Академический Документы
Профессиональный Документы
Культура Документы
43
David Deterding and Christine Lewis
Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 786
Features of Pronunciation for ELF-Based Teaching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787
Vowel Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 789
Vowel Reduction in a Corpus of ELF Speech Misunderstandings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 789
The Incidence of Vowel Reduction in a Corpus of ELF Speech . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 790
Misunderstandings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 793
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
Abstract
Taking into consideration the nature of English as a lingua franca (ELF), a native-
speaker model for pronunciation is no longer crucial in international English
classrooms. Therefore, it is essential to reevaluate which features of pronunciation
English teachers should prioritize in order to ensure that their pupils develop a high
level of intelligibility in international settings. We specifically consider the usage of
reduced vowels, especially [ə], that occur in the weak forms of many function words
and the unstressed syllables of polysyllabic words. Through analysis of 40 1-min
extracts from a corpus of interactions between speakers from Southeast Asia, we try
to determine the frequency of reduced vowels, and we attempt to estimate what
impact the relative absence of vowel reduction has on intelligibility. Using the same
corpus, this chapter also explores some other variant pronunciations which led to
misunderstandings. Finally, we conclude the chapter with a discussion about how
teachers of English in ELF contexts should approach vowel reduction and how they
should handle the absence of a fixed model of pronunciation in ELF-based teaching.
Keywords
Pronunciation · Reduced vowels · Word stress · Intelligibility ·
Misunderstandings · English as a lingua franca (ELF)
Introduction
less agreement about exactly which areas of pronunciation should be the focus of
attention. In addition to the features mentioned in the introduction of this chapter,
Cruttenden (2014) suggests that there are some other details of native-speaker
pronunciation that learners of English do not need to imitate. Under this category,
he proposes that learners should not be required to learn the use of syllabic conso-
nants in words such as bottle and button, as [əl] and [ən] at the end of these words are
perfectly acceptable. He also states that there is no need to mimic the devoicing of
final consonants in prepausal position in words such as leave, breathe, and peas
(Cruttenden 2014, p. 193), as these words can be pronounced with the underlying
[v, ð, z] with no detriment to intelligibility. In addition, he suggests that it may
be acceptable for the consonants at the start of words such as thin and this to be
pronounced as plosives. Furthermore, he notes that the functional load for the
distinction between /ʃ/ and /ʒ/ is low, as few words of English are distinguished by
means of a contrast between just these two sounds, so it may be acceptable to use [ ʃ ]
in place of /ʒ/ (Cruttenden 2014, p. 342). Indeed, it seems unlikely that [juːʃəli] could
be heard as anything other than usually, and pleasure could only be misheard as
pressure if the /l/ in the initial cluster is pronounced as [r].
The proposals of Jenkins (2000) are rather more radical than those of Cruttenden.
She has suggested a Lingua Franca Core (LFC) of just those features of pronunci-
ation that she claims are essential for maintaining international intelligibility. She has
argued that non-core features should be free for learners of English to realize as they
choose, noting that variable pronunciation of the non-core features allows speakers
the flexibility to maintain their own distinct accent while at the same time ensuring
that they are highly intelligible (Jenkins 2007). The LFC proposed by Jenkins
includes all consonant sounds (except the dental fricatives), aspiration on /p, t, k/,
maintenance of initial and medial consonant clusters, a clear distinction between
long and short vowels, and standard use of tonic (nuclear) stress. In contrast, non-
core features that do not need to be taught include the dental fricatives, vowel
quality, the weak forms of function words, stress-timed rhythm, and the exact
pitch movement associated with intonation (Jenkins 2007, pp. 23–24). While she
originally regarded word stress as “a gray area” (Jenkins 2000, p. 150), she subse-
quently included it as part of the non-core (e.g., Jenkins 2007, p. 24), and Cruttenden
(2014, p. 352) notes that the importance of word stress represents one of the key
differences between his proposals and those of Jenkins.
Of particular relevance here is the exclusion by Jenkins (2007) of the weak forms
of function words from the LFC. Cruttenden (2014, p. 345) similarly suggests that it
is unnecessary for learners of English to use the weak forms of English, and he
argues that use of vowel reduction in the unstressed syllables of polysyllabic words
is not essential. While Roach (2009, p. 89) agrees that nonnative speakers of English
can be understood perfectly well when they only use the strong forms of function
words, he suggests (2009, p. 72) that the contrast between strong and weak syllables
is essential for the intelligibility of polysyllabic words, so he disagrees with Jenkins
and Cruttenden in this respect.
Here, we will consider the vowel reduction which may occur in two contexts: in
the weak forms of function words such as and, at, for, and to and in the unstressed
43 Pronunciation in English as Lingua Franca 789
syllables of words such as balloon and calendar. But before considering some data
that evaluates the frequency of occurrence of vowel reduction in some ELF data and
its role in maintaining intelligibility in those interactions, we should elaborate on
what is meant by vowel reduction.
Vowel Reduction
A reduced vowel is articulated in the center of the mouth, and it only occurs in
unstressed syllables. The most frequent reduced vowel is the schwa, /ə/, which is a
mid-central lax vowel that is produced without much energy (Roach 2009, p. 65);
however, /ɪ/ and /ʊ/ can also function as reduced vowels (Cruttenden 2014, p. 273).
One crucial role for a reduced vowel is to ensure that the syllable in which it occurs is
not prominent. This serves to enhance the salience of syllables that have full vowels and
thereby allows the key words of an utterance to be identified (as most function words
have reduced vowels). It also facilitates the perception of word stress which is crucial (at
least for native speakers) in enabling words to be successfully identified. In this respect,
we might note that word stress is signaled by a number of factors, including pitch
movement, duration, and loudness in addition to vowel quality (Roach 2009, p. 74).
However, Cutler (2015) notes that very few contrasts in English, such as that between
INsight and inCITE, are signaled solely by means of the pitch, duration, and loudness of
the syllables, as nearly all contrasts between words with initial and non-initial stress
involve vowel quality contrasts. For example, the noun/verb pairs CONvert vs conVERT
and also RECord vs reCORD differ over whether the first syllable has a reduced vowel
or not. Indeed, Richards (2016, p. 2) notes that native speakers tend to focus primarily on
the quality of the vowels in determining the stress pattern of words.
Though the incidence of reduced vowels seems to play a crucial role in perception
for native speakers of English, it is not clear if their role is as important in many new
varieties of English. For example, Deterding (2007, p. 28) has noted that the first
syllables of words such as adventure and compare tend to have a full vowel in
Singapore English, and Deterding and Salbrina (2013, p. 53) reported that every
single one of the 53 speakers of Brunei English that they investigated had a full
vowel in both that and had in the phrase “that had just escaped.”
The question then arises about the importance of vowel reduction in maintaining
intelligibility in international settings. How frequent is the use of full vowels in
function words and the unstressed syllables of polysyllabic words in ELF conver-
sations? And does this use of full vowels lead to any loss of intelligibility?
In Token 1, a male speaker from Nigeria had [ʌ] in the first syllable of attend, and
a female listener from Brunei was unable to suggest what the word might be; and in
Token 2, a female speaker from Malaysia had [æ] in the first syllable of agenda, and
a listener from Taiwan heard agent.
One thing that these results do not tell us is how frequent lack of vowel reduction
is in ELF conversations. We now consider a new corpus that has been collected
particularly to focus on polysyllabic words, and we investigate how frequent lack of
vowel reduction is in function words and the unstressed syllables of polysyllabic
words in 1-min extracts from the 40 recordings in this corpus.
2009, p. 90). The three other tokens of the preceding a vowel all have [ðə]: “the
electric,” “the electronic,” and “the almost.”
In contrast with the pronunciation of articles, reduced vowels are rare in other
monosyllabic function words in this ELF data. The pronunciation of a range of
function words that have a weak form with [ə] in native-speaker pronunciation
(Roach 2009, pp. 90–95) is shown in Table 3. For that, only its use as a subordinator
was considered, as in native-speaker pronunciation there is no weak form for that
when it occurs as a demonstrative (Wells 2008, p. 818).
In native-speaker pronunciation, to is generally pronounced as [tu] before a vowel
(Roach 2009, p. 93). In this data, only four of the tokens of to with a full vowel occur
before a vowel (“to express,” “to attend,” “to attach,” “to a policeman”). All the
remaining 37 tokens of the full vowel for to occur before a consonant (e.g., “to
leave,” “to sell,” “to name,” “to buy”) or before a pause.
Clearly, reduced vowels are largely avoided in monosyllabic function words apart
from a and the in this data, and only can exhibits a sizeable number (6 out of 38, i.e.,
16%) with a schwa. We might note that use of a reduced vowel in can might be
encouraged, especially for those with American usage, as it helps to differentiate can
from can’t. However, there seems little advantage in using weak forms of the other
function words listed in Table 3.
Now, let us consider polysyllabic words which would typically have a schwa in
the first syllable in native-speaker pronunciation. Table 4 shows the pronunciation of
the first syllable of a range of such words in the current data.
Most of these words have a schwa in the first syllable, especially about and above.
However, a full vowel occurs more often in the first syllable of alarm, another
(though all three tokens of another with a full vowel in the first syllable occur with
the same speaker, a male from Thailand), control, and remote.
792 D. Deterding and C. Lewis
Finally, let us consider the second syllable of polysyllabic words that would
typically have a reduced vowel in native-speaker accents. Table 5 shows a range of
such words. The words father, mother, and daughter have been excluded from this
list as they always end with a schwa (even if it can be quite long), for there is no other
way to say them. In these data, difference and conference are both treated as
bisyllabic words.
We can see that difference, person, and seven all tend to have a reduced vowel in
their final syllable and Japanese also usually (but not always) has a reduced vowel
in its second syllable. In contrast, the final syllable of children, sofa, and ticket tends
to have a full vowel and so does the second syllable of calendar.
To summarize, in the 40 min of data analyzed here, there is usually a reduced vowel
in the articles a and the, but a full vowel is far more common in other monosyllabic
function words. In the unstressed syllables of polysyllabic words, a reduced vowel
sometimes occurs, especially in the first syllable of about and the final syllable of
difference, but the use of full vowels in unstressed syllables occurs quite often in other
words, particularly the first syllable of alarm, control, and remote and the second
syllable of calendar, children, and sofa. We will now consider the impact of the
relative absence of vowel reduction in ELF interactions for intelligibility.
43 Pronunciation in English as Lingua Franca 793
Misunderstandings
the consonant at the end of tooth, and this seemed to work. We might note that
Jenkins (2000) has proposed that realization of the TH sounds may be variable in
ELF interactions without impacting on intelligibility, but the evidence from example
(2) suggests that only some replacements work: while [f] can be understood as a
realization of voiceless TH, a glottal stop [ʔ] is more problematic.
Both examples (1) and (2) involve the unexpected pronunciation of consonants. Now
let us consider vowels. An example in which a distinction between long and short
vowels caused the problem is extract (3), as the female participant from Indonesia
pronounced the vowel in peaks as [ɪ] rather than the expected /iː/ and the male from
Thailand heard “pics” (something he confirmed in subsequent feedback). In this case,
the Indonesian successfully resolved the problem by paraphrasing “peak” as “top of the
mountain,” illustrating the importance of skills at paraphrasing in international interac-
tions. (<1> and </1> in this extract show the start and end of overlapping speech.)
(3) FIndo: I think I can see three peaks [pɪks] of mountains (0.9) <1> three
peaks of mountains </1>
MThai : <1> er: </1> in the
FIndo: Yup
MThai: <2> the </2>
FIndo: <2> the top </2> of the mountain i can see three tops of mountains
MThai: Okay okay
In the examples so far analyzed, there have been no instances in which the use of
a full vowel in a monosyllabic function word gave rise to a problem. Pronouncing at
as [æt] or of as [ɒv] does not seem to have any impact on intelligibility in these ELF
interactions, and it might even make the speech more intelligible for speakers from
Southeast Asia compared to use of a reduced vowel in these function words.
However, in some instances the unexpected use of a full vowel in a polysyllabic
word may have contributed to a misunderstanding, especially when it resulted in a
perception of shifted stress. Consider, for example, extract (4), an instance of
misunderstanding that has been discussed in Lewis and Deterding (2018). (In this
extract, “(.)” indicates a short pause lasting less than half a second.)
(4) MViet: How about the BALloon? [bʌlʊn]
FIndo: <1> the? </1>
MViet : <1> that </1> i have the (.) er two BALloon (.) s (1.2) two
BALloons
FIndo : BalLOONS? <2> no </2>
MViet: <2> yeah </2> (.) you don’t have it?
FIndo: No
In this extract it sounds like the first syllable of balloon said by the male from
Vietnam was stressed, and then his interlocutor, a female from Indonesia, could not
understand this word. We might note that the male participant was able to fix the
grammar, by adding a final -s onto balloons; but he did not know how to change the
pronunciation in order to make himself understood. He just kept repeating the word
with [ʌ] in the first syllable, though after the third instance, his interlocutor managed to
figure it out.
43 Pronunciation in English as Lingua Franca 795
In (4), the stress shift is leftward, to an earlier syllable in the word than that
expected in standard pronunciation. Extract (5) shows an example in which a
misunderstanding occurred because of the use of a full vowel instead of a reduced
vowel and the consequent rightward stress shift. The female participant from Laos
subsequently stated that she heard orchestra bands spoken by the male from
Cambodia as “a crossta man.” We note here lexical issues might also have contrib-
uted to the problem: orchestra bands is a somewhat unusual collocation. Further-
more, the picture just shows three people playing music, and describing three people
as an orchestra is a little unexpected. Nevertheless, the main problem seems to be the
use of a full vowel and the resulting perception of stress on the second syllable of
orchestra, though we should note that not only is it a full vowel but it is also an
unexpected full vowel, [ʌ] instead of the [e] that might be expected and is actually
accepted in one possible realization of orchestra (Wells 2008, p. 568).
(5) MCamb: Er they are (.) orCHEStra [ɒˈkʌstrə] bands (.) they are s- (.) er
performing in front of er holiday hotel (0.9) do you have (.) er that picture?
FLaos: I have er the picture? of a man? (.) climbing up the stairs? of the
holiday hotel?
However, we should emphasize that a misunderstanding did not occur in all
instances of a full vowel occurring instead of the reduced vowel expected in
native-speaker pronunciation and the consequent perception of shifted stress. In
extract (6), the male from Laos pronounced calendar with a full vowel and most
prominence on the final syllable, but there is no indication that any misunderstanding
occurred in this instance.
(6) MLaos: A:nd (1) i <1> also have </1>
FCamb: <1> maybe </1>
MLaos: A cali- calenDAR [kælɪnˈdɑː] (0.5) hm (0.5)
FCamb: <2> is it also thing july? right? </2>
MLaos: <2> and uh the mother </2>
The ongoing investigation will try to determine how often the unexpected use of
full vowels in polysyllabic words caused a misunderstanding to occur in these ELF
interactions. We now consider the implications of these preliminary findings for ELT.
Discussion
It seems clear that there is no need for English teachers to insist on close adherence to
a native-speaker model of pronunciation in ELF-based teaching. There are many
minor allophonic variations in native-speaker accents, and it is not necessary for
learners of English to mimic them all. Indeed, there is little point on spending
valuable classroom time encouraging learners of English to develop these patterns
when they do not enhance intelligibility in international settings. For example, native
speakers tend to vary their pronunciation of the and to depending on whether the
next word begins with a consonant or a vowel, but most ELF speakers in Southeast
796 D. Deterding and C. Lewis
Asia appear not to have this alternation; and it seems to offer little benefit in
enhancing intelligibility, so there is no need for teachers to focus on it.
The speakers in the current study avoid vowel reduction in the overwhelming
majority of their tokens of monosyllabic function words such as and, at, from, of,
that, and to, and there is no evidence that this resulted in any detriment to the
intelligibility of their speech. If use of full vowels in function words does not have
any impact on intelligibility, there seems to be no need for teachers to encourage
their use; and in some cases, full vowels in most function words (though perhaps not
in can) may actually enhance intelligibility in ELF contexts. Moreover, learners of
English who pronounce both of and have with full vowels are unlikely to write
“could of” instead of “could have,” a mistake that is nowadays rather common
among many native speakers of English.
Although close imitation of native-speaker patterns is claimed to be unnecessary,
pronunciation remains vitally important for learners of English, and some features of
pronunciation, including simplification of initial clusters or omission of a final [d] in
a word such as cloud, can be problematic, so teachers of English should try to ensure
that their pupils do not exhibit patterns such as these. Furthermore, Cutler (2015)
notes that perceived stress is important for identifying words in English, so if the
absence of vowel reduction leads to a perception of shifted stress, this can result in
misunderstandings occurring. Field (2005) suggests that rightward stress shift is
more problematic than leftward stress shift, though we might note that two of
the examples given here are counter-examples, as BALloon in extract (4) involves
a leftward shift and it was misunderstood, while calenDAR in extract (6) exhibits a
rightward stress but it was not problematic. While the effects of different patterns of
stress shift need to be investigated further in data from ELF conversations, it does
seem that shifted stress can sometimes be an issue. Furthermore, if use of reduced
vowels in the unstressed syllables of polysyllabic words following native-speaker
norms can help avoid misunderstandings arising from the perception of shifted
stress, then such use of vowel reduction might be encouraged in the ELF classroom.
Jenkins (2000) has suggested that one advantage of avoiding close mimicking of
all native-speaker pronunciation habits is that it can free up classroom time for other
more valuable activities, and we can consider what kind of things this might involve.
One exceptionally useful activity is to have students record themselves reading a text
or having a conversation. In general, learners of English are often not aware of
problems in their pronunciation, and when a misunderstanding occurs, while they
can sometimes fix the grammar, they frequently do not know what to do to enhance
their pronunciation. If they knew more about their own patterns of pronunciation,
this might enable them to resolve issues when they find that they are not understood.
Though this awareness can develop from students revisiting their recordings after
direct feedback from teachers, it is also beneficial for speakers to have the opportu-
nity to evaluate themselves. Students can become more mindful of their pronunci-
ation as they listen for specific pronunciation features in their recorded speech or
attempt to transcribe their conversations.
Using classroom time for activities that require students to understand one
another can also be of great benefit. Information gap exercises, find-the-difference
tasks, dictations, and communicative pronunciation games that use minimal pairs or
43 Pronunciation in English as Lingua Franca 797
near minimal pairs require students to understand one another in order to accomplish
the tasks. Some examples of engaging tasks that focus on consonant cluster minimal
pairs might include dictations in which students have to hear whether their partner said
“the boy is on the sand” or “the boy is on the stand,” an information gap in which
learners have to find out if students are “paying with money,” or “playing with
money,” a find-the-difference in which one person has an image of someone wearing
a T-shirt with a cow on it while the other person has a T-shirt with cloud on it, or a
BINGO or matching game in which students have to call out words such as pan, plan,
planned, and, sand, stand. Even in monolingual settings, speakers will have to
differentiate between these words in their pronunciation when they participate in
such activities. When breakdowns in intelligibility occur, students can try to work it
out or ask the teacher for feedback to determine the cause. When activities focused on
minimal pairs become routine in the classroom, students can see that variant pronun-
ciation can give rise to misunderstandings and they can then work on improving their
communication skills to ensure that they are understood in international contexts.
In addition, students should be given tools to help adapt their language to their
audience. Changing one’s pronunciation to the needs of one’s listener is one kind of
accommodation, and development of skills in accommodation can be promoted in
the English classroom. Students should not just learn how to fix their pronunciation
when they find that they are misunderstood, but they might also be encouraged to
develop skills in paraphrasing what they say, maybe adopting simpler vocabulary or
else finding other ways to express themselves. These are valuable skills that can be
encouraged in the ELF-based classroom as we move away from closely imitating
increasingly irrelevant native-speaker norms of pronunciation (Walker 2010).
We also need to acknowledge that pronunciation is not the only source of
difficulties, as poor listening skills can often be the cause of misunderstandings,
and an essential skill for learners of English in ELF settings is to be able to handle a
wide range of different accents. Teachers should not just make their students familiar
with a single target accent, as they should expose them to a wide variety of materials
produced by speakers from different countries around the world. This will enable
learners of English to develop the listening skills that will help them function
successfully in the kind of ELF interactions in which they are likely to be involved
in their future lives. Language learning websites, such as English Listening Lesson
Library Online (www.elllo.org), which present speakers from a variety of countries,
can provide invaluable ELF-based pronunciation materials; and an increase in
resources that celebrate diverse varieties of English will encourage teachers and
students to develop listening skills appropriate for ELF contexts.
Finally, these suggestions are more difficult if teachers themselves do not feel that
they are empowered to speak in their own variety of pronunciation (Kirkpatrick
2007). With current pronunciation materials which generally focus on native-
speaker norms, teachers who elect ELF-based instruction face the problem of
sounding different from the model speakers they present in the class. Consequently,
there is a need for teacher training, support, and substantially more ELF-based
resources. Walker (2010) explores materials and strategies for ELF-based pronunci-
ation teaching that could be a starting place for teacher development, but there is an
urgent need for further research and resource development.
798 D. Deterding and C. Lewis
Cross-References
References
Cruttenden A (2014) Gimson’s pronunciation of English, 8th edn. Routledge, New York
Cutler A (2015) Lexical stress in English pronunciation. In: Reid M, Levis JM (eds) The handbook
of English pronunciation. Wiley Blackwell, Malden, pp 106–124
Deterding D (2007) Singapore English. Edinburgh: Edinburgh University Press.
Deterding D (2010) Norms for pronunciation in Southeast Asia. World Englishes 29(3):364–367
Deterding D (2013) Misunderstandings in English as a lingua franca: an analysis of ELF interac-
tions in South-East Asia. De Gruyter, Berlin
Deterding D, Kirkpatrick A (2006) Emerging South-East Asian Englishes and intelligibility. World
Englishes 25(3/4):391–409
Deterding D, Salbrina S (2013) Brunei English: a new variety in a multilingual society. Springer,
Dordrecht
Field J (2005) Intelligibility and the listener: the role of lexical stress. TESOL Q 39(3):399–423
Gardiner IA, Deterding D (2017) Pronunciation and miscommunication in ELF interactions: an
analysis of initial clusters. In: Jenkins J, Baker W, Dewey M (eds) The Routledge handbook of
English as a Lingua Franca. Routledge, London, pp 224–232
Jenkins J (2000) The phonology of English as an international language. Oxford University Press,
Oxford
Jenkins J (2007) English as a lingua franca: attitude and identity. Oxford University Press, Oxford
Kirkpatrick A (2007) World Englishes: implications for international communication and English
language teaching. Cambridge University Press, Cambridge, UK
Lewis C (2017) Word stress in non-native English: evidence from English as a lingua franca.
Retrieved from http://fass.ubd.edu.bn/news/lewis.html
Lewis C, Deterding D (2018) Word stress and pronunciation teaching in English as a lingua franca
contexts. CATESOL J 30(1):161–176
Mesthrie R, Bhatt RM (2008) World Englishes: the study of new linguistic varieties. Cambridge
University Press, Cambridge, UK
Richards MG (2016) Not all word stress errors are created equal: validating an English word stress
error gravity hierarchy (Unpublished doctoral dissertation). Iowa State University, Iowa
Roach P (2009) English phonetics and phonology: a practical course, 4th edn. Cambridge
University Press, Cambridge, UK
Seidlhofer B (2011) Understanding English as a lingua franca. Oxford University Press, Oxford
Walker R (2010) Teaching the pronunciation of English as a lingua franca. Oxford University Press,
Oxford
Wells JC (2008) Longman pronunciation dictionary, 3rd edn. Longman, Harlow