Content uploaded by Jasper Hong Sim
All content in this area was uploaded by Jasper Hong Sim on Mar 04, 2019
Content may be subject to copyright.
“But you don’t sound Malay!” Language
dominance and variation in the accents of English-
Malay bilinguals in Singapore
University of Cambridge
This study examined the English accents of English-Malay bilinguals in Singapore to
ascertain whether language dominance was a determinant of accent variation in
Singapore English, with a hypothesis that a Malay-dominant bilingual would have more
ethnic-specific features than an English-dominant one. Ten English-Malay bilinguals –
five English-dominant and five Malay-dominant – who differed greatly in their language
dominance took part in this study. In an ethnic discriminability task that involved 60
naïve raters, Malay-dominant bilinguals were significantly more often correctly identified
as ethnically Malay and were rated as having a significantly more perceivable Malay-
accented English accent, while those who were English-dominant had an English accent
that lacked ethnic-specific features so much so that naïve raters, including raters who
were English-Malay bilinguals, were less able to identify the speakers as ethnically
Malay. The results of this study indicate that early sequential bilinguals or simultaneous
bilinguals of the same two languages need not have similar accents. The findings also
suggest that language dominance is a determinant of accent variation in Singapore
English, at least for the English-Malay bilinguals.
Keywords: Singapore English, accent variation, language dominance, English-Malay
bilinguals, ethnic identification
It is widely assumed that bilinguals cannot isolate languages and process them like monolinguals, and
that the two linguistic systems necessarily interact (Grosjean 2008). As regards phonological systems,
such interactions have been extensively documented, particularly in studies of cross-linguistic transfer
in many multilingual contexts (e.g. Rasier and Hiligsmann 2007; Pytlyk 2008) and in descriptions of
distinctive accents of World Englishes (e.g. Gramley and Pätzold 2004: 314–346; Deterding and
Kirkpatrick 2006). Yet, as bilinguals vary in how the languages are used according to the different
contexts and needs, they are not equally bilingual and are typically dominant in one language (Grosjean
2010). The definition of language dominance varies and may include different dimensions of language
use and experience. It is sometimes regarded as the language which the bilingual hears or uses more
frequently (e.g. Yavas 2007; Grosjean 2010), or which the bilingual is more proficient in (e.g. Gathercole,
Mueller, and Enlli Môn 2009; Hernàndez-Chávez, Burt, and Dulay 2013). Language dominance has also
been described more broadly to include language use, proficiency and language history (e.g. Goral,
Campanelli, and Spiro 2015). In addition, some studies found parameters like language attitudes (e.g.
Miller 2017) and age of acquisition and attrition (Birdsong 2014) to be associated with shifts in language
dominance. Language dominance is therefore defined here as consisting of four sub-components. The
first is “proficiency”, which follows Hulstijn’s (2010: 186) definition as the “largely implicit,
unconscious knowledge in the domains of phonetics, prosody, phonology, morphology, and syntax”
and “the largely explicit, conscious knowledge in the lexical domain (form-meaning mappings)”. In
much simpler terms, this refers to the attainment and mastery of the syntax, grammar, lexis, and
pronunciation of a language that are required for understanding, speaking, reading, and writing. It is
recognised here that, albeit often conflated, proficiency and dominance are distinct concepts (Birdsong
2006: 49), with proficiency as an essential component of dominance, but not the sole defining factor.
The second component is “use”, which echoes Harris, Gleason, and Aycicegi’s (2006: 264) assertion that
language dominance refers to the language most accessible and most highly activated – “the default
language for speaking and thinking”. The third is “attitudes”, in that fluency of either language can be
gained, maintained, or lost as a result of shifting language attitudes (Hakuta and D’Andrea 1992).
Factors such as cultural identification (Marian and Kaushanskaya 2004) affect such attitudes, and
therefore language dominance. The final component is “history”, which refers to parameters such as
age of arrival (AoA), age of first exposure, length and type of language instruction, and home
language(s). Following Grosjean (2008: 37–40), dominance is seen as a continuum rather than a
Studies have shown that language dominance has an effect on cross-linguistic interactions and
thereby affects production and various language processes (e.g. Pallier, Colomé, and Sebastián-Gallés
2001; Sebastián-Gallés et al. 2006; Caffarra, Zimnukhova, and Mancini 2016). In particular, several
studies on perceived foreign accent (e.g. Flege, Frieda, and Nozawa 1997; Piske and MacKay 1999;
Guion, Flege, and Loftin 2000) have revealed the effects of variation in L1 use on L2 production of
bilinguals, where a self-reported more frequent use of the L1 is correlated with the perception of a more
L1-accented L2 speech. In their study, Flege, Frieda, and Nozawa (1997) measured the accent ratings of
adult native Italians who immigrated to Canada with similar chronological age and AoA (range = 2.6–
9.6 years). The “High-L1 use” group (N = 20) and the “Low-L1 use” group (N = 20) differed in their
self-reported Italian use (36 per cent versus 3 per cent of the time), and a stronger Italian accent in
English was found in the “High-L1 use” group. Piske and MacKay (1999) replicated this study to
include adult late bilinguals, finding the same effect in both early (AoA mean range = 6.9–8.3) and late
bilinguals (AoA mean = 19.5). Guion, Flege, and Loftin (2000) investigated Quichua-Spanish adult early
bilinguals living in Otavalo, Ecuador, who began learning Spanish between the ages of 5–8 and also
differed in their use of Quichua (L1). They found that similar to Flege, Frieda, and Nozawa’s Italian-
speaking subjects, the high Quichua use group had significantly stronger foreign accent in their Spanish
(L2) speech. A study on adult early Korean-English bilinguals with the same reported average AoA (11
years) showed a similar L1 use effect. These studies (see also Flege, Yeni-Komshian, and Liu 1999; Flege,
MacKay, and Piske 2002) suggest that the amount of continued L1 use, and thus activation, influences
production capabilities in an L2, and age of onset is not the only factor. They also show that bilinguals
who speak the same two languages do not necessarily have the same degree of L1-accented L2 speech.
Language attitudes can also influence production. Languages hold much socioeconomic value and
may serve as a symbol of identity for their speakers (Ricento 2005). An example of how speakers utilise
phonetic segments of a particular variety for identity purposes is Labov’s (1972) Martha’s Vineyard
study, which showed that residents use a particular vowel sound to distinguish themselves from
tourists who regularly visit their island. Schilling-Estes (2004) similarly demonstrated that speakers
varied their accent to emphasise ethnic distinctiveness through linguistic distancing. In another study
of two women of German origin in southern Brazil, Zilles and King (2005) also found that their
participants maintained certain German-linked features and also used Brazilian features to reflect both
aspects of their identities.
Such variation in accents in these small bilingual communities warrants an investigation of larger
and more complex multilingual societies, particularly in the context of contact varieties of English. In
countries such as Singapore, India, and Hong Kong, where English and other languages serve
important functions or are official languages, bilinguals differ greatly in their language dominance in
the dimensions outlined above, and the attempts to delineate the general accents of these New
Englishes may have neglected the potential and important variation that may exist due to the effects of
language dominance. This paper aims to fill this gap through examining the bilinguals in Singapore by
ascertaining whether language dominance is a determinant of accent variation in Singapore English
1.1 Language situation in Singapore
English in Singapore is acquired and used alongside ethnic mother tongues (EMTs) that include
Mandarin, Malay, and some Indic or Dravidian languages in what is known as “English-knowing
bilingualism” (Kachru 1983: 42). The language policy requires Singaporeans to be formally instructed
in English first, with an EMT as a second subject according to one’s ascribed race, regardless of the
variety/ies spoken at home. Due to its pragmatic motivations, the language policy has been premised
as the “functional polarization” of languages (Pendley 1983): English was to serve as the primary
working language of education, science, technology, law, and commerce, and the local inter-ethnic
lingua franca. The EMTs, contrastingly, were meant as a means to demarcate and embody the traditional
roots and culture of Singapore, acting as a cultural ballast and at the same time as an anchor for
Singaporeans against Western influences (Rubdy 2001: 342; see Ng  for a detailed review of the
bilingual policy). Consequently, individual bilingualism and societal multilingualism are the norm in
Singapore; even domestic multilingualism is wide-spread and not an exception (Gupta 1994: 49). Since
the introduction of the bilingual policy and full English-medium schools in 1987, all Singaporean
children have been required to receive formal instruction for both English and an EMT from ages 6 to
16 (or 18 for some A-level candidates). Most children will also have been exposed to English and at
least one EMT at home, at preschool or in kindergarten.
However, like many members of multilingual communities, Singaporeans vary considerably in
their language dominance. With regard to language use, for example, the distribution of EMTs and
English for many is not diglossic. While English is the working language, there is variation with respect
to the language preferred in personal domains, such as at home or for social purposes. Since the
initiation of the bilingual policy, the use of English as a home language has become more prevalent,
increasing from 11.6 per cent in 1980 (Foley 1998: 221) to 29.8 per cent in 2010 (Tan, P. 2012: 129),
especially amongst younger and more educated Singaporeans (Bokhorst-Heng 1999: 239). The General
Household Survey of 2005 also states that about 27.9 per cent of residents aged 15–29 use English at
home. Despite the rising use of English, EMTs are still used, especially with parents or grandparents,
or even exclusively for personal domains (see Tan 2014 for a detailed discussion). There is similar
disparity in language attitudes. Since English is a language with greater social prestige, younger
Chinese Singaporeans are displaying a lack of interest in learning Mandarin (Ng 2014). Similarly, even
though Malay plays an important role in the Malay community, especially in the religious domain, an
increasing number of young Malay Singaporeans indicated English as the language they would identify
as their own and that they would prefer to speak English with their child, if they were to have one (Tan
1.2 Studies on accents of Singapore English
Since Singaporeans are bilinguals of English and different EMTs, accentual variation between ethnic
groups exists. Though such inter-ethnic variation has been studied, intra-ethnic variation that may exist
due to differences in language dominance has not. In fact, most work on the phonology of SgE tends to
describe pronunciation features monolithically, typically referring to them as features characteristic of
Colloquial SgE (e.g. Wee 2004), as those of a mesolectal variety (e.g. Deterding 1994), or as belonging to
an “idealised Singaporean speaker” (Brown and Deterding 2005: 2). Studies on ethnic discriminability
(e.g. Deterding and Poedjosoedarmo 2000; Lim 2000; Tan, Y. Y. 2012) demonstrate the inadequacy of
such monolithic descriptions as inter-ethnic variations or ethnic-specific features persistently
distinguish speakers of different EMTs. Apart from ethnicity, no other determinant of accent variation
in SgE has been thoroughly examined. Clearly, as regards accents of SgE, more or other variables are at
1.3 The Malay speech community in Singapore
The Malay community in Singapore is well-suited for the present study. Unlike the Chinese and Indian
communities, which have several other languages and dialects, the Malay community is linguistically
more homogenous, as the Malay language is in fact, in the linguistic sense, their mother tongue (Chong
and Seilhamer 2014: 364). The ethnic Malays make up 13.4 per cent of the population (Singapore
Department of Statistics 2010), and of all EMTs Malay is described as more resilient to language shift,
as 82.7 per cent of Malay residents aged 5 years and over still use Malay at home. However, Malay is
not only spoken by ethnic Malays; the community of English-Malay bilinguals (EMBs) in Singapore
includes ethnic Malays as well as minorities, i.e. people of Javanese, Boyanese, Arab, or Indian descent.
The variant of Standard Malay spoken in Singapore is Bahasa Melayu, which is the same as Bahasa
Malaysia, the variety spoken by the educated in Peninsular Malaysia (Deterding and Poedjosoedarmo
1998). Apart from Bahasa Melayu, Malay-lexified pidgin languages such as Bazaar Malay and Baba
Malay exist, though the use of both is less common today, especially with younger Singaporeans (see
Cavallaro and Serwe 2010). For convenience, the sub-variety of SgE used by these speakers is termed
Malay Singapore English (MSE).
1.4 Current study
Whilst the effects of language dominance on accents are known – as is the fact that disparity in the
language dominance of Singaporeans exists – it is surprising that language dominance has hitherto not
been studied as a possible determinant of accentual variation in SgE. To this end, an ethnic
discriminability task was performed in the present study, testing the following specific research
hypothesis: educated EMBs who are dominant in Malay have a more Malay-accented SgE accent than
educated EMBs whose dominant language is English.
2.1 Assessing language dominance
Language dominance of EMBs was first assessed to recruit suitable participants. For this purpose, self-
reporting methodology was used instead of linguistically-based tasks. Self-reporting tools take less time
as they do not require elaborate test items or complex scoring. At the same time, self-reporting is a
reliable methodology as bilinguals are able to self-assess language experience and abilities that
correspond with behavioural measures of linguistic performance (e.g. Lim et al. 2008: 399; Gertken,
Amengual, and Birdsong 2014: 218). Moreover, self-reporting can also be used to assess non-linguistic
factors, such as language attitudes. There exist several validated self-reporting instruments to measure
bilingual language dominance: the Language Experience and Proficiency Questionnaire (Marian,
Blumenfeld, and Kaushanskaya 2007), the Self-Report Classification Tool (Lim et al. 2008), the Bilingual
Dominance Scale (Dunn and Fox Tree 2009), and the Bilingual Language Profile (Gertken, Amengual,
and Birdsong 2014). Eventually, the Bilingual Language Profile (BLP) was selected for its fit. This is
because the BLP measures all four components of language dominance defined above, is adequately
comprehensive (e.g. has at least three items for each of the components), provides a continuous
dominance score (as opposed to descriptive profiling, or discrete/dichotomous groups), and presents
no problematic issues when applied to the local context. All four components received equal weighting
in the calculation of dominance. The questionnaire, in both Google Form format and hardcopy
(Birdsong, Gertken, and Amengual 2012), was adapted for use. The dominance scores were
automatically tabulated, and possible scores range from -218 (Malay-dominant) to +218 (English-
The BLP was administered to 95 Malay male and female students conveniently sampled from the
National Institute of Education, Nanyang Technological University, Singapore Institute of
Management, National University of Singapore, and general public spaces. At the end of the survey,
participants were asked whether they object to being contacted for the second part of the experiment
(i.e. the speech recording). Variables that might affect the comparability and validity of the results were
controlled. Specifically, all participants were born and raised in Singapore, had Malay (Bahasa Melayu)
as their EMT, were educated in Singapore, and had completed post-secondary or pre-university
education. Since educated EMBs above 18 years of age have undergone different phases of language
policies (see Tan 2014: 326), EMBs of 18–29 years of age were studied, as this age range coincides with
the phase in which the bilingual policy and full English-medium education were established. They
should also have been exposed to both English and Malay at least by five years old. From the
respondents who met the criteria, the five with the most negative BLP scores (Malay-dominant group,
MD), and the five with the most positive BLP scores (English-dominant group, ED) were recruited for
the recording (see Tables 1 and 2).
Table&1. Profiles of ED subjects
Table&2. Profiles of MD subjects
Table 3 shows the key results of the BLP survey. The two groups differ considerably in their
language dominance but are still functioning bilinguals to varying extents. One notable difference lies
in the age of acquisition: EDs were exposed to both Malay and English languages since birth, but MDs
only started learning English at about 4;0. This could be attributed to the linguistic shifts that occurred
as a result of the bilingual policy. In the last two decades, EMTs and dialects were used predominantly.
In fact, in the 1980s, only 11.6 per cent of the population used English as a home language (Foley 1998:
221). Therefore, MDs may have been raised in Malay-speaking families, and only began learning
English formally in preschool or kindergarten.
2.3 Recording procedure
During the recording process, only the author, who is ethnically Chinese, was present. This minimised
variation that might result from speech accommodation towards a Malay research assistant.
Divergences from the speech of the Chinese interviewer, albeit possible, may be less likely given that
the researchers of many studies on MSE were ethnically Chinese, and yet MSE features were still
exhibited. Reference was made to the guidelines for good acoustic recording detailed by Ladefoged
(2003). The recordings were made in quiet rooms with good acoustic support. They were done at a
sampling rate of 48 kHz/16 bit, and a sound check was done with the participants using sample
sentences prior to the actual recording. To ensure that the subjects were at ease, a short casual
conversation was held at the start. They were told to speak and read in a comfortable and natural way
prior to the recording and were reminded that it was not a test of their speaking ability.
Table&3. Key BLP results for both MD and ED subjects
(20 is maximum)
Age of learning
Years spent in a family where
language is spoken
Talking to yourself
(out of 6)
(out of 6, X=Malay or English)
I feel like myself when I speak X
I identify with an X-speaking
It is important to eventually master
I want others to think that I’m a
proficient speaker of X
Note: Standard Deviations in parentheses.
Materials of two styles – read and spontaneous speech – were elicited in controlled environments. Read
speech was elicited using the Wolf Passage (Deterding 2006). A conversation with the author was
conducted for use as spontaneous speech data. Neutral questions were asked, such as what the
participants did during the term break, or what their favourite holiday destination was. Similar
ethnic/accent identification tasks from Lim (1996), Lau (2002) and Flege, Frieda, and Nozawa (1997)
were adapted for use as the speech perception task. From the elicited material, one extract from the read
passage and one from the conversation were used as stimuli for each speaker. The part of the read
passage used was: “As soon as they heard him, the villagers all rushed from their homes, full of concern
for his safety, and two of his cousins even stayed with him for a short while”.
Extracts from the conversation were carefully selected to ensure that the content (e.g. lexis,
grammar) was ethnically and culturally neutral. The length of the extracts was kept within 15–20
seconds. Speech samples of two Mandarin-English bilinguals and two Tamil-English bilinguals were
also collected for use as distractors in the ethnic identification task.
2.5 Rating task
A total of 60 naïve educated Singaporean respondents recruited based on convenience sampling took
part in the ethnic identification experiment. There were equal numbers of male and female respondents,
and they were from the three main ethnic groups. Due to limitations, an equal number of informants
per ethnicity/EMT could not be recruited, and therefore results from this task may be biased towards
the perceptions of Chinese Singaporeans.
The rating experiment was hosted on an online survey platform, and the total time needed for the
task was kept within 10–15 minutes to prevent raters’ fatigue. The task was divided into two parts: Part
1 consisted of read speech extracts, while Part 2 comprised those from spontaneous speech. Raters were
asked the same questions for both parts; for each token, raters were asked to identify the ethnic group
of the speaker. Thereafter, they had to indicate how Chinese/Malay/Indian-sounding (according to the
choice for the first question) the English accent of the speaker was, on a scale of zero to ten – zero being
“very slight”, and ten being “very strong”. In both parts, the samples were presented in a randomised
order, and a practice task always preceded the actual rating tasks to help calibrate the respondents’
judgements. Raters were also reminded to only rely on accentual features to make their judgements. In
Part 1, the extract from the read passage was shown before any rating was done. In Part 2, the
corresponding question to each extract was shown for each token (see Figure 1 for an example).
Figure&1. Sample question in the ethnic identification task
2.6 Analysis of accentual features
Analyses of accentual features in both read and spontaneous speech were done in Praat (v6.039;
Boersma and Weenink 2018). The specific techniques used for the measurement of each feature will be
described in more detail in Section 3.
The results for the identification task are divided into two parts: responses for the ethnic identification
question will be discussed first, followed by the degree of perceived EMT-accentedness. According to
the hypothesis outlined above, MD speakers will be more readily identified as Malay than ED speakers,
i.e. yield higher accurate identification rates, and the recordings of MD speakers (MD tokens) will also
be rated with a higher degree of perceived Malay-accentedness than the recordings of ED speakers (ED
tokens). Findings from the analysis of accentual features are then presented.
3.1 Ethnic identification
Table 4 presents the ethnic identification rates in percentages for both read and spontaneous speech,
which are also illustrated in Figures 2 and 3, respectively. It is apparent that MD speakers were more
readily identified as Malay (range = 61.7–98.3 per cent) than ED speakers (range = 6.7–48.3 per cent) for
both read and spontaneous speech. Also, especially for spontaneous speech (Figure 3), the distribution
of the data indicates that the informants were more in agreement in deciding the ethnic group for MD
speakers, but less so for ED speakers; this strongly suggests that the features that were found in MD
tokens but not in ED tokens may indeed be Malay-specific. Except for ED5, all ED subjects, conversely,
were rated highly as Chinese (range = 45.0–75.0 per cent). This implies that ED speakers could have
exhibited Chinese-specific features, or that, more likely, as Tan (2010: 582) suggests, the Chinese accent
may have been regarded as the default SgE accent, and that the speakers would be labelled as Chinese
in cases where no prominent Malay or Indian features were detected. There is at present no compelling
explanation for ED5 being perceived as Indian by most raters, however, until a phonetic study that
includes Indian-accented SgE is performed. A Pearson’s chi-square test was conducted to examine the
relationship between MD/ED and being rated as Malay/Non-Malay. All expected cell frequencies were
greater than five. There was a statistically significant association between MD speakers being rated as
Malay and ED speakers being rated as Non-Malay (d.f.(1) = 523.301, p < 0.001). This supports the first
prediction that MD speakers will be more readily identified as Malay than ED speakers.
Table&4. Responses for ethnic identification (N = 60)
Figure&2. Responses for the ethnic identification task (read speech)
Figure&3. Responses for the ethnic identification task (spontaneous speech)
A way to further the claim that MD speakers sound more Malay is to look at the accuracy of
identification of informants who also have Malay as an EMT. Speakers of the same language may be
expected to exhibit in-group identification, seeing that they are, perhaps, more sensitive to nuances that
would not have been picked up by raters who speak other EMTs. Table 5 presents the rates of accurate
identification of the 17 EMB informants. It can be seen that the accuracy of identification is much higher
when participants rated MD speakers (range = 64.7–100 per cent) than when rating ED speakers (range
= 5.8–47.0 per cent). A Pearson’s chi-square test was conducted to examine the relationship between
MD/ED speakers and being rated correctly/wrongly. All expected cell frequencies were greater than
five. There was a statistically significant association between MD speakers being rated correctly and
ED speakers being rated wrongly (d.f.(1) = 139.149, p < 0.001). The results further support the fact that
ED speakers exhibited fewer/no perceivable ethnic-specific markers such that even other EMBs could
not accurately identify them.
Table&5. Rates of correct ethnic identification by 17 raters with Malay as EMT
% of raters with accurate identification
3.2 Degree of Malay-accentedness
Table 6 presents the descriptive statistics for the degree of Malay-accentedness of ED and MD tokens
for both read and spontaneous speech. It can be observed that the means for MD tokens for read (range
= 5.61–7.51) and spontaneous speech (range = 7.40–8.47) are higher than those for ED tokens for read
(range = 2.80–5.73) and spontaneous speech (range = 3.50–6.16). Most of the medians and modes for
MD tokens are also larger than those of ED tokens. The difference was tested statistically. Since the data
is ordinal in nature, a Mann-Whitney U test was performed for each context. For both data sets, the
distribution of ratings was not similar, as assessed by visual inspection. The read data ratings for ED
tokens (mean rank = 90.49) and MD tokens (mean rank = 140.34) were statistically significantly different
(U = 6,065, z = 3.789, p < 0.001). The spontaneous speech ratings for ED tokens (mean rank = 96.61) and
MD tokens (mean rank = 195.18) were also statistically significantly different (U = 15,723, z = 7.544, p <
0.001). In other words, even though some ED speakers were rated as Malay, their English accents were
regarded as less Malay-accented than the MD speakers who were also rated as Malay. This confirms
the second prediction, namely that MD tokens will be rated with a higher degree of Malay-accentedness
than ED tokens.
Table&6. Responses for perceptions of Malay-accentedness
3.3 Analysis of accentual features
An acoustic analysis was conducted to identify accentual features that raters could have relied on to
make their judgements. This section highlights a few of these prominent features.
3.3.1 Word-initial voiceless plosives
A common property that distinguishes stop consonants of languages is Voice Onset Time (VOT).
English voiceless stop plosives /p t k/ are generally aspirated at word-initial position with a VOT of
50–60 milliseconds for /k/, and slightly less for /t/ and /p/ (Ladefoged 2012: 139). The opposite is
true for the Malay language, where word-initial /p t k/ are unaspirated (Othman and Atmosumarto
1995: 7), such that listeners whose language is an aspirating variety may perceive them to be voiced,
like the English plosives /b d g/ (Deterding and Poedjosoedarmo 1998: 45). VOT was thus investigated
to ascertain whether the word-initial voiceless plosives in MD speech lack aspiration or whether they
are aspirated less in both read and spontaneous speech. VOT was measured using the waveform as the
main reference (Ladefoged 2003: 94), where the onset was measured from the spike that indicated a
burst to the first complete vibration of the vocal folds (Cho and Ladefoged 1999: 215), as signalled by
the first upward-going zero crossing of the regular sinusoidal curve on the waveform. Spectrographic
information was used to aid in the decision whenever the indications on the waveform were unclear.
In this case, the onset was also at the burst, and the offset was defined as the beginning of F2 striation of
the following vowel (Thomas 2011: 117). The tokens that were measured are presented in Tables 7 and
8. Unfortunately, because of the length of extracts, there was no /p/ to be measured for the read speech,
and the number of tokens for all plosives for each speaker varied in the spontaneous extracts. Therefore,
only group means are reported in Table 9.
Table&7. Words measured for VOT in the read passage
Wor d -initial plosive
Table&8. Sample words measured for VOT in conversational speech
Wor d -initial plosive
people, Pilates, patrol
tone, town, time
competing, cook, care
Table&9. VOT group means. All measurements in milliseconds (ms)
Note: Standard Deviations in parentheses.
The voiceless plosives in ED tokens have higher VOT means than those in MD tokens for all plosives
and across all contexts, which suggests that ED speakers aspirate more for word-initial plosives than
MD speakers – almost twice as much in many circumstances. Particularly for conversation data, the
VOTs of voiceless plosives for many MD speakers were below the typical thresholds for English
aspirated stops (e.g. Byrd 1993: 6; Ladefoged 2012: 139), and are in fact representative of Malay
unaspirated voiceless stops, which could have been perceived as English voiced stops by raters.
3.3.2 Syllable-final /l/
An unexplored EMB feature that surfaced in this analysis is that of the nature of syllable-final /l/. In
many varieties of English, the lateral approximant /l/ has two allophones – clear and dark: a clear [l]
is found at pre-vocalic or intervocalic positions, while a dark [ɫ] is found at syllable-final or pre-
consonantal positions (Roach 1991: 7). These two allophonic variants differ in their realisation: dark [ɫ]
is velarised, or even pharyngealised (Thomas 2011: 126). In Malay, however, /l/ is clear in all positions
(Yunus Maris 1980: 71; Clynes and Deterding 2011: 262). The /l/ of interest in this analysis is that at the
Acoustically, [ɫ] and [l] can be differentiated by the second formant of the spectrogram: [ɫ] has a
substantially lower F2 that is almost similar to that of [w] (Thomas 2011: 127). Due to the lengths of
extracts, only a few instances of syllable-final /l/ could be measured. Additionally, syllable-final /l/
can be vocalised into back rounded vowels, and many speakers of SgE do so, where syllable-final /l/
is often replaced with [ʊ] (Brown and Deterding 2005: 12) or dropped. Nonetheless, auditorily, syllable-
final clear /l/ could be a distinguishing feature to raters. Based on spontaneous speech data, there were
many instances of syllable-final clear /l/ in MD tokens (see Table 10), but never in ED tokens.
Table&10. Instances of syllable-final clear /l/ in spontaneous speech of MD
It’s very small right, so…
… take care of the hotel, they…
… a real life…
… a real drama…
… the world alone…
… the world…
… people there are…
Prosodic features were reported to be one of the most used criteria by raters in many ethnic
identification tasks (e.g. Lim 2000), and this is unsurprising, given the very distinct prosodic features
that characterise each ethnic group. One feature that differentiates the two groups is rhythm. Prosodic
rhythm is here discussed in terms of syllable-timing and stress-timing, where the distinction is the
presumed isochrony of feet versus syllable duration (Abercrombie 1965). In ideally stress-timed
languages, the stress-foot, which consists of a stressed syllable and all unstressed syllables before the
next stress, is isochronous. In a language with syllable-timing, contrastingly, the duration of each
syllable is relatively uniform. It is, however, more accurate to treat such a categorisation as continuous
rather than as a neat dichotomy (Yaeger-Dror and Fagyal 2011: 120), since empirical investigations have
failed to prove such absolute regularity in isochrony (Laver 1994: 523–524). On a scale reflecting a
continuum between stress-timing and syllable-timing, the results from past studies can be interpreted
as indicating that British English (BrE) is stress-timed (Grabe and Low 2002), SgE (of Singaporean
Chinese subjects) is less stress-timed than BrE (Low, Grabe, and Nolan 2000; Deterding 2001), and
Standard Malay is also thought to be less stress-timed than BrE (Deterding and Poedjosoedarmo 1998:
106; Grabe and Low 2002; Deterding 2011).
The durations of vocalic and intervocalic segments of the extract of the read passage were measured
according to the normalised pairwise variability index (nPVI) outlined in Grabe and Low (2002). Vocalic
segments were defined as the intervals between vowel onset and offset, regardless of the type and
number of vowels. Intervocalic segments were defined as the interval between vowel offset and vowel
onset, regardless of the type and number of consonants, but excluding pauses or hesitations. These
segments were identified solely based on acoustic signals, rather than based on phonetic or
phonological criteria. For the text measured, the number of syllables was found to be comparable for
both ED and MD tokens (range = 34–36). In complex phonological environments, such as nasal-vowel
and glide-vowel sequences, observable transitions in the formant structure guided the segmentation.
The nPVI scores for vocalic and intervocalic segments were then computed separately in an Excel
template (Grabe 2005), and values were plotted on a scatterplot using SPSS (see Figure 4). The results
in Figure 4 clearly indicate that MD tokens have lower vocalic nPVI (VnPVI) scores than ED tokens,
and they also have slightly lower intervocalic nPVI (InPVI) scores. This suggests that ED tokens have a
more stress-timed rhythm than MD tokens.
Figure&4. nPVI scores
Although the absolute values of VnPVI scores are not directly comparable between related studies
because of differences in the materials and methods (e.g. see Ong, Deterding, and Low 2005), several
conclusions can still be drawn. Deterding (2011) compared the rhythm of Standard Malay and BrE and
found the VnPVI scores of BrE (mean = 58.52) to be higher than those of Malay (mean = 44.37). Grabe
and Low (2002: 544) also found the VnPVI of a single SgE speaker (mean = 52.3) to be much closer to
that of Standard Malay (mean = 53.6), than to that of BrE (mean = 57.2). Therefore, SgE and Standard
Malay are comparable in terms of rhythm, and both are regarded as less stress-timed than BrE.
Following these studies, two inferences can be drawn if broad generalisations are made from the results:
first, in contrast to ED, the rhythm of MD speech is less stress-timed, like that of Standard Malay and
SgE. Second, ED speech may be more stress-timed than that of typical Singaporeans. Tan and Low
(2014), who compared the rhythm of Malaysian EMBs and Singaporean EMBs, using the same passage,
found the Malaysian EMBs (mean = 41.21) to be less stress-timed than the Singaporean EMBs (mean =
47.30). As possible reasons for this result, they had posited the differences in educational and social
environments of these two countries (Tan and Low 2014: 211); Malaysia is largely a Malay-dominant
society, where Malay is the L1 of most Malaysians and the medium of instruction at all levels of
education, which all points to the possible effects of language dominance that could further explain this
In terms of melody, Tan (2010: 180–182) described the global pitch curves of MSE as being characterised
by two distinct contours – one at sentence-initial and one at sentence-final position – and a long medial
stretch of mid-high tone with little pitch movement. Tan (2010: 182) found the same patterns in the
subjects’ Malay utterances, leading her to conclude that they were features transferred from the Malay
language. Having pitch accents at the initial and final words of an utterance has also been reported as
typical of the Malay language (Deterding and Poedjosoedarmo 1998: 179; Hamzah and German 2014).
Such stretches of mid-level tones have been found in the spontaneous speech of MD speakers but are
absent in the speech of ED speakers. Examples occur in the speech of MD1 and MD5, as shown in
Figures 5 and 6.
Figure&5. Pitch contour illustrating mid-level tones found in the speech of MD1
Figure&6. Pitch contour illustrating mid-level tones found in the speech of MD5
Tan (2010: 179) also found that for EMB, there was a tendency for the last syllable of a two- or three-
syllable word to be characterised by a fall-rise-fall tune, especially in statements and wh-questions, and
regardless of the location of the lexical stress. An example is found in the speech of MD3, in the word
station, as shown in Figure 7.
Figure&7. Pitch contour illustrating a fall-rise-fall tune in the speech of MD3
3.3.5 Peak alignment
Finally, Lim (1996, 2000) found that for Malay subjects, the F0 peak of the final lexical item was aligned
later than that of the Chinese and Indian participants. In fact, when the alignment of the peak was
calculated as a ratio of the total duration of the syllables in two-syllable utterance-final words, the F0
peak was so much later that it fell on the final syllable (Lim 2000: 17). Figure 8 illustrates an example
found in the speech of MD2, in the word drama.
Figure&8. Pitch contour illustrating an alignment of the F0 peak found in the speech of MD2
The results of this study indicate that language dominance is a determinant of accent variation, at least
for EMBs in Singapore. The participants were ten EMBs who differed considerably in their language
history, use, proficiency, and attitudes. In an ethnic discriminability task, the EMBs who were more
Malay-dominant were correctly identified as ethnically Malay significantly more often and were rated
as having a significantly more perceivable Malay-accented English accent. Contrastingly, th o s e w h o
were English-dominant had an English accent that lacked ethnic-specific features so much so that naïve
raters, including raters who were EMBs, were less able to identify the speakers as ethnically Malay. The
acoustic analysis of accentual features also revealed several differences between the two groups.
However, when interpreting the results of the analysis, it is important to note that, according to
previous studies (e.g. Lim 2000), raters may rely on different features to make their judgements, and
thus features that are regarded as distinctively “Malay” by some raters may not be so for others. Further,
raters could have also relied on a combination of these features, or other features that have not been
described here, to help make their judgements.
The findings corroborate earlier studies in that bilinguals who speak the same two languages do
not necessarily have the same L1-accented L2 pronunciation, and that accentedness is influenced by
language dominance. However, a major difference in this study is that language dominance was
broadly defined as comprising four components; it is uncertain whether all are correlates of accent
variation, or whether some are more predominant than others, although this question does not reflect
the aim of the present study.
Guion, Flege, and Loftin (2000), in their study of Quichua-Spanish and Korean-English bilinguals,
explained that bilinguals use a common phonological space for the production of their two languages.
This corresponds to Flege’s (1995) single-system hypothesis, where neither of the languages can be fully
deactivated, and the amount of recent L1 use affects the strength of the influence on the L2. This could
explain the differences found in this study, given that MD speakers use Malay more regularly (40–90
per cent with friends, 60–100 per cent with family, 10–70 per cent at work) than ED speakers (0–20 per
cent with friends, 0–50 per cent with family, 0–10 per cent at work). However, ED4 uses Malay
frequently with his family (40 per cent) and ED5 does so exclusively, and so the amount of use of the
non-dominant language cannot fully account for the performance of these EMBs.
The differences in age of acquisition between MD and ED subjects as reported in the BLP survey
may also be a reason for the differences in performance. Studies on sequential bilinguals have shown
that age of acquisition may influence L1-L2 influence. Darcy and Krüger (2012) studied 9-year-old
Turkish-German bilingual children who started to acquire German from two to four years of age and
found that they had difficulties with certain vowel contrasts. Some (e.g. Flege 1995) suggest that the
differences are likely due to neural plasticity or maturation constraints, as L1 phonemic categories
become set with age. Since MD speakers frequently use Malay during their early developmental years,
it may have become harder to acquire new L2 (English) categories when they are in kindergarten.
However, McCarthy et al. (2014), in their study on Sylheti-English sequential bilingual children, have
shown that although the perception and production of English plosives were influenced by the
children’s L1 when they were 52 months old, the production matched that of their monolingual peers
a year later, after starting school. In other words, children refine their phonemic categories of two
languages with experience. It is thus difficult to explain how more than 15 years of formal English
education have not significantly influenced the MD speakers’ L1 and L2 phonemic categories.
Moreover, there is insufficient data for us to conclude that the ED speakers’ performance is due to early
input by their parents, since the SgE variant used by their caretakers was not specified. In fact, Pakir
(1994: 176) related that Colloquial SgE is used for most social interactions, and their parents would have
exhibited more prominent ethnic features (Deterding and Poedjosoedarmo 2000).
One reason why MD speakers might be resistant to change could lie in the community that each
group subscribes to and the social identity that they wish to construct. According to the BLP results,
MD speakers belong to more Malay-dominant families and social circles, and they identify more with
a Malay-speaking culture and much less with an English-speaking one. For MD speakers, then,
exhibiting ethnic-specific markers or adopting and maintaining a shared Malay-accented English could
be a way of indicating ethnic membership. Similarly, for ED speakers, their stronger sense of
membership with an English culture could motivate them to approximate and maintain a SgE accent
that would be less ethnically marked. That said, it is important to note that ethnicity and language are
distinct entities, albeit often conflated in Singapore’s context. The differences in the accents observed
are largely a result of speaking Malay rather than being Malay. In other words, we would expect ethnic
Chinese or Indians who are also English-Malay bilinguals and dominant in Malay to have similar
results as MD participants in the accent-perception tasks. Surely, as complicated as the linguistic
landscape in Singapore is, there is no one reason for the performance observed in this study.
The results of this study bear very important implications for future research on accents of SgE as
well as those of other multilingual contexts. It is known that describing the accents of SgE has largely
been an elusive task because of the existence of intricate internal variability that is said to limit the
efficacy of the research tools, regardless of the theoretical approach taken (Bao 2001: 75). It is no wonder
variationists often underscore the fact that descriptions of SgE accents only reflect tendencies.
Consequently, these variables have to be considered in any sociolinguistic work. The findings of this
study have shown that SgE speakers with contrasting language dominance can exhibit considerably
different accentual features. Being a possible determinant of accent variation, language dominance
should be considered as part of any experimental design of any sociolinguistic work focusing on SgE,
or in fact, of descriptive studies that involve bilinguals in similar contexts. The findings of the present
study also support the fact that a monolithic description of SgE is less than ideal. This leads us to the
following question: what does it mean to be an “idealised” Singaporean speaker or to sound typically
“Singaporean” if there is such observable variation across bilinguals of different EMTs and even
bilinguals of the same EMT?
This study has introduced language dominance as a potential parameter of variationist descriptions
of SgE. A preliminary diagrammatic representation of this sociolinguistic variation is proposed in
Figure 9. This model is similar to that described by Deterding and Poedjosoedarmo (2000: 7), who
described ethnic variation in terms of register (i.e. formal/informal).
Figure&9. Accent variation in SgE according to language dominance
Figure 9 reflects that the variation in the English accents of EMT-SgE bilinguals can be described
along the cline of language dominance (left). Bilinguals who are highly dominant in their EMT may
find more ethnic-specific features in their English accent, or may produce these ethnic-specific features
more consistently and prominently, such that they are more identifiably Malay/Chinese/Indian, as
depicted by the diverging peaks at the bottom. Contrastingly, those that are ED share more ethnic-
neutral features with speakers of other EMTs, as indicated by the overlapping areas of the triangles.
The convergence of the triangles as English dominance increases also denotes that the ethnicity of the
speakers becomes less identifiable. Finally, the dotted outlines of the triangles for Chinese and Indian
bilinguals indicate that their variation with respect to language dominance is only a prediction – the
model can only be finalised when future work is done on the two other speech communities.
Considering the effects of language dominance on language processing and production, and the
lack of such research in Singapore, future research should also be conducted to see whether this model
can be extended to other aspects of linguistic variation, such as grammar and lexis. For instance, some
studies (e.g. Tan 2005) have looked at the influence of the EMT on the learning of English in Singapore
schools. It would be important to see whether children who come from largely EMT-dominant families
and are thus likely to be EMT-dominant themselves perform differently compared to those who are ED
in terms of all linguistic aspects, and not only with respect to accent alone.
4.1 A h omogenising SgE accent?
A common conception of SgE accents was that, in the increasing stabilisation of SgE, accentual
variations between ethnic groups had diminished and homogenised, and that “it [was] impossible to
study ethnic varieties of English” (Bloom 1986: 416) or that ethnic features were “becoming rarer,
particularly among the younger generation” (Platt and Weber 1980: 46). Indeed, today, the stabilisation
of SgE sees pan-Singaporean accentual features across ethnic groups (e.g. Deterding 2005). Since the
introduction of the bilingual policy, English in Singapore has become more prevalent as a home
language. Following the postulation in this study, if English dominance amongst Singaporeans
increases, their SgE accents will also be observed to become less EMT-accented. If this is indeed the
case, then this conception holds even more truth for the future. Statistics have shown that, in 2010,
English, for the first time, became the dominant language of Singaporean Chinese children aged 5–14,
replacing Mandarin (Singapore Department of Statistics 2010: 11). The BLP scores of EMBs surveyed in
this study also showed that Malays are now mostly ED. Further, Chong and Seilhamer (2014) also found
that younger, educated Malays now use predominantly English in many domains, although attitudes
towards Malay are still highly positive. Perhaps, in the next two decades, ethnic-specific features may
indeed become rarer.
Abercrombie, David. 1965. Studies in Phonetics and Linguistics. London: Oxford University Press.
Bao, Zhiming. 2001. “Two Issues in the Study of Singapore English Phonology”. In Vincent B. Y. Ooi, ed. Evolving
Identities: The English Language in Singapore and Malaysia. Singapore: Times Media Private Limited, 69–78.
Birdsong, David. 2006. “Dominance, Proficiency, and Second Language Grammatical Processing”. Applied
Psycholinguistics 27: 46–49.
Birdsong, David. 2014. “Dominance and Age in Bilingualism”. Applied Linguistics 35: 374–392.
Birdsong, David, Libby Gertken, and Mark Amengual, eds. 2012. Bilingual Language Profile: An Easy-to-Use
Instrument to Assess Bilingualism. COERLL: University of Texas at Austin.
Bloom, David. 1986. “The English Language and Singapore: A Critical Survey”. In Basant K. Kapur, ed. Singapore
Studies. Singapore: Singapore University Press, 337–458.
Boersma, Paul, and David Weenink. 2018. “Praat: Doing Phonetics by Computer [Computer Program] (Version
6.0.39)”. <http://www.praat.org/> (accessed April 1, 2018).
Bokhorst-Heng, Wendy. 1999. “Singapore’s Speak Mandarin Campaign: Language Ideological Debates and the
Imagining of the Nation”. In Jan Blommaert, ed. Language Ideological Debates. Berlin: De Gruyter, 235–265.
Brown, Adam, and David Deterding. 2005. “A Checklist of Singapore English Pronunciation Features”. In David
Deterding, Adam Brown, and Low Ee Ling, eds. English in Singapore: Phonetic Research on a Corpus. Singapore:
Byrd, Dani. 1993. “54,000 American Stops”. UCLA Working Papers in Phonetics 83: 97–116.
Caffarra, Sendy, Svetlana Zimnukhova, and Simona Mancini. 2016. “What Usage Can Do: The Effect of Language
Dominance on Simultaneous Bilinguals’ Morphosyntactic Processing”. Linguistics Vanguard 2: 43–53.
Cavallaro, Francesco, and Stefan Karl Serwe. 2010. “Language Use and Language Shift among the Malays in
Singapore”. Applied Linguistics Review 1: 129–170.
Cho, Taehong, and Peter Ladefoged. 1999. “Variation and Universals in VOT: Evidence from 18 Languages”. Journal
of Phonetics 27: 207–229.
Chong, Euvin Loong, and Mark Seilhamer. 2014. “Young People, Malay and English in Multilingual Singapore”.
World Englishes 33: 363–377.
Clynes, Adrian, and David Deterding. 2011. “Standard Malay (Brunei)”. Journal of the International Phonetic
Association 41: 259–268.
Darcy, Isabelle, and Franziska Krüger. 2012. “Vowel Perception and Production in Turkish Children Acquiring L2
German”. Journal of Phonetics 40: 568–581.
Deterding, David. 1994. “The Characteristics of Singapore English Pronunciation”. Review of Educational Research
and Advances for Classroom Teachers 1.
Deterding, David. 2001. “The Measurement of Rhythm: A Comparison of Singapore and British English”. Journal
of Phonetics 29: 217–230.
Deterding, David. 2005. “Emergent Patterns in the Vowels of Singapore English”. English World-Wide 26: 179–197.
Deterding, David. 2006. “The North Wind versus a Wolf: Short Texts for the Description and Measurement of
English Pronunciation”. Journal of the International Phonetic Association 36: 187–196.
Deterding, David. 2011. “Measurements of the Rhythm of Malay”. Proceedings of the 17th International Congress of
Phonetic Sciences: 576–579.
Deterding, David, and Andy Kirkpatrick. 2006. “Emerging South-East Asian Englishes and Intelligibility”. World
Englishes 25: 391–409.
Deterding, David, and Gloria Poedjosoedarmo. 1998. The Sounds of English: Phonetics and Phonology for English
Teachers in Southeast Asia. Singapore: Prentice Hall.
Deterding, David, and Gloria Poedjosoedarmo. 2000. “To what Extent Can the Ethnic Gro up of Young
Singaporeans Be Identified from their Speech?” In Adam Brown, David Deterding, and Low Ee Ling, eds. The
English Language in Singapore: Research on Pronunciation. Singapore: Singapore Association for Applied
Dunn, Alexandra, and Jean E. Fox Tree. 2009. “A Quick, Gradient Bilingual Dominance Scale”. Bilingualism:
Language and Cognition 12: 273–289.
Flege, James E. 1995. “Second-Language Speech Learning: Theory, Findings, and Problems”. In Winifred Strange,
ed. Speech Perception and Linguistic Experience. Timonium: York Press, 233–277.
Flege, James E., Elaina Frieda, and Takeshi Nozawa. 1997. “Amount of Native-Language (L1) Use Affects the
Pronunciation of an L2”. Journal of Phonetics 25: 169–186.
Flege, James E., Ian MacKay, and Thorsten Piske. 2002. “Assessing Bilingual Dominance”. Applied Psycholinguistics
Flege, James E., Grace Yeni-Komshian, and Serena Liu. 1999. “Age Constraints on Second Language Learning”.
Journal of Memory and Language 41: 78–104.
Foley, James E. 1998. “The New Englishes: Language in the Home”. In Joseph Foley, Thiru Kandiah, Bao Zhiming,
Anthea Fraser Gupta, Lubna Alsagoff, Ho Chee Lick, Lionel Wee, Ismail Talib, and Wendy Bokhorst-Heng,
eds. English in New Cultural Contexts: Reflections from Singapore. Singapore: Prentice Hall, 218–243.
Gathercole, Virginia C. Mueller, and Thomas Enlli Môn. 2009. “Bilingual First-Language Development: Dominant
Language Takeover, Threatened Minority Language Take-Up”. Bilingualism: Language and Cognition 12: 213–
Gertken, Libby, Mark Amengual, and David Birdsong. 2014. “Assessing Language Dominance with the Bilingual
Language Profile”. In Pascale Leclercq, Amanda Edmonds, and Heather Hilton, eds. Measuring L2 Proficiency:
Perspectives from SLA. Bristol: Multilingual Matters, 208–225.
Goral, Mira, Luca Campanelli, and Avron Spiro. 2015. “Language Dominance and Inhibition Abilities in Bilingual
Older Adults”. Bilingualism: Language and Cognition 18: 79–89.
Grabe, Esther. 2005. “Esther Grabe: Personal Website”. <http://www.phon.ox.ac.uk/files/people/grabe>
(accessed April 15, 2018).
Grabe, Esther, and Low Ee Ling. 2002. “Durational Variability in Speech and the Rhythm Class Hypothesis”. In
Carlos Gussenhoven, and Natasha Warner, eds. Laboratory Phonology VII. Berlin: De Gruyter, 515–546.
Gramley, Stephan, and Kurt-Michael Pätzold. 2004. A Survey of Modern English (2nd ed.). London: Routledge.
Grosjean, François. 2008. Studying Bilinguals. Oxford: Oxford University Press.
Grosjean, François. 2010. Bilingual. Life and Reality. Cambridge: Harvard University Press.
Guion, Susan, James E. Flege, and Jonathan D. Loftin. 2000. “The Eff ect of L1 Use on Pronunciation in Quichua-
Spanish Bilinguals”. Journal of Phonetics 28: 27–42.
Gupta, Anthea Fraser. 1994. The Step-Tongue: Children’s English in Singapore. Clevedon: Multilingual Matters.
Hakuta, Kenji, and Daniel D’Andrea. 1992. “Some Properties of Bilingual Maintenance and Loss in Mexican
Background High-School Students”. Applied Linguistics 13: 72–99.
Hamzah, Diyana, and James S. German. 2014. “Intonational Phonology and Prosodic Hierarchy in Malay”. In H.
Li, and P. Ching, eds. Proceedings of Interspeech 2014. International Speech Communication Association, 106–
Harris, Catherine L., Jean Berko Gleason, and Ayse Aycicegi. 2006. “When Is a First Language More Emotional?
Psycho Physiological Evidence from Bilingual Speakers”. In Aneta Pavlenko, ed. Bilingual Minds: Emotional
Experience, Expression, and Representation. Clevedon: Multilingual Matters, 257–283.
Hernàndez-Chávez, Eduardo, Marina K. Burt, and Heidi C. Dulay. 2013. “Language Dominance and Proficiency
Testing: Some General Considerations”. NABE Journal 3: 41–54.
Hulstijn, Jan. H. 2010. “Measuring Second Language Proficiency”. In Elma Blom, and Sharon Unsworth, eds.
Experimental Methods in Language Acquisition Research. Amsterdam: Benjamins, 185–200.
Kachru, Braj B., ed. 1983. The Other Tongue: English across Cultures. Oxford: Pergamon Press.
Labov, William. 1972. “On the Mechanism of Language Change”. In John J. Gumperz, and Dell Hymes, eds.
Directions in Sociolinguistics. New York: Holt, Rinehart and Winston, 312–338.
Ladefoged, Peter. 2003. Phonetic Data Analysis. An Introduction to Fieldwork and Instrumental Techniques. Oxford:
Ladefoged, Peter. 2012. Vowels and Consonants (2nd ed.). Oxford: Blackwell Publishing.
Lau, Wen Li. 2002. “Distinguishing a Malay speaker of Singapore English: A Phonetic Study”. B.A. Honours
Dissertation, National University of Singapore.
Laver, John. 1994. Principles of Phonetics. Cambridge: Cambridge University Press.
Lim, Lisa. 1996. “Prosodic Patterns Characterising Chinese, Indian, and Malay Singapore English”. Ph.D.
Dissertation, University of Reading.
Lim, Lisa. 2000. “Ethnic Group Differences Aligned? Intonation Patterns of Chinese, Indian and Malay Singaporean
English”. In Adam Brown, David Deterding, and Low Ee Ling, eds. The English Language in Singapore: Research
on Pronunciation. Singapore: Singapore Association for Applied Linguistics, 10–21.
Lim, Valerie, Susan J. Rickard Liow, Michelle Lincoln, Chan Yiong Huak, and Mark Onslow. 2008. “Determining
Language Dominance in English-Mandarin Bilinguals: Development of a Self-Report Classification Tool for
Clinical Use”. Applied Psycholinguistics 29: 389–412.
Low, Ee Ling, Esther Grabe, and Francis Nolan. 2000. “Quantitative Characterizations of Speech Rhythm: Syllable-
Timing in Singapore English”. Language and Speech 43: 377–401.
Marian, Viorica, and Margarita Kaushanskaya. 2004. “Self-Construal and Emotion in Bicultural Bilinguals”. Journal
of Memory and Language 51: 190–201.
Marian, Viorica, Henrike K. Blumenfeld, and Margarita Kaushanskaya. 2007. “The Language Experience and
Proficiency Questionnaire (LEAP-Q): Assessing Language Profiles i n Bilinguals and Multilinguals”. Journal of
Speech, Language, and Hearing Research 50: 940–967.
McCarthy, Kathleen M., Merle Mahon, Stuart Rosen, and Bronwen G. Evans. 2014. “Speech Perception and
Production by Sequential Bilingual Children: A Longitudinal Study of Voice Onset Time Acquisition”. Child
Development 85: 1965–1980.
Miller, Lauren. 2017. “The Relationship between Language Proficiency and Language Attitudes”. Spanish in Context
Ng, Patrick. 2011. “Language Planning in Action: Singapore’s Multilingual and Bilingual Policy”. Ritsumeikan Asia
Pacific Journal 30: 1–12.
Ng, Patrick. 2014. “Mother Tongue Education in Singapore: Concerns, Issues and Controversies”. Current Issues in
Language Planning 15: 361–375.
Ong, Peter, David Deterding, and Low Ee Ling. 2005. “Rhythm in Singapore English: A Comparison of Indexes”.
In David Deterding, Adam Brown, and Low Ee Ling, eds. English in Singapore: Phonetic Research on a Corpus.
Singapore: McGraw-Hill, 74–85.
Othman, Zaharah, and Sutanto Atmosumarto. 1995. Colloquial Malay. London: Routledge.
Pakir, Anne. 1994. “Education and Invisible Language Planning: The Case of English in Singapore”. In T. Kandiah,
and J. Kwan-Ter ry, e ds . English and Language Planning: A South-East Asian Contribution. Singapore: Times
Academic Press, 158–181.
Pallier, Christophe, Angels Colomé, and Núria Sebastián-Gallés. 2001. “The Influence of Native-Language
Phonology on Lexical Access: Exemplar-Based versus Abstract Lexical Entries”. Psychological Science 12: 445–
Pendley, Charles. 1983. “Language Policy and Social Transformation in Contemporary Singapore”. Southeast Asian
Journal of Social Science 11: 46–58.
Piske, Thorsten, and Ian R. A. MacKay. 1999. “Age and L1 Use Effects on Degree of Foreign Accent in English”. In
John H. Ohala, ed. Proceedings of the 14th International Congress of Phonetic Sciences. San Francisco: University of
Platt, John, and Heidi Weber. 1980. English in Singapore and Malaysia: Status, Features and Functions. Oxford: Oxford
Pytlyk, Carolyn. 2008. “Interlanguage Prosody: Native English Speakers’ Production of Mandarin Yes-No
Questions”. In Susie Jones, ed. Proceedings of the 2008 Annual Conference of the Canadian Linguistic Association.
Rasier, Laurent, and Philippe Hiligsmann. 2007. “Prosodic Tra ns fe r from L1 to L2. Theoretical and Methodological
Issues”. Nouveaux cahiers de linguistique française 28: 41–66.
Ricento, Thomas. 2005. “Considerations of Identity in L2 Learning”. In Eli Hinkel, ed. Handbook of Research in Second
Language Teaching and Learning. Mahwah: Lawrence Erlbaum, 895–910.
Roach, Peter. 1991. English Phonetics and Phonology: A Practical Course (2nd ed.). Cambridge: Cambridge University
Rubdy, Rani. 2001. “Creative Destruction: Singapore’s Speak Good English Movement”. World Englishes 20: 341–
Schilling-Estes, Natalie. 2004. “Constructing Ethnicity in Interaction”. Journal of Sociolinguistics 8: 163–195.
Sebastian-Gallés, Núria, Antoni Rodríguez-Fornells, Ruth de Diego-Balaguer, and Begoña Díaz. 2006. “First- and
Second-Language Phonological Representations in the Mental Lexicon”. Journal of Cognitive Neuroscience 18:
Singapore Department of Statistics, ed. 2010. Census of Population 2010: Statistical Release 1: Demographic
Characteristics, Education, Language and Religion. Singapore: Department of Statistics, Ministry of Trade and
Tan, Charlene. 2005. “English or Singlish? The Syntactic Influence of Chinese and Malay on the Learning of English
in Singapore”. Journal of Language and Learning 3: 156–179.
Tan, Peter. 2012. “English in Singapore”. International Journal of Language, Translation and Intercultural Communication
Tan, Rachel, and Low Ee Ling. 2014. “Rhythmic Patterning in Malaysian and Singapore English”. Language and
Speech 57: 196–214.
Tan, Ying Ying. 2010. “Singing the Same Tune? Prosodic Norming in Bilingual Singaporeans”. In Madalena Cruz-
Ferreira, ed. Multilingual Norms. Frankfurt: Peter Lang, 173–194.
Tan, Ying Ying. 2012. “Age as a Factor in Ethnic Accent Identification in Singapore”. Journal of Multilingual and
Multicultural Development 33: 569–587.
Tan, Ying Ying. 2014. “English as a ‘Mother Tongue’ in Singapore”. World Englishes 33: 319–339.
Thomas, Erik. 2011. Sociophonetics: An Introduction. UK: Palgrave Macmillan.
Wee , Li on el. 2004. “Singapore English: Phonology”. In Edgar W. Schneider, Kate Burridge, Bernd Kortmann,
Rajend Mesthrie, and Clive Upton, eds. A Handbook of Varieties of English. Berlin: De Gruyter, 1017–1033.
Yaeger-Dror, Malcah, and Zsuzsanna Fagyal. 2011. “Analyzing Prosody”. In Marianna Di Paolo, and Malcah
Yaeger-Dror, eds. Sociophonetics: A Student’s Guide. London: Routledge, 119–130.
Yavas, Mehmet. 2007. “Multilingual Speech Acquisition”. In Sharynne McLeod, ed. The International Guide to Speech
Acquisition. Clifton Park: Thomson Delmar Learning, 96–100.
Yunus Maris. 1980. The Malay Sound System. Selangor: Penerbit Fajar Bakti.
Zilles, Ana. M. S., and Kendall King. 2005. “Self-Presentation in Sociolinguistic Interviews: Identities and Language
Variation in Panambi, Brazil”. Journal of Sociolinguistics 9: 74–94.
Theoretical and Applied Linguistics
Faculty of Modern and Medieval Languages
University of Cambridge
CB5 8BL, United Kingdom