ROBIN WALKER AND WAFA ZOGHBOR
For most English language teachers the current goal of pronunciation teaching is either (near-) native-speaker competence or “comfortable intelligibility” (Kenworthy 1987). Both goals assume a native-speaker listener and nobody seriously questioned the dependence of pronunciation teaching on NS norms until the publication of empirical data obtained from the study of interactions between non-native speakers (Jenkins 1998). These data led Jenkins to challenge the validity of native-speaker judgments of intelligibility when English was being used as a lingua franca (ELF). Her research also led to the development of the Lingua Franca Core (LFC), a set of key segmental and suprasegmental pronunciation features present in all NNS English regardless of the speaker’s accent (Jenkins 2000).
This chapter will explore the origins of the pronunciation for English as a lingua franca, before going on to detail how an ELF approach to teaching pronunciation can be put into practice. Key issues that will be dealt within the chapter include:
Kachru (1985) conceptualized the spread of English in three circles. The inner circle represents the countries where English is used as a native language (ENL) by those who have traditionally been described as its native speakers (the United States, the United Kingdom, Canada, Ireland, Australia, and New Zealand). The outer circle represents countries where English has a colonial history and is now used as a second language (ESL) alongside other official languages (Malaysia, Singapore, and Kenya, for example). The expanding circle represents the remaining countries where English is used as a foreign language (EFL), that is to say, in order to facilitate communication with the language’s native speakers in the inner circle countries.
English as a lingua franca (ELF) is not another circle to be added to Kachru’s model. Rather it refers to the ways in which English is now being used within the three circles. Most frequently, English is being used as a lingua franca between members of two or more expanding circle countries who do not share the same first language (Jenkins 2007; Seidlhofer 2005). ELF interactions are not normally defined to include, although they do not exclude, native speakers. Moreover, ELF interactions can occur within the inner circle itself, as is the case when communication is between non-native speakers visiting, working or studying in an inner circle country.
Table 24.1 summarizes the four settings. The first three (ENL, ESL, and EFL) correspond to the use of English in Kachru’s three circles. The fourth shows how ELF differs from the other three.
Table 24.1 Differences in using English in ENL, ESL, EFL, and ELF contexts.
ENL | ESL | EFL | ELF | |
Users | Traditionally referred to as the native speakers | Speakers of English as a second language | Speakers of English as a foreign language | Users of English as a lingua franca |
Location | Kachru’s inner circle | Kachru’s outer circle | Kachru’s expanding circle | No specific location; mainly involves speakers from the expanding and outer circles |
Norms | Norm providing: has long been the model for speakers from the expanding circle | Norm developing: has developed its own varieties; does not depend on NS norms and models | Norm-dependent: correctness in pronunciation based on NS norms | Uses most features of NS Englishes, but accepts all intelligible variations |
Function | Communication between native speakers | Communication in English within an ESL context | Communication with the NSs of English | International communication, predominantly between NNSs |
These different settings have implications on teaching/learning pronunciation in three dimensions:
Speakers of the same language vary in the way they speak. This variation might be due to geographical distance, social variables, or through the language evolving to meet the needs of its speakers. If variation is found across the entire linguistic system – grammar, vocabulary, pragmatics, and pronunciation – we refer to it as a dialect. Accent, in contrast, refers to variation exclusively in pronunciation. It is perfectly feasible for two speakers to use the same dialect of English with different accents. Standard English, for example, is widely spoken throughout the United Kingdom by speakers from upper-class and upper-middle-class backgrounds. However, as Trudgill points out, “[p]erhaps 9%–12% of the population of Britain … speak Standard English with some form of regional accent” (Trudgill 1999: 118).
Accents, then, are a natural and inevitable outcome of language variation. They are also one of the most salient aspects of variation, to the point that they are frequently used to classify speakers, either geographically, socially, or both. In this respect, it is not uncommon to hear language teachers and lay people refer to learners as having a foreign accent when referring to differences in NNS pronunciation compared to an NS norm. However, such a reference is problematic when English is being used as a lingua franca, since ELF, by definition, has no foreigners. Thus, whilst the goal of a great deal of pronunciation teaching in EFL, explicit or otherwise, is the (near) elimination of the learner’s foreign accent, the goal of pronunciation for learners of English as a lingua franca is mutual intelligibility across an ever-widening range of accents.
Although intelligibility has long been acknowledged as a key issue in pronunciation, linguists do not subscribe to a single definition. Smith and Nelson (1985) referred to intelligibility as the ability of the listener to recognize individual words or utterances, whilst defining comprehensibility as the listener’s ability to understand the meaning of the word or utterance in its given context. In contrast, Derwing and Munro (Munro and Derwing 1995; Derwing and Munro 1997) define intelligibility as the extent to which a speaker’s utterance is actually understood, whilst for them comprehensibility refers to the listener’s estimation of the difficulty or ease of understanding an utterance.
The work of Derwing and Munro, like that of Smith and Nelson, stresses the importance of the distinction between intelligibility and comprehensibility, though, for both groups, being able to do well in one of the two areas does not ensure success in the other (Derwing 2006; Smith and Nelson 1985). Nelson points out that “comprehensibility can fail even when the degree of intelligibility between participants is high” (Nelson 2008: 302). Zielinski (2004), for example, found that listeners could identify individual words accurately but puzzle over the whole message. Matsuura, Chiba, and Fujieda (1999), on the other hand, found that although Japanese listeners could easily understand the utterances in their study, they could not transcribe the words correctly, transcription being a standard test of intelligibility.
In their work on intelligibility, Derwing and Munro (1997) referred to the term “accentedness”, which they use in order to indicate the degree to which a particular accent differs from a local norm. Accentedness, they concluded, is quite different from intelligibility: “One very robust finding in our work is that accent and intelligibility are not the same thing. A speaker can have a very strong accent, yet be perfectly understood” (Derwing and Munro 2008: 1). The distinction they make between accent and intelligibility is crucial to ELF pronunciation, given the goal of mutual intelligibility across the range of accents that characterize ELF encounters.
A key assumption of the research on intelligibility is the belief that it is not a one-way process, that the burden to make him- or herself intelligible to the listener(s) does not lie exclusively with the speaker. For Smith and Nelson, for example, “intelligibility is not speaker- or listener-centered, but is interactional between speaker and listener” (Smith and Nelson 1985: 333). For too long, listening has been described as a receptive skill, when in practice speaker–listeners make themselves intelligible to each other by co-constructing meaning. Jenkins (2000) places great importance on this process of negotiation of meaning, and in particular highlights the role of accommodation as a central skill for successful ELF exchanges.
As with intelligibility, accommodation is not a simple concept. Communication Accommodation Theory (CAT) (Giles and Coupland 1991) holds that people’s verbal and nonverbal behavior can change according to the setting, the topic, or the interlocutor. CAT interprets the way people attune to others during an interaction by using three strategies: convergence, whereby individuals shift towards each other’s communicative behaviors; divergence, which refers to how speakers accentuate speech and nonverbal differences between themselves and others; and maintenance, whereby interactants preserve their speech patterns and other communicative behaviors in order to maintain their group identity (Giles, Coupland, and Coupland 1991). Jenkins (2000) offers an excellent introduction to accommodation theory, and in particular describes how phonological accommodation may be motivated by:
Motivation 3 does not require speakers to make any changes in their pronunciation, and so is of no interest here. Motivation 1 is interesting, but is only likely to come about when speakers work or live together on a long-term basis. In practice, however, most ELF discourse occurs through short-term interactions between interlocutors who are not yet fully competent. In such situations Motivation 2, communicative efficiency, is the driving force behind any attempts at accommodation.
In ELF interactions, changes that are deliberately made to a speaker's pronunciation constitute an important accommodation strategy, and Jenkins’ data (2000) offer s clear examples of speakers modifying their pronunciation in order to make themselves more intelligible to their interlocutors. In particular, they employed phonological accommodation in order to converge on a common ground of mutually intelligible English, and it is to what constitutes this common phonological ground that we now turn our attention.
Pronunciation targets adopted in English language teaching are generally derived from native-speaker varieties of English, principally the standard British and American English varieties of Received Pronunciation (RP) and General American (GA) respectively. In an attempt to provide similar targets for ELF pronunciation, Jenkins identified empirically which phonological features caused breakdowns in NNS–NNS communication, and revised the contents of existing pronunciation syllabuses to generate the Lingua Franca Core (LFC), the segmental and supra-segmental features required for intelligible spoken communication among NNSs (Jenkins 2000). The inclusion or exclusion of certain features from the LFC was based essentially on two criteria: their influence on intelligibility among NNS interlocutors and the concept of teachability–learnability. For Dalton and Seidlhofer (1994) “[s]ome things, say the distinction between fortis and lenis consonants, are fairly easy to describe and generalize – they are teachable” (1994: 72–73). In contrast, other areas of English pronunciation, because of their complex nature, are only learnable, i.e., acquirable, outside the classroom. For example, “the attitudinal use of intonation is something that is best acquired through talking with and listening to English speakers” (Roach 1991: 169).
Table 24.2 compares EFL pronunciation targets with the LFC. Column B lists the generally agreed EFL pronunciation targets, whilst column C indicates the impact of these features in ELF communication. Column D shows the targets for ELF pronunciation.
Table 24.2 Pronunciation targets for teaching EFL and ELF (from Zoghbor 2011a, modified from Jenkins 2005).
# | A | B | C | D |
Aspect of pronunciation | EFL targets | Impact on ELF intelligibility | ELF targets | |
1 | The consonant inventory | All sounds | √ but not all | All sounds except /θ/ and /ð/ and the allophone [ł] |
RP nonrhotic /r/ GA rhotic /r/ | Not clear | Rhotic /r/ preferred | ||
RP intervocalic [t] GA intervocalic [ɾ] | Not clear | Intervocalic [t] preferred | ||
2 | Phonetic requirements | Rarely specified | √ but not all | Aspiration after /p/, /t/, and /k/. Appropriate vowel length before fortis/lenis consonants. |
3 | Consonant clusters | All word positions | √ but not all | Word initially, word medially |
4 | Vowel quantity | Long–short contrast | √ | Long–short contrast |
5 | Vowel quality | Close to RP or GA | X | Consistent L2 regional qualities acceptable |
6 | Weak forms | Important | X | Unhelpful to intelligibility |
7 | Other features of connected speech | Important | X | Inconsequential or unhelpful |
8 | Stress-timed rhythm | Important | X | Inconsequential |
9 | Word stress | Important | X | Unnecessary/can reduce flexibility |
10 | Nuclear (tonic) stress | Important | √ | Important |
11 | Pitch movement | Important | X | Inconsequential |
The core features of the LFC are:
All the consonant sounds of the RP/GA syllabus are core features of EFL pronunciation. One very significant exception to this are the dental fricatives /θ/ and /ð/. Absent from many of the world’s languages, as well as from a number of NS varieties and regional accents, these two consonants are especially resistant to classroom teaching (Jenkins 2000; Pennington 1996). Moreover, certain substitutions of /θ/ and /ð/ are found to be fully intelligible in ELF contexts. The commonest of these are the dental plosives [t̪] and [d̪] and the labio-dental fricatives /f/ and /v/. A third substitution is that of /s/ for /θ/ and /z/ for /ð/. A preference, rather than an exception, is that of the RP intervocalic [t] over the GA intervocalic flap /ɾ/ when <t> occurs intervocalically, as in words like “water” or “matter”. This is because of the proximity of the GA variant to /d/, which can result in “matter”, for example, sounding like “madder”.
Aspiration. In the absence of aspiration following the fortis plosives /p/, /t/, and /k/ in the initial position in a stressed syllable, the listener will find it more difficult to identify /p/, /t/, and /k/ as voiceless. An unaspirated /p/ may be mistaken for /b/, unaspirated /t/ for /d/, and /k/ for /g/, with “peach” sounding like “beach”, and so on (Jenkins 2000; Osimk 2009; Rajadurai 2006; Zoghbor 2011b).
Vowel length. There is a marked shortening of vowel length before fortis consonants (fortis clipping). The long vowel /iː/ is shorter in “seat” than in “seed”, for example, and the /æ/ is shorter in “back” than in “bag”. This phonetic effect is seldom included in EFL pronunciation syllabuses, perhaps because it is considered an aspect of pronunciation for advanced-level learners only, but is a core feature for ELF.
Consonant clusters in the word-initial position are a core feature. Speakers from L1s with relatively few clusters, or with an underlying consonant–vowel syllable structure, simplify clusters, either by the addition of a vowel or by deletion of one or more of the consonants. Of these two strategies, addition is less damaging to intelligibility in ELF than deletion. Thus, while the pronunciation of “sting” as [eˈstɪŋ] or “stone” as [sɪˈtəʊn] is not found to threaten ELF intelligibility, the deletion of one of the consonants does, since it produces either [sɪŋ] or the nonsense word [tɪŋ], which might be understood as “thing”.
The maintenance of the contrast between long and short vowels, such as those found in “feel” and “fill” or “pool” or “pull”, is a core feature. Vowel quality, in contrast, is non-core. In ELF contexts, variations in vowel quality that are consistent are found to be intelligible, and are seen as a regional variation on a par with that which exists in the vowel qualities of different NS Englishes. Jenkins (2000) highlights one exception. In her data, the quality as well as the quantity of the long central vowel /ɜː/ was found to be important for ELF intelligibility.
The nuclear stress carries the most salient part of the speaker’s message, and thus the focus of the listener’s attention. Deviations in the placement of the nucleus have the potential to affect the listener’s processing of entire chunks of the message. Jenkins (2000) gives the example of a Swiss speaker explaining to her Taiwanese interlocutor how many cigarettes she smoked a day. The Taiwanese speaker responded:
you smoke more than i DO
The speaker was comparing her smoking habits with those of her interlocutor, and her failure to place the nuclear stress on “I” to signal contrast meant that she had to repeat the utterance several times before her interlocutor was able to understand it. Jenkins argues that the rules of unmarked and contrastive are simple enough for learners to master, can easily be integrated receptively and productively into classroom work, and operate at a more conscious level than the other aspects of the intonation system such as pitch movement.
A number of features have been described as having no influence on intelligibility for ELF speakers (Deterding 2010; Jenkins 2000; Zoghbor 2011b).
Certain features of NS pronunciation, particularly weak forms, schwa, and vowel reduction, are not only unnecessary for intelligibility in ELF settings; they can actually have a negative impact. Weak forms, for example, hold potential problems of “recoverability” where NNSs interlocutors are unable “to work backward from the surface form through a derivation to obtain the unique underlying representation” (Weinberger 1987: 404). The arguments against teaching schwa and vowel reduction are similar. Avery and Ehrlich (1992) pointed out that the relative absence of reduced vowel sounds did not seem to cause any misunderstanding, whilst Deterding (2010) argued that “avoiding reduced vowels is the norm in new varieties of English around the world, and speakers of such Englishes find that the use of full vowels in function words can enhance intelligibility” (2010: 9).
Although the goals and prioritized features of an ELF approach to pronunciation are clear, as we have seen in the previous two sections of the chapter, a key issue in an ELF-based approach to pronunciation teaching is the choice of a model. In EFL, with an NS accent as the goal, the model is a speaker of that accent. Choosing a model for an ELF approach is more complex; NS standard accents are not directly relevant to ELF goals and ELF is spoken with a vast range of accents. Thus, for ELF, the term “model” encompasses any speaker, with any accent, who at a minimum is competent in the features of the LFC. This speaker can be a native speaker of English, but given the demographics of ELF, is most likely to be a non-native speaker.
At a purely pragmatic level, three options are available to teachers when providing a classroom model within an ELF approach:
In practice, until ELT materials are available with competent ELF users as models, teachers will struggle with Option 2, although Walker (2010) indicates a number of ways to alleviate the problem. This leaves teachers with Options 1 and 3. Of these, although a less confident teacher will probably be more comfortable with the former, there are good reasons for promoting the third option. Clarifying a common misinterpretation of an ELF approach to pronunciation, Jenkins insists that the model “is not the LFC but the local teacher whose accent incorporates both the core features and the local version of the non-core features” (Jenkins 2007: 25). One significant outcome of Option 3 is the empowerment it supposes of the local (bilingual) NNS teacher, who is placed on (at least) equal footing with NS teachers of English.
An ELF approach to teaching pronunciation centers around two areas – competence in the LFC and good accommodation skills – and two major teaching situations – multilingual groups, with students from a range of different L1 backgrounds, and monolingual groups, where the L1 background is shared by the students, and usually the teacher.
A great deal of what can be found in existing pronunciation manuals can be applied directly to the teaching of pronunciation for ELF, although certain features of English pronunciation that are central to ELF phonological competence are often only considered suitable for advanced learners of English as a foreign language. This is the case with the aspiration of the voiceless (fortis) plosives /p, t, k/ or with fortis-clipping (the shortening of vowels when followed by voiceless consonants). It is also largely true for the treatment of word-initial consonant clusters or nuclear stress placement, both of which would need to be tackled early in an ELF approach.
Another feature of teaching pronunciation for ELF is the extent to which the learner’s L1 phonology can be brought to bear on the business of achieving competence in the LFC. Perceived as an obstacle to good pronunciation because of the negative impact L1 phonological transfer has on the target NS accent, the learner’s mother-tongue phonology has traditionally been seen as an impediment to successful learning. However, since an NS accent is not the goal for ELF, the value of the L1 phonology changes significantly, as we will see later in this chapter.
With a whole chapter dedicated to this area in The Phonology of English as an International Language, Jenkins (2000) signaled the importance of accommodation skills from the outset. In the intervening years interest has continued to grow; Deterding underscored their importance when he stated that “the emphasis of the ELF proposals on developing accommodation skills … is exceptionally constructive and valuable for English language teaching” (Deterding 2011: 94).
Depending on the levels of competence of the speaker(s)/listener(s) involved, accommodation will be need to be either receptive (with adjustments made to deal with incoming speech), productive (with adjustments made to the speaker’s own pronunciation), or both. Jenkins (2000: 187–194) and Walker (2010: 88–92) described ways in which both receptive and productive accommodation skills can be taught in the classroom, and more recently Hancock has produced teaching materials addressing the same goal (Hancock 2012, 2013). These activities include:
The vast majority of ELT takes place in classes where learners, and very often their teachers, share the same L1. These classes cannot recreate the multilingual backgrounds found in language schools in the inner circle countries, and are necessary for the activities outlined in the previous section. As a result, it is fair to say that “much thought will have to be given to the problem of accommodation in groups containing members of the same L1” (Jenkins 2000: 193). This is because of the convergence on the L1 influenced forms that characterize attempts at increased intelligibility when interlocutors share a common mother tongue, as illustrated in Table 24.3.
Table 24.3 Communication tasks and pronunciation with multilingual and monolingual classes (from Walker 2010, adapted from Jenkins 2000).
Multilingual pair/group | → | Desire to communicate; convergence on common pronunciation | → | Replacement of unintelligible features from the mother tongue by items in the LFC | → | Intelligibility and reinforcement of items from the LFC |
Monolingual pair/group | → | Desire to communicate; convergence on common pronunciation | → | Convergence on mother-tongue pronunciation | → | Intelligibility but reinforcement of mother-tongue accent |
In addition to the problem of L1 convergence, learners in a monolingual environment will receive only limited exposure to the range of accents that are commonplace in a multilingual class. This will reduce opportunities for learning to deal with accent variation and for needing to modify their own output.
In an examination of the reality of working on ELF pronunciation with monolingual groups, Walker (2001b) chose to focus on the benefits of such a situation in terms of both learner goals and the enhanced role of the NNS teacher. He later went on to suggest the use of student recordings as one way to ameliorate, although not resolve, the issue of convergence on the L1 phonology, and the subsequent divergence from internationally intelligible LFC forms (Walker 2005).
Improvements in receptive phonological accommodation are, fortunately, much less problematic for monolingual groups. Even though the group is geographically situated in a monolingual environment, technology today makes it easy for classes to access a multitude of ELF accents. Walker (2010: 95–96) mentions a number of websites that can be used to access accents from all over the world, exposure to which will help students to at least increase awareness of the problem of dealing with different accents.
The almost total absence of activities for accommodation skills training for monolingual groups stands in inverse proportion to the importance of such training for ELF users, who “must be prepared both to cope with major pronunciation differences in the speech of their different-L1 partners and to adjust their own pronunciation radically for the benefit of their different-L1 hearers” (Jenkins 2000: 194). In this respect, developing ways of improving phonological accommodation skills for such groups constitutes an important challenge for ELF pedagogy.
The Contrastive Analysis Hypothesis (CAH) (Lado 1957) held that second language (L2) phonology is filtered through the learner’s first language. Similarity between the L2 and L1 phonologies is thought to lead to positive transfer, which equates with ease in the acquisition, whilst the difference leads to negative transfer and difficulty in acquisition. Negative transfer is commonly referred to as “interference”, and while researchers today minimize the role that transfer plays in other areas of language acquisition, most agree that it operates strongly in L2 pronunciation. Given this, some approaches to teaching pronunciation for English as a foreign language openly make use of the L1 and L2 phonologies when determining priorities. In order to generate an inventory of phonological features for learners in a specific L1 context, for example, Brown (1992) suggests listing the phonemes and allophones of the L1 and L2 and determining the distributional restrictions on the L1 and L2 allophones and phonemes. Similarly, beginning her excellent summary of the processes of L1 phonological transfer (Jenkins 2000), Jenkins insists that a teaching syllabus for ELF “must be based on an understanding of the process of phonological transfer and its effects” (2000: 99).
There is a fundamental difference, however, between the value of the L1 in the teaching of pronunciation for English as a foreign language and its value for ELF. For the former, the learner's L1 is a root cause of error and L1 transfer is to be eliminated, or at least minimized, where it does not coincide with the NS target features. For the pronunciation of ELF this is not the case, since the goal here is not NS competence in the target features but intelligibility as determined by other NNSs, which, as we saw earlier, does not automatically equate with NS performance. This difference in goals allows us to view the learner's L1 phonology in a different light. Walker, for example, sees the learner’s L1 as a resource rather than an obstacle and suggests that “[b]y openly starting from the learner’s L1, we not only contemplate the reduced, achievable set of goals identified by the LFC. Equally importantly, we switch the emphasis towards what our learner CAN do (it is already part of their L1), and away from what they supposedly canNOT do” (2001b: 5). Jenkins puts the case for using the learner’s L1 phonology more strongly: “[p]honological transfer is deep-rooted and can be of benefit to learners; it is not – and should not – be abandoned easily or willingly, unless there is very good reason to do so” (Jenkins 2000: 119).
There are two ways in which the L1 phonology can be of benefit to learners: the use of L1 allophones, accents, or related languages in order to achieve competence in target features of ELF pronunciation and the fine-tuning of the LFC to a specific L1 background. Both benefits can be optimized when the teaching takes place in a monolingual setting in which the learners have the same L1 background and in which the instructors are competent in the phonetics and phonologies of both English and the L1.
With regard to the first of these, Walker (2001b) demonstrated how allophones and L1 accents allow Spanish-L1 learners to achieve ELF-intelligible pronunciations of a number of phonemes, including /z/, /ŋ/, /ʃ/, and /ʒ/: [z] is an allophone of /s/ in Spanish, occurring naturally in words like “asno” or “mismo”; [ŋ] is an allophone of /n/ before /k, g/, and is found in words like “banco” or “tengo”; /ʃ/ is a phoneme in various regional languages in Peninsular Spain; /ʒ/ is found in the “y” or “ll” of Argentinian Spanish, as in the words “yo” or “calle”.
The approach works for other L1s. Arabic-L1 speakers, for example, have problems with /p/ in English, which often sounds like /b/. However, [p] exists in Arabic as an allophone of /b/ as in [sʌpt] (“Saturday”) and [kʌpt] (“depression”). Since the allophonic variation of /b/ in this case is similar to English /p/, teachers can draw their students’ attention to this /p/-like sound and then use this variant to train learners to pronounce a similar /p/ in their English.
Loan words are a related L1 resource that teachers can draw upon. Berger (in Walker 2010) suggests the use of Adaggio from Italian to help German-L1 speakers to pronounce /dʒ/. Similarly, in the Gulf states the Chinese origin loanword “chai” /tʃaɪ/ is commonly used when referring to tea, thus providing easy access to /tʃ/ for Arab-L1 learners from this area, whilst in Malay word-final /tʃ/ can be accessed through the loanword “Mac” (March), where the sound is word-final.
The second issue to consider with respect to the use of the learner’s L1 phonology is the fine-tuning of the LFC. Jenkins (2000), for example, considers the quality of /ɜː/ a core feature; in her study, the Japanese speaker replaced /ɜː/ with /ɑː/ and was unintelligible to her interlocutor. However, Zoghbor (2011b) found that when Arab leaners replaced /ɜː/ by /eɪ/, this did not cause intelligibility problems, suggesting that the quality of /ɜː/ is a non-core feature for Arab learners. In contrast, an empirical study of Arab learners’ word stress on words of more than two syllables (Zoghbor 2011b) suggests that this is a core feature for ELF intelligibility for this particular learner group.
Fine-tuning can also reveal gaps in the LFC. In a small-scale study of the intelligibility of Brazilian students, da Silva Sili (1999) found that the while the /r/-/h/ conflation typical of Brazilian speakers was problematic, the most common difficulty came from listeners not hearing or failing to identify the last syllable in words like “gazing”, “happen”, “patches”, or “fancy”. This led him to conclude that “the strong reduction of final syllable vowels by the speakers is not included by Jenkins in her ‘core areas’, but must definitely be considered a major area for error elimination in the speech of Brazilian students” (da Silva Sili 1999: 24).
Surveys and questionnaires have revealed that both learners and teachers harbor concerns about ELF pronunciation, whilst some linguists have accused ELF of leading to a lowering of standards. Keys and Walker (2002), Jenkins (2007: 22–28), and Walker (2010: 49–61) offer a full treatment of these and other concerns and misinterpretations, and what follows is restricted to learner and teacher preferences.
A number of studies indicate that most learners want to sound like a native speaker. Dalton-Puffer, Kaltenboeck, and Smit (1997) surveyed 132 university students of English and asked them to rate unidentified native (RP, near RP, and GA) and non-native (Austrian) accents of English. The majority rated RP as their favorite model for pronunciation. The Austrian English accent was rated lowest, whilst in general the ratings reflected the respondents’ familiarity with a given accent.
Timmis (2002) surveyed 400 teachers and students, exploring preferences for native-speaker norms not just for pronunciation but also for written and spoken grammar. In terms of pronunciation goals, two-thirds of his respondents showed a preference for native speaker competence. Grau (2005) asked first-year university students of English what the objective should be in German schools regarding pronunciation. Results showed that 65% opted for international intelligibility, as opposed to near-nativeness, but 59% then went on to say that neither /s/ nor /d/ were acceptable substitutions of the interdental fricative “th”, despite the fact that Jenkins (2000) argued both variations are internationally intelligible.
Scales et al. (2006) analyzed the perceptions that 37 English language students and 10 NNS undergraduate students had of four accents: GA, BrE, Chinese English, and Mexican English. They found that “[w]hen asked to choose between wanting to be easily understood and having a native accent, the majority (62%) of English learners stated that their goal was to sound like a native speaker, compared with 38% who listed intelligibility as their pronunciation goal” (2006: 723). Interestingly, though, only 29% of the respondents were able to actually identify the American accent when asked to do so. In a blind listening task, a subject who had stated that her Asian classmates were less intelligible, chose the Chinese accent “as the easiest to understand and the one she liked most” (2006: 734). In general, Scales and her colleagues found the respondents to be unaware of the issues behind the choice of accent in a world where English has become the lingua franca.
Jenkins attributes some of the above contradictions to the linguistic insecurity that NNSs have as an outcome of the “negative stereotyping of their English by the NS community” (2004: 39). In this respect, it is interesting to note that in Timmis’ study, where two-thirds of the respondents had shown a preference for an NS accent, this figure was actually reversed when students from outer circle countries were analyzed separately. This could be accounted for by the fact that outer circle countries are endonormative regarding English, and consequently possess greater linguistic security with respect to their own English accents and pronunciation. When Tokumoto and Shibata surveyed Japanese, Korean, and Malaysian university students, they found that while the Japanese and Koreans preferred an NS accent, the Malaysian students highly valued their accented English (Tokumoto and Shibata 2011).
What is clear at the present time is that the vast majority of learners know very little about English as a lingua franca. It may be that once they are “in full possession of the socio-linguistic facts” (Jenkins 2004: 36), and once teachers see ELF intelligibility as a legitimate goal, learner preferences will shift towards ELF in those contexts in which ELF reflects the dominant use a group will make of their English.
The results of teacher surveys largely parallel those of learners. In a survey of teachers in Spain (Walker 1999), almost two-thirds said they would choose either RP or a standard British accent for their own pronunciation, with 75% selecting an NS accent. A study involving NS and NNS teachers working in Greece and the UK (Hannam 2006) found that “the majority of British participants were very critical of stigmatized British accents such as Liverpool and Belfast” (2006: 4) and would not use either in the classroom. In contrast, almost all of the Greek participants were positive about both accents, “with 100% saying they would use the Liverpool accent in the classroom and 75% the Belfast accent” (2006: 4). The Greek participants were much more critical of their own English accent, however, “with only 50% saying they would be happy to use this as a model” (2006: 4), whereas all of the British participants were positive about the idea of using the Greek English accent as a classroom model.
Jenkins carried out extensive research into teachers’ attitudes to ELF in general and ELF accents in particular (2007), and found that with regard to NNS teacher preferences “NS accents, and particularly UK and US accents, [were] preferred in all respects by this large group of expanding circle respondents” (2007: 186).
Overall, English language teachers, especially NNS teachers, value NS accents highly. One explanation for this might be related to the prestige that an NS accent can give a teacher. Good teachers want to display very high levels of competence in the language they teach, grammatical, lexical, and phonological, and for the moment phonological competence is still seen in terms of proximity to a native-speaker accent. This argument is put forward by Wach (2011) on analyzing the results of a survey of 234 Polish students, who, as English majors, are destined to become teachers of English.
There are, of course, external restraints that condition the desirability and legitimacy of ELF accents. Although individual teachers may feel drawn to ELF, they could find it difficult to implement this desire in their classrooms. Colleagues might object on the basis that they are using a traditional NS accent as their model and so are worried that an ELF approach might confuse learners. Similar objections might also come from Directors of Studies or Principals. This is especially likely in private language schools, where marketing is frequently articulated around the employment of NS teachers and models.
These sorts of pressures help to explain why teachers who responded positively to the concept of ELF accents at a theoretical level, “did not think that it would be feasible to implement the teaching of ELF accents in classrooms in their own countries or even, in most cases, to use their own proficient NNS accents as pronunciation models” (Jenkins 2007: 224).
The situation is further complicated by the fact that many international exam boards assess pronunciation in terms of the presence or absence of a foreign accent. Until this changes, teachers preparing learners for such exams are obliged to take NS accents into account. It is possible, however, that the overall attitude to ELF will change in the future. Referring to the situation for assistant English teachers in Japan, Sutherland (2008) suggests that “as awareness of ELF increases, students, their parents and other interested parties will realize that Japanese teachers should not be characterized as NNESs, with all the negative associations implied by that term, but should instead be seen as proficient ELF speakers” (2008: 10). In the meantime, teachers can become agents for raising awareness in their local environment, beginning with their colleagues, Directors of Studies, and Principals.
The teaching of the pronunciation of English as a lingua franca is a complex business, and this chapter has only been able to provide a brief first contact. For some, this will be a first and last contact, since ELF is a subject that generates sometimes fierce opposition, something the authors are fully aware of. However, everything points to English being the world’s leading lingua franca for some time to come, and on a daily basis anyone operating in this brave new world will come across examples of successful spoken communication despite decidedly non-standard, non-native speaker accents. How can this be? How can communication succeed with pronunciations so far removed from the native-speaker norm? And yet it does. Research into ELF pronunciation attempts to understand how it does and how to convert these findings into coherent pedagogy.