive commitment’. While this approach
has the drawback of using relatively undefined metalanguage, having a scale
of only four terms makes it a workable system. Similarly, the hierarchical
placement on a single scale of assertiveness is enlightening in that it puts for-
ward an inherent relationship between the particles themselves: rather than
being randomly borrowed from substrate languages, they seem to fulfil, in
the view proposed here, the purpose of a complete set of discourse particles.
This would link Gupta’s (1992) scale with Bao’s (2005) system transfer (in
his case with regard to the aspect system), but also to Lim (2007: 463), who
sees in at least some of the particles (see p. 90) the ‘wholesale importing [of]
an entire functional category’. Gupta however does not make any such claim,
and limits herself to providing a framework within which the particles can be
analysed as part of a coherent system. It should be noted here that she is also
the only one to mention the particle geh (Jyutping ge2), which is classed, to-
gether with leh, as ‘maximally assertive’ (1992: 42). This typically Cantonese
particle does not seem to enjoy wide currency among CSE speakers, and it
does not seem to occur in other corpora of SgE.
The hierarchical arrangement of Figure 4.7 does not mean that there
is no overlap between particles. Lah, for instance, ‘covers the full range
within the assertive continuum (Gupta 1992: 42). Similarly, mah and what
are not distinguished functionally. This analysis is similar to those of others
(Wong 2004; Ler 2006; Lim 2007) who regard particularly the particle lah as
fulfilling different semantic and pragmatic functions (see Table 4.3). While
Lim considers various tone patterns to account for these different functions,
she acknowledges (2007: 463) that tonal distinctions may be lost over time
‘as they become more fully assimilated into’ SgE. This will ultimately result
in the picture presented by Ler (2006) and Wong (2004), where lah is simply
treated as a single particle taking on various functions.7 With regard to mah
and what, however, Gupta (1992) fails to make the distinction that other
authors do (Wee 2004; Lim 2007), who tend to note an additional dimension
to what: while mah simply marks information as obvious, what is said to
contain an element of contradiction in addition to marking obviousness.
Lah
This is the most stereotypical particle, with a high level of awareness on the
part of speakers. The following metalinguistic use illustrates this point: when
asked by an acquaintance of my host family what my research was on, my
answer ‘Singlish’ prompted the simple reply ‘lah!’. It is extremely common
in everyday speech, used primarily sentence-finally. Classified as ‘assertive’
by Gupta (1992: 37), it ‘indicates [the] speaker’s mood/attitude and appeals
to [the] addressee to accommodate [to that mood]’ (Wee 2004: 125). It is
pronounced [lA] and usually spelt lah, although some older scholarly works,
including Gupta (1992), use la. Its origins are disputed, and Lim (2007: 464)
mentions Malay, Hokkien, or Cantonese as equally likely sources.
Ah
Used more frequently than lah, ah is, however, much less overtly commented
on in public discourse on ‘Singlish’. Scholarly accounts, of course, abound
(Gupta 1994; Low and Brown 2005), and it is used for diglossic/stylistic
purposes. While Ho and Platt (1993: 10) call it a ‘Y/N question particle’,
Gupta (1992: 37) calls it a ‘tentative’ particle and situates it at the bottom end
7
Even though usually considered void of tone, which is not phonemic in SgE, the final particles
are clearly subject to overall prosodic patterns, and can be seen as continuing the utterance
intonation.
of her ‘scale of assertiveness’. Ah also serves as an important prompt when

confirming a question, as exemplified in (21).
(21) R: What did you do?
3: What did I do ah?
R: Yeah.
3: Basically, I watch TV. (iii.I.3.f )
The same uncertainty as for lah exists with respect to ah’s etymology: any one
of the three languages Malay, Hokkien, or Cantonese is a possible candidate
(Lim 2007: 464).
Leh
Situated in the middle field of Gupta’s (1992) ‘scale of assertiveness’, leh
‘marks a tentative suggestion or request’ (Wee 2004: 125). It can, however,
also be used non-pragmatically, as an anaphoric reference in ‘x-interrogatives
without wh-words’ (Gupta 1992: 36), thus taking on the function of StBE
‘what about’. In example (22) below, the group of informants is attempting
to name a type of boat used at their holiday destination. Unconvinced by the
previous guess, iii.C.2.f makes a tentative suggestion, using leh.
The spelling with h presumably indicates a long vowel (an unnecessary
indication, vowel length being non-phonemic, see Section 3.1.1) – alternative
renderings include lei (Gupta 1992), le, and sometimes ler, possibly by
analogy with lor. Lim (2007: 463) stipulates a Cantonese origin for leh, where
a similar le558 exists that ‘indicates “what about?”’ (2007: 461).
(22) 3: Bamboo boat.
2: That’s not bamboo boat leh, but very small.
4: Sampang. (iii.C.gr)
Meh
Meh indicates scepticism (Wee 2004: 121), and is highest in Gupta’s ‘as-
sertive’ category (1992: 37), just before ‘contradictory’ particles. Like most
other particles, it occurs clause-finally, and more often than not, utterance-
finally. A spelling variant is me, and, if irony is intended, mare (the
pronunciation of all of these, of course, is [mE], see Section 3.1.1). The
source of this particle, according to Lim (2007: 463), is the Cantonese parti-
cle me55, used to indicate surprise in a question and as a marker of rhetorical
questions. It is unrelated to the American English meh described in Zimmer
(2006).
8
Non-Jyutping tone marking scheme: 5 = high tone, 1 = low tone; thus 35 = high rising,
55 = high level. See Chao (1930).
Lor
At the other end of Gupta’s ‘assertive’ category, we find lor, which ‘indicates
obviousness or a sense of resignation’ (Wee 2004: 125), see example (23)
below. Alternative spellings are lo and loh. Here two substrate particles,
both from Cantonese, may have triggered its emergence in CSE: Cantonese
lo3 ‘indicates obviousness, inevitability and irrevocability’, while lo1 ‘points
out what appears to be obvious’ (Lim 2007: 461). While the former seems
to more closely match the function of the CSE particle, the existence of the
latter in the substrate may have helped the emergence of the new CSE lor.
(23) Because she wants to sing mah. So she want to use, she want to
join to sing, so we just groom her lor. (ii.C.4.m in ii.C.rm)
Hor
According to Wee (2004: 125), hor ‘asserts and elicits support for a proposi-
tion’, nicely exemplified by (24) below. Gupta (1992: 37) places it among her
‘tentative’ particles. Again, Lim (2007: 463) argues for an origin in Cantonese.
(24) Then there’s another issue, if you marry later, er, woman tend to
have more complicated pregnancies hor. (iii.I.3.f in iii.I.dia.23)
Mah
Unrelated to the Mandarin question particle ma (嗎), mah is at the top of
Gupta’s (1992: 37) ‘scale of assertiveness’, one of two ‘contradictory’ parti-
cles. Wee, who uses the spelling ma, notes that it ‘indicates information
as obvious’ (2004: 125), which is certainly what mah achieves in (23) above.
Its likely source is Cantonese (Lim 2007: 462), where a homophone particle
fulfils the same function.
Hah
Another ‘tentative’ particle (Gupta 1992: 37), hah (sometimes spelt without
final h) is described by Wee (2004: 125) as a ‘question particle’. It can occur
as a stand-alone utterance in its own right, where it becomes a request for
clarification, or an invitation to repeat a previous utterance.
What
The second of Gupta’s ‘contradictory’ particles (1992: 37), what, has nothing
to do with its English homograph, neither etymologically, nor functionally.
To underline this fact, many authors commonly use a variety of alterna-
tive spellings, including wot, wut, wo, and wat (my choice of using
what herein is motivated by it being native speakers’ spelling of choice).
Pragmatically, it contradicts a preceding utterance, and, to paraphrase Wee
(2004: 125), implies obviousness. Here Lim (2007: 464) does not identify a
single substrate particle as the likely source, but postulates a calque on a
Chinese particle ma which occurs in Cantonese and Hokkien as well as in

Mandarin.
(25) 4: Wait wait wait. No money, wanna go everywhere.
3: No, thousand two convert to Thai Baht is quite a lot what,
in a way, really. . .
1: Yeah, is about millions. (ii.I.gr)
Example (25) shows this process at work in a conversation. The group had
been listing activities which they would enjoy doing during their holiday
in Thailand, when ii.I.4.m interrupts and reminds them of the existence
of a budget. The retort by ii.I.3.f contradicts this by explaining that the
exchange rate will multiply their budget, thereby obviously expanding their
possibilities. Factual inaccuracies aside (purchasing power ought to be taken
into consideration too), the statement implies obviousness. The next turn,
by ii.I.1.f, reinforces this with agreement and further (slightly exaggerated)
information.
Etymology
Various approaches have been taken to analyse the grammatical class of what
I termed earlier discourse particles. Thus Lim (2007) analyses their structural
and sociolinguistic etymology, while Ler (2006) takes a relevance-theoretic
approach. Other approaches work within the frameworks of semantics and
pragmatics (Gupta 1992; Wee 2004), Natural Semantic Metalanguage (Wong
1994), epistemic modality (Gupta 2006a), epistemic/deontic authority and
illocutionary force (Kim and Wee 2009), language acquisition (Kwan-Terry
1991), or even cultural studies (Wong 2004). More general introductory
works (such as Ho and Platt 1993; Wee 1998; Low and Brown 2005) usually
limit themselves to a list of particles with their general semantic/pragmatic
scope.
The discourse particles investigated here are of a special kind. While the
English ones referred to above (well, like, etc.) are also part of the SgE sys-
tem, they are not the ones that authors usually refer to when discussing
SgE. (Even though at least Gupta (1992: 38) explicitly acknowledges their
presence, they are often overlooked.) Those particles that are of interest
here are those that have their origin in a language other than English (for
a list of particles, see Table 4.3) – for this reason they are treated here, as
elsewhere, as a separate category. Gupta explains their different theoretical
status with the fact that the first group (English-derived particles) exhibits
different usage patterns across sociolects: these particles are not markers of
either SSE or CSE, unlike the second group (substrate-derived particles).
‘Thus,’ she says, ‘varietal distinction (as well as historical origin) justifies their
separation’ (Gupta 1992: 38). Indeed, discourse particles such as the stereo-
typical lah (but also lor, leh, ma, etc., see Table 4.3) appear very sparingly
indeed in SSE, and they are very obvious and widely recognised markers of
Singlish.
The origins of the particles lie undoubtedly in the substrate languages
involved in the emergence of CSE (primarily southern varieties of Chinese).
While this view is now generally accepted (Lim 2007), some sources put
forward alternative origins for individual particles – sometimes, as Gupta
puts it (1992, 1994), those varieties the researcher happens to know best (lah
from Malay (Platt and Weber 1980), ah from Tamil (Baskaran 1988), among
others). The problem is, of course, that more than one substrate language fea-
tures particles similar in form and function. In an attempt at a comprehensive
account of the particles’ history, Lim (2007: 460–1) gives a thorough overview
of their counterparts in substrate languages, noting similarities in form (such
as CSE lah – Bazaar Malay la – Mandarin la – Cantonese laa3 /laa1 , or CSE
lor – Hokkien lô – Cantonese lo4 /lo1 – Mandarin luo, etc.) and in pragmatic
function (such as ‘softens command’, ‘slightly exclamatory’, ‘indicates em-
phasis’, ‘marks a question with expectation of agreement’, etc.). She further
distinguishes segmental from suprasegmental form (tone). Her conclusion is
that while at a superficial level, many similarities can be observed between
substrate particles and CSE ones (Cantonese, in particular, appears to have
a counterpart for all her eight CSE particles, but Bazaar Malay only two),
at the suprasegmental level things are less straightforward: while the CSE
particles carry tone (or are, at least, stressed), this is not the case for either
Hokkien or Mandarin. She illustrates this with the example of lor, which
has in Cantonese lo3 a perfect counterpart in both tone and function (Lim
2007: 463). The same can be said for hor, leh, meh, and mah. As far as the
stereotypical lah is concerned, however, her analysis remains noncommittal
and she acknowledges that Malay, Hokkien, and Cantonese are all ‘likely
sources’ (Lim 2007: 464) – which might have played an important role in
reinforcing the emerging CSE lah. Lim’s is a much more careful approach
than that taken by Kwan-Terry, who notes that ‘a substantial proportion of
the particles [used by her informant] in speaking Cantonese were also used
when he spoke English, and for similar functions[, which] points clearly to
the Chinese origin of these particles’ (Kwan-Terry 1991: 175). This claim is
then somewhat put into perspective when she later concedes that ‘one cannot
conclude from this that [he] must have transferred these particles from the
Cantonese he spoke to his English’ (Kwan-Terry 1991: 180). Nonetheless,
she reaches the conclusion that her informant’s order of acquisition of, in
particular, the particle hor ‘definitely points to transfer from Cantonese to
English’ (1991: 180).
The least obvious CSE particle is what, which, despite its phonetic and
orthographic form,9 is clearly unrelated to StBE what (even in its archaic
9
See p. 88 for alternative spellings, including the more phonemic wot.
form as an interjection, as in ‘Nice day, what?’,10 Gupta 1992: 42), but has
a correspondence in form in Cantonese wo4 – the problem here is that the
functions only partially overlap: in Cantonese it signals a noteworthy discov-
ery, and in CSE, contradiction of a preceding statement (Gupta 1992: 42;
Lim 2007: 464). A more satisfactory solution, Lim argues, would be to regard
it as a calque on ma as used cross-linguistically in Chinese varieties, where
it ‘gives emphasis as well as reflects an enhanced degree of disapproval or
even annoyance on the part of the speaker’ (2007: 464), thus having the same
pragmatic qualities as CSE what. Nonetheless, the fact that other particles
have a stipulated Cantonese origin, and that there exists a particle wo4 in Can-
tonese, seems too much of a coincidence to ignore when looking at possible
etymologies for the CSE/Singlish particle what.
Semantics and pragmatics

A closer look at the semantic and pragmatic definitions of these particles is
now warranted. The wealth of research published on the topic (see, inter
alia, Platt 1987; Gupta 1992, 2006a; Wong 1994, 2004; Wee 2004; Ler 2006;
Lim 2007) makes for difficult choices as to which is the most appropriate.
Despite Wong’s valid objection about the less than ideal use of ‘obscure meta-
language’ (2004: 749), Table 4.3 nonetheless tries to summarise the major
definitions given by a selection of authors (without necessarily endorsing
them). There is of course a lot of overlap, but many authors have tried to
find new ways in which to define them: while Gupta (1992) and Wee (2004)
follow the traditional practice of using adjectives such as assertive, obvious, or
tentative, Wong rejects this as imprecise and calls these definitions ‘obscure
metalinguistic statements’ (2004: 749), and provides definitions in Natural
Semantic Metalanguage (see below), which, he claims, explains the actual
semantic dimension of each particle more accurately. This is relevant in that
it departs from seemingly subjective formulations such as ‘more persuasive’
(Lim 2007: 460) and ‘solidarity’ (Wee 2004: 125), or clumsy ones such as ‘meh
signals to the hearer that an assumption recently manifest in the external
environment challenges an existing one in the cognitive environment of the
speaker’ (Ler 2006: 164).
The overview given in Table 4.3 also highlights the vast differences in
scope of the various authors describing CSE particles. For example, Gupta
(1992) lists eleven particles, Lim (2007) seven, and Ler (2006) only three.
Not all of these studies purport to be exhaustive, and neither does the list
herein: while Gupta’s was primarily concerned with proposing a systemic
analysis of all particles, Lim’s is a much more diachronic account focussing
on a smaller set. Ler, too, restricts herself to three particles, but subjects
them to extensive analysis.
10
In sense B.I.3. of the Oxford English Dictionary (2nd edition 1989).
Table 4.3. Definitions of CSE particles by various authors. Particles are listed alphabetically based on the spellings used as headers in this section,
the spellings within the table follow those used by the respective author, and shaded areas indicate particles not addressed by the author in question. The
definitions from Lim (2007: 460–1) are verbatim
Part. Platt (1987) Kwan-Terry (1991) Gupta (1992) Wee (2004) Ler (2006) Lim (2007)
ah a24 marks a question a21 in interrogatives: (1) punctuates an a24 signals

where the speaker is seeks confirmation utterance, occurs continuation (in
uncertain about the (echo-questions) mid-clause or between narratives or
answer main and dependent explanations) and keeps
clause in complex interlocutors in contact;
a22 marks a question narratives softens command;
with a definite marks a question
answer/echo (2) tentative, checking expecting agreement
question truth value of a
statement a21 marks a question
requiring response
geh assertive; expresses a
commitment that an
interlocutor is expected
to act upon
hah ha24 in declaratives/ question marker (1) marks wh or

imperatives: emphatic declarative
marker question
(2) asks for

repetition of
previous
utterance
hor hõ24 marks a hoh24 in asserts and elicits sup- assertive; marks a question
question where the interrogatives: solicits port for a proposition garners support asserting a
speaker is uncertain agreement or support proposition and
about the answer trying to garner
support for the
proposition
leh le22 indicates le33 in declara- marks a tentative tentative; marks a question
disagreement with a tives/imperatives: suggestion or request softens a involving comparison
suggestion; conveys emphatic marker request or a
information command
assumed to be new le55 in interrogatives:
turns an NP or a VP
le55 equivalent to into a question
‘what about’
lah la24 when the la33 in declara- assertive; indicates the solidarity Any one of the draws attention to
matter discussed is tives/imperatives: speaker’s marker following: mood or attitude and
known to the persuasive, ‘convey[s] mood/attitude and appeals for
addressee an attitude of appeals to addressee to (1) solidarity accommodation;
acquiescence or accommodate (2) emphasis indicates solidarity,
la22 friendlier than concession’ (3) obviousness familiarity, informality
la24, indicates (1991: 176) (4) persuasion
information (5) friendliness la24 is more persuasive
assumed to be new; (6) hostility
of which the speaker la21 is more
is certain matter-of-fact
lor lo22 used to ‘put loh55 in declara- indicates obviousness indicates indicates a sense of
down’ the addressee tives/imperatives: or a sense of resignation obviousness or a obviousness as well as
indicates obviousness, sense of resignation
lo55 indicates concession resignation
obviousness
(cont.)
Table 4.3. (cont.)
Part. Platt (1987) Kwan-Terry (1991) Gupta (1992) Wee (2004) Ler (2006) Lim (2007)
mah ma22 ma33 in declaratives/ indicates information as indicates indicates obviousness

interchangeable with imperatives: obvious obviousness
what; ‘speech group obviousness
marker’ (basilectal/
Chinese)
what what22 objection to indicates information as indicates indicates that

an action or point of obvious obviousness and information is obvious,
view of the contradiction contradicting
addressee something previously
asserted
meh indicates scepticism indicates Various functions: marks a question
scepticism (1) questions a involving scepticism
presupposition
(2) expresses
surprise
(3) means the
opposite of what was
thought to be true
na used for rebukes

It may be worth briefly including here an explanation of the particle lah’s

meaning as described by Wong (2004). Wong uses Natural Semantic Meta-
language (NSM) to explain the particles’ semantics. NSM, first developed
by Wierzbicka (1972), is a minimalist metalanguage using a limited (but
growing as the theory evolves) number of semantic primitives (or semantic
primes) used for semantic analysis. These primitives, currently numbering
around sixty and thought to be universal, constitute an ‘irreducible core’
which can be used to ‘understand all complex thoughts and utterances’ in a
given language (Wierzbicka 1998: 113). They include quantifiers (one, some,
all), substantives (I, you, something), speech (say, words), mental predicates
(think, want, know), logical connectors (not, if, because), and many more
(see e.g. Goddard 2002). The way in which the primitives are combined
follows precise rules, thus giving rise to something ‘like a natural language in
miniature’ (Thornsteinsson 2008). Wong’s definitions of three types of lah
illustrate the workings of this approach more succinctly:
(26) a. Impositional lah21:
1. I think something now
2. I think you don’t think like this
3. I don’t want this
4. I want you to think like this
5. I think if I say something, you can think like this
6. I say it now
b. Propositional lah51:
2. I think you don’t think like this
3. I think you could have thought like this before
4. I think it will be good if you think like this
6. I say it now
c. Persuasive lah35:
2. I think you don’t think it is good to think like this
3. I don’t want this
4. I think it will be good if you think like this
6. I say it now
(Wong 2004: 766, 768–9, 772)
The advantage of semantic explications such as these is that they explain,
step by step, the reasoning behind the use of the particle. This is more precise,
arguably, than the ‘obscure metalanguage’ (Wong 2004: 749) employed by
the other authors quoted above: rather than giving ill-defined and culture-
specific definitions such as ‘tentative’, ‘contradictory’, or ‘obvious’, an NSM
explanation rests on terms that are thought to be universal. This cultural

independence is a major argument in Wong’s preference for NSM in the case
of CSE particles: the terms just mentioned (‘tentative’, etc.) are culturally
defined, he argues, and, as such, ill-suited for a description of a CSE particle
by scholars from non-Singaporean cultures (Wong 2004: 751–4). Whether
or not the convoluted definitions in (26) help to understand the semantics of
Singlish particles is, of course, debatable. Certainly the charge that ‘obscure’
non-NSM definitions are culturally loaded is acceptable only to a certain
extent: linguists both from Singapore and from elsewhere (i.e. belonging
to different ‘cultures’) have used them, usually with at least superficially
comparable meaning.
Summary
To sum up, the discourse particles, regardless of their exact semantic and
pragmatic meaning, are usually restricted to Singlish, of which they are
stereotypical. Especially the particle lah has become something of an icon of
Singlish and of Singaporean-ness. Throughout her diglossic analysis of SgE,
Gupta (1989, 1994, 2001) classifies the particles as good diagnostic markers
of CSE as opposed to SSE. It is certainly true that particles are absent in the
standard speech heard, for instance, on television or in parliament. Nonethe-
less, even in settings of this kind, it would not be completely out of place
to use a particle, if the context and the intended use fit: a jocular reference
to Singlish or a politician’s attempt to bring himself closer to his electorate
(to create social capital, in Alsagoff’s (2007) terms) are very plausible
indeed. Singlish discourse particles are excellent indexes of a range of social
meanings, and can be used as such even in otherwise standard discourse.
4.2.2 Lexical semantics

Whereas Section 3.2 focussed in particular on the etymology of SgE lexical
items that had their origin in languages other than English, this short section
will take a closer look at the differences in the meanings covered by lexical
items common to both SgE and other Standard Englishes. The standard
used for comparison here is Standard British English, although American
English will be considered if and when relevant.
Innovations
There are words, particularly compounds, that are English in form, but
created locally, usually with highly localised meanings. Examples include
White Horse (usually capitalised), designating the sons of politicians or other
influential persons, especially during National Service (Lau 2004), perhaps
originating from the white clothes worn by members of the ruling party.
Another example is hawker centre, which refers to a purpose-built covered
Table 4.4. A selection of SgE initialisms
Initialism Definition Notes
CTE Central Expressway

CPF Central Provident Fund A compulsory savings scheme
HDB Housing Development Board Refers to the government agency as
well as to individual flats
JC junior college Can be combined with the college’s
name, e.g. NJC for National Junior
College
KJE Kranji Expressway
MOE Ministry of Education
MRT Mass Rapid Transit The metro system
NIE National Institute of Education
NS National Service Can be suffixed to give NS-man
‘conscript’
NTU Nanyang Technological University
NUS National University of Singapore
PIE Pan-Island Expressway
PSLE Primary School Leaving Examination
PUB Public Utilities Board
ROM Registry of Marriages Government agency as well as the
actual (civil) wedding
SAF Singapore Armed Forces
SMU Singapore Management University
area harbouring the stalls of street hawkers (who have thus become seden-
tary); they feature sitting areas and public amenities. Hawker centres can
hold dozens of stalls serving a variety of local, regional, and international
food. Another term is void deck, which designates ‘the empty space kept for
social functions on the ground floor of most apartment blocks’ (Deterding
2007a: 76).
Under this heading we might want to include the wealth of initialisms
found in Singapore, and described in some detail by Deterding (2007a: 77–
9). While not true innovations as such, the concatenation of the initial (often
three) letters of the individual words of a phrase is, in the cases listed in
Table 4.4, restricted to Singapore, unlike, for instance, CBD ‘central busi-
ness district’, which is used in Singapore as well as in most other cities
where English is used. As noted by Deterding (2007a: 78), the existence of
initialisms such as the ones listed in the table is nothing remarkable per se,
but their use in common everyday language makes them worthy of inclusion
here. It is rare, for instance, to hear American students say they attend HS
(for High School), or for a British student to say that she studies at CCC
(for Corpus Christi College), whereas their peers in Singapore would rarely
choose to say Junior College or Anglo-Chinese School, opting for JC and ACS
instead.
Semantic change
There are, in SgE, several lexical items whose use differs from that in other
English-speaking locales. Some were mentioned in Section 3.2.1, like to send
‘to give a lift’, to follow ‘to take a lift’, to keep ‘to put away’, to renovate ‘to
redecorate’, and to upgrade ‘to renovate’. To this list Deterding (2007a: 80–2)
adds to bring in example (27), where other varieties would have to take, and to
fetch, used to ‘describe the action of taking someone to a particular location’,
as in (28). He also mentions the use of to stay to denote a place of permanent
residence (to live in SBrE), and Low and Brown (2005: 81–2, 86) mention
spoilt ‘damaged (of machinery)’, to hack ‘to remove carefully (tiles, plaster,
etc.)’, and to bathe ‘to have a shower’.
(27) Well if you wanna go overseas, know like, I mean you can actually
bring your family along also, right. (i.I.4.m in i.I.dia.34)
(28) We only have two cars, for the entire extended family, so I don’t
think it will be, it will be wise to carry, to fetch everyone by car.
(i.C.1.m)
Formal/archaic words
It is a point often made (Brown 1999; Deterding 2007a) that some words,
which have lost some currency in other English-speaking countries, or are
used there only in formal speech, are commonly used in SgE. Deterding
mentions sibling, alight (from a bus or train), and thrice, all of which he
quotes from Brown (1999), but exemplifies with occurrences from his own
data. Example (29) shows that at least sibling is still very much in use:
(29) a. I have two little siblings at home. (ii.C.3.f )
b. Because there’s one vacation where I went out with my whole
family, with my cousins and uncles, and my father has nine
siblings, so it’s really a lot. (i.M.2.f )
c. The age difference between me and my siblings are quite
big, so we don’t really go for holidays together. (i.C.4.m)
Summary
The extent to which the lexical items listed here are used throughout the SgE
speech community differs. The terms here differ from the ones of Hokkien
or Malay origin, discussed in the previous chapter – because of their form
and recognisable non-English origin, they tend to be more easily associated
with Singlish and with code-switching in general. It is less clear whether
the purely English terms here (such as bath ‘shower’) are used with stylistic
variation; certainly bring in example (27) and fetch in (28) are used by educated
youngsters in settings that can be considered formal. With this in mind, an
indexical approach to their use would simply regard them as indexes of a
Singaporean identity – but one that is rarely consciously used in order to
4.3 Conclusion 99
index this identity, perhaps even more rarely than the phonological variables
discussed in Section 3.1.1.
4.3 Conclusion
This chapter attempted an overview of the major aspects of SgE grammar.
The first section on syntax and morphology dealt in some detail with fea-
tures relating to the verb, with existential constructions, and with aspect
marking, whereas the second section on semantics and pragmatics took an
in-depth look at the famous Singlish discourse particles and at features of SgE
lexical semantics. The case for indexicality was most prominently made in
Section 4.1.1 and its analysis of verbal features, where it was seen not only
that the distribution of Standard and Singlish variants were not in line with
diglossic expectations, but also that their use may be conditioned by co-text
as well as by several social factors. The following chapter will now take a
closer look at the concept of indexicality, and undertake to apply it to the
variation found in SgE.
5 Indexicality: a model for Singapore?
As the previous chapters have shown, the variation in Singapore English be-
tween what has, traditionally, been called ‘Standard English’ on the one hand
and ‘Singlish’ on the other hand is less than straightforward. The diglossia
approach, presented in Section 2.1.3, took a domain-based distribution of
two relatively clearly identified varieties as its premise, in which the choice
of one variety over the other was motivated, essentially, by the nature of
the exchange: university lectures, all but the least formal writing, public
speaking, school and church settings, all called for the high (H) variety,
Standard English, whereas everyday conversations, peer-group interaction
in the playground, and generally all informal speech was carried out in the
low (L) variety, Singlish. That this view of variation in Singapore English is
unsatisfactory was subsequently explained at length, drawing mostly on data
showing the problematic interplay of features ostensibly ‘belonging’ to both
H and L in the same utterance.
Alternatives were found in Alsagoff’s cultural orientation model (see
Section 2.2.1), a view of the Singlish–Standard variation that takes into
account the potential social meanings encoded in ‘switches’ from H to L
and/or vice versa. Still, the view there was largely one that took a supposed
‘variety’ of English (either Standard or Singlish) as its unit of analysis (a
point further addressed in Chapter 6), and was also largely concerned with
cultural orientation (thence the name of the model), rather than with an
overarching theory that could explain all of the social meanings potentially
indexed through the various features used. This chapter presents the in-
dexical approach in more detail, and, building on the previously mentioned
elements in Section 2.2.2, argues that indexicality has the potential to more
thoroughly explain the variation at work in Singapore’s linguistic landscape.
5.1 Indexicality as an analytic tool

The indexical approach to variation is one that is situated within ‘third-wave’
variationist studies (Eckert 2005; see also the discussion in Section 2.2.2).
Its founding principles, as mentioned by Ochs (1992), have been formulated
by the likes of Bakhtin (1981) and Silverstein (1976). While the former, like
100
5.1 Indexicality as an analytic tool 101
Bourdieu (1986), is situated outside the realm of linguistics proper (being

concerned more with the philosophy of language, and, in the case of Bour-
dieu, with anthropology), Silverstein is an anthropological linguist, and works
within a tradition of ‘research on speech events and speech activities’ (Ochs
1992). A central point of Silverstein (1976) is the distinction between refer-
ential and non-referential indexicality: referential indexicality contributes to
the semantic value of an utterance, while non-referential indexicality does
not. This can be shown with the examples below, where (1a) is an example of
an utterance that makes reference to entities (books and pages) and states (the
containing of pages) regardless of context: the utterance is true irrespective
of the speaker, the time of the utterance, the addressee, etc. In (1b), there
is a ‘referential indexical’, called by Silverstein a ‘shifter’, in the form of the
pronoun I, which has both a referential meaning (first person singular) and a
contextual, or deictic, meaning (which depends on who speaks and therefore
who I refers to). Example (2) is fully deictic, in that the meanings of we, here,
and in two hours are all irrecoverable without personal, spatial, and temporal
context.
(1) a. Books have pages.
b. I have several books.
(2) We shall meet again here in two hours.
Non-referential indexicality is what sociolinguistics is most interested in:
for instance, pronunciation variants such as a glottal stop for the intervocalic
/t/ in words like butter do not add anything to the semantics of the lexical
item or to the truth value of the utterance – such variants are, after all, called
allophones of a single phoneme, thus highlighting their status as variants that
are unable to create minimal pairs. Nonetheless, at a sociolinguistic level, the
pronunciation of butter as [b2P@] most definitely carries (or indexes) some
social meaning. It is the description of this social meaning that a sociolin-
guistic analysis grounded in indexicality seeks to address. Ochs (1992: 339),
for instance, gives the example of how American pre-adolescent children use
pitch range to index gender, at an age when their actual, physical voice is as
yet undifferentiated.
An addition to the theory of indexicality is the concept of indexical orders
(Silverstein 2003). Indexical speech acts can be conceptualised as consisting
of a first level of indexicality, which is essentially pragmatic in nature, of
a second, metapragmatic order, and possibly even of ‘higher’ orders of in-
dexicality, with conventionally recognised discourses such as ‘life-style em-
blematisation’ (Silverstein 2003: 222). To illustrate this ranking, a first-order,
pragmatic index can be found in the distinction, made in many languages,
between formal and informal second-person singular pronouns, the so-called
T/V distinction (from French tu and vous, being respectively the informal
and the formal version of the second-person singular subject pronoun). In
such languages, the choice of one or the other pronoun can be regarded as a
choice indexing solidarity and power differentials (the categories introduced
in Brown and Gilman’s (1960) analysis of T/V deference indices), such that
a pupil addressing his or her teacher using V indexes a perceived higher po-
sition of power on the part of the teacher. Similarly, gender indices such as
the third-person singular pronouns in English can act as first-order indices,
in that they simply add pragmatic information to the utterance, signalling
that the referent is male, female, or non-human.
Second-order indexicality is found in metapragmatic meanings such as
social class indices: for instance, the intervocalic glottalling discussed above
might well index, within the larger speech community of Britain, a lower so-
cial class than the alveolar variant, or all sorts of other personal characteristics
that are seen to be indexed by this variable (level of education, friendliness,
authenticity, etc.). Similarly, in cases where T/V distinctions become ‘en-
registered’ (Silverstein 2003: 108), the social meanings go beyond ‘simple’
pragmatics by taking on metapragmatic meanings such as ‘formal’ for V and
‘informal’ for T, or other dyads such as ‘public’ vs ‘intimate’. Thus ‘V-talk’
(Silverstein 2003: 209) can index a stance of lessened intimacy, in situations
where such a highlighting of distance might be desirable.
Finally, ‘higher’-order indexicality is a phenomenon argued by Silverstein
(2003) to occur when certain ways of talking associated with a particular so-
cietal group, activity, etc., are reused to index certain properties seen to be
inherent in that group or activity. These are ‘rarefied indexical orders of
commoditised identity’ (Silverstein 2003: 222). His example is that of ‘wine-
talk’, dubbed oinoglossia, the speech act of commenting on and discussing
the look, smell, taste, ‘body’, etc. of wine by upwardly socially mobile young
urbanites (‘yuppies’). The lexicon and discourse structure employed are
those of professional wine critics, the ‘evaluative terms’ index ‘connoisseur-
ship’ and are identifiable as belonging to this ‘micro-context’ of prestigious
wine-tasting professionals (Silverstein 2003: 226). This micro-context is in-
dexically linked to one at the macro-sociological level, ‘locatable’ because
of ‘its enregisterment in a cultural schema of sociolinguistic differentiation’
(Silverstein 2003: 226), such that the use of such wine-talk indexes a range of
stances, from ‘refined’ to ‘well-bred’ and ‘elite’. The association is one that
is conventionally recognised, or ‘enregistered’, so that once one engages in
the practice of oinoglossia, the social meanings and stances indexed are easily
recognised within the speech community.
These layers (orders) of indexicality combine to create meaning located
well beyond the referential level, and even beyond the mere pragmatic or
first-wave sociolinguistic level. That these orders can co-occur within the
same indexical process is easily demonstrated: imagine a speaker engaging
in oinoglossia (to stay with Silverstein’s example) in the context of a dinner
party. A first-order (pragmatic) index may be found in linguistic features
such as the T/V pronouns used in the course of the interaction, as well as
in affect indices (indexing emotional states, e.g. rhythm, voice pitch range,
etc.). At the second (metapragmatic) level of indexicality, the speaker may
be (consciously or not) using a certain accent associated with a given social
image that is deemed appropriate in this context. This may include linguistic
features that index a higher social class (such as the avoidance of intervocalic
glottalling or an RP-style pronunciation of the goat diphthong, for instance),
a higher level of education (e.g. the avoidance of sentence-final prepositions),
and so forth. Whether or not these features actually (empirically) correlate
with social class, level of education, etc. is less relevant than the fact that they
come to be associated with these social meanings, and thus take on indexical
meaning which can be used agentively. Finally, a ‘higher’ order of indexicality
is achieved by the oinoglossic discourse itself, which, as discussed at length
in Silverstein (2003), evokes a whole range of social meanings, chief among
them the claim to be part of a well-educated, sophisticated segment of the
population who ‘know their wines’.
It should be clear by now that an indexical approach is usually concerned
with a more micro-sociolinguistic view of variation. The indexing processes
observed are rather complex, and formulating a theory of variation for a
whole speech community is less straightforward than analysing features
in given interactional settings. Research to date has focussed on gender
indexing (Ochs 2004), discourse analysis (Blommaert 2007), and local-dialect
stereotypes (Johnstone and Kiesling 2008). In all of these, indexicality is
presented as a new analytical tool that has a lot to offer to complement
previous approaches. In Johnstone and Kiesling (2008), for instance, the
pronunciation of the diphthong in words like house as a monophthong [a:] by
residents of Pittsburgh, often described as a local feature, is in fact shown to be
perceived with a much wider array of attitudes than traditional sociolinguistic
approaches would suggest: the disjunction between individual production
and perception, coupled with speakers’ own reflections on the variable, offers
insights into the layered nature of indexical processes operating in spoken
interaction, as well as into the multiplicity of social meanings indexed by the
same variable. It is this more inclusive nature and this potential to explain
the complex interplay of many social meanings and linguistic variables that
make the indexical approach such a powerful tool.
5.2 Eckert’s indexical field

The indexical approach presented in Section 2.2.2 and further refined in
Section 5.1 offers the opportunity to formulate a new model for Singapore
English (SgE) variation that builds on previous models’ shortcomings. First
and foremost, it bypasses the inconvenience of having to define subvarieties of
SgE. Secondly, it is firmly entrenched in third-wave approaches to language
variation, an angle from which SgE has not, thus far, been truly analysed –
with the possible exception of Alsagoff (2007).
The problems with subvarieties (or ‘codes’) of SgE, further discussed in

the next chapter, revolve around the large degree of grammatical overlap in
the two putative varieties ‘H’ and ‘L’. Therefore, a model seeking to explain
SgE variation should have, as its central objective, neither the definition
of subvarieties, nor the labelling of strings of utterances as being of such
a subvariety. These enterprises, while justifiable in first-wave approaches,
fail to account for the true complexity of the linguistic repertoire of SgE
speakers. However, it is important for the model to identify features (or
variables) of what may be considered, in an attempt to simplify the discussion,
codes associated with the local vernacular (Singlish) and the/an international
standard (StdE).
Such a model, which combines elements of stylistics and indexicality
(Coupland 2007; Eckert 2008; Rampton 2009) with the more traditional,
binary diglossic view of SgE (Gupta 2006b; Alsagoff 2007), may not neces-
sarily result in a graphic model as aesthetically pleasing as the ones proposed
by Platt (1975, 1977), Pakir (1991), or Poedjosoedarmo (1995). A simple list
of features associated with Singlish and StdE might well be enough as a
starting point. Similarly, as far as the social meanings indexed are concerned,
the formulation of an indexical field, as per Eckert (2008), while not a pre-
requisite, would inform the understanding of the stances potentially indexed
by means of the aforementioned features. Unlike Alsagoff (2007) and the
orientations of her COM, however, I do not feel the need for these indexical
stances to come in dyadic form: some of them will do, of course, such as
formal–informal, but others need not.
The features, then, consist largely of the variables described in Chapters 3
and 4, as well as those reported in other descriptions of SgE. Features of
Singlish include, but are not restricted to, discourse particles, existential
constructions with got, the where got collocation, non-inflected verb phrases
and noun phrases, pro-drop, be-deletion, and the absence of conjunctions in-
troducing conditional clauses. Features that mark StdE in Singapore include
inversion in interrogatives, the use of certain auxiliaries, and the presence
of verbal and nominal inflexions. While these features are seen, in this new
model, as being indicative of Singlish and StdE respectively, that is, as ‘be-
longing’ to their respective codes, they do not, by their mere occurrence in a
given utterance, mark that utterance as being ‘in Singlish’ or ‘in StdE’. This
is an important point, as it solves the issues encountered in previous models
with regard to speakers using separate homogeneous codes and constantly
switching from one to the other: this model suggests that features recog-
nised as diagnostic of Singlish and StdE are potentially used consciously by
speakers in order to index a particular social meaning.
It is these social meanings that need further clarification. Alsagoff (2007)
proposes a list of cultural orientations that are indexed by the use of a local
or an international variety of SgE. This list, given in Section 2.2.1, is repro-
duced here in Table 5.1 for ease of reference. Alsagoff’s model is the first
Table 5.1. ‘Features of the two orientations in the

cultural orientation model’ (Alsagoff 2007: 39, table 1)
ISE LSE
Globalism Localism
(a) Economic capital Sociocultural capital

(b) Authority Camaraderie
(c) Formality Informality
(d) Distance Closeness
(e) Educational attainment Community membership
to take an indexical approach to SgE variation, even though she does not
explicitly label it thus. The model sees the speakers as orienting themselves
towards one of two cultural orientations, globalism and localism, and as
expressing this orientation by choosing features associated with ISE (Inter-
national Singapore English) or LSE (Local Singapore English) respectively.
The innovative element of this approach is that a mere orientation towards
one or the other does not necessarily render the entirety of the speakers’
utterance ISE or LSE. Rather, the possibility of combining features from
both of these codes enables the speakers to situate themselves between the
two orientations, along a continuum of cultural orientation which allows for
an expression of local identity couched in a discourse on the advantages of
globalisation, for instance.
That the expression of local identity is important to Singaporeans is
beyond doubt. As described by Alsagoff (2007) and explained in Section 2.2.1,
Singaporeans live in an environment where cultural expression is complex
and multipolar. Besides the relatively straightforward cultural policies of
the government, which equate ethnicity with cultural affiliation, there is the
presence of so-called ‘Western’ culture(s) as well as the cultures surrounding
the island-state, and, last but not least, the Singaporean culture itself. There
is still considerable uncertainty as to what this culture actually encompasses,
but there is also an increasing awareness of Singapore’s local distinctiveness in
a globalised world. The quote in (3), taken from an online discussion forum,
is a contribution from an ostensibly Chinese Singaporean. It is a recognition
of the independent status of Singapore (and its ‘quirks’, or culture), yet at the
same time acknowledging its origins outside of the country – using a Tourist
Board slogan ‘uniquely Singapore’ followed by the ubiquitous lah. There is
an expression of pride in the local varieties of English and Mandarin, often
criticised by policy-makers as not matching the exogenous standards still
seen, by some, as the ideals towards which Singaporeans should strive.
(3) SG is NOT Beijing or Shanghai or Fujian or Canton, or UK
or USA . . . we’re uniquely Singapore lah!! And as a born and
bred Singaporean . . . I really think locals should be proud of
Sociocultural capital Economic potential
Closeness Local Global
Community membership Educated
Camaraderie Important Serious
Friendly Distance
Relaxed
Informal Formal
Uneducated Authority
Mocking
Rude
Pretentious
Figure 5.1. Indexical field of SgE. Bold = cultural orientation, italic =
stances.
their unique regional quirks, including Singlish. So what if we

can’t enuniciate [sic] perfect Queen’s English, so be it. Ditto
Beijing-perfect Mandarin. (SG Chinese 2009: #22)
In a COM approach, this message is entirely in ISE, as it is directed at an
international audience (the forum in question being used predominantly by
‘Western’ expatriates in Singapore). The only LSE element is lah, which is
clearly used to highlight Singaporean identity. Notwithstanding its stereo-
typical character, it very effectively indexes a local stance, or, in COM terms,
takes a localist orientation towards sociocultural capital.
Cultural orientation, therefore, should form an important part of the
indexical system of SgE. More stances can be listed, however, as mentioned
earlier: the ones in Table 2.11 should be taken into consideration as well. It
should be noted here that there can be an indexical field for each variable (see
Eckert’s example of /t/ release in Figure 2.2). The attempt here is to give a
non-exhaustive list of stances that can be achieved by using the Singlish or
StdE features described above. Figure 5.1 gives an impression of how these
stances could be arranged following Eckert’s (2008) styling of an indexical
field.
As mentioned above, these stances need not be dyadic, although many are.
Figure 5.1 maintains Alsagoff’s (2007) COM features (slightly reworded),
and situates them around the larger cultural orientations of ‘global’ and ‘local’.
In addition, the stances described earlier as found in our data are added to the
field, and placed in the approximate neighbourhood of the cultural orientation
with which they may be associated. Thus, for instance, an ‘important’ stance
2.
Feature selection
Speaker chooses the
feature(s) that appropri-
ately index(es) the desired
stance
Discourse particle
1.
Stance selection
Speaker chooses from
Community
the indexical field the membership
stance(s) he/she wishes
to take
de ‘OK lah, we go to
co Genting ah.’ (i.C.4.m) 3.
din
g by Production
hea The utterance is pro-
rer
duced with the fea-
ture(s) selected
Figure 5.2. Indexing process with an example
is likely to be indexed with the same linguistic variables as an expression of

‘economic potential’, and ‘camaraderie’ is related to ‘relaxed’ in linguistic
production. Besides this horizontal arrangement, there is also a vertical one,
which is broadly reflected in the positive and negative associations evoked by
the wording of the stances. ‘Pretentious’ can be indexed with ISE features,
but evokes very different connotations from, say, ‘educated’, indexed with
the same features.
The indexing process itself can be usefully illustrated with a diagram such
as the one in Figure 5.2. Any act of linguistic identity expression starts with
a decision (1), on the speaker’s part, as to which stance he or she wishes to
take and to convey to the addressee(s). The linguistic resources available are
then searched (2) to identify features that index the desired stance. Finally
(3), these features are used in the utterance in order to fully index the
stance taken. This utterance and the indexical features it contains are then
decoded by the addressee(s), who identify the stance taken. In the example in
Figure 5.2, the speaker (i.C.4.m) wishes to express community membership,
selects discourse particles as the appropriate feature to index this stance, and
consequently uses two in his utterance.
The advantages of such a model are twofold. Firstly, it removes the
problems inherent in the identification and definition of subvarieties of SgE,
a concern central to existing models and their major flaw. The indexicality
model does not postulate a switch from one homogeneous code or variety to
another when a speaker uses a particular variable: it is the features of one of
two codes that are seen as fulfilling the task of indexing a particular stance –
not of making that particular utterance conform to a given variety. Thus,
the example in Figure 5.2 is not Singlish as such, it simply uses Singlish
features to index a stance of community membership. Secondly, this model
emphasises speaker agency. As opposed to models based on Platt (1975), the
use of Singlish features is not, in the indexical model, based on speakers’
level of education, socio-economic status, or another measure of English
proficiency, but rather on their knowledge of the features’ potential to index
certain stances. As shown in the examples given throughout the preceding
chapters, and especially in Section 2.2, there is a high level of awareness
of the social meaning of the linguistic variables used. This awareness was
recognised by Gupta (1994), who attributes the diglossic distribution of CSE
and StdE to speakers’ knowledge of when to appropriately use one or the
other. The indexicality model takes this a step further by refuting the idea
of codes and code-switching between H and L, and by further underlining
the importance of linguistic variables not as diagnostic of a particular variety
being used, but as indexing a particular stance.
5.3 Indexicality in Singapore

By way of an interim summary, Table 5.2 presents a brief overview of the
models examined in more detail in Chapter 2. I then consider how these mod-
els can explain the data presented in the preceding two chapters. While the
continuum and diglossia are the two major models, the others (in particular
Alsagoff’s COM) are also important and need to be considered here.
Example (4) below, from Leimgruber (2012b: 9), is given as a snippet of
data against which these models can be tested. The extract is from the group
recording of the Malay informants at the polytechnic (ii.M.gr). The group
consists of two males (ii.M.1.m and ii.M.4.m, 1 and 4 below) and two females
(ii.M.2.f and ii.M.3.f, 2 and 3 below). Features identified by Gupta (1994)
as Singlish have a straight underline, and features identified as StdE have a
wavy underline.
::::::::::::::
(4) 3: Fifteen minute ah, is it? Aiyoh, don’t know what to say ah.
::::::::::
1
4: Eh after this ∅ finish ah? 2
2: No. One of you have to go walk around the school, meet your 3
friends and talk.
::::::
4
4: :::
Are::::
you sure ah? 5
2: Because he want to see how we all talk, normally. . . 6
3. (***) stupidity. 7
2: That’s: what him1 say to us just now. 8
4: Serious? So dumb. 9
1
Colloquial feature not included in Gupta (1994).
Table 5.2. Overview of existing variationist models for SgE
Terminology (number
SgE internal organisation of subvarieties) Reason(s) for speaker variation
Post-creole continuum continuum with StdE at the acrolectal and acrolect/basilect (∞) level of education and stylistic choice
(Platt 1975) Singlish at the basilectal end
Polyglossia (Platt 1977) diglossic relationship between Singlish and Hx /Lx (2/∞) home language, situational context
StdE but also between English and other
languages
Diglossia (Gupta 1989) StdE and Singlish in diglossic relationship H/L (2) situational context
with occasional code-switching
Proficiency (Pakir 1991) StdE and Singlish at extremes of a continuum SSE/CSE (∞) language proficiency and choice of level
linked to proficiency and formality of formality
Modified triangles same as Pakir (1991) but formality/proficiency acrolect/basilect (∞) language proficiency and choice of level
(Poedjosoedarmo link only for acrolectal speakers of formality
1995)
Cultural orientation StdE and Singlish at extremes of a continuum ISE/LSE (∞) cultural orientation (global or local)
model (Alsagoff used to signal cultural orientation
2007)
3: Do you realise that he’s

::::::::::::: :
actually quite nervous? 10
4: I know! 11
2: Yeah. He :::will::::
edit it lah, I think. 12
1: So confirmation right, we will be Bangkok.
::::::
13
4: (***) 14
2: Yeah. 15
3: [laugh] 16
2: Ok lah, confirm ah, Bangkok. Let’s have a try ah. . . It’s: more 17
like a survival things2 ah, with three hundred. 18
4: (***), eat like rice? 19
3: No ah, that one is :
only for accommodation. But actually, if you 20
go there right, my brother say for just going there is: actually 21
five hundred :::::
bucks ah. 22
4: No lah. 23
2: Minimum! 24
3: Ng-ng. My brother, my brother one3 ::::: went there and he ::::
took 25
the– 26
2: Budget one? 27
3: Budget airline and then he come back with normal, the normal 28
airline, then the total is :
hundred. But never mind, we can 29
always. . . 30
4: But if you take, yeah, my friend say ah, because he ∅ going 31
back Thailand, (***) don’t know right, Chiang Mai, Chiang 32
Rai? 33
1: Chiang Mai, Chiang Mai. 34
4: He say if he :::::
takes Silk Air, it’s:
five hundred, to go. 35
2: To go only. 36
3: Hah? 37
2: Wah lau eh! 38
4: But if you take Jet Star, can ∅ eight ::::::dollars. From4 ah, from 39
eight ::::::
dollars onwards. But then got the airport tax. 40
2: That one the eight dollar is a plus plus thing, you know. 41
4: Yeah lah. 42
(ii.M.gr)
In Platt’s (1975) continuum, we would expect these informants to be con-

versing in a relatively informal atmosphere, signalled by the use of Singlish
features (in particular the discourse particles), which outweigh the StdE
ones 32 to 19. They would therefore be using, much of the time, a lower
mesolectal variety on the SgE continuum. However, as Platt’s continuum
consists of two dimensions, we would also need to locate the speakers on the
socio-economic scale in Figure 2.1. This is more complicated, as the type of
2
Hypercorrect form.
3 4
Colloquial feature not included in Gupta (1994). Emphasis.
5.3 Indexicality in Singapore 111
school attended is not in itself enough to decide on the informants’ socio-

economic status (see Section 1.3), although this was Platt’s measure. One
could of course decide to use the adolescent informants’ parents’ occupation
(see Appendix B), and, indeed, the group seems to come from comparable
family backgrounds. This, however, does little to explain the actual inherent
variability of the discourse: why, for instance, does ii.M.3.f use, in the one
turn in lines 20–2, variants in the sequence L–H–L–H–H–L? Of course the
absence of a clear method of identification for the continuum’s subvarieties is
regrettable, but surely we cannot explain this apparent alternation between
acrolect and basilect merely by referring to the speaker’s socio-economic
status, or, indeed, the formality of the setting. It is hardly conceivable that
situational setting and levels of formality are switching at such a quick rate.
Diglossia, too, does little to explain this example (I subsume here Platt’s
polyglossia to general diglossia – there are no instances of English–Malay
code-switching in (4)). Clearly the exchange is neither fully H nor fully L;
perhaps the notion of a ‘leaky’ diglossia (Gupta 2006b: 22) is of help. The L–
H–L–H–H–L sequence in ii.M.3.f’s turn certainly seems to use both features
recognised by Gupta (1994) as H (nominal inflexion) and as L (particles).5 In
order for ‘leaks’, or switches, to be identified, however, it would be necessary
to first identify which code is dominant and which one is being switched into
(‘matrix’ and ‘embedded’ language respectively, in Myers-Scotton’s (1997)
terminology). But this is far from simple. In fact, the two subvarieties are
so close that it is not possible, unlike in more straightforward examples of
classic diglossia, to be entirely sure which code is the ‘matrix’ in a given
utterance. This is particularly so because of the nature of the features used
to identify H or L. Take noun plural inflexion, for instance. Its absence is
diagnostic of L, but its presence, while a requirement of H, is also possible
in L. This makes a clear distinction between Singlish and StdE difficult.
Clearly, then, a diglossic approach to (4) is of little help, since, as shown by
the highlighting throughout the example, H and L variants are interwoven
to an extent that makes the identification of individual subvarieties close
to meaningless. The triangular models of Pakir (1991) and Poedjosoedarmo
(1995) (see Section 1.2.2) may well offer a better explanation, although they
suffer from problems similar to those of Platt (1975). The proficiency levels
of the informants were not tested by any means other than through sampling
in schools that have particular entry requirements, including appropriate
results in English Language. That dimension of the triangles is therefore
impossible to test here. More crucially shaky, however, is the nature of the
linguistic continuum stipulated by Pakir and Poedjosoedarmo. Of course
5
Gupta (1994: 13) notes that ‘[CSE] nouns are not normally inflected . . . while this is cate-
gorical in StdE’. Thus, inflexions are not as such markers of H, since they can optionally
occur in L. This, however, does not invalidate the point made here. A further example,
by speaker ii.M.2.f in line 12, shows alternation of H and L features that are diagnostic of
their respective subvarieties (modal auxiliary and particles respectively): ‘He :::
will :::
edit it lah,
I think.’
the acrolect is equated, in both cases, with SSE or StdE, and the basilect
with CSE/Singlish, but the subvarieties linking these two are not defined
in great detail (unlike, for instance, DeCamp’s (1971) implicational scale for
the Jamaican continuum) – a shortcoming also found in Platt. Thus, it is as
hard to locate a speaker, or even a single utterance, in a given position on
the triangles, as it is to locate that speaker or utterance on Platt’s continuum,
because of the seemingly incoherent use of Standard and Singlish variants
within the same utterances. Given that formality is, in all these models, a
major factor, an utterance such as the one in lines 20–2 of (4) would be very
hard indeed to classify along such a scale. And to suggest that its variant
sequence of L–H–L–H–H–L represents successive switches from informal
to formal is as unsatisfactory as the suggestion of successive switches from
one diglossic subvariety to the other.
Of the models presented in Table 5.2, it may well be Alsagoff’s (2007)
cultural orientation model (COM) that is most suited to explain the data
presented here. Its shortcoming, perhaps, lies in the name of the model: it
might be too restrictive to limit its applicability to cultural orientation alone.
While it is conceivable that ii.M.3.f’s observation in line 10 (‘Do you realise
that he’s actually quite nervous?’), which presents a mild criticism of the
researcher, is framed in ISE – evidenced by the use of inversion and [+be] –
because of the referent’s status as an outsider, thus forcing the utterance into
a cultural orientation of ‘globalism’, it is hard to see how ii.M.2.f’s ‘That’s
what him say to us just now.’ (line 8) is different enough to warrant LSE.
Overall, however, it does seem to be the case, at least in example (4), that
StdE features prevail in exchanges where the researcher is part of the topic.6
Other components of COM, such as the dyads authority–camaraderie and
educational attainment–community membership (see p. 50) can also be used
to explain many cases of apparent H–L alternation. The main strength of
COM resides in the fact that, much like Eckert’s indexicality, features of H
and L (or ISE and LSE, in Alsagoff’s terms) can combine to accommodate
several orientations in the same utterance:
A speaker may choose to stress authority and yet signal community
membership. As a consequence, while they may decide to orientate
towards ISE to indicate authority, they might also exhibit some degree
of Singlish features to indicate a local perspective in order to stress
membership in the community. (Alsagoff 2007: 40)
5.4 Conclusions
Despite its advantages, explained above, the indexical model has some lim-
itations. While arguably solving the conundrum of synchronic variation in
6
Compare a similar instance in the previous chapter’s example (1), where there also was a
switch from LSE to ISE when the researcher was addressed (via the microphone).
5.4 Conclusions 113
SgE, indexicality is probably not suited to explain in much detail the changes
in progress in the English spoken in Singapore. I would tend to follow the
argument of Trudgill (2004: 156–7) that linguistic expressions of identity are
much more likely to be a result of language change (and language contact,
in the case of SgE) than its cause. It would be hard to argue that discourse
particles, for instance, were deliberately selected by Singaporeans, who then
transferred them from the substrate into the emerging Singlish in order to
adequately mark a Singaporean identity. It is much more plausible that this
feature only became a marker of identity once it was established in Singlish.
Similarly, it is important to mention that unlike, for instance, Platt’s (1975)
model, language proficiency is not a variable in this model. It seems reason-
able to assume that the SgE speech community is composed of both native
speakers (see Gupta 1994: 13–32 and Bokhorst-Heng et al. 2007 for an ex-
ploration of this concept) and non-native speakers. The indexicality model,
as applied to SgE, presupposes a respectable command of both Singlish and
StdE, or rather, of which code individual features belong to. It makes sense,
therefore, to restrict its validity to the younger generation, whose use of
English has been cemented by now compulsory education in the English
language.7
Likewise, the model does not contribute extensively to recent typological
work on World Englishes (Kortmann and Schneider 2008; Sharma 2009;
Lim and Gisborne 2009). While the post-creole continuum and the diglossia
hypotheses grew out of models proposed for other varieties, thereby situating
SgE in a wider sociolinguistic typological context, the indexicality model is
limited to a highly localised explanation of SgE variation. While the model
itself is potentially universally applicable, it requires a level of local adjust-
ments (in the linguistic variables as well as the social stances indexable) that
make it unique to – and limited to – the speech community at hand.
Despite these shortcomings, the indexicality model has the unrivalled
potential to explain the seemingly random variation observed in SgE. The
primary reason for this success is to be found in the high level of detail that the
model takes into account, as well as its departure from ill-defined concepts
such as code and subvariety (of which more in the concluding chapter). Its
focus is less on the average production of speakers’ SgE than on their actual
usage of variables associated with Singlish or StdE. The resulting stance-
taking is an integral part of the model, serving as the explanation for actual
variation.
7
English was made the sole medium of education in 1987. Compulsory education (for the six
years of primary school) was introduced in 2003, although education was near-universal by
the end of the 1990s (MOE 2009).
6 Conclusion: the variety as a
structural unit
The preceding chapters have highlighted the problems of analysing the

interaction between Singlish and Standard English as diglossic, with similar
problems encountered in the continuum approach. Furthermore, the data
show that the usual variety-based approach to this variation is flawed, or at
least unsatisfactory in its explanatory power. This chapter recapitulates these
issues, and thematises the use of the ‘variety’ as a structural unit in fields
such as sociolinguistics, contact linguistics, and World Englishes.
6.1 Codes, varieties, and code-switching

So far, terms such as code, variety, and subvariety have been used as near-
synonyms when talking about SgE. Matthews (2007) offers definitions for
each, saying that a variety is ‘any form of language seen as systematically
distinct from others’, and that a code is ‘any language or distinct variety of a
language, whether or not it is actually thought of as a code in any illuminating
sense’. This may need a bit of clarification. In sociolinguistics, variety is
a term that is used widely, and came about initially to avoid the loaded
terms accent, dialect, and language. The problems in identifying language
boundaries along a dialect continuum, for instance, are well known. It is,
therefore, convenient to use variety not just for every dialect making up
the continuum, but also for the superimposed standardised ones. The same
holds true for code, but works at an even more abstract level: it can be applied
not just to varieties of language as we normally understand it, but also to any
means of communication, such as, for example, the Morse code (Matthews
2007: 62). There is a good overview of these considerations in Wardhaugh
(2002: 25–6). What is of concern here is the applicability of these terms when
modelling variation in SgE.
In a classic diglossic situation, there seems to be general agreement about
what each variety stands for. Take the Swiss diglossia as a convenient ex-
ample. Standard German is H, Swiss German is L: two varieties, used by
members of the speech community depending on the context (as defined by
Ferguson 1959). The situation, however, is not quite as simple as that. L is
not a uniform variety at all. Leaving the usual sociolinguistic variations (class,
114
6.1 Codes, varieties, and code-switching 115
age, sex, etc.) aside, there is immense geographical variation in L: in fact,

what is commonly called ‘Swiss German’ consists of several varieties of High
Alemannic, Highest Alemannic, and Low Alemannic (in order of number of
speakers), which themselves present internal geographic variation. The High
Alemannic varieties of Frauenfeld (Thurgovia) and of Kriens (Lucerne) are
identifiably different. This poses the problem of having a considerable num-
ber of varieties, all of which co-exist, collectively, in a diglossic relationship
as L with Standard German, H. This may warrant the term subvariety, to
highlight their common trait as a Swiss German diglossic variety.
It does not stop there, however. H, too, is not the uniform variety that the
theory might wish it to be. There are well-known differences in the stan-
dard varieties of Germany and of Switzerland (Standard German and Swiss
Standard German respectively), which are also codified in (pan-)German
dictionaries. Many are lexical, partly due to the different political systems of
the countries where they are used, others are orthographic (e.g. the absence of
ß in the Swiss variety) or morphological, others again are phonological. But
even given a complete codification – if that were possible – of these two stan-
dard varieties, in production many Swiss would still exhibit various types of
H. This may be a result of various levels of proficiency, of exposure to spoken
H, or of willingness to use the spoken language of a neighbouring country,
but more significantly, of L1 interference, an L1 which, as explained in the
previous paragraph, is everything but uniform. Given a sample of ten Swiss
German speakers from ten geographically distinct varieties, it is likely that
the H form produced by each would contain strong elements (pronunciation,
particularly) of their respective L.
In this case, the idea of two uniform varieties co-existing harmoniously in
a diglossic relationship is little more than a theoretical idealisation. The same
can be said of other classic diglossic situations, and the same can be said of
SgE. Authors such as Platt (1975) recognised this, stipulating a continuum of
subvarieties, linking a prestige variety (StdE) with the local basilect Singlish.
Besides the recognition of more than two varieties, however, this model
suffers from a shortcoming similar to the diglossia model: the identification
of subvarieties. Despite grounding his work in DeCamp’s (1971) post-creole
continuum, a ‘true’ continuum without discrete subvarieties, Platt suggests
an identifiable variety in the basilect, where ‘bunching’ (Platt 1975: 366)
occurs.
The main problem, common to all these approaches, is the assumption
that speakers, in any one utterance, are using a homogeneous code in a given
context. Granted, Gupta allows for ‘leaks’ of one code into the other, but
maintains nonetheless that variables ‘tend to constellate . . . and most utter-
ances will be identifiable as one or the other variety’ (Gupta 1994: 8). Clearly,
as the data presented here have shown, things are not as straightforward. If
we were to analyse the exchange in example (4) on p. 108 using Platt’s con-
tinuum, speakers would be seen to be continuously gliding up and down
116 Conclusion: the variety as a structural unit
the SgE continuum, switching from one subvariety to the other, presumably
triggered by situational context and topic. Not only would this presuppose a
redefinition of the term to border on style-switching (as done by Alsagoff),
it would also imply a constant reevaluation of situational context or topic
within a given utterance, triggering the appropriate switch. Instances of ho-
mogeneous subvarieties are rare in the data, seeming to suggest that this is
the wrong explanation. Similarly, an analysis of (4) in a diglossic framework
would see speakers constantly mixing H and L, switching several times in a
single utterance – an explanation not far from that of the continuum, except
that instead of moving up and down a linear continuum, speakers move from
one diglossic subvariety into the other.
There is no denying that certain variables can be seen as being associated
with a given code. For example, discourse particles are clearly and almost
unambiguously considered to be part of the local vernacular (i.e. Singlish,
or L), and do-support is a good marker of StdE (H). However, it seems
reasonable to argue that Singlish and StdE have more similarities than dif-
ferences: the bulk of English grammar is common to both, with exceptions
carefully listed in descriptions of SgE. So it becomes possible to analyse an
example such as (1) in a more enlightening way: the features of L and of H,
highlighted by straight underlines and :::::
wavy ::::::::::
underlines respectively, can be
seen as marking Singlish and StdE, and, therefore, presenting a departure
from this common (grammatical) core. It is not the aim of this approach to
stipulate three codes (e.g. ‘neutral’, StdE, and Singlish), but much rather to
view the features usually ascribed to StdE or Singlish as being recognised by
the speakers, who then exploit them to create social meaning.
(1) a. But budget airline∅, how much ah ∅ they cost, how much
do they cost?
:::::::::::
b. Ah, how much would it cos[th ] (.) to travel there? (i.I.4.m)
::::::::::::::
In (1a), i.I.4.m requests information from his peers about the cost of
budget airlines’ flights to Malaysia. There are features of L in the topic (an
uninflected noun phrase) and in the question (a discourse particle and missing
do-support). The question is then immediately reformulated, this time with
two (related) features of H (inversion and do-support). Such instances of
putative self-correction are frequently encountered in the data, and may
have several explanations, one of them being a sudden awareness of being
recorded, and thus switching to a more formal style, by marking it with H
features.
Likewise, (1b) shows how speakers are aware of the social meanings of
these (H and L) features. This question is asked several turns after (1a), by
the same speaker. Not only is he using an auxiliary construction, but also a
released final /t/, something that hardly ever happens in spoken SgE (see
Section 3.1). Here the /t/ is, furthermore, strongly aspirated, followed by a
slight pause enabling the speaker to finish the question. The desired effect is
clearly achieved when his co-informants respond with laughter.
It would appear, therefore, that rather than switching from one code to
another (or, rather than H ‘leaking’ into L or vice versa), features of an
appropriate code are used at a given point in the utterance in order to index a
social meaning. Clearly, the homogeneous use, over a complete utterance, of
an L code, for instance, is very rare indeed: it is much more useful to think of
the first half of (1a) not as being in L as such, but as having three L features.
Given the near-impossibility of satisfactorily defining a subvariety ‘L’ and
‘H’ of SgE, the idea of features of L and H being used for social indexical
purposes brings us closer to a viable model for SgE variation.
6.2 English in Singapore and Southeast Asia

These considerations of the concept of the variety need to be seen against the
sociolinguistic reality in Singapore and the wider Southeast Asian region.
There is, firstly, a clear sense in which the English language has been quite
intentionally promoted in Singapore through various policies. These poli-
cies target the promotion of Standard English as the ‘working language’ of
the country on the one hand, and the demotion of the vernacular, Singlish.
Secondly, Singapore’s location within Southeast Asia (and, specifically, just
across the straits from Malaysia) places it in a region where English is increas-
ingly used for various purposes and in various forms. At the diplomatic level,
the Association of Southeast Asian Nations (ASEAN), for instance, uses
English as its only working language. However, regular citizens of ASEAN
nations are also experiencing an increase in use of English, be it as a lingua
franca used when travelling, or as one of several (near-)native languages in
some educated households.
6.2.1 Language policies and Singapore English

A very brief overview of Singapore’s language policies was given in
Section 1.1.3. Besides the constitutional provision of four official languages
(Malay, Mandarin, Tamil, English) and one ‘national language’ (Malay), two
of these languages have benefited from extra attention from policy-makers:
Mandarin, through the Speak Mandarin Campaign, which aggressively pro-
moted its use by Chinese Singaporeans and actively demoted the use of
non-Mandarin varieties (the ‘dialects’) (see e.g. Bokhorst-Heng 1998; Leim-
gruber 2013), and English, having retained, after independence, the special
status of the language of public administration, government, business, and
generally enjoying the semi-official status of ‘working language’. The impor-
tance of English is regularly highlighted by government officials, not least
by Lee Kuan Yew himself, as recently as September 2011 (Ramesh 2011):
English is seen as the tool of Singapore’s economic success, and because it
is so vital to the survival of the country, it needs to be taught, Singapore-

ans need to be proficient in it, and deviations from the norms of Standard
English need to be kept in check. This is where the Speak Good English
Movement (SGEM) comes in, a yearly campaign started in 2000 with the
aim of improving standards. The launch speech by then Prime Minister Goh
Chok Tong again underlines the economic advantages of English, and at the
same time argues that Singlish diminishes this competitive edge Singapore
has over its neighbours:
The ability to speak good English is a distinct advantage in terms of
doing business and communicating with the world. This is especially
important for a hub city and an open economy like ours. If we speak
a corrupted form of English that is not understood by others, we will
lose a key competitive advantage. My concern is that if we continue to
speak Singlish, it will over time become Singapore’s common language.
(SGEM 2011)
The SGEM has received some attention from scholars (Rubdy 2001, 2007;
Rappa and Wee 2006; Wee 2006, 2010a; Leimgruber 2013); the main pol-
icy issue is the upkeep of proficiency in Standard English, without which
the country’s service-based economy is seen to be at risk. ‘Good English’
is in opposition to Singlish, or ‘bad English’. The focus is on eliminating
(mostly grammatical) features of Singlish from Singaporeans’ speech. The
campaign takes on various forms, including writing competitions in schools,
banners and advertising material posted on the public transport network,
TV adverts, and the online media (including the campaign website, where
one can take English quizzes). The stance taken by the government is quite
clear, and in some ways, harsher towards Singlish than towards the Chinese
‘dialects’: while diglossia seems to be at least implicitly acknowledged (with
Standard English and Singlish perceived as two obviously identifiable vari-
eties), ‘co-existence is not an option’ (Rappa and Wee 2006: 95). The mere
existence of Singlish is seen to be a threat to competence in Standard English.
The argument is that the ability to code-switch between the two diglossic
varieties is in fact confined to the educated segment of the population, and
that therefore, the existence of Singlish puts those at the lower end of the
educational spectrum at a disadvantage – removing Singlish from the equa-
tion would result in just the standard available, even to low achievers.
The SGEM is just the more proactive side of language policy. For instance,
there are very clear regulations on what is allowed on national television, as
this extract from the TV Programme Code shows:
Standard English, which is grammatically correct, should be used for
programmes such as news, current affairs and info-educational pro-
grammes. Local English, which is also grammatically correct but pro-
nounced with a Singaporean accent and which may include local terms
and expressions, could be used for programmes like dramas, comedies

and variety shows . . . Singlish, which is ungrammatical local English, and
includes dialect terms and sentence structures based on dialect, should
not be encouraged and can only be permitted in interviews, where the
interviewee speaks only Singlish. The interviewer himself, however,
should not use Singlish. (MDA 2004: 10, my emphasis)
These regulations, therefore, make a distinction between grammatically cor-
rect and grammatically incorrect English, with the latter being proscribed
on television (though this is not always strictly enforced). The grammar and
some lexis of Standard Singapore English are acceptable, as is a Singaporean
pronunciation, though it is restricted in use to certain genres. Finally, the
‘serious’ programmes are to be broadcast in Standard English only (though
it is not specified with which accent).
The target of SGEM is, therefore, clearly Singlish. None of the English-
language policies in place in Singapore, however, are meant to encourage a
shift away from the ‘mother tongues’, which are so important for ‘cultural
grounding’ (see Section 2.2). The mother tongues (Mandarin, Malay, Tamil)
have important roles to fulfil in the language policy of the government;
additionally, their economic value is sometimes also highlighted (especially
for Mandarin, less often for Malay, rarely for Tamil). Nonetheless, the status
of English within the country has meant a steady increase in the proportion of
people who claim it as their dominant home language, as shown in Figure 6.1.
Especially the importance of the language in the education system, where
it has been the only available medium of instruction since 1987, has had
an impact on language shift: parents have deliberately increased their use of
English in the home, hoping to give their children a head start at school. After
all, while the mother tongue is weighted at a respectable 25 per cent of the
year’s final mark, so is English, and the two other crucial topics, mathematics
and science (25 per cent each), are all taught in English. Knowledge of and
proficiency in the language is, therefore, essential for academic, and, later,
socio-economic success. Needless to say, this has resulted in a massive age-
graded language shift towards English, with now more than 54 per cent
of Primary 1 students (Tan 2009) coming from homes where English is the
dominant home language – children who are then educated and who socialise
in English, and for whom it is quite clearly a native language.
6.2.2 English in Southeast Asia

With regard to the degree of nativisation of English, Singapore stands out
from its regional context. Of the other nine ASEAN member states, only
the Philippines has English as a co-official language. Singapore is alone in
having made the language of its former colonial masters the sole medium of
instruction in its state school system. On the other hand, English does play
60
50
40 English
Mandarin
30 Chinese dialects
Malay
20 Tamil
Others
10
0
1980 1990 2000 2010
Figure 6.1. Language most frequently spoken at home, as a percentage

of the resident population. Data for 1980 from Foley (1998: 221) citing
Lau (1993: 6), for 1990 and 2000 from Leow (2001), and for 2010 from
Wong (2011).
an important role in Malaysia, where it was co-official until 1967 and is still
used extensively in the official domain – the Language Act 1967 specifies,
for instance, that legislation has to be in both Malay and English (with the
former being authoritative), that court proceedings may be in English if so
desired by any party involved, and that members may address Parliament in
English.
Historically and linguistically, Malaysia and Singapore are in fact very
close neighbours. Both were under British rule (and were briefly part of the
same independent country1 ), both have similar ethnic compositions (though
with different majorities), and both are characterised by widespread multi-
lingualism. The languages involved are similar: Malay, Min Nan Chinese
and Cantonese, Tamil and several other South Asian languages, and, of
course, English. The colloquial English spoken in Malaysia and Singapore
are similar in more than one respect, and were described, by early scholars
(Tongue 1974; Platt 1977; Platt and Weber 1980; Platt et al. 1983), as a
unit. More recently, the two have been analysed separately, with Malaysian
English having received some attention (Nair-Venugopal 2000; Baskaran
2005, 2008b,a; Rajadurai 2007), though it seems Singapore English is more
1
The Federation of Malaya became independent in 1957, while Singapore was granted self-
government in 1959. Singapore joined the renamed Federation of Malaysia in 1963 together
with Sabah and Sarawak, but left it again to become an independent country in 1965.
thoroughly described. Both Englishes (that is, their vernacular form) share
features in phonology (monophthongs in gate and goat, non-reduced vow-
els, final consonant devoicing, non-rhoticity including absence of linking /r/,
syllable-timed rhythm, etc.) and in grammar (ellipsis of articles, pronouns,
and verbs, variable lack of inversion, variable past tense and third person
singular marking, etc.), as well as features of the lexicon (cut ‘to overtake’,
lah (see Section 4.2.1), kiasu ‘afraid to lose out’, hubby ‘husband (in formal
contexts)’, witness ‘experience, see (in informal contexts)’, etc.).
Of course, the sociolinguistic situation differs on both sides of the straits.
English has been promoted extensively among the entire population since
even before independence in Singapore, whereas Malaysia focussed on mak-
ing Malay (or Bahasa Malaysia ‘Malaysian Language’, see footnote 7 on p. 6)
its national language. The education system in Singapore works solely in
English, whereas in Malaysia Malay is the main medium of instruction, with
English used for mathematics and science lessons (a policy introduced in
2003 and revoked in 2012). English is the language of politics and public ad-
ministration in Singapore, whereas in Malaysia it co-exists in a subordinate
position with Malay. It is perhaps the linguistic interethnic relations that
show most differences: Singapore’s policy-makers have always emphasised
English not just as an important economic asset (see above), but also as an
important internal and interethnic lingua franca: Singaporeans of all races
communicate with each other in English. While this is also true, to a certain
extent, in Malaysia, it is not the result of an articulated language policy. The
national language Malay is more likely to fulfil the role of a national lingua
franca.
The distinction between Singapore English and Malaysian English is,
therefore, mostly a sociolinguistic one – and a political one, seeing as the
labels modifying English are the countries’ names. The distinction is not
purely linguistic, as such a distinction would need to take a close look at
what characterises the two ‘varieties’, and it would seem that there are, in
fact, more features in common between these two forms of English than
features that could be used to discriminate them. This further underlines the
problematic concept of the ‘variety’: while it is hard to argue that Singapore
English and Malaysian English are part of a dialect continuum, they are
indeed highly mutually intelligible.
The same cannot be said for Singapore English and Estuary English,
for instance. Deterding (2005) shows how Singaporean speakers of English,
when presented with an Estuary English recording featuring th-fronting in
three (i.e. [fri:]), transcribed it unanimously as free, even when this does
not make sense in context. On the other hand, other variables, such as the
diphthongs in gate and goat, or rhoticity in American English speakers, do
not seem to present problems. Conversely, the merger in, for instance, ship
and sheep also has the potential to reduce mutual intelligibility, although, of
course, context will help.
Within the Southeast Asian context, however, such problems arise less
often. They are certainly absent between speakers of Singapore English and
Malaysian English, but they have also been shown to be comparatively in-
frequent among speakers from a range of ASEAN countries. Deterding and
Kirkpatrick (2006), who collected conversation data from groups of four in
which each speaker came from a different ASEAN country, found that there
were few instances of miscommunication, all associated with idiosyncratic
pronunciations by speakers from Myanmar (Burma), Laos, and Vietnam,
such as [S] for the first sound in sauce. On the other hand, the authors
identified a list of seven (phonological) features shared by speakers from at
least four countries. They were: dental fricative /T/ as [t], non-aspiration of
initial plosives, monophthongal gate and goat, bisyllabic triphthongs, lack
of reduced vowels, stressed pronouns, and heavy end-stress (Deterding and
Kirkpatrick 2006: 395). Building on the assumption that some of these fea-
tures actually promote mutual intelligibility (in particular the monophthongs
in /eI/ and /@U/ and the break-up of triphthongs), they further argue against
the teaching of a British pronunciation target, which only sounds ‘preten-
tious and insincere’ in a local context (Deterding and Kirkpatrick 2006: 406),
and which, at any rate, is inappropriate for judgements of intelligibility in an
intraregional context (Seidlhofer 2005; Deterding 2010).
On the question whether there is evidence for the emergence of a South-
east Asian variety of English, Deterding and Kirkpatrick (2006) point to the
shared features above, which seem to indicate at least some level of con-
vergence. It is, however, worth highlighting (as Deterding and Kirkpatrick
do) that the English language does have quite different statuses in ASEAN
countries: while it is a native language of Singapore and firmly entrenched
in Malaysia, the Philippines, and Brunei, it is little more than an important
foreign language in the other six countries, Indonesia, Thailand, Vietnam,
Laos, Cambodia, and Myanmar (despite the latter’s colonial history). In
Kachruvian terms, we are in the presence of a group of countries that are
found spread over the outer and the expanding circles. As far as the extents
and domains of use of English in these countries are concerned, therefore,
they are all but equal.
The nature of ‘ASEAN English’ points to a considerable amount of inter-
nal variation. We are obviously in the presence of a lingua franca, a concept
much discussed (see e.g. Kirkpatrick 2007, 2010), which, notwithstanding
the evidence of convergence reported by Deterding and Kirkpatrick (2006),
shows extensive variation according to geography, educational category, eth-
nolinguistic background, and many other aspects. In fact, given the human
and physical geography of ASEAN, such variation is expected: the region has
a population of more than half a billion, as per the breakdown in Table 6.1, and
covers an area of around 5.8 million square kilometres (2.2 million sq miles).
There are eleven official languages in the ten countries, but 1,447 more spo-
ken throughout the region (half of these in Indonesia). That a single form of
Table 6.1. Population, area, and languages of the ten ASEAN member states in 2009.
Population and area from UNDESA (2010), languages from Lewis (2009)
Country Population Area (km2 ) Official language(s) (+others)
Brunei 400,000 5,797 Malay (+14)

Cambodia 14,805,000 200,068 Khmer (+22)
Indonesia 229,965,000 2,804,451 Indonesian (+718)
Laos 6,320,000 52,231 Lao (+83)
Malaysia 27,468,000 1,017,333 Malay (+136)
Myanmar 50,020,000 602,651 Burmese (+110)
Philippines 91,983,000 299,619 Filipino, English (+169)
Singapore 4,737,000 683 English, Mandarin, Malay, Tamil (+17)
Thailand 67,764,000 513,364 Thai (+73)
Vietnam 88,069,000 331,086 Vietnamese (+105)
Total 581,531,000 5,827,283 11 (+1,447)
English should appear in such a large and diverse ecology seems unlikely, at
least at this stage.
These caveats are important when discussing ‘Asian English’ (McArthur
2003), the emergence of a regional standard, or some sort of levelling hap-
pening. Notwithstanding these issues, it is the case that English is becoming
more important in the region: ASEAN has had it as a working language
since its inception (though it was only formally enshrined in its charter as
such in 2007), other international diplomatic ties function in English rather
than Malay (the language with the second largest geographical distribution
in the region, especially when including Indonesia), and the significance of
English in the education systems keeps increasing. It is also at the grassroots
level that English is gaining momentum. Thanks, in part, to visa-free travel
between member states for their citizens, intra-ASEAN tourism is booming,
and the lingua franca used between Filipino, Vietnamese, and Indonesian
hosts and their Malaysian, Singaporean, and Thai visitors is English. Busi-
ness exchanges, at all hierarchical levels and through all industries (though
more so in some than in others), are in English. In short, Southeast Asian
people are getting closer, are meeting more often, and therefore they are
speaking with each other more often. The use of English as a lingua franca
in this context serves practical needs, and may well result, given continuing
contact and face-to-face interaction, in some form of convergence.
6.3 ‘Varieties’ of English

To come back to the issue of ‘varieties’ of English in this context, it is
worth pointing out that while scholars in the field of World Englishes are
generally happy to use Singapore, Malaysian, and Philippines as modifiers
of English, resulting in a label such as Malaysian English, which suggests
the status of an independent ‘variety’ with identifiable features and some

form of local recognition, few are equally happy to use English preceded
by Thai, Indonesian, or Cambodian, for instance. Here the preferred term
is English in X, where X = any country. Terms such as Singapore English
are well established, and although other forms are sometimes proposed (e.g.
Singaporean English (Platt and Weber 1980: 1)), the existence of a ‘variety’
Singapore English is rarely questioned. We might argue that, after all, English
is spoken in Singapore, and the form of English spoken there is different from
the forms of English spoken in the UK or in the USA, so it is worthwhile
to point out this difference by a label. In fact, the same procedure applies
to the more traditional, ‘inner-circle’ Englishes as well: there is Australian
English, South African English, Scottish English, even Somerset English and
Birmingham English. In all these instances, a label is used to define the unit
of analysis: a ‘variety’ of language found in a particular geographical locale
(which may be a nation-state, a subnational entity, a county, or just a city).
The challenge for linguists is that these varieties, these forms of English,
are defined not in linguistic terms, but in geographical terms. Of course, it
is convenient to think of language as being closely tied to place: traditional
dialectology also took a similar approach, by plotting forms of speech on actual
maps and showing how they differed one from another. That the nation-state
is very much the level of analysis in World Englishes has probably more to
do with a concern for convenience; after all, what better way to describe the
forms of English found in New Zealand, Singapore, and Ghana, for instance,
than as New Zealand English, Singapore English, and Ghanaian English? It
makes immediate sense.
However, traditional dialectology has also shown that it can be quite hard
to apply labels such as these to individual manifestations of language. The
dialect continuum is the clearest way of showing that language changes grad-
ually from one geographical locale to the next, as illustrated in Figure 6.2: in
a given geographical area, here abstracted to a one-dimensional line, every
related topolect (i.e. a variety spoken in a given location, represented by
a ) is mutually intelligible with and differs minimally from its immediate
neighbours. With increasing (Euclidean) distance, differences between va-
rieties increase, whereas mutual intelligibility decreases, such that while a
speaker of d 1 may still understand a speaker of d 7 , this may not be the case
for a speaker of d n . An interesting case is that of two adjacent locales and
their topolects, which find themselves, by some language-external process,
on opposite sides of a language border, as in d 7 and d 8 in Figure 6.2. The
classic example is that of the continental West Germanic topolects, which
can be spread on such an imaginary line stretching from the coast of the
Netherlands to the eastern part of Austria: with the emergence of European
nation-states, borders such as those between the Netherlands and Germany
became important psychological barriers, also as far as language planning was
involved, with the result that any topolect within the geopolitical construct
border
d1 d7 d8 dn
Figure 6.2. Dialect continuum
‘Germany’ was now called a ‘German dialect’, whereas those on the other
side of the border were called ‘Dutch dialects’, without changing their actual
linguistic form.
The dialect continuum serves to highlight the fluidity of linguistic varia-
tion in geographical space, coupled with divisions such as political borders,
which may be arbitrary from a linguistic perspective, but nonetheless have
an impact on the naming conventions of the varieties affected. That alone
might be enough to question geographic labels in World Englishes. What the
continuum does not deal with, however, is the sociolinguistic dimension of
language variation. A sociolinguistic approach to language, in any location,
deals with the variation found between different categories of speakers and
between different registers and speech forms used by the same speaker, and
is interested in the rules governing the choice of what forms are used by
whom under what conditions. From a sociolinguistic perspective, there is
no uniform ‘variety’, since even single speakers vary in their linguistic per-
formance depending on situational context, interlocutor, level of education,
peer group, etc. As a result, saying that someone speaks ‘British English’ is
vague at best: their speech is probably non-rhotic (unless they hail from the
Southwest or from Scotland), they might use pro-predicate do,2 and they
will favour lift over elevator (and favour over favor). But beyond such
generalities, any number of features might be present, and zooming in to a
lower geographical level of analysis (for instance England, West Midlands,
Birmingham), while it adds precision (i.e. /U/ in cup, a short /æ/ in path),
does not give a full picture of the speaker’s actual performance: do they
have glottal replacement or h-dropping, and do their sentences end with
high rise terminals? Are there features of non-standard grammar, such as
negative concord, and how much lexical admixture from non-English lan-
guages does their speech feature? All these are features that are relevant in
any description of a variety, and to leave them out is to miss some crucial
information about the speaker’s performance. Of course, it is defensible to
list the features all (or a majority) of speakers in a particular geographical
locale have in common. The problem with this approach is that some of
these features will inevitably also be found outside this locale – think of
2
As in the following exchange:
(i) A: Are you coming to the party tonight?
B: I might do.
/U/ in cup in Birmingham English (‘Brummie’), a feature shared with most

accents in the North of England, or short /æ/, which is also used in most of
North America.
The alternative, then, can be found in a closer analysis of idiolectal speech,
an analysis which departs from the generalisations needed in a varietal ap-
proach. The interaction of geographical features with social features, as well
as their social meanings in production, can provide a more complete pic-
ture of the language being used in situations such as these. Included in
this are elements from ‘varieties’ or codes not traditionally associated with
the geographical locale where the speaker happens to live, but which are
nonetheless part of their repertoire. The importance of this latter point is
evident in a place like Singapore, where English has been in intense contact
with languages that are themselves non-local in origin (with the exception of
Malay), but it is equally valid elsewhere, as few places are completely void
of external linguistic influences. Especially in the field of World Englishes,
as the cases of Singapore and Southeast Asia show, the interaction between
English and other languages is important and the focus of much scholarly
attention. What is less obviously taken into account, however, is the fact that
these interactions cross borders and are to be found in similar shape out-
side the geographical unit that labels the English under investigation. If the
brief discussion of Singapore and Malaysian English, above, has anything to
offer, it is this argument for a more integrated approach, where geography,
linguistics, and sociolinguistics all have their role to play.
6.4 Conclusion
This final chapter has made the case for a departure from the often ill-
defined concept of the ‘variety’ as the basic structural unit of analysis in the
field of World Englishes. That the ‘variety’ is becoming increasingly less
appropriate for sociolinguistic analyses in an increasingly globalised world,
and especially so in super-diverse (Vertovec 2007) urban settings, has been
argued before (Seargeant and Tagg 2011). The alternative is an approach
taking into account individual variation, concepts such as truncated reper-
toires (Blommaert 2010), and a move away from distinct ‘varieties’ and the
purely linguistic features associated with them, towards linguistic resources,
which may originate from any imaginable language or variety, but which are,
in production, used for their social meaning. Such an integrated approach,
combining traditional World Englishes methodology with contemporary so-
ciolinguistic theories of indexicality, has the potential to more satisfactorily
explain the variation observed in contact situations such as the ones found in
Singapore and Southeast Asia in general.
Appendix A: Chinese romanisation
A.1 Mandarin: Pı̄nyı̄n

Consonants Vowels
Pı̄nyı̄n IPA Pı̄nyı̄n IPA
b p a a, E3
c tsh e 7, @
ch tùh i i, 14
d t o o5
f f u u, y6
g k ü y
h x
j tC
k kh
l l
m m
n n
ng N Tones
p ph
q tCh Pı̄nyı̄n IPA
r õ1
s s Ă
ā £ (55, high level)
sh ù á Ę£ (35, rising)
t th ǎ ŁŘ£ (214, falling–rising)
w w, ∅2 à Ď£ (51, falling)
x C
y j, ∅2
z ts
zh tù
Notes
1
r varies between [õ] and [ü].
2
w and y are the approximants /w/ and /j/ except in the sequences wu and
yi, where they are /∅/, and in the syllable yu, which is /y/. Before u, y
is generally [4].
3
a is /E/ when preceded by a /j/ and followed by /n/; thus yan = /jEn/,
yuan = /4En/, tian = /th jEn/.
4
i is /1/ after c, ch, r, s, sh, z, zh.
5
o is /uo/ after b, p, m, f.
6
u is /y/ after j, q, x, y.
127
128 Chinese romanisation
A.2 Cantonese: Jyut6ping3

Consonants Vowels
Jyut6ping3 IPA Jyut6ping3 IPA
b p a 5
c tsh aa a
d t, t^ in syllable coda e e
f f eo 8
g k i i, y2
gw kw o o
h h oe œ
j j1 u u
k kh , k^ in syllable coda yu y
kw kwh
l l
m m
n n
ng N
p ph , p^ in syllable coda
s s
t th
w w
z ts
Tones
Jyut6ping3 IPA
Ă
1 £ (55, high level) or Ć£ (53, high falling)
2 Ę£ (35, mid rising) or Ğ£ (25, low rising)
3 Ă£ (33, mid level)
4 Ą£ (21, low falling) or Ă£ (11, low level)
5 Ę£ (13, low rising) or Ě£ (23, low rising)
6 Ă£ (22, low level)
Ă
1 (7) £ (5, high level)
3 (8) Ă£ (3, mid level) in syllables ending in /p, t, k/
6 (9) Ă£ (2, low level)
Notes
1
j is /∅/ when preceding yu.
2
i is /y/ in the sequence eoi (/8y/).
A.3 Hokkien: PĲeh-ōe-jı̄ 129
A.3 Hokkien: PĲeh-ōe-jı̄

Consonants Vowels
PĲeh-ōe-jı̄ IPA PĲeh-ōe-jı̄ IPA
b b a a1
ch tC e e
chh tCh i i
h h, P in syllable coda o2 o
j dz, ý o. O
k k, k^ in syllable coda u u
kh kh , k^ in syllable coda
n
nasalisation3
l l, d
m m¯
n n Tones
ng N
p p, p^ in syllable coda
ph ph PĲeh-ōe-jı̄ IPA
s s, C
Ă
th th a £ (55, high level)4
t t, t^ in syllable coda á Ď£ (51, high falling)
à Ę£ (35, mid rising)
a Ą£ (35, mid falling)4
â Ę£ (24, low rising)
ā Ă£ (33, mid level)
Ă
Ĳa £ (44, high level)
Notes
1
a is /E/ when between /i/ and /n/.
2
o is /u/ in diphthongs and triphthongs.
3 n
indicates optional nasalisation, e.g. an /a/ or /ã/. All six vowels can be
nasalised.
4
The two unmarked tones can be distinguished by the syllable coda: when it is
a plosive, the syllable carries a high level tone, if it is a nasal, it carries a mid
falling tone.
Appendix B: Informants
In each school a group of four Chinese, four Malay, and four Indian students
were selected. One group was usually recorded per day. Each informant was inter-
viewed individually first, followed by a researcher-led discussion in pairs, then by
a task-based, unsupervised group recording, and finally by an unsupervised radio-
microphone recording in a location deemed as informal, in each case the school
canteen (see Table B.1).
The time allocated for each recording was fifteen minutes. The total time of all
72 recordings was 16h32m12s, subsequently transcribed into 110,422 words.
Table B.1. Number of recordings, by school and ethnic group
Type of recording Per group Per school Total
Individual interview 4 12 36
Dialogue interview 2 6 18
Group recording 1 3 9
Radio-microphone recordings 1 3 9
Informants were of an average age of 17.5 years, with twenty females and sixteen
males. Due to the structure of the interviews and the sampling procedure, they
came in equal numbers from the three main ethnic groups (twelve Chinese, twelve
Malays, and twelve Indians), as well as from three schools, taken to represent educa-
tional achievement (junior college (highest), polytechnic (mid-range), and vocational
training school (lowest)). Table B.2 lists informants and gives their identifier (used
throughout this book to refer to them), their school, ethnic, and gender affiliation,
their date of birth, the occupations of their parents, and the interview date.
130
Table B.2. Informants and their demographic and socio-economic background
Identifier School Ethnicity Sex Date of birth Father’s occupation Mother’s occupation Interview
i.C.1.m JC Chinese M 01.09.1990 Driver Manager 01.08.2007

i.C.2.f JC Chinese F 25.06.1990 Computer analyst Housewife 01.08.2007
i.C.3.f JC Chinese F 08.12.1990 Chemical engineer Tax officer 01.08.2007
i.C.4.m JC Chinese M 02.03.1990 Teacher Teacher 01.08.2007
i.M.1.f JC Malay1 F 23.04.1989 Insurance agent Administrator 25.07.2007
i.M.2.f JC Malay2 F 28.06.1990 Unemployed Librarian 25.07.2007
i.M.3.m JC Malay M 17.05.1990 Dance instructor Admin officer 25.07.2007
i.M.4.m JC Malay3 M 21.11.1990 Driver Housewife 25.07.2007
i.I.1.f JC Indian4 F 12.05.1990 Engineer Clerk 25.07.2007
i.I.2.f JC Indian5 F 23.07.1990 Civil engineer Accountant 25.07.2007
i.I.3.m JC Indian4 M 17.03.1990 Unemployed Clerk 25.07.2007
i.I.4.m JC Indian4 M 21.11.1990 Lecturer Clerk 25.07.2007
ii.C.1.f Poly Chinese F 09.12.1988 Taxi driver Secretary 07.11.2006
ii.C.2.m Poly Chinese M 08.07.1989 Taxi driver Bus driver (part-time) 07.11.2006
ii.C.3.f Poly Chinese F 20.04.1989 Taxi driver Draughtsman 07.11.2006
ii.C.4.m Poly Chinese M 20.06.1989 Teacher Homemaker 07.11.2006
ii.M.1.m Poly Malay M 26.06.1987 Retiree Housewife 08.11.2006
ii.M.2.f Poly Malay F 14.02.1989 Self-employed Housewife 08.11.2006
ii.M.3.f Poly Malay 6 F 03.11.1989 Postman Housewife 08.11.2006
(cont.)
Table B.2. (cont.)
Identifier School Ethnicity Sex Date of birth Father’s occupation Mother’s occupation Interview
ii.M.4.m Poly Malay M 12.12.1989 Deceased Clerk 08.11.2006

ii.I.1.m Poly Indian7 M 11.07.1988 Security guard University lecturer 10.11.2006
ii.I.2.f Poly Indian4 F 19.10.1988 Cashier Salesperson 10.11.2006
ii.I.3.f Poly Indian8 F 22.09.1988 Engineer Nurse 10.11.2006
ii.I.4.f Poly Indian4 F 04.10.1989 Manager Bank executive 10.11.2006
iii.C.1.m ITE Chinese M 13.12.1988 Restaurant manager Clerk 20.09.2006
iii.C.2.f ITE Chinese F 16.02.1987 Odd jobs Housewife 20.09.2006
iii.C.3.m ITE Chinese M 11.03.1987 Electrical officer Nurse 20.09.2006
iii.C.4.m ITE Chinese M 22.02.1988 Mechanic Housewife 20.09.2006
iii.M.1.f ITE Malay F 06.12.1988 Equipment operator Medical devices service specialist 21.09.2006
iii.M.2.f ITE Malay F 10.08.1989 Senior technician Housewife 21.09.2006
iii.M.3.m ITE Malay6 M 29.05.1989 Unemployed Housewife 21.09.2006
iii.M.4.m ITE Malay M 16.08.1988 Shipyard technician Cleaner 21.09.2006
iii.I.1.f ITE Indian4 F 26.11.1987 Delivery driver Massage therapist 31.10.2006
iii.I.2.f ITE Indian4 F 03.11.1988 Equipment controller Housewife 31.10.2006
iii.I.3.f ITE Indian4 F 15.03.1988 Taxi driver Housewife 31.10.2006
iii.I.4.f ITE Indian4 F 29.10.1987 Customer service officer Accountant 31.10.2006
Notes: 1 Mother Javanese, 2 Mother Chinese, 3 Boyanese, 4 Tamil, 5 Telugu, 6 Javanese, 7 Punjabi, 8 Malayalee.
Appendix C: Chinese languages
S INO -T IBETAN
T IBETO -B URMAN Southern
C HINESE Hakka
Northern Yue
Mandarin Guangfu
Northeastern HK Cantonese
Zhongyuan Siyi
Beijing Gaoyang
Std Mandarin Goulou
Lanyin Wuhua
Jilu Min
Southwestern Nan (Southern)
Jiaoliao Xiamen
Jinghuai Amoy
Central Hokkien
Wu Chaozhou (Teochew)
Taihu Hainan
Wuzhou Bei (Northern)
Shanghainese Dong (Eastern)
Xiang Zhong (Central)
Gan Pu-Xian
Figure C.1. Phylogenetic tree of Sino-Tibetan. The main subdivisions

are from Norman (1988, cited in Thurgood and LaPolla 2003) and the
subgroups from Lewis (2009), except for Yue (LAC 1988). The tree is
incomplete, concentrating on the varieties relevant for this book.
133
References
Abdulaziz Mkilifi, Mohammed H. 1978. Triglossia and Swahili-English bilingualism

in Tanzania. In Fishman, Joshua A. (ed.), Advances in the Study of Societal
Multilingualism, 129–52. The Hague: Mouton.
Abu Baker, Jalelah. 2009. One generation – that’s all it takes ‘for a language to
die’. The Straits Times, 8 March 2009, Letters. www.asiaone.com/News/
Education/Story/A1Story20090306-126699.html. Accessed 22 January 2010.
Aijmer, Karin, and Simon-Vandenbergen, Anne-Marie. 2006. Pragmatic Markers in
Contrast. Oxford: Elsevier.
Aitchison, Jean. 2001. Language Change: Progress or Decay? 3rd edn. Cambridge
University Press.
Alleyne, M. 1980. Comparative Afro-American. Ann Arbor: Karoma Press.
Alsagoff, Lubna. 2007. Singlish: negotiating culture, capital and identity. In Vaish,
Viniti, Gopinathan, Saravanan, and Yongbing, Liu (eds.), Language, Capital,
Culture, 25–46. Rotterdam: Sense Publishers.
2010. English in Singapore: culture, capital and identity in linguistic variation.
World Englishes, 29, 336–48.
Alsagoff, Lubna, and Ho Chee Lick. 1998. The grammar of Singapore English. In
Foley et al. 1998: 127–51.
Ansaldo, Umberto. 2004. The evolution of Singapore English. In Lim 2004: 127–49.
Ansaldo, Umberto, and Matthews, Stephen J. 1999. The Minnan substrate in Baba
Malay. Journal of of Chinese Linguistics, 27, 38–68.
Bakhtin, Mikhail M. 1981. Unitary language. In Holquist, Michael, and Emerson,
Caryl (eds.), The Dialogic Imagination: Four Essays, 269–95. Austin: University
of Texas Press.
Bao, Zhiming. 1995. Already in Singapore English. World Englishes, 14, 181–8.
1998. The sounds of Singapore English. In Foley et al. 1998: 152–74.
2001. The origins of empty categories in Singapore English. Journal of Pidgin and
Creole Languages, 16, 275–319.
2005. The aspectual system of Singapore English and the systemic substratist
explanation. Journal of Linguistics, 41, 237–67.
2009. One in Singapore English. Studies in Language, 33, 338–65.
Bao, Zhiming, and Hong, Huaqing. 2007. Diglossia and register variation in Singa-
pore English. World Englishes, 25, 105–14.
Bao, Zhiming, and Wee, Lionel. 1998. Until in Singapore English. World Englishes,
17, 31–41.
134
References 135
Baskaran, Loga. 2005. A Malaysian English Primer: Aspects of Malaysian English

Features. Kuala Lumpur: University of Malaya Press.
2008a. Malaysian English: morphology and syntax. In Mesthrie 2008: 610–23.
2008b. Malaysian English: phonology. In Mesthrie 2008: 278–91.
Baskaran, Lohanayahi. 1988. Aspects of Malaysian English syntax. Ph.D. thesis,
Department of Phonetics and Linguistics, University College London.
Baxter, Alan N. 1988. A Grammar of Kristang. Canberra: Australian National
University.
Beeching, Kate. 2001. Repair strategies and social interaction in spontaneous spoken
French: the pragmatic particle enfin. French Language Studies, 11, 23–40.
2002. Gender, Politeness and Pragmatic Particles in French. Amsterdam: John Ben-
jamins.
Bickerton, Derek. 1973. On the nature of a creole continuum. Language, 49,
640–69.
Blommaert, Jan. 2007. Sociolinguistics and discourse analysis: orders of indexicality
and polycentricity. Journal of Multicultural Discourses, 2, 115–30.
2010. The Sociolinguistics of Globalisation. Cambridge University Press.
Bokhorst-Heng, Wendy. 1998. Language planning and management in Singapore.
In Foley et al. 1998: 287–309.
Bokhorst-Heng, Wendy, Alsagoff, Lubna, McKay, Sandra, and Rubdy, Rani. 2007.
English language ownership among Singaporean Malays: going beyond the
NS/NNS dichotomy. World Englishes, 26, 424–45.
Bourdieu, Pierre. 1981. Questions de sociologie. Paris: Les éditions de minuit.
1986. The forms of capital. In Richardson, John G. (ed.), Handbook of Theory and
Research for the Sociology of Education, 241–58. New York: Greenwood Press.
2001. Langage et pouvoir symbolique. Paris: Éditions Fayard.
Brown, Adam. 1999. Singapore English in a Nutshell: An Alphabetic Description of Its
Features. Singapore: Federal Publications.
Brown, Adam, Deterding, David, and Ling, Low Ee (eds.). 2000. The English Lan-
guage in Singapore: Research on Pronunciation. Singapore Association for Applied
Linguistics.
Brown, Charles Cuthbert. 1983. Sĕjarah Mĕlayu or Malay Annals: An Annotated
Translation. Kuala Lumpur: Oxford University Press.
Brown, Roger, and Gilman, Albert. 1960. The pronouns of power and solidarity. In
Sebeok, Thomas A. (ed.), Style in Language, 253–76. Cambridge, MA: MIT
Press.
Bruthiaux, Paul. 2003. Squaring the circles: issues in modelling English Worldwide.
International Journal of Applied Linguistics, 13, 159–78.
Bucholtz, Mary, and Hall, Kira. 2005. Identity and interaction: a sociocultural lin-
guistic approach. Discourse Studies, 7, 585–614.
Canagarajah, A. Suresh. 1999. Resisting Linguistic Imperialism in English Teaching.
Oxford University Press.
Chambers, Jack K. 2003. Sociolinguistic Theory. 2nd edn. Oxford: Blackwell.
Chang, Rachel, and Hussain, Zakir. 2010. S’poreans split on Mother Tongue weight-
ing. The Straits Times, 11 May 2010, Education. www.asiaone.com/News/
Education/Story/A1Story20100511-215458.html. Accessed 29 October 2010.
Chao, Yuen Ren. 1930. @ sistim @v ‘toun-let@z’. l@ mE:tr@ fOnetik, 30, 24–7.
136 References
Chaudenson, Robert, and Carayol, Michel. 1979. Essai d’analyse implicationnelle

d’un continuum linguistique. In Wald, P., and Manessy, G. (eds.), Plurilin-
guisme: normes, situations, stratégies, 129–72. Paris: Harmattan.
Cheng, Chin-Chuan. 1996. Quantifying dialect mutual intelligibility. In James,
Huang Cheng-Teh, and Audrey, Li Yen-Hui (eds.), New Horizons in Chinese
Linguistics, 269–72. Dordrecht: Kluwer Academic Publishers.
Chew, Phyllis Ghim-Lian. 1995. Lectal power in Singapore English. World Englishes,
14, 163–80.
Chiang, Ming Yu. 2009. Shame to let dialects die out in Singapore. The Straits
Times, 16 March 2009, Letters. www.asiaone.com/News/Education/Story/
A1Story20090316-128717.html. Accessed 22 January 2010.
Coupland, Nikolas. 2007. Style: Language Variation and Identity. Cambridge Uni-
versity Press.
Crystal, David. 1995. The Cambridge Encyclopedia of the English Language. Cambridge
University Press.
2006. English worldwide. In Hogg, Richard, and Denison, David (eds.), A History
of the English Language, 420–39. Cambridge University Press.
Davie, Sandra. 2010. Mother tongue weighting in PSLE could be cut. The Straits
Times, 23 April 2010, Education. www.asiaone.com/News/Education/Story/
A1Story20100421-211605.html. Accessed 29 October 2010.
DeCamp, David. 1971. Towards a generative analysis of a post-creole speech contin-
uum. In Hymes, Dell (ed.), Pidginisation and Creolisation of Languages, 349–70.
Cambridge University Press.
DeFrancis, John. 1986. The Chinese Language: Fact and Fantasy. Honolulu: Univer-
sity of Hawai‘i Press.
Deterding, David. 2003a. An instrumental study of the monophthong vowels of
Singapore English. English World-Wide, 24, 1–16.
2003b. Tenses and will/would in a corpus of Singapore English. In Deterding
et al. 2003: 31–8.
2005. Listening to Estuary English in Singapore. TESOL Quarterly, 39, 425–40.
2007a. Singapore English. Edinburgh University Press.
2007b. The vowels of the different ethnic groups in Singapore. In Prescott, David
(ed.), English in Southeast Asia: Varieties, Literacies and Literatures, 2–29. New-
castle upon Tyne: Cambridge Scholars Publishing.
2010. Norms for pronunciation in Southeast Asia. World Englishes, 29, 364–77.
Deterding, David, and Kirkpatrick, Andy. 2006. Emerging South-East Asian En-
glishes and intelligibility. World Englishes, 25, 391–409.
Deterding, David, and Poedjosoedarmo, Gloria R. 2000. To what extent can the
ethnic group of young Singaporeans be identified from their speech? In Brown
et al. 2000: 1–9.
Deterding, David, Ling, Low Ee, and Brown, Adam (eds.). 2003. English in Singa-
pore: Research on Grammar. Singapore: McGraw-Hill.
Dixon, Quentin L. 2005. The bilingual education policy in Singapore: implications
for second language acquisition. In 4th International Symposium on Bilingualism,
625–35. Somerville, MA: Cascadilla Press.
Doraisamy, Theodore C. 1969. 150 Years of Education in Singapore. Singapore:
Teachers’ Training College.
References 137
Eckert, Penelope. 1989. Jocks and Burnouts: Social Categories and Identity in the High
School. New York: Teachers College Press.
2000. Linguistic Variation as Social Practice. Malden, MA: Blackwell.
2005. Variation, convention, and social meaning. Paper presented at the Annual
Meeting of the Linguistics Society of America, Oakland, CA, 7 January 2005.
www.stanford.edu/∼eckert/EckertLSA2005.pdf. Accessed 18 June 2009.
2008. Variation and the indexical field. Journal of Sociolinguistics, 12, 453–76.
Eckert, Penelope, and McConnell-Ginet, Sally. 1992. Think practically and look
locally: language and gender as community-based practice. Annual Review of
Anthropology, 21, 461–90.
Erb, Maribeth. 2003. Moulding a nation: education in early Singapore. In Bun,
Chan Kwok, and Kiong, Tong Chee (eds.), Past Times: A Social History of
Singapore, 17–32. Singapore: Times.
Fasold, Ralph W. 1984. The Sociolinguistics of Society. Oxford: Blackwell.
Ferguson, Charles A. 1959. Diglossia. Word, 15, 325–40.
Ferguson, Will, and Ferguson, Ian. 2003. How to be a Canadian. Vancouver: Douglas
& McIntyre.
Fernández, Mauro. 1993. Diglossia: A Comprehensive Bibliography 1960–1990.
Amsterdam: John Benjamins.
Fischer, Kerstin (ed.). 2006a. Approaches to Discourse Particles. Oxford: Elsevier.
2006b. Towards an understanding of the spectrum of approaches to discourse
particles: introduction to the volume. In Fischer 2006a: 1–20.
Fishman, Joshua A. 1967. Bilingualism with and without diglossia; diglossia with
and without bilingualism. Journal of Social Issues, 23, 29–38.
1972. Sociolinguistics: A Brief Introduction. Rowley, MA: Newbury House.
1980. Prefatory notes. In Nelde, Peter Hans (ed.), Languages in Contact and Con-
flict, xi. Wiesbaden: Franz Steiner.
Foley, Joseph A. 1998. The new Englishes: language in the home. In Foley et al.
1998: 218–43.
Foley, Joseph A., Kandiah, Thiru, Bao, Zhiming, Gupta, Anthea Fraser, Alsagoff,
Lubna, Ho, Chee Lick, Wee, Lionel, Talib, Ismail L., and Bokhorst-Heng,
Wendy (eds.). 1998. English in New Cultural Contexts: Reflections from Singapore.
Singapore: Oxford University Press.
Frangoudaki, Anna. 1992. Diglossia and the present language situation in Greece: a
sociological approach to the interpretation of diglossia and some hypotheses on
today’s linguistic reality. Language in Society, 21, 365–81.
Giles, Howard, Taylor, Donald M., and Bourhis, Richard. 1973. Towards a theory of
interpersonal accommodation through language: some Canadian data. Language
in Society, 2, 177–92.
Goddard, Cliff. 2002. The search for the shared semantic core of all languages. In
Goddard, Cliff, and Wierzbicka, Anna (eds.), Meaning and Universal Grammar:
Theory and Empirical Findings, vol. 1: 5–40. Amsterdam: John Benjamins.
Goh, Hwee. 2010. Mother Tongue weighting for PSLE to remain. ChannelNews Asia,
11 May 2010, Home. www.channelnewsasia.com/stories/singaporelocalnews/
view/1055797/1/.html. Accessed 29 October 2010.
Gumperz, John. 1964. Linguistic and social interaction in two communities. American
Anthropologist, 66, 137–53.
138 References
Gupta, Anthea Fraser. 1986. A standard for written Singapore English? English
World-Wide, 7, 75–99.
1989. Singapore Colloquial English and Standard English. Singapore Journal of
Education, 10, 33–9.
1992. The pragmatic particles of Singapore Colloquial English. Journal of Prag-
matics, 18, 31–57.
1994. The Step-Tongue: Children’s English in Singapore. Clevedon: Multilingual
Matters.
1998. The situation of English in Singapore. In Foley et al. 1998: 106–26.
2001. English in the linguistic ecology of Singapore. In The Cultural Politics of
English as a World Language. Freiburg i.B., Germany: GNEL/MALVEN.
2006a. Epistemic modalities and the discourse particles of Singapore. In Fischer
2006a: 244–63.
2006b. Singlish on the web. In Hashim, Azirah, and Hassan, Norizah (eds.), Vari-
eties of English in SouthEast Asia and Beyond, 19–37. Kuala Lumpur: University
of Malaya Press.
Heller, Monica. 2006. Linguistic Minorities and Modernity. 2nd edn. London:
Continuum.
Ho, Mian-Lian, and Platt, John T. 1993. Dynamics of a Contact Continuum: Singa-
porean English. Oxford University Press.
Hu, Jianhua, and Pan, Haihua. 2007. Focus and the basic function of Chinese
existential you-sentences. In Comorovski, Ileana, and von Heusinger, Klaus
(eds.), Existence: Semantics and Syntax, 133–45. Dordrecht: Springer.
Hussain, Zakir. 2006. How to use ‘aiyah’? Look it up online, lah. The Straits Times,
10 February 2006, 1–2.
IMF, International Monetary Fund. 2011. World Economic Outlook Database,
April 2011 Edition. Website. www.imf.org/external/pubs/ft/weo/2011/01/
weodata/index.aspx. Accessed 5 February 2012.
Johnstone, Barbara, and Kiesling, Scott F. 2008. Indexicality and experience: ex-
ploring the meanings of /aw/-monophthongisation in Pittsburgh. Journal of
Sociolinguistics, 12, 5–33.
Kachru, Braj B. 1982. The Other Tongue: English across Cultures. Urbana: University
of Illinois Press.
1985. Standards, codification and the sociolinguistic realism: the English language
in the Outer Circle. In Quirk, Randolph, and Widdowson, Henry G. (eds.),
English in the World: Teaching and Learning the Language and Literatures, 11–
30. New York: Cambridge University Press.
(ed.). 1992. The Other Tongue: English across Cultures, 2nd edn. Urbana: University
of Illinois Press.
Kazazis, Kostas. 1993. Dismantling Greek diglossia. In Fraenkel, Eran, and Kramer,
Christian (eds.), Language Contact – Language Conflict, 7–26. New York: Peter
Lang.
Kerswill, Paul. 1996. Children, adolescents and language change. Language Variation
and Change, 8, 177–202.
Kim, Chonghyuck, and Wee, Lionel. 2009. Resolving the paradox of Singapore
English hor. English World-Wide, 30, 241–61.
Kirkpatrick, Andy. 2007. World Englishes: Implications for International Communica-
tion and English Language Teaching. Cambridge University Press.
References 139
2010. English as a Lingua Franca in ASEAN: A Multilingual Model. Hong Kong

University Press.
Kor, Kian Beng. 2010. Race option for kids. The Straits Times, 13 January 2010, A1,
A6.
Kortmann, Bernd, and Schneider, Edgar W. 2008. Varieties of English. Berlin: Mou-
ton de Gruyter.
Kwan-Terry, Anna. 1991. Through the looking glass: a child’s use of particles in
Chinese and English and its implications on language transfer. In Kwan-Terry,
Anna (ed.), Child Language Development in Singapore and Malaysia, 161–83.
Singapore University Press.
Labov, William. 1963. The social motivation of a sound change. Word, 18, 1–42.
1966. The Social Stratification of English in New York City. Washington, DC:
Centre for Applied Linguistics.
1969. Contraction, deletion and inherent variability of the English copula. Lan-
guage, 45, 715–62.
1982. Objectivity and commitment in linguistics science: the case of the Black
English trial in Ann Arbor. Language in Society, 11, 165–202.
LAC, Australian Academy of the Humanities and the Chinese Academy of Social
Sciences. 1988. Language Atlas of China. Hong Kong: Longman.
Lau, Cheow. 2004. US to Send Task Force to Learn from SAF. Website. www.
talkingcock.com/html/article.php?sid=1336. Accessed 9 January 2011.
Lau, Kak En (ed.). 1993. Singapore Census of Population, 1990. Singapore: SNP
Publishers.
Le Page, Robert B., and Tabouret-Keller, Andrée. 1985. Acts of Identity: Creole-based
Approaches to Language and Ethnicity. Cambridge University Press.
Lee, Jack Tsen-Ta. 2004. A Dictionary of Singlish and Singapore English. Website.
www.singlishdictionary.com. Accessed 27 February 2010.
Lee, Nala Huiying, Ling, Ai Ping, and Nomoto, Hiroki. 2009. Colloquial
Singapore English got: functions and substratal influences. World Englishes, 28,
293–318.
Leimgruber, Jakob R. E. 2005. Interviewer influence and linguistic variation in the
English of twenty younger educated Singaporeans. Lic. Phil. thesis, Department
of English, University of Fribourg, Switzerland.
2009a. Ethnicity as a variable in Singapore English. Paper presented at the 3rd
International Conference on the Linguistics of Contemporary English, Institute
of English Studies, University of London, 15–17 July 2009.
2009b. Modelling variation in Singapore English. D.Phil. thesis, Faculty of
English, University of Oxford.
2012a. Singapore English. Language and Linguistics Compass, 5, 47–62.
2012b. Variation in Singapore English: the case for an indexical approach. World
Englishes, 31, 1–14.
2013. The management of multilingualism in a city-state: language policy in
Singapore. In Siemund, Peter, Gogolin, Ingrid, Schulz, Monika, and Davy-
dova, Julia (eds.), Multilingualism and Language Contact in Urban Areas: Ac-
quisition, Development, Teaching, Communication, 229–57. Amsterdam: John
Benjamins.
Leow, Bee Geok (ed.). 2001. Census of Population 2000. Singapore: Department of
Statistics.
140 References
Ler, Soon Lay Vivien. 2006. A relevance-theoretic approach to discourse particles

in Singapore English. In Fischer 2006a: 149–66.
Lewis, M. Paul (ed.). 2009. Ethnologue: Languages of the World. 16th edn. Dallas:
SIL International.
Lim, Lisa S. 2000. Ethnic group differences aligned? Intonation patterns of Chinese,
Indian and Malay Singaporean English. In Brown et al. 2000: 10–21.
(ed.). 2004. Singapore English: A Grammatical Approach. Amsterdam: John
Benjamins.
2007. Mergers and acquisition: on the ages and origins of Singapore English
particles. World Englishes, 26, 446–73.
2010. Peranakan English in Singapore. In Schreier et al. 2010: 327–47.
Lim, Lisa S., and Gisborne, Nikolas (eds.). 2009. The Typology of Asian Englishes.
Special issue of English World-Wide, 30.
Lim, Sonny. 1975. Baba Malay: the language of the ‘Straits-born’ Chinese. MA
thesis, Department of Linguistics, Monash University.
Low, Ee Ling, and Brown, Adam. 2005. English in Singapore: An Introduction.
Singapore: McGraw-Hill.
Mackey, William F. 1993. History of diglossia. In Fernández, Mauro (ed.), Diglossia.
A Comprehensive Bibliography 1960–1990, xiii–xx. Amsterdam: John Benjamins.
Mair, Victor H. 1991. What is a Chinese ‘dialect/topolect’? Reflections on some key
Sino-English linguistic terms. Sino-Platonic Papers, 29, 1–31.
Marçais, William. 1930. La diglossie arabe. L’Enseignement Public, 97, 20–39.
Matthews, Peter H. 2007. Oxford Concise Dictionary of Linguistics. 2nd edn. Oxford
University Press.
McArthur, Tom. 2003. English as an Asian language. English Today, 19, 19–22.
MDA, Media Development Authority. 2004. Free-to-air television programme code.
PDF document. www.mda.gov.sg/Documents/PDF/industry/Industry TV
ContentGuidelines FTATVProgCode.pdf. Accessed 7 March 2011.
Mesthrie, Rajend (ed.). 2008. Varieties of English: Africa, South and Southeast Asia.
Berlin: Mouton de Gruyter.
MITA, Ministry of Information, Communication and the Arts. 2003. Keynote
address by Deputy Prime Minister, Mr Lee Hsien Loon at the Network
Conference 2003. Online government press release. http://web.archive.org/
web/20120207063549/http://stars.nhb.gov.sg/stars/public/viewHTML.
jsp?pdfno=2003050301. Accessed 18 July 2008.
MOE, Ministry of Education. 1995. Press release 02995: Performance indicators of
Junior Colleges 1995. Singapore: Ministry of Education.
2006. Education statistics digest. PDF document. http://web.archive.org/web/
20061201011010/www.moe.gov.sg/esd/ESD2006.pdf. Accessed 2 February
2007.
2008. Enhanced programmes by Special Assistance Plan schools to enrich stu-
dents’ learning of Chinese language and values. Press release. www.moe.gov.
sg/media/press/2008/02/enhanced-programmes-by-special.php. Accessed
25 October 2010.
2009. Compulsory Education. Website. www.moe.gov.sg/initiatives/compulsory-
education/. Accessed 1 September 2009.
References 141
2010a. 2010 Primary One registration exercise: application form. PDF

document. http://web.archive.org/web/20100926214730/http://moe.gov.
sg/education/admissions/primary-one-registration/files/p1-phase1.pdf. Ac-
cessed 29 October 2010.
2010b. Returning Singaporeans: Mother-Tongue Language Policy. Website.
www.moe.gov.sg/education/admissions/returning-singaporeans/mother-
tongue-policy/. Accessed 29 October 2010.
MOH, Ministry of Health. 2005a. Flu Pandemic Guide. Singapore: Ministry of
Information, Communications and the Arts.
2005b. Instructions to patients with confirmed or suspected dengue. PDF
document. http://web.archive.org/web/20060528003719/www.moh.gov.sg/
cmaweb/attachments/publication/3691dfe8b6Lu/Dengue patient advisory
- English version.pdf. Accessed 10 April 2006.
MrBrown.com. 2007. The magical shrinking cup of iced tea. Blog post, 6 Octo-
ber 2007. www.mrbrown.com/blog/2007/10/the-magical-shr.html. Accessed
30 November 2011.
Muysken, Pieter. 1995. Code-switching and grammatical theory. In Milroy,
Leslie, and Muysken, Pieter (eds.), One Speaker, Two Languages: Cross-
disciplinary Perspectives on Code-switching, 177–98. Cambridge University
Press.
Myers-Scotton, Carol. 1997. Duelling Languages: Grammatical Structure in
Codeswitching. 2nd edn. Oxford: Oxford University Press.
2000. Explaining the role of norms and rationality in codeswitching. Journal of
Pragmatics, 32, 1259–71.
Nair-Venugopal, Shanta. 2000. English, identity and the Malaysian workplace. World
Norman, Jerry. 1988. Chinese. Cambridge University Press.
Ochs, Elinor. 1992. Indexing gender. In Duranti, Alessandro, and Goodwin, Charles
(eds.), Rethinking Context: Language as an Interactive Phenomenon, 335–58.
Ochs, Elinor. 2004. Narrative lessons. In Duranti, Alessandro (ed.), A Companion to
Linguistic Anthropology, 269–89. London: Blackwell.
Pakir, Anne. 1986. A linguistic investigation of Baba Malay. Ph.D. thesis, University
of Hawai‘i.
1991. The range and depth of English-knowing bilinguals in Singapore. World
2001. Bilingual education with English as an official language: sociocultural im-
plications. In Alatis, James E., and Tan Ai-Hui (eds.), Georgetown University
Round Table on Language and Linguistics, 1999: Language in Our Time, 11–30.
Washington, DC: Georgetown University Press.
Parliament of Singapore. 1990. Official Report of Parliamentary Debates, 14 March
1990.
2007. Singapore Parliament Reports.
2010. Singapore Parliament Reports: Oral Answers to Questions, 12 January 2010.
Patrick, Peter L. 1999. Urban Jamaican Creole: Variation in the Mesolect. Amsterdam:
John Benjamins.
142 References
Pennycook, Alistair. 1994. The Cultural Politics of English as an International Lan-

guage. London: Longman.
2007. Global English and Transcultural Flows. Abingdon: Routledge.
Platt, John T. 1975. The Singapore English speech continuum and its basilect
‘Singlish’ as a ‘creoloid’. Anthropological Linguistics, 17, 363–74.
1977. A model for polyglossia and multilingualism (with special reference to
Singapore and Malaysia). Language in Society, 6, 361–78.
1987. Communicative functions of particles in Singapore English. In Steele, Ross,
and Threadgold, Terry (eds.), Language Topics: Essays in Honour of Michael
Halliday, 391–401. Amsterdam: John Benjamins.
Platt, John T., and Weber, Heidi. 1980. English in Singapore and Malaysia. Petaling
Jaya: Oxford University Press.
Platt, John T., Weber, Heidi, and Ho Chee Lick. 1983. Text Volume of Singapore and
Malaysian English. Amsterdam: John Benjamins.
Poedjosoedarmo, Gloria R. 1995. Lectal variation in the media and the classroom: a
preliminary analysis of attitudes. In Ching, Teng Su, and Mian-Lian, Ho (eds.),
The English Language in Singapore: Implications for Teaching, 53–67. Singapore:
Singapore Association for Applied Linguistics.
Poirier, Claude. 2008. Base de données lexicographiques panfrancophone. Website. www.
bdlp.org. Accessed 29 January 2013.
Prapañca, Mpu. 1995. Deśawarn.ana (Nāgarakr.tāgama). Leiden: Kroninklijk Insti-
tuut voor Taal-, Land- en Volkenkunde.
Psichari, Jean. 1885. Essais de grammaire néo-grecque. Annuaire pour
l’Encouragement des Études Grecques en France, 19, 1–288.
Pullum, Geoffrey K. 2007. Fifty ways to lose your lover: Edinburgh’s street names.
Language Log post. http://itre.cis.upenn.edu/∼myl/languagelog/archives/
004903.html. Accessed 25 July 2008.
Rajadurai, Joanne. 2007. Sociolinguistic perspectives on variation in non-native va-
rieties of English: the case of Malaysian English. Multilingua, 26, 409–26.
Ramesh, S. 2011. Choosing English as working language ensured S’pore’s survival:
Lee Kuan Yew. ChannelNews Asia, 6 September 2011, Singapore News.
www.channelnewsasia.com/stories/singaporelocalnews/view/1151407/1/.
html. Accessed 24 February 2012.
Rampton, Ben. 1995. Crossing: Language and Ethnicity among Adolescents. London:
Longman.
2006. Language in Late Modernity: Interaction in an Urban School. Cambridge
University Press.
2009. Interaction ritual and not just artful performance in crossing and stylisation.
Language in Society, 38, 149–76.
Ramsey, Robert S. 1987. The Languages of China. Princeton University Press.
Rappa, Antonio L. 2000. Surviving the politics of late modernity: the Eurasian
fringe community of Singapore. Southeast Asian Journal of Social Science, 28,
153–80.
Rappa, Antonio L., and Wee, Lionel. 2006. Language Policy and Modernity in
Southeast Asia: Malaysia, the Philippines, Singapore, and Thailand. New York:
Springer.
Rash, Felicity. 1998. The German Language in Switzerland: Multilingualism, Diglossia
and Variation. Bern: Peter Lang.
References 143
Rauniomaa, Mirka. 2003. Stance accretion. Paper presented at the Language, Interac-
tion, and Social Organization Research Focus Group, University of California,
Santa Barbara.
Rhoı̈dis, Emmanouēl D. 1885. [Title unknown]. Akropolis (Newspaper), 16–17 April
1885, 1056/1057.
Rickford, John R. 1977. The question of prior creolization in Black English. In
Valdman, Albert (ed.), Pidgin and Creole Linguistics, 190–221. Bloomington:
Indiana University Press.
1987. Dimensions of a Creole Continuum. Stanford University Press.
Roach, Peter. 2000. English Phonetics and Phonology: A Practical Course. 3rd edn.
2004. British English: Received Pronunciation. Journal of the International Pho-
netics Association, 34.2: 239–45.
Rubdy, Rani. 2001. Creative destruction: Singapore’s ‘Speak Good English Move-
ment’. World Englishes, 20, 341–55.
2007. Singlish in the school: an impediment or a resource? Journal of Multilingual
and Multicultural Development, 28, 308–24.
Savage, Victor R., and Yeoh, Brenda S. A. 2005. Toponymics: A Study of Singapore
Street Names. 2nd edn. Singapore: Eastern Universities Press.
Schneider, Edgar W. 2007. Postcolonial Englishes. Cambridge University Press.
Schreier, Daniel, Trudgill, Peter, and Schneider, Edgar W. (eds.). 2010. The Lesser-
Known Varieties of English: An Introduction. Cambridge University Press.
Seargeant, Philip, and Tagg, Caroline. 2011. English on the internet and a ‘post-
varieties’ approach to language. World Englishes, 30, 496–514.
Sebba, Mark. 1997. Contact Languages: Pidgins and Creoles. London: Macmillan.
Seidlhofer, Barbara. 2005. Language variation and change: the case of English as
a lingua franca. In Dziubalska-Kołaczyk, Katarzyrna, and Przedlacka, Joanna
(eds.), English Pronunciation Models: A Changing Scene, 59–75. Bern: Peter
Lang.
SG Chinese. 2009. Do people know that Singapore and China are different countries?
ExpatSingapore message board contribution, posted 25 August 2009. www.
expatsingapore.com/forum/index.php?topic=55265.15. Accessed 26 August
2009.
SGEM, Speak Good English Movement. 2011. Speak Good English Movement. Web-
site. www.goodenglish.org.sg. Accessed 6 March 2011.
Sharma, Devyani. 2009. Typological diversity in New Englishes. English World-
Wide, 30, 170–95.
Siegel, Muffy E. A. 2002. Like: the discourse particle and semantics. Journal of
Semantics, 19, 35–71.
Silverstein, Michael. 1976. Shifters, linguistic categories, and cultural description. In
Basso, Keith H., and Selby, Henry A. (eds.), Meaning in Anthropology, 11–56.
Albuquerque: University of New Mexico Press.
2003. Indexical order and the dialectics of sociolinguistic life. Language and Com-
munication, 23, 193–229.
Simpson, Andrew. 2007. Singapore. In Simpson, Andrew (ed.), Language and Na-
tional Identity in Asia, 374–90. Oxford University Press.
Simpson, John, and Weiner, Edmund (eds.). 2000. Oxford English Dictionary. 2nd
edn. Oxford University Press.
144 References
Singapore Land Authority. 2003. Singapore Street Directory. 21st edn. Singapore:
SingTel Yellow Pages.
Singstat, Department of Statistics. 1999. Yearbook of Statistics: Singapore. Singapore:
Department of Statistics, Ministry of Trade and Industry.
2000. Resident working persons aged 15 years and over by occupation, ethnic
group and sex. PDF document. www.singstat.gov.sg/pubn/popn/c2000sr3/
t5-6.pdf. Accessed 29 January 2013.
2010. Key indicators on marriages and divorces, 2005–2010. PDF doc-
ument. http://web.archive.org/web/20120103135040/www.singstat.gov.sg/
stats/themes/people/marriages.pdf. Accessed 5 February 2012.
SMC, Speak Mandarin Campaign. 2009. 讲华语运动. Website. www.mandarin.org.
sg/2009. Accessed 22 January 2010.
Suzanna, Bte Hashim, and Brown, Adam. 2000. The [e] and [æ] vowels in Singapore
English. In Brown et al. 2000: 84–92.
Tan, Amelia. 2009. Refine bilingual policy. The Straits Times, 29 December 2009,
Letters. http://wanderingchina.org/2009/12/30/. Accessed 29 January 2013.
Tan, Charlene. 2005. English or Singlish? The syntactic influences of Chinese and
Malay on the learning of English in Singapore. Journal of Language and Learning,
3, 156–79.
Tan, Chor Hiang, and Gupta, Anthea Fraser. 1992. Post-vocalic /r/ in Singapore
English. York Papers in Linguistics, 16, 139–52.
Tan, Ludwig. 2003. Topic-prominence and null arguments in Singapore Colloquial
English. In Deterding et al. 2003: 43–53.
Teo, Peng Suan. 1995. The syntax and semantics of got in colloquial Singapore
English. BA thesis, Department of English Language and Literature, National
University of Singapore.
Thornsteinsson, Einar. 2008. NSM Semantics in Brief. Website. www.une.edu.au/
bcss/linguistics/nsm/semantics-in-brief.php. Accessed 23 April 2008.
Thurgood, Graham, and LaPolla, Randy J. 2003. The Sino-Tibetan Languages. Lon-
don: Routledge.
Tongue, Ray K. 1974. The English of Singapore and Malaya. Singapore: Eastern
Universities Press.
Trudgill, Peter. 1974. The Social Differentiation of English in Norwich. Cambridge
University Press.
1995. Sociolinguistics: An Introduction to Language and Society. London: Penguin.
1999. Standard English: what it isn’t. In Bex, Tony, and Watts, Richard J. (eds.),
Standard English: The Widening Debate, 117–28. London: Routledge.
2004. New Dialect Formation: The Inevitability of Colonial Englishes. Edinburgh
University Press.
Trudgill, Peter, and Hannah, Jean. 1994. International English: A Guide to Varieties
of Standard English. London: Arnold.
Turnbull, Constance Mary. 1996. A History of Singapore: 1819–1988. 3rd edn.
Singapore: Oxford University Press.
UNDESA, United Nations Department of Economic and Social Affairs, Population
Division. 2010. World Population Policies 2009. New York: United Nations.
Vertovec, Steven. 2007. Super-diversity and its implications. Ethnic and Racial Stud-
ies, 29, 1024–54.
References 145
Wardhaugh, Ronald. 2002. An Introduction to Sociolinguistics. 4th edn. Oxford: Black-

well.
Wee, Lionel. 1998. The lexicon of Singapore English. In Foley et al. 1998: 175–200.
2004. Reduplication and discourse particles. In Lim 2004: 105–26.
2006. The semiotics of language ideologies in Singapore. Journal of Sociolinguistics,
10, 344–61.
2008. Singapore English: phonology. In Mesthrie 2008: 259–77.
2010a. ‘Burdens’ and ‘handicaps’ in Singapore’s language policy: on the limits of
language management. Language Policy, 9, 97–114.
2010b. Eurasian Singapore English. In Schreier et al. 2010: 313–26.
Wells, John C. 1982. Accents of English. Cambridge University Press.
Wenger, Etienne. 1998. Communities of Practice. Cambridge University Press.
Wierzbicka, Anna. 1972. Semantic Primitives. Frankfurt am Main: Athenäum.
1998. The semantics of English causative constructions in a universal-typological
perspective. In Tomasello, Michael (ed.), The New Psychology of Language:
Cognitive and Functional Approaches to Language Structure, 113–53. Mahwah,
NJ: Lawrence Erlbaum.
Wijeysingha, Eugene. 1963. A History of Raffles Institution 1823–1963. Singapore:
University Education Press.
Willemyns, Ronald. 1987. The investigation of ‘language continuum’ and ‘diglos-
sia’: a macrological case study and a theoretical model. In Blanc, Michel, and
Hamer, Josiane F. (eds.), Problèmes théoriques et méthodologiques dans l’étude
des langues/dialectes en contact aux niveaux macrologiques et micrologiques. Pro-
ceedings of the International Conference DALB/CIRB/ICSBT (University of
London, 23–26 May 1985), 30–49. Québec: Centre international de recherche
sur le bilinguisme.
Winford, Donald. 1993. Predication in Caribbean English Creoles. Amsterdam: John
Benjamins.
2003. An Introduction to Contact Linguistics. Oxford: Blackwell.
Wong, Jock Onn. 1994. A Wierzbickan approach to Singlish particles. MA the-
sis, Department of English Language and Literature, National University of
Singapore.
2004. The particles of Singapore English: a semantic and cultural interpretation.
Journal of Pragmatics, 36, 739–93.
Wong, Wee Kim (ed.). 2011. Census of Population 2010. Singapore: Department of
Statistics.
Wouk, Fay. 1999. Gender and the use of pragmatic particles in Indonesian. Journal
of Sociolinguistics, 3, 194–219.
Zimmer, Benjamin. 2006. Meh-ness to society. Language Log post,
8 June 2006. http://itre.cis.upenn.edu/∼myl/languagelog/archives/003238.
html. Accessed 30 November 2011.
Author index
Abdulaziz Mkilifi, Mohammed H. 43 DeFrancis, John 3

Abu Baker, Jalelah 8 Deterding, David 21, 51, 60, 64–6, 80, 97,
Aijmer, Karin 83, 84 98, 121, 122
Aitchison, Jean 23 Dixon, Quentin L. 11
Alleyne, M. 28 Doraisamy, Theodore C. 4
Alsagoff, Lubna xii, 11, 12, 16, 19, 21, 46–51,
58, 62, 63, 80, 96, 103–6, 109, 112, 113 Eckert, Penelope xi, 16, 20, 52–5, 58, 63,
Ansaldo, Umberto 34, 35, 44, 62 100, 104, 106
Erb, Maribeth 9
Bakhtin, Mikhail M. 49, 100
Bao, Zhiming xii, 3, 7, 8, 21, 64, 66, 77, 80, Fasold, Ralph W. 39, 43, 63
81, 85 Ferguson, Charles A. xii, 18, 26, 27, 39–41,
Baskaran, Loga 120 44, 45, 63, 114
Baskaran, Lohanayahi 90 Ferguson, Ian 82
Baxter, Alan N. 35 Ferguson, Will 82
Beeching, Kate 82, 84 Fernández, Mauro 38
Bickerton, Derek 30 Fischer, Kerstin 83, 84
Blommaert, Jan 46, 51, 52, 103, 126 Fishman, Joshua A. 35, 39, 43
Bokhorst-Heng, Wendy 11, 21, 45, 47, 113, Foley, Joseph A. 7, 120
117 Frangoudaki, Anna 40
Bourdieu, Pierre 50, 101
Bourhis, Richard 37 Giles, Howard 37
Brown, Adam 18, 19, 65, 82, 86, 89, 98 Gilman, Albert 102
Brown, Charles Cuthbert 1 Gisborne, Nikolas 113
Brown, Roger 102 Goddard, Cliff 95
Bruthiaux, Paul 14 Goh, Hwee 12
Bucholtz, Mary 55 Gumperz, John 43
Gupta, Anthea Fraser xii, 9, 16, 18, 21, 34,
Canagarajah, A. Suresh 15 35, 38, 39, 41–5, 47, 51, 57, 58, 60, 62, 63,
Carayol, Michel 30 66, 67, 71–7, 82, 84–91, 96, 104, 108–11,
Chambers, Jack K. 52, 63 113, 115
Chang, Rachel 12
Chao, Yuen Ren 87 Hall, Kira 55
Chaudenson, Robert 30 Hannah, Jean 41, 66
Cheng, Chin-Chuan 3 Heller, Monica 51
Chew, Phyllis Ghim-Lian 21, 51 Ho Chee Lick 48, 67, 80, 120
Chiang, Ming Yu 8 Ho, Mian-Lian 15, 16, 21, 30, 31, 33, 44, 52,
Coupland, Nikolas 62, 104 75, 80, 82, 84, 86, 89
Crystal, David 13, 14, 82 Hong, Huaqing 21
Hu, Jianhua 78
Davie, Sandra 12 Hussain, Zakir 12, 18, 33, 45
DeCamp, David 17, 26–33, 35, 46, 63, 112,
115 IMF, International Monetary Fund 6
146
Author index 147
Johnstone, Barbara 103 Pennycook, Alistair 46, 51

Platt, John T. xi, xii, 9, 15–19, 21, 28, 30–8,
Kachru, Braj B. 13, 14 41, 44, 46, 47, 49, 50, 52, 62, 67, 75, 78,
Kazazis, Kostas 40 80, 82, 84, 86, 89–91, 104, 108–11, 113,
Kerswill, Paul 23 115, 120, 124
Kiesling, Scott F. 103 Poedjosoedarmo, Gloria R. xi, 19, 20, 49, 51,
Kim, Chonghyuck 89 60, 62, 104, 109, 111
Kirkpatrick, Andy 122 Poirier, Claude 82
Kor, Kian Beng 12 Prapañca, Mpu 1
Kortmann, Bernd 113 Psichari, Jean 38
Kwan-Terry, Anna 21, 89, 90, 93 Pullum, Geoffrey K. 11
Labov, William 21, 53, 75 Rajadurai, Joanne 120

LAC, Australian Academy of the Ramesh, S. 117
Humanities and the Chinese Academy of Rampton, Ben 23, 61, 62, 104
Social Sciences 133 Ramsey, Robert S. 3
LaPolla, Randy J. 133 Rappa, Antonio L. 11, 35, 118
Lau, Cheow 96 Rash, Felicity 39
Le Page, Robert B. 27 Rauniomaa, Mirka 55
Lee, Jack Tsen-Ta 34, 56, 65, 79 Rhoı̈dis, Emmanouēl D. 38
Lee, Nala Huiying 77–9 Rickford, John R. 26–8, 30, 32
Leimgruber, Jakob R. E. xiii, 16, 60, 65, 66, Rubdy, Rani 21, 113, 118
77, 108, 117, 118
Ler, Soon Lay Vivien 82, 84, 86, 89, 91 Savage, Victor R. 1, 10
Lim, Lisa S. 16, 60, 69, 82, 85–92, 113 Schneider, Edgar W. 14, 113
Lim, Sonny 16 Seargeant, Philip 126
Ling, Ai Ping 77–9 Sebba, Mark 28, 45
Low, Ee Ling 18, 19, 82, 86, 89, 98 Seidlhofer, Barbara 122
SG Chinese 106
Mackey, William F. 38, 39 SGEM, Speak Good English Movement
Mair, Victor H. 3 118
Marçais, William 39 Sharma, Devyani 113
Matthews, Peter H. 75, 114 Siegel, Muffy E. A. 82
Matthews, Stephen J. 35 Silverstein, Michael 16, 20, 21, 100–3
McArthur, Tom 123 Simon-Vandenbergen, Anne-Marie 83, 84
McConnell-Ginet, Sally 54 Simpson, Andrew 45
McKay, Sandra 21, 113 Singapore Land Authority 10
MDA, Media Development Authority 119 Singstat, Department of Statistics 6, 9, 11
MITA, Ministry of Information, SMC, Speak Mandarin Campaign 8
Communication and the Arts 47 Suzanna, Bte Hashim 65
MOE, Ministry of Education 9, 12, 21, 22,
113 Tabouret-Keller, Andrée 27
MOH, Ministry of Health 6, 7 Tagg, Caroline 126
MrBrown.com 79 Tan, Amelia 119
Muysken, Pieter 69 Tan, Charlene 60
Myers-Scotton, Carol 69, 111 Tan, Chor Hiang 66
Tan, Ludwig 3
Nair-Venugopal, Shanta 120 Taylor, Donald M. 37
Nomoto, Hiroki 77–9 Teo, Peng Suan 77, 78
Norman, Jerry 133 Thornsteinsson, Einar 95
Thurgood, Graham 133
Ochs, Elinor 100, 101, 103 Tongue, Ray K. 15, 16, 46, 120
Trudgill, Peter 3, 21, 40, 41, 44, 52, 66, 113
Pakir, Anne xi, 12, 16, 18, 19, 21, 35, 43, 46, Turnbull, Constance Mary 1, 3, 4, 15
47, 49, 50, 52, 62, 104, 109, 111
Pan, Haihua 78 UNDESA, United Nations Department of
Parliament of Singapore 6, 11 Economic and Social Affairs, Population
Patrick, Peter L. 27 Division 123
148 Author index
Vertovec, Steven 52, 126 Wijeysingha, Eugene 3

Willemyns, Ronald 43, 45, 63
Wardhaugh, Ronald 39, 114 Winford, Donald 27, 39, 45
Weber, Heidi 15, 21, 32, 67, 78, 80, 90, 120, Wong, Jock Onn 86, 89, 91, 95,
124 96
Wee, Lionel 11, 12, 65, 66, 77, 80, 81, 84, Wouk, Fay 82
86–9, 91, 118
Wells, John C. 64 Yeoh, Brenda S. A. 1, 10
Wenger, Etienne 54
Wierzbicka, Anna 95 Zimmer, Benjamin 87
Subject index
acrolect, 17, 19, 26, 27, 28, 29, 32, 47 definition of, 39
acronym, see initialism leaky, 43, 63, 77, 111
ah, see discourse particles discourse particles
already, 80 ah, 86–7
ang moh, 65, 67 hah, 88
ASEAN, 117, 119, 122–3 hor, 88
auxiliary verbs, 60, 73–5, 116 lah, 86, 95, 105
leh, 69, 87
Baba, see Peranakan lor, 88
Baba Malay, 16, 35 mah, 88
basilect, 17–18, 19, 26, 27, 32, 115 meh, 87
Bazaar Malay, 9, 16, 34, 37, 38, 90 what, 88–9
Cantonese, see Chinese Cantonese education system, 3, 11, 12, 15, 16, 21–3
Chinese Estuary English, 82, 121
Cantonese, 3, 7, 38, 67, 69, 87, 88, 89, 90, ethnic varieties, 51, 60–2
91, 120 ethnicity
ethnic group, 3, 7 and culture, 48, 51, 61, 62, 105
Hakka, 3, 8 and language, 11, 12, 47, 61, 121
Hokkien, 3, 7, 8, 9, 38, 65, 67, 68, 69, classification, 11
78–9, 89, 90 mixed groups, 34
Mandarin, 6, 8, 9, 69, 80, 88, 89, 90 multiple ethnicities, 11
Southern Min, 7, 120 Eurasians, 12, 34
Teochew, 3, 7 expanding circle, see Kachruvian paradigm
varieties of, 3
code-switching, 27, 50, 51, 58–9, 60, 63, hah, see discourse particles
68–70, 116 Haitian Creole French, 26, 39
and diglossia, 43, 69, 116 Hakka, see Chinese Hakka
concentric circles, see Kachruvian Hokkien, see Chinese Hokkien
paradigm hor, see discourse particles
continuum
Jamaican, 26–30 indexical field, 54–5, 103–8
post-creole, 26 indexical orders, 101–3
copula deletion, 75–6 indexicality, 20–1, 56–62, 70, 99, 100–3, 112,
creole, 26, 27, 28, 32, 34, 35 113, 126
creoloid, 16, 26, 34–5 referential and non-referential, 101
crossing, 61 Indian
ethnic group, 60
dialect continuum, 125 Indonesian, see Malay language
diglossia, 18, 27, 35, 38–46, 74, 75, 76, 96, inflexions, 71–3
111–12, 114–15 initialism, 97
broad, 39 inner circle, see Kachruvian paradigm
classic, 39, 114 inversion, 73–5
149
150 Subject index
Jamaican Creole, 27 Peranakan, 16, 34, 35

Javanese, 1 permanent quality, 54
place-names, 10–11
Kachruvian paradigm, 13–15, 122 polyglossia, 35–8
pragmatic particles, see discourse particles
lah, see discourse particles pro-predicate do, 125
language policy, 6, 10, 11, 12, 117–19
leh, see discourse particles race, see ethnicity
lor, see discourse particles referential indexicality, see indexicality
referential and non-referential
mah, see discourse particles road names, see place-names
Malacca, 1, 16, 35
Malay social meaning, 54, 62, 104–8
ethnic group, 1, 3, 4, 6, 9, 16 social type, 54
language, 3, 6, 8, 9, 10, 37, 67, 90, 120, Speak Good English Movement, 118–19
121, 123 Speak Mandarin Campaign, 8, 45, 117
Indonesian, 6, 123 Sranan Tongo, 28
national language, 7, 37 stance, 54, 55, 56, 102, 104, 106–8, 113
Malaysia, 120–121 stance accretion, 55
Malaysian English, 35, 120, 121, 122, 123 Swiss German, 18, 39, 44, 114–15
Mandarin, see Chinese Mandarin system transfer, 80–2
meh, see discourse particles
mesolect, 19, 26, 32 Tamil, 3, 6, 7, 8, 9, 11, 21, 37, 90, 120
mother tongue, 6, 8, 9, 11, 12, 48 Teochew, see Chinese Teochew
mutual intelligibility, 3, 33, 44, 121, 122, 124 th-fronting, 66, 121
third-wave sociolinguistics, 52–4, 103
Natural Semantic Metalanguage, 91–6 traditional dialectology, 124
non-referential indexicality, see indexicality
referential and non-referential unidimensionality, 26, 27, 33
Nonya, see Peranakan
NSM, see Natural Semantic Metalanguage what, see discourse particles
where got, 79, 104
oinoglossia, 102 World Englishes, 51, 113, 123, 124, 125,
orders of indexicality, see indexical orders 126
outer circle, see Kachruvian paradigm varieties of, 121, 123, 124

Singapore English - Structure, Variation, and Usage PDF

Uploaded by

Copyright:

Available Formats

Singapore English - Structure, Variation, and Usage PDF

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

Singapore English - Structure, Variation, and Usage PDF

Uploaded by

Copyright:

Available Formats

more information - www.cambridge.

jakob r. e. leimgruber is Lecturer in the English Department at the

Already published in this series:

Earlier titles not listed are also available

This publication is in copyright. Subject to statutory exception

First published 2013

Library of Congress Cataloguing in Publication data

isbn 978-1-107-02730-5 Hardback

Cambridge University Press has no responsibility for the persistence or

List of figures page xi

1 Introduction: Singapore and its Englishes 1

1.1 Location of Singapore within Asia page 2

1.1 Language most frequently spoken at home page 7

This book builds on a D.Phil. thesis (Leimgruber 2009b) submitted to the

Financial support was given throughout my studies by the Pembroke

1.1 Historical and present-day background

introduced a massive public housing construction scheme, which now houses

1.1.2 Linguistic diversity

1980 1990 2000 2010

English 11.6 % 18.8 % 23.0 % 32.3 %

1.1.3 The place of English

I don’t think in Malay . . . My thoughts are mostly in English, so when I

Figure 1.3. Kachru’s concentric circles of World Englishes (Kachru

1.2 The Englishes of Singapore and their models

constantly having to contrast them against British or American English,

Further and more fundamental influences on the education medium were

Singapore speech SE speech con-

Figure 1.4. ‘Relation between socio-economic factors and the usage of

franca in inter-ethnic group communication’ (Platt 1975: 372). The concept,

Intimate CSE Rudimentary

Figure 1.5. ‘Expanding triangles of English expression’ (Pakir

indicator of proficiency) will have access to a smaller range of stylistic varia-

RP Pronunciation (media & American Pronunciation (media an-

Syntactically & lexically standard

Cline of formality Cline of

Stylistic range for

Colloquial style for many educated speak-

Pidgin form: uneducated speakers of other lan-

Figure 1.6. Modified triangles (Poedjosoedarmo 1995)

economic capital, informality vs formality, and community membership vs

Indexical analysis takes into account several layers of reference – ‘orders

1.3 Data and methods

Figure 1.7. Post-secondary enrolment in 2005 (data from MOE

by extension, of English language proficiency: clearly the academic standard

r Individual interview: each informant was singly interviewed for 15 min-

2.1 Early models and diglossia

The Jamaican continuum

non-discreteness – the impossibility of subdividing the continuum into iden-

The conditions listed by DeCamp for the emergence of a post-creole

Table 2.2. Speakers from Table 2.1 rearranged

Table 2.3. Speaker ranking from Table 2.2 with

The Singaporean continuum

Singapore speech SE speech con-

Figure 2.1. ‘Relation between socio-economic factors and the usage of

possible, as in Jamaica according to DeCamp (1971), to predict the behaviour

to the colloquial form (‘Singlish’), there is an intermediate ‘semi-formal’ one,

While this may be a recent phenomenon – and certainly not a widespread

Table 2.4. Proposal of a sociolinguistic typology

Table 2.5. ‘A model for polyglossia with