Nativenessand Intelligibility in Pronunciation Assessment: A Saudi Context IbrahimAlfaifiUniversityof Reading 1. Literature Review1.1.
Nativenessand Intelligibility in Pronunciation ResearchAccording toLevis (2005) and echoed by Harding (2017), teaching and, by extension,assessing second language (L2, henceforth) pronunciation can be characterizedas the tension between two ‘contradictory principles’— ‘nativeness’ and ‘intelligibility’. The nativeness principle positsthat nativelike (or unaccented) pronunciation is both a primary goal of pronunciationlearning and a standard for pronunciation assessment. However, theintelligibility principle holds that the chief goal of pronunciation learningis for learners to be understood by their interlocutors, and consequently intelligibility,instead of nativeness, emerges as an appropriate assessment criterion. Thenativeness principle has held a strong position in pronunciation teaching andassessment for a long time. However, it has regularly been criticized bypronunciation experts and applied linguists for reasons such as theimpracticality of attaining a native-speaker (NS, henceforth) accent for mostlearners (Jenkins, 2000; Munro & Derwing, 2011; Singleton, 2005), with somelearners speaking the L2 with a strong foreign accent even after years of L2immersion and only very few advanced learners being able to sound nativelike(Abrahamsson & Hyltenstam, 2009; Moyer, 1999). Also, pronunciation is thearea of language with the largest individual variation in performance, comparedto grammar or vocabulary (Abrahamsson & Hyltenstam, 2009; Moyer, 1999),where there is variation within native English varieties in segmental andsuprasegmental aspects of speech, especially in vowels (Lindemann, 2017), thus, this leads to theproblem of deciding on who should be considered an NS (Cook, 1999).
Finally, thereis a concern that nativeness involves aspects of identity (Giles, 1979; Rindal,2010; Walker, 2010), where learners may wish to retain identifiable traces of their national orfirst language (L1, henceforth) identity whenthey speak (Walker, 2010).1This hascaused some scholars to suggest that speakers should develop their own distinctvariety of speech but with the focus on comprehensible output (Jenkins, Cogo, & Dewey, 2011). This is related to theargument suggesting that what an L2 speaker needs is to be understood ratherthan sounding as if (s)he is an NS (Derwing & Munro, 2009b; Isaacs, 2013).This position is further supported by research indicating that a strong foreignaccent does not necessarily hinder intelligibility (Derwing & Munro, 1997; Derwing& Munro, 2015; Munro andDerwing, 1995; Munro& Derwing, 1999).
The claim suggesting that foreign accent can sometimeshinder communication and lower intelligibility is associated with manyunconscious negative stereotypes (Gluszek & Dovidio, 2010). The advantageof intelligibility over nativeness is further supported by empiricallydata showing that the two most widely taught English accents, RP (Received Pronunciation)and GA (General American), are less intelligible to non-native speakers (NNS(s),henceforth) than other NNS accents (Smith, 1992). Also, the heavy bias towards NS accents inpronunciation teaching was described as conformity to NSs’ norms rather thanaccuracy in creating a relationship between form and meaning (Willis, 1996). Lastly,observation has shown that in one NS discourse plus one NNS discourse, bothparticipants modify their discourse, rather than one speaker behaving normallyand the other speaking like a foreigner (Riley, 1989, as cited in Jenkins, 2000).2Ballard andWinke (2017) investigated the interplay between speakers’ accent andcomprehensibility (degree of listeners’ understanding) and their acceptabilityas EFL teachers, focusing on non-native listeners. The findings show that non-nativelisteners do not seem to readily label accented speakers as unacceptableteachers.
Instead, listeners associate speakers’ acceptability as EFL teacherswith their perceptions of these speakers’ comprehensibility. These findings areechoing previous work by Derwing and Munro (2009a), which showed a similar resultfor non-native English speaking engineers in an English-medium workplacesetting. Overall, it may not be meaningful or reliable to talk about nativenessas a measure of pronunciation. Instead, as many have argued, there is a needfor emphasis on intelligibility, which does not necessarily correlate withperceptions of non-native accent (Munro & Derwing, 1995).
1.1.1. Linguistic Features Important for Intelligibility(or Comprehensibility)It used to be thought thatintelligibility was a one-way process where NNSs are trying to make themselvesunderstood while NSs have the prerogative to decide what is intelligible or not(Bamgbose, 1988). However, this view is not any longer tenable despite beingheld widely by NSs and many teachers of English (Jenkins, 2000), and by researchin the area of pronunciation comprehensibility and intelligibility (e.g.Anderson-Hsieh, Johnson, & Koehler, 1992; Crowther, Trofimovich, Isaacs, & Saito, 2015; Derwing & Munro, 2009a;Derwing, Munro, & Wiebe, 1998; Hahn, 2004; Isaacs& Trofimovich, 2012; Kang, 2010; Kang, Rubin, & Pickering, 2010; Munro & Derwing, 1995; Munro & Derwing,1999; Munro & Derwing, 2006; Pickering, 2001; Saito, Trofimovich, , 2015; Winters & O’Brien, 2013; Zielinski, 2008).
According to studies such asSmith (1992), and Smith and Nelson (1985), intelligibility is a complex issuewhere an NS is not necessarily a perfect judge of what is intelligible or not,nor is (s)he necessarily more intelligible than an NNS. Also, according toLindemann (2017), assessing intelligibility comes with its own challenges, anddoes not eliminate the possible effects of bias against non-nativeness. Definingnon-native pronunciation as an error would promote a standard language ideology(Lippi-Green, 2012), that privileges those whose language is viewed as standardin spite of variation in that language, to the detriment of those who are notviewed as standard speakers, in some cases based purely on their appearance.3Consequently, sometest developers and researchers have explored other approaches. For example,Brown and Lumley (1998, as cited in Jenkins & Leung, 2013) created a testof English proficiency where the NS was not set as the ‘ideal’; they tried toinclude appropriate local cultural content and English language usage. There wasalso research by Harding (2012a) on the advantages of shared first language inlanguage testing. Harding examined whether test takers with a particular L1background gain advantage when listening to English passages given in their ownaccents. However, the use of English involves speakers from diverselinguacultural backgrounds, including NSs and NNSs of English, hence this necessitatesthe adoption of an alternative approach that considers this diversity.
The lingua franca approach aimsto accommodate this variation among L2 varieties, and, by extension, variationacross NS contexts which are themselves extremely diverse as a result ofmigration and inequality, among other factors. We are all, to a large extent,lingua franca users, even if we live in so-called monolingual environments(Sewell, 2017). Marckwardt (1958,as cited in Smith, 2015) suggested that NSs (not only NNSs) who speak differentvarieties of English may not understand one another and should modify theirspeech to communicate successfully. Furthermore, Field (2004, as cited in Sewell, 2017)observed that a range of different ‘standard’ accents from around the world ismore appropriate than the uncritical adoption of ‘local’ accents. Therefore, thelingua franca approach comes to prioritize adaptability and flexibility overthe ability to reproduce a predefined system, while at the same time itrecognizes the continuing importance of intelligibility.
Also, if variation occurspervasively across interactions within native English and non-native Englishvarieties, this would suggest that such variation should be accepted, and therewould then be less relevance attached to whether this variation is accuratelyperceived.A landmark text in the developmentof the lingua franca approach to intelligibility was Jenkins (2000), one of thefirst to study intelligibility in interactions between NNSs of English. Jenkins’suse of the term ‘intelligibility’ was that of Smith and Nelson (1985); however,she approached the term in the spirit of researchers such as Bansal and Ufomatawhere, accordingly, successful communication is based on phonology, not only ininternational contexts, but in intra-national ones as well. Bansal (1990)identified a number of phonetic features which are inclined to affect theintelligibility of spoken English in India, for example: lack of cleararticulation, vowel or consonant substitution, and accent on the wrong syllable.Ufomata (1990) argued that mutual intelligibility is particularly related toaccent. Overall, Jenkins (ibid.
) argued that intelligibility involves theproduction and recognition of words and utterances and, particularly, theability to produce and receive phonological form. In the late1980s Jenkins carried out a large-scale empirical research project to identifywhich features of Received Pronunciation (RP) or General American (GA) werenecessary for intelligibility in ELF communication, and which were unnecessaryor even damaging to intelligibility. The data was acquired from NNSs of Englishwith a large number of first languages interacting with each other in a widerange of contexts, both educational and social, and was analysed to identifywhich intelligibility problems could be traced directly back to pronunciation.The items that emerged as necessary for intelligibility she labelled the LinguaFranca Core, or LFC.According to Jenkins (2000), the firstto establish a phonological common core for mutual intelligibility was theAmerican linguist Hockett; however, his interest was only in communicationamong NSs, and it was descriptive rather than prescriptive. Later, thephonetician Bryan Jenner came up with a core that was motivated specifically byproviding a list of pronunciation teaching priorities that would guaranteeintelligibility for NNS learners of English. However, intelligibility is acomplex phenomenon and cannot be guaranteed by pronunciation alone (Jenkins,2000).
One more problem with Jenner’s core is his perspective towards the NS asboth producer and receiver of intelligible pronunciation. According to Jenkins (2000), thecurrent phonological orthodoxy is that suprasegmental errors are of a majoreffect on intelligibility while segmental errors are of a rather less seriouseffect. However, those who hold such orthodoxy tend to perceive communication tobe between NNSs and NSs without considered the implications of ELF norconducting investigations to support their claims; plus, as far as teaching isconcerned, some aspects of English suprasegmental system are not teachable(ibid.). Thus, the LFC seeks to create balance between suprasegmental andsegmental appropriate to ELF.1.1.
1.1. The LFC4As regardssegementals in the LFC, there are 24 consonant sounds in RP and GA in commonvital for phonological intelligibility; thus, elision or substitution of thesesounds cause a loss of intelligibility.
However, there are two consonants ofthe LFC entitled to substitution, the dental fricative pair /?/ and /ð/;substitution of these phonemes do not lead to phonological unintelligibility.The other omission from the LFC is related to a phonetic rather than phonemicfeature; it is the substitution of dark /l/ with regular substitutions such as clear /l/ or /?/. Where RP and GA differ, the GA rhoticvariant, the retroflex approximant ?, rather than the RP post-alveolar approximant ? is opted for in the LFC. However, the LFCfollows RP in terms of the consonant /t/, excluding its aspiration when itoccurs word-initially and its potential for elision when it occursword-finally, in contrast to the GA use of it where it becomes the voiced flap? once it occursintervocalically. However, the LFC includes two phonetic features: theaspiration ? following the fortisplosives (/p/, /t/ and /k/) once they take place at the beginning of a stressedsyllable; and the shortening effect of the fortis consonants on a vowel soundpreceding them in contrast to the lenis consonants which maintain the length ofa preceding vowel sound. Addition can sometimes be problematic tointelligibility, especially in the case of epenthesis in a stressed syllable.Consonant cluster simplification in the case of consonant deletion is more of athreat to intelligibility, which is not considered elision since elision isgoverned by rules in contrast to consonant deletion. The situation forconsonant cluster simplification is the same for both RP and GA except for theintervocalic cluster ‘-nt-‘ when it occurs before an unstressed syllable, wherethe core follows RP in which it does not allow for elision of the /t/ while GAdoes.
Overall, the LFC requires, as a rule, learners to produce consonantsounds similar to those of the RP and GA rather than to imitate them. There arehowever two considerations as regards vowel sounds in the LFC: quality andquantity; vowel quality is concerned with tongue and lip position while vowelquantity is concerned with length. Vowel quality is fairly stable acrossvarieties of English while vowel quantity is not; however, whatever qualitiesL2 speakers use, except the vowel sound /??/ in RP which must beincluded, they must be consistent in using their preferred vowel qualities. Asregards vowel quantity, L2 speakers must manage length distinctions amongstvowel sounds properly. The same argument holds good with diphthongs; the lengthmust be maintained and the quality must be used consistently regardless ofwhich qualities L2 speakers use.
As for suprasegmentals inthe LFC, word stress is not essential for the intelligibility of individualwords, but since it has implications for nuclear stress and soundidentification, the LFC recommends teaching it to learners but in the form ofgeneral guidelines while minding the many exceptions of word stress rules.According to the LFC, nuclear stress is the most important key to the speaker’sintended meaning, whether unmarked (on the last content word in a word group)5or contrastive (somewhere else); it is the salient part that indicates wherethe listener should pay attention. Thus, English speakers should include onenuclear syllable (although complex word groups include more than one nucleus)in each meaningful unit (or chunk) of their utterances.
1.1. Research Investigated the LFC ExperimentallyThere havebeen remarkably few published replications of Jenkins’s original study perhaps partlysince pronunciation has always tended to be of less interest than otherlinguistic levels among applied linguists and even English teachingprofessionals, and perhaps partly since orientations towards Englishpronunciation, more than to other linguistic levels, remain dominated by nativeEnglish language ideology (Jenkins, Cogo, & Dewey, 2011). The scholars who havereplicated the ELF pronunciation research have found that their findingssupport those of Jenkins (e.g.
Da Silva, 1998; Deterding, 2013, as cited in Sewell,2017; Deterding & Kirkpatrick, 2006; Kirkpatrick, 2010; Osimk, 2009; Rajadurai,2007). Forinstance, the overall conclusion of Deterding’s corpus-based study was that themost significant impact on intelligibility came from consonants, which isconsistent with the LFC proposals. There is also a study conducted by Zoghbor (2011) where she found similarresults out of comparing English learners’ performance following two differentsyllabuses where one syllabus was based on the LFC, but her findings wereinsignificant.However,Pickering (2009) found results contradictory to the LFC where she producedexperimental evidence demonstrating that pitch cues may have a role to play inELF communication.
Furthermore, Field (2005) have reached an opposite conclusion in termsof Jenkins’s conclusion that word stress does not affect intelligibility. He gavean example where the change of stress from the first to the last syllable inthe word second (i.e., “seCOND”) had reduced understanding in both native andnon-native listeners.
2. Arguments for and against the LFCAccordingto Jenkins (2009), the advantage of Lingua Franca Core is far more relevant toELF interaction contexts rather than approximating an NS’ accent; the LFC is meant forcommunication on a global level rather than with NSs only.The LFC is supposed to promotebetter intelligibility among ELF interlocutors than many NS varieties. Also,the LFC allows NNSs the same sociolinguistic rights as those enjoyed by L1speakers by legitimating NNS accents (Jenkins, 2005; Jenkins, 2007), and thus maintainingtheir social identity (walker, 2010).
This can be explained by theCommunication Accommodation Theory (CAT) (Giles & Coupland, 1991; Giles, Coupland,& Coupland, 1991). Accordingly, people’s speech might change in keepingwith its setting, the topic of discourse, and the type of person involved. Thetheory explains the way people accustom themselves to others during interactionusing three strategies: convergence, where individuals adopt to each other’scommunicative behaviours concerning a broad range of linguistic, prosodic, andnonverbal features; divergence, where speakers emphasize speech and non-verbaldifferences between themselves and others; and maintenance, whereinterlocutors preserve their speechpatterns and other communicative behaviours in order to maintain their groupidentity (Giles et al., 1991). Thus, by using such speech adjustments, NNSinterlocutors evoke the addressee’s social approval, promote communicativeefficiency between interlocutors, and maintain positive social identity (Beebe& Giles, 1984).Kubota (2001), Wells (2005), and Yamaguchi(2002) pointed out that even NSs modify their English and use simplified,sometimes ungrammatical, speech leading to a register known as ‘Foreigner Talk’to facilitate their communication.
The purpose of referring to Foreigner Talkis that what is required in ELF communication is to accommodate interlocutorsin the way NSs do it using Foreign Talk. Also, some empirical studies (e.g. Smith& Nelson, 2006; Smith & Rafiqzad, 1979) have showed that NNSs might bemore intelligible to their NNS counterparts than NSs. However, there are somestudies that have found that NSs are easier to understand than NNSs (e.g. Bent& Bradlow, 2003; Major, Fitzmaurice,Bunta, & Balasubramanian, 2002).But, for Matsuura, Chiba, and Fujieda (1999) and Rajadurai (2007), studiesrevealing lower intelligibility ratings attributed to NNSs might be assigned tofactors other than NNS phonology: for example, tolerance and attitudes towardsthe speaker.
There is also an argument suggesting that theLFC might cause diversification in language use and, as a result, inunintelligible varieties (Dziubalska-Ko?aczyk, 2005; Lee & Ridley, 1999; Tarone,1987; Yamaguchi, 2002). However, Jenkins (2000), Smith (1992), and Widdowson(1994) argued that this is rarely likely to occur. According to the languageuniversals theory (Anderson, 1987; Jakobson, 1941), there is a universality ofsolutions and substitution of sounds that are used by interlocutors when L2 featuresdo not exist in L1. For example, the dental fricatives /?/ and /ð/ in English arecommonly substituted in L2 by restricted alternatives — /t/ and /d/, /s/ and/z/ or, less commonly, /f/ and /v/ — and, thus, L1 transfer will not reduceintelligibility (Jenkins, 2000). This is also one of the reasons behindJenkins’s (2000) and (2002) suggestion that learners of ELF pronunciationshould be exposed to speakers of English with different L1s (along with NSs) inorder that they recognize the alternatives used by those speakers for somephonemes. Another argument concerns the claim of shiftingthe ownership of English (Smith, 1983; Widdowson, 1994).
Sobkowiak (2005)describes this position as “highly emotional, even hysterical” (P. 136), whichis probably a result of confusing linguistics and political or ideologicalmatters. For Jenkins (2007), this claim neglects the fact that the greatmajority of English speakers around the world are NNSs. Also, accordingto Jenkins et al. (2011), those who have criticized the LFC, e.g.
Sobkowiak(2005) and Nelson (2008), have tended not to support their criticisms withempirical evidence but only with their intuitions and even personal dogmas. Finally, since the LFC is derived mainly fromthe ELF debate (or based on the lingua franca approach) and the need to aim atintelligibility rather than NS pronunciation, the arguments on the advantage ofintelligibility over nativeness, and the ones on the lingua franca approachmentioned above can all lend support to the LFC. 1.2. Research QuestionsTrofimovich and Isaacs (2017) concludedtheir volume with questions suggested for future research; one of these questionsis ‘how dodifferent stakeholders perceive assessments of pronunciation in formal andinformal contexts?’. Thus, given this question and the significant advantage ofintelligibility over nativeness in pronunciation demonstrated above, and thelimitedness of research in pronunciation assessment in general (Harding, 2012b;Munro & Derwing, 2015), my research will come to look into the attitudes tonativeness and intelligibility in pronunciation assessment in a formal, untouchedbefore, context — a Saudi context. Trofimovichand Isaacs (2017) also stated that assessment research targeting multilingual linguafranca L2 users in non-Western contexts is lacking; thus, given thislimitation, my research will come to share in filling this literature gap bylooking at pronunciation assessment given to L2 users using English forcommunication with other L2 users with different L1s (e.g.
, Arabic, Hindi,Urdu).Furthermore, taking intoconsideration that what is more relevant to both language researchers andteachers is which linguistic dimensions are relatively more important forcomprehensibility, compared to other dimensions (Saito, Trofimovich, Isaacs, & Webb, 2017), and what is mentionedabove on the advantages of the lingua franca approach and the LFC, I willconsider the LFC items as measures for intelligibility. However, because of someof the experimental studies that replicated the ELF pronunciation research,mentioned above, have found results contradictory to some of the LFC items, Iwill exclude those items from the LFC when considering the measures forintelligibility in my study to guarantee more reliable findings. Overall, theguiding questions to my study are the following:1. What are EFL NS and NNS teachers’ beliefsabout nativeness and intelligibility in pronunciation assessment in SaudiArabia?2. To what extent do EFL NS and NNS teachersapply nativeness and intelligibility in their assessment of pronunciation inSaudi Arabia? 3. To what extent is the pronunciationassessment materials based on nativeness and intelligibility in Saudi Arabia?1 More on non-nativeness and maintaining identityis to come in section 1.
126.96.36.199. Arguments for and against the LFC.2Section 1.
188.8.131.52. Argumentsfor and against the LFC below includes more detailsregarding interlocutors’ modification of their speech.
3Section 184.108.40.206.2. Argumentsfor and against the LFC below tells more about thisideology. 4All the details of the LFC were extracted from Jenkins (2000).
5 Aword group are words which together form a meaningful unit, and it is separatedfrom the preceding and/or following word group(s) by a pause at the boundariesor, less commonly, by a change in an overall pitch level or rhythm (Jenkins,2000).