Teachers of English to Speakers of Other Languages, Inc.

(TESOL)

Second Language Accent and Pronunciation Teaching: A Research-Based Approach Author(s): Tracey M. Derwing and Murray J. Munro Reviewed work(s): Source: TESOL Quarterly, Vol. 39, No. 3 (Sep., 2005), pp. 379-397 Published by: Teachers of English to Speakers of Other Languages, Inc. (TESOL) Stable URL: http://www.jstor.org/stable/3588486 . Accessed: 19/12/2011 10:44
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Teachers of English to Speakers of Other Languages, Inc. (TESOL) is collaborating with JSTOR to digitize, preserve and extend access to TESOL Quarterly.

http://www.jstor.org

SecondLanguageAccent and Pronunciation Teaching: A Research-Based Approach

TRACEY M. DERWING
University Alberta of Canada Edmonton, Alberta,

MURRAY MUNRO J.

SimonFraserUniversity Canada British Vancouver, Columbia,

Empirical studies are essential to improvingour understanding of the relationshipbetween accent and pronunciation teaching. However,the study of pronunciation has been marginalized within the field of applied linguistics.As a result,teachers are often left to rely on their own intuitions with little direction. Although some instructorscan successfullyassist their students under these conditions, many others are reluctant to teach pronunciation. In this article we call for more research to enhance our knowledge of the nature of foreign accents and theireffects communication. Research of thistypehas much to on offerto teachers and studentsin termsof helping them to set learning goals, identifying appropriate pedagogical prioritiesfor the classroom, and determiningthe most effective approaches to teaching. We discuss these possibilitieswithina framework which mutual intelligibility in is the primary of consideration,although social ramifications accent must also be taken into account. We describe several problem areas and identifysome misconceptions about pronunciation instruction. In addition, we make suggestionsfor futureresearch that would address functionalload, computer-assisted intelligibility, language learning,and the role of the listener.Finally,we recommend greater collaboration between researchers and practitioners, such that more classroomrelevantresearch is undertaken.

phenomenon that we call a foreign accent is a complex aspect of language that affects speakers and listeners in both perception and production and, consequently, in social interaction. Although second language (L2) accent has long been a topic of discussion and speculation, only the last few decades have seen a systematic effortto investigate

The

TESOL QUARTERLY 39, No.3, September Vol. 2005

379

the impact of L2 accented speech on communication. Even now, much less research has been carried out on L2 pronunciation than on other skills such as grammar and vocabulary,and instructionalmaterials and practices are stillheavilyinfluenced by commonsense intuitivenotions. Though no one would question the value of intuitionbased on practical experience, complete reliance on anecdotal evidence and personal impressionsin language pedagogy has serious drawbacks. In particular, these sources cannot resolvemanyof the criticalquestions thatface classnor do theyalwayslead to valid, productive classroom room instructors activities. Therefore,the need forempirical,replicable studies to inform pronunciation instructionis clear. The focus of this article is on English as a second language (ESL) contexts such as North America, Australia, Britain, and New Zealand, where English is the language of the majority and where mutual is intelligibility a crucial concern. Given the currentlevels of immigration in these English-speaking countries,the potential for miscommunihas increased. More than cation and even language-based discrimination ever before, an understandingof accent is needed, not only on the part of instructors and applied linguists, but also by the general public. We note that in English as an internationallanguage (EIL) contexts, fromthose that issues concerning pronunciation can be quite different has argued that EIL arise in ESL environments. Jenkins (2000, 2002) learners should not have to adapt to native speaker (NS) norms but should adjust their speech to suit an audience of primarilynonnative speakers (NNSs). She has proposed a lingua franca core for pronunciation instructionthat takes into account aspects of mutual intelligibility. JenkinssuggeststhatNNS phonological phenomena that are unlikelyto cause comprehension problems for other NNSs (such as interdental fricatives)fall outside the core. She also argues against the excessive reliance on intuitionrather than empirical evidence in the creation of pronunciation syllabi (Jenkins, 2002). Like Jenkins, we argue that is mutual intelligibility the paramount concern for second language ESL learners have to make themselvesunderstood to learners; however, a wide range of interlocutorswithin a context where their L2 is the language forcommunication and where, in manycases, NSs are primary In the majority. addition, the purposes forcommunication mayvaryto a extent when immigrantsintegratesociallyin the targetculture, greater which is an importantdifferencefromEIL environments.

WHAT COUNTS AS RESEARCH?

If it is to be useful,research on pronunciation must be subject to the same constraintsas research in any other domain of second language
380 TESOL QUARTERLY

acquisition. Data collection should be rigorous and neutral ratherthan selective and biased. For example, classroom observation of particular error patterns, accompanied by analysts' commentaries on learner output, are valid waysof documenting aspects of pronunciation,but they do not in themselves constitute sufficientevidence about learners' on abilitiesor about the effects instruction learners' language output. of One problem with this approach is selection bias. As Schachter (1974) notes, observing only errors that occur in natural productions may conceal underlyingprocesses because of learner avoidance strategies. A second serious problem is the potential for observerbias in the evaluation of classroom performance. Suppose, for example, that a new pedagogical technique results in an immediate improvement on a the researcher's perparticular pronunciation feature. Unfortunately, ception maybe clouded by the desire to see studentsimprove.He or she may also overlook the possibilitythat, even if the learners' output did change, it may not have long lasting or communicative value. Still another problem is that researchers cannot assume that their own observationsof studentscoincide withthose of the community members withwhom L2 studentscommonlyinteract.In fact,researchers' perceptions of an improvement may not be shared with others. For these reasons, useful, generalizable conclusions about learners' speech are with possible only through assessmentsthat are unbiased by familiarity the speaker or the teaching situation. At present, the best available technique for achieving this goal is the collection of blind responses from community members with whom L2 speakers interact. These communities often include both NSs and NNSs, whose responses may take a wide range of forms,including ratings,transcriptions, sentence or verifications, answersto comprehension questions. The use of listeners'judgments in the evaluation of L2 speech is not withoutproblems. Their responses to an utterancemaybe influenced by such factorsas their experience with accented speech or personal bias against particularaccents or voices. Thus, such ratingshave some degree of subjectivity. studies (e.g., Derwing & Nevertheless,rating-judgment Munro, 1997; Derwing,Munro, & Wiebe, 1997, 1998; Munro & Derwing, across groups 1995, 1999, 2001) have shown a high degree of reliability of listeners,such that some shared sense of what constitutesintelligible versus unintelligibleL2 speech is possible. If thiswere not so, that is, if individual responses to L2 speech were entirelysubjective,pronunciation teaching would be pointless because the effectsof changing a for speaker's pronunciation would be different everylistener. Another limitationof the research on responses to accented speech is that insufficient workhas been done withNNS listeners.EIL research is now beginning to address this issue (see Jenkins,2002). Furthermore, studies indicate that NNSs often find understandingan L2 preliminary
SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING 381

speaker from their own L1 background easier than understanding someone froma different background (Major, Fitzmaurice,Bunta, & L1 Balasubramanian, 2002; Smith & Bisazza, 1982). However, additional work comparing the responses of NSs and NNSs is needed to develop a more complete understandingof L2 speech intelligibility.

THE MARGINALIZATION OF PRONUNCIATION WITHIN APPLIED LINGUISTICS

An extensive,growingliteratureon L2 speech has been published in journals that focus on speech production and perception, for example, and Journal Phonetics, Language Society America, of of of Acoustical Journal the Yet and Speech. thisworkis rarelycited or interpretedin teacher-oriented publications. Researchers may not be aware of this literature in part because it is inaccessible to those without specialized knowledge of phonetics. Moreover, some of the research may not be perceived as practical because it has been carried out under strictlaboratoryconditions, so that it is not immediatelyclear how the findingsapply to the classroom. However, the consequences of ignoring this importantbody of work are serious. Levis (1999), for instance, presents the disturbing observation that "present intonational research is almost completely divorced from modern language teaching and is rarely reflected in teaching materials"(p. 37). This problem can be resolved onlyifapplied technical research forpedafor linguiststake responsibility interpreting gogical specialists and incorporating pertinent findings into teacher trainingmaterialsand studenttexts.Despite teachers' increased interest in pronunciation in recent years,as evidenced by the establishmentof a TESOL interestsection and a proliferationof pronunciation materials for learners, it remains a verymarginalized topic in applied linguistics. Consider, for example, The Handbook of Second Language Acquisition (Doughty & Long, 2003), in which the authors do not mention pronunciation research. An informalsurveyof recent applied linguistics journals intended forteachers also revealsfewpapers on pronunciation,and an examination of many recent general ESL teacher texts shows only minimal attentionto pronunciation,ifany (a notable exception is CelceMurcia, 2001). Nunan (1999), forinstance,devotes only 2.5 pages to the topic, most of which is concerned with describing the critical period. Other teacher preparationbooks have even less information(e.g., Davies & Pearse, 2000; Hedge, 2000; Lightbown & Spada, 1999; Willis, 1996). Although Harmer (2001) gives more attention to pronunciation than others do, he cites no research. In the past, traditional textbooks designed to prepare ESL teachers to teach pronunciation did not
382 TESOL QUARTERLY

mention anyresearch findings(e.g., EnglishLanguage Services, explicitly 1966, 1967). In their comprehensive texts on pronunciation teaching, Dalton and Seidlhofer (1994) and Celce-Murcia,Brinton,and Goodwin (1996) introduced an important change to the field by including references to research that provide useful background information. Despite thissignificant step, the need remainsfor textsthatmake direct research findingsand waysto address specificproblems. links between The lack of attentionto pronunciation teaching in otherwiseauthoritative texts has resulted in limited knowledge about how to integrate into second language classrooms. appropriate pronunciation instruction Reputable programsoftenuse materialsbased on perceptionsof successful practice,withlittleor no referenceto research findingsand without empirical evidence of improved outcomes (Breitkreutz,Derwing, & Rossiter,2002). This situationthus creates a twofoldproblem: relatively little published research on pronunciation teaching and very little reliance on the research that does exist.

UNDERSTANDING FOREIGN ACCENTS AND THEIR EFFECTS ON THE LISTENER

It is well established thatforeignaccents are a normal consequence of second language learning. Numerous studies have shown that most people who acquire a second language afterearlychildhood are likelyto exhibitnonnativepatternsof pronunciation.The strength and nature of such accents varywithL1, withthe age when the L2 learning began, with L1-L2 use and L2 experience, and with motivationalfactors (see Piske, MacKay,& Flege, 2001, for a comprehensive review). At the same time,nativeand nonnativelistenersare surprisingly adept at noticingwhen speech differs fromtheirown variety. Flege (1984), for instance, found that phonetically untrained listeners could identify nonnative speakers on the basis of short samples of speech, including phrases,words,individualsegments,and even segmentportionsof about 30 millisecondsin length. Clearly,nativelistenersare extremely sensitive to nonnative productions.Presumablytheysucceed at detectingforeignaccented speech by relyingon multiple cues, such as segmental variations and prosodic factors.Munro, Derwing, and Burgess (2003) found that listenerscould detect accentedness even in a single word presented backward. This findingsuggests that,along with speech characteristics voice qualitymay help listeners commonlynoted in pronunciation texts, to recognize NNSs, as proposed by Esling and Wong (1983).

SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING

383

HOW RESEARCH CAN HELP SettingGoals

Research can help teachers and learners set realisticgoals. In commuis nicatively oriented ESL settings, improved intelligibility generally identifiedby pedagogical specialists as the most importantoutcome of Abercrombie (1949), forexample, statedthat pronunciation instruction. learners need no more than a comfortably intelligible pro"language nunciation" (p. 120). This view of second language speech has been reaffirmed Gilbert(1980), Penningtonand Richards (1986), Crawford by (1987), and Morley (1991), among others, yet empirical research is rarelyif ever cited to support this goal (Hahn, 2004). In fact,teachers show a great deal of confusion about what is possible and what is desirable in pronunciation instruction,despite some quite definitive research findingson this topic. In a studyof 100 adult ESL learners in Canada, Derwing (2003) found that the overwhelming majority considered speaking with perfectly nativepronunciation to be a desirable goal. It is sometimesargued (e.g., if Harmer, 2001) that learners should not be denied this possibility that is what they want, but this reasoning is incongruent with available research evidence on ultimate attainmentin pronunciation. Although some work suggests that adult L2 learners may occasionally achieve to native-like speech patterns,thisphenomenon appears to be restricted a verysmall number of highlymotivatedindividuals (Moyer,2004) and to those withspecial aptitude (Ioup, Boustagi, El Tigi, & Moselle, 1994). Studies of ultimateattainmentin general suggestthatnative-like pronunciation among those who acquire an L2 afterearlychildhood is actually exceedingly rare (Flege, Munro, & Mackay,1995; Scovel, 2000). In fact, there is no reason to believe that this goal is achievable in typicalESL classrooms. Indeed, we know of no studydocumenting a link between pronunciation instruction and the elimination of a foreign accent. Rather, most learners who strivefor nativeness are likely to become disheartened. Though all learners should be encouraged to reach their full potential, which may well exceed the minimum required for basic it intelligibility, may do more harm than good for teachers to lead learners to believe that theywill eventuallyachieve nativepronunciation or to encourage them to expend time and energyworkingtowarda goal that they are unlikelyto achieve. Thus we, and the research, disagree withHarmer (2001), who assertsthat"itwould be churlishto deny them [L2 students] such an objective" (p. 184). Out of concern forthe welfare of L2 students,teachers should help them to set realistic goals on the basis of currentresearch findings.
384 TESOL QUARTERLY

Setting Pedagogical Priorities In determining it to pedagogical priorities, is essential have an of accurate system. understanding thetarget language's phonological facts are work Although many ofEnglish phonology established, remains about tobe doneto test of interpretations assumptions native speakers' As out, English prosody. Levis(2002)haspointed somesuprasegmental to as phenomena maynot be as important meaning is sometimes claimed.He found, example, for thatnative listeners distinguished in three five of intonation contours. Before speakers L2 meanings only the aretaught patterns English of researchers teachers and must speech, ensure theinformation provide accurate. that is they an Within ESL setting,foreign a accent several has for consequences the speaker(Flege,1988). On the positive to side,it signals an NS interlocutor an L2 learner nonnative may that is and therefore require modified side, (Gass& Varonis, 1984).On thenegative an accent input in bothNS-NNSand NNS-NNS interactions mayreduce intelligibility andmay as serve a basis negative for social evaluation discrimination and of to 1997; Munro,2003). The reactions a listener (Lippi-Green, accented are complex may understood many and be at levels. speech
Munro and Derwing (1995) focused on threeaspects of foreign-accented speech (see Table 1): (a) the extent to which the speaker's intended utterance is actually understood by a listener (intelligibility), (b) the listener'sperception of the degree of difficulty encountered when trying to understand an utterance (comprehensibility), and (c) how much an from the varietyof English commonly spoken in the L2 accent differs community(accentedness). Though it is often assumed that greater accentedness automatically entails reduced intelligibility and comprehensibility, situationis not the
TABLE 1 and Intelligibility, Comprehensibility, Accentedness Term Intelligibility Definition Measure task Transcription % wordscorrect Scalarjudgment task 1 = extremely easy to understand 9 = extremely difficult to understand Scalarjudgment task 1 = no accent 9 = extremely accent strong

The extentto whicha listener an actuallyunderstands utterance Comprehensibility A listener'sperceptionof how difficult is to understandan it utterance A listener'sperceptionof how different speaker'saccent is a fromthatof the L1 community

Accentedness

SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING

385

so straightforward. fact, one of the most robust findingsin studies In the relationships among these dimensions is that they are examining partially independent. Althoughlistenerswho findspecificL2 utterances to be both unintelligible and incomprehensible always perceive such samples as heavily accented, the reverse is not necessarilytrue. Thus, listenersoften assign good comprehensibility ratingsto speech samples that they have also rated as heavily accented. Moreover, transcription studies indicate that some heavily accented speech samples are completely intelligible, while others are not (Derwing & Munro, 1997; Munro & Derwing, 1995). The gap between what is unintelligibleand what is merely heavily accented but still understandable might be explained in part by the fact that listeners use context to interpret speech. In a studyby Suenobu, Kanzaki, and Yamane (1992), English words produced byJapanese speakers were presented to 48 American scores increased English listeners for transcription.The intelligibility from 42% to 67% when the words were presented in their original sentence context as opposed to being presented in isolation. Although is research indicates that intelligibility an achievable goal, a greater understandingis needed of the relationshipbetween accent phenomena withmeaning. Assumingequal contextualinformation, and interference is one utteranceunderstandable and another unclear? An answerto why that question could be pedagogically useful because it would allow teachers to prioritize the aspects of pronunciation covered in their lessons. For instance, it is widelyaccepted that suprasegmentalsare very but importantto intelligibility, as yetfewstudies support this belief. At least three categories of studies would help elucidate the factors that interferemost with intelligibility. First, listener tasks (e.g., judgments, transcriptions,sentence verification) can show how different error types contribute to intelligibility (Anderson-Hsieh,Johnson, & Koehler, 1992; Munro & Derwing, 1995, 1999). Second, before- and after-teaching experimentsthat focus on a particularaspect of learners' can show whetherlistenersare affectedby a change in the productions students' speech (Derwing, Munro, & Wiebe, 1997, 1998; Perlmutter, 1989). Third, psycholinguisticexperiments limited to particular segments or prosodic elements can provide insightsinto the processing of accented speech (Tajima, Port, & Dalby, 1997; Wingstedt& Schulman, 1987). Hahn (2004) conducted a studyof the thirdtype,in which a suprasegwas manipulated.She playedKorean accented minilectures mentalfeature to three groups of undergraduate college students. Each lecture was stress (which Hahn identical, except withrespect to assignmentof nuclear She then measured the listeners' comprehension of calls primary stress). the lecturesand collected theirreactionsto the passages. The group who heard the appropriate assignment of nuclear stress understood
386 TESOL QUARTERLY

more of the lecture and rated the speaker more favorably significantly than the other groups. Although errorsin nuclear stressconstitute just one aspect of a foreign accent, Hahn's work indicates that they can seriously affect comprehensibility.It follows that suitable instruction should be given to L2 speakers who do not assign nuclear stressappropriately.This studyand Pennington and Ellis's (2000) research on the role of attentionin the acquisition of prosody are preciselythe type of work needed to confirm or refute approaches emphasizing suprasegmentals as suggested by pronunciation specialists (e.g., Firth, 1992; Gilbert, 1993). Ultimately,the teachers' goal should be to analyze students' speech and help them select areas for practice based on empirical findings.

Effective Determining TeachingFoci

It is somewhat surprising that so few studies have evaluated the of efficacy pronunciation teaching,given the general acknowledgement that many L2 speakers need support to improve theirintelligibility. is It all the more remarkablebecause a popular currentparadigm in applied linguisticsis the assessment of various approaches designed to have an impact on learners' productionsin other areas of language development (e.g., task-based language learning with a focus on form; Long & Crookes, 1993). Evidence suggests that acquiring pronunciation is no differentfrom acquiring syntax in this respect-students need help noticing what they are doing (Flege & Wang, 1989). We are not suggestinga returnto a comparison of methods of the typecarriedout in the 1960s and 1970s. Rather,we are concerned with matching instructional content to ESL speakers' needs. That is, we would ask whetherthe are aspects of a learner's speech thatcause problems forintelligibility the focus of instruction, regardlessof the teaching methods employed. Among the fewESL studies of pronunciation instructionare Couper (2003); Derwing,Munro, and Wiebe (1997, 1998); Macdonald, Yule, and Powers (1994); and Perlmutter (1989). These studies all suggest that pronunciation teaching can improveL2 learners' oral productions (even in individuals who have lived in an English-speakingenvironmentfor several years), but they give only limited insight into the range of changes that can be effected.Furthermore,only Derwing, Munro, and Wiebe (1997, 1998) and Perlmutter (1989) assessed improvementsin untrained listeners. intelligibility per se using a pool of linguistically more research on the effectsof pronunciation instructionis Although needed, especiallylongitudinalstudies thatcan determinehow long the improvementslast, the studies identifiedhere suggestthat teaching can be beneficial.Justas studentslearning certain grammarpoints benefit
SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING 387

frombeing explicitlyinstructed (Spada, 1997) to notice the difference between their own productions and those of Li speakers (Schmidt, 1990), so studentslearning L2 pronunciation benefitfrombeing explicitly taught phonological form to help them notice the differences between their own productions and those of proficientspeakers in the L2 community. If we accept thatpronunciation instruction can make a difference, the next step is to identifyways to tailor it to the students' needs. One important issue addressed by researchers concerns the relationship between perception and production. Numerous studies have suggested that many L2 production difficulties rooted in perception. Evidence are also indicates thatappropriate perceptual trainingcan lead to automatic improvement in production. For instance, work by Bradlow, Pisoni, Akahane-Yamada, and Tohkura (1997) has shown that when Japanese speakers are trained to perceive the /r/-/l/ distinction,their productions may automaticallyimprove, even when no production trainingis provided. This empirical findingsupports the intuitivepractice of using perceptual training tasks such as discrimination and identification exercises in the classroom (Gilbert, 1993). Another example of how research can informclassroom practice is illustrated in Derwing, Munro, and Wiebe (1998). In that study,two groups of ESL students, all of whom evidenced both segmental and received either global (mainly suprasegmental production difficulties, and were prosodic) or segmental content in their classroom instruction compared with an uninstructed control group. Both experimental groups showed improvement in perceived accentedness and comprewhen reading sentences aloud, but only the studentsin the hensibility group showed any improvementon an extemporaneous picture global more comprehennarrativetask: They were judged to be significantly In sible afterinstruction. a follow-upanalysisof the participants'productions,Derwing and Rossiter (2003) determined thatthe studentsin both experimentalgroups learned what theywere taught.The participantsin the segmental group showed considerable improvementin theirproducin tion of individual sounds, but the overall effecton comprehensibility was negligible. This observation does not lead extemporaneous speech to the conclusion that segmentsshould not be taught.Rather,Derwing, Munro, and Wiebe (1998) propose that In concernsbenefits ESL students. to attention both global and segmental a breakdown caused by a mispronunciation, the case of a communication be able to focuson the student who has receivedsegmental might training in On form a self-repetition. theotherhand,globalinstrucmispronounced skills the with thatcan be appliedin extempotionseemsto provide learner to raneousspeechproduction, despitetheneed to allocateattention several (p. speechcomponents. 407)
388 TESOL QUARTERLY

SOME CURRENT PROBLEMS AND MISCONCEPTIONS

As a resultof pronunciation's marginalized status,manyESL teachers have no formalpreparationto teach pronunciation.Breitkreutz, Derwing, and Rossiter (2002), for instance, reported that 67% of ESL teachers surveyedin Canada had no trainingat all in pronunciation instruction. This phenomenon is not limitedto NorthAmerica: Burgess and Spencer (2000) also called for more pronunciation training for teachers in Britain. MacDonald (2002) cites several studies in Australia indicating that many teachers do not teach pronunciation "because they lack confidence, skills and knowledge" (p. 3). The general lack of teacher in of preparation maypartially explain the findings another survey which 8 of 100 adult intermediateESL learners indicated that theyhad only received any pronunciation instruction, despite having been enrolled in ESL programsforextended periods of time (Derwing & Rossiter, 2002). Relyingon experiences and intuitionssometimesservesteacherswell. Those who have strong observation skills and who are phonologically aware may address learners' needs satisfactorily. These same teachers criticalevaluation skillsso thattheygain a sense of whatwill maydevelop and will not work for their students. Their intuitions may well be confirmed by research findings. However, expecting teachers to rely solely on intuitionis unrealisticand unfair.Other aspects of pedagogy receive extensiveattentionin teacher preparation courses and materials, but in many instances L2 instructorsare apparently left to teach themselveshow to address pronunciation with theirstudents. The consequences of inadequate teacher preparation are many. In some instances, studentssimplydo not receive any instruction, they or are directed to focus on the most salient characteristics theiraccent, of For regardlessof theirinfluence on intelligibility. example, Derwingand Rossiter (2002) found that, of students who were able to identify any pronunciation problem in their own speech, 90% identifiedindividual most frequent response (see segments, with th as the overwhelmingly comments later on functional load). Another concern is thatthose untrained instructors who do choose to teach pronunciation may rely too heavily on pronunciation textbooks and softwarewithout regard for their own students' problems. This strategydoes not work for several reasons. First,most materials have been designed withouta basis in pronunciation research findings.This omission precludes teachers' understanding of the rationale for the content and activities suggested and thus does not allow forappropriate matchingto students'requirements.Second, teacherswho do not have a foundation in either linguisticsor pronunciation research may not be able to make wise choices with respect to uses of computer software, whetherit is specifically intended for pronunciation instructionor not.
SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING 389

For example, Sustarsic (2003) recommends using Dragon NaturallySpeaking (ScanSoft, 1997) software(a program that automaticallytranscribes speech) as a means of providing feedback to learners on their pronunciation. Of course, if computer softwarecould actually provide useful, individualized feedback to learners on their pronunciation, the teacher's burden would be dramaticallyreduced. However,as observed by Derwing, Munro, and Carbonaro (2000), that softwarewas not intended forthispurpose. Nor does it respond to L2 users' speech in the same waythathuman listenersdo, and any adjustmentsthat the learners make to accommodate the software may be useless or even counterproductive in real interactions. A third serious problem caused by teachers' lack of knowledge of phonetics has been discussed byWang and Munro (2004), who note that ESL learners sometimesexperience pedagogical misdirectionwhen they are taught the English /i/-/I/ distinction(e.g., beatvs. bit).Hillenbrand and Clark (2000) observed that North American English speakers on distinguishbetween /i/ and /I/ primarily the basis of vowel quality ratherthan length.Yet other research (Bohn, 1995) reveals thatlearners of English from many Li backgrounds tend to perceive /i/ as a long in vowel and /I/ as a shortvowel withlittleor no difference quality.This tell studentsthat the problem is reinforcedby teachers who mistakenly important distinction between these two vowels is length. Wang and Munro (2004) showed that, with only a small amount of perceptual training,ESL learners who had received misleading instructionon this in point could learn to focus more on the differences vowel qualityand to largelyignore length. ESL instructorswho have not had opportunities for professional development in pronunciation teaching may develop some teaching strategiesthat actuallyhave littleor no value or thatmay be counterproductive. For example, in a teacher-orientedpublication, Usher (1995) claimed that the distinctionbetween /b/ and /p/ should be taught to studentsas a differencein breathing,such that /b/ requires inhalation while /p/ requires exhalation. This assertion directlycontradicts the well-established fact that normal English speech sounds are never produced while inhaling (e.g., Cruttenden, 1994). Similarly,having students hold a pencil between their nose and upper lip has been advocated as an all-purpose exercise for improvingEnglish pronunciation (Stuparyk,1996). We know of no evidence to suggest that such a technique has any value. Rather,such activitiesnot only waste students' time and money,but also may cause more problems than theysolve. is One misapprehension about L2 speech instruction thattechnology is a panacea forcorrectingpronunciation. Among the greatestpotential benefitsof computer-assisted language learning (CALL) are the opportunitiesit could provide for individualized instructionand for exposure
390 TESOL QUARTERLY

to a wide range of voices and contexts through extended listening is practice. At present,however,it seems that most available software of the "one size fits all" variety,designed to appeal to a mass market. Moreover, much of the recent CALL softwareappears to exploit the impressive multimedia capabilities of computers, rather than present content that is linguisticallyand pedagogically sound. For instance, Breitkreutzet al. (2002) found that the most popular pronunciation softwareprograms in Canadian ESL classrooms focused exclusivelyon segmentalsratherthan prosody,and thatsome had inaccurate representations of allophonic variation. Merely presenting a large number of phonological contrasts has little value with no indication of what is important,either in termsof the learner's needs (depending on L1 and on individual differences)or of what mattersfor intelligibility. To avoid these negativeoutcomes, it is importantfirst, thatinstructors have opportunitiesto learn about pronunciation pedagogy and, second, thatsuch teacher preparationbe grounded in research findings. Teacher trainersneed to help instructors the skills to critically evaluate develop materialsand curriculumon the basis of empirical research.

FUTURE RESEARCH DIRECTIONS

With the heightened interestin pronunciation now evident among pedagogical professionals, applied linguistsneed to undertake programmatic research that will extend knowledge of pronunciation learning. We see several directionsin which thiswork should proceed. to First,more research should be cgnducted on intelligibility establish the most effective of assessing it and to identifythe factors that ways contributeto it. No single approach to intelligibility assessmentcan take into account all the subtleties that might influence a listener (see Zielinski,2004). However,researchersseem to agree thatthe intelligibility of L2 utterances can at least be estimated using a varietyof techtask (e.g., Brodkey,1972; Burda, niques, one ofwhich is the transcription Scherz, Hageman, & Edwards, 2003; Suenobu, Kanzaki & Yamane, and other methods of collecting 1992). Furtherworkusing transcription listeners' responses will help establish the bases of intelligibility; in researchers need to explore how several factors interact at addition, prosodic, segmental,and voice qualitylevels. Closely linked to the intelligibility question is the role of functional load. The gravity certain typesof errorsis believed to differ, of depending on the functionalload of the phonological contraststhatthe learner has incorrectlyproduced. For instance it has been proposed that the substitutionof /t/ for /0/ (ting for thing)is less important than the substitution /b/ for/p/ (bat forpat). Although importanttheoretical of
SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING 391

work has already discussed the issue of functional load (Brown, 1991; Catford, 1987), more systematic experimental studies are needed that test predicted hierarchies,along the lines of Wingstedtand Schulman (1987). Further work on CALL softwareshould take into account research and functional load to yield materials with findings on intelligibility motivatedcontent thatmeets students' needs. Pennington appropriately and Ellis (2000) have shown that softwarecan help L2 learners acquire This prosodic patternsifit focuses theirattentionon crucial distinctions. Of course, other is a promisingarea that deserves further development. pedagogical materials should also take into consideration research findingsto provide useful instructionfor learners. with foreignA wide range of listener factors, such as familiarity accented speech, willingnessto communicate, and attitudestoward L2 (Gass & speakers,contributeto the success of any communicativeactivity & Donovan, 2003; Rubin, Varonis, 1984; MacIntyre, Baker, Clement, 1992). These phenomena require extensive furtherresearch. Preliminary evidence suggeststhat native listenerscan benefitfromtrainingto & improve theirskillsat listeningto accented speech (Derwing,Rossiter, More attention should also be focused on the mutual Munro, 2002). of intelligibility NNSs of English. This is especially importantat a time when English is increasinglyused as a lingua franca around the world (Jenkins,2000; Kachru, 1992). One of the most important challenges in the coming years is an emphasis on greatercollaboration between researchersand practitioners research. It is not reasonable to to encourage more classroom-relevant all teachers to have the expertise,time,and resources to conduct expect the type of research that we see as critical.Although some individuals setsuccessfullybalance teaching and research (usually in university work of the ESL programsin which manyinstructors tings),the structure makes it difficultto assume both roles. Ideally, teacher preparation background to programs should provide ESL teachers with sufficient enable them to assess their students' pronunciation problems and to evaluate research findings,materials,and techniques to detercritically mine theirapplicabilityfor theirstudents.At the same time,researchers need to understand classroom dynamics and students so that theycan workin concert withteachers to ensure appropriate research methodolwithan ogy and meaningfulfindings.In the meantime,applied linguists ESL teacher preparation interestin pronunciation should ensure that programs offer courses in pronunciation pedagogy firmlyrooted in existing research. Researchers and teachers owe this to ESL students, many of whom view pronunciation instructionas a priority.

392

TESOL

QUARTERLY

ACKNOWLEDGMENTS
to Our researchdescribedhere wassupportedbygrants bothauthorsfromtheSocial Sciences and HumanitiesResearch Council of Canada. We would like to thanktwo and on anonymousreviewers JohnLevis fortheirhelpfulcomments an earlierdraft to of thisarticle.We are also grateful Marian Rossiterand Ron Thomson for their ESL students, who first interested in us we input.Finally, are indebtedto our former and intelligibility. issues of pronunciation

THE AUTHORS
Tracey M. Derwing is a professorin the TESL program in the Departmentof in at of Educational Psychology the University Alberta, Edmonton,Alberta, Canada, of and and codirector the PrairieCentreofExcellence forResearchon Immigration include issues related to L2 accent as well as Integration.Her research interests refugeesettlement. in at Munrois a professor linguistics Simon FraserUniversity Vancouver, of MurrayJ. BritishColumbia, Canada. As a specialistin applied phonetics,he has carriedout research on age and foreignaccent, L2 speech learning,and the perception of foreignaccented speech.

REFERENCES
D. Abercrombie, (1949). Teaching pronunciation. 3, English LanguageTeaching, 113122. between Anderson-Hsieh, Johnson,R., & Koehler, K. (1992). The relationship J., nativespeakerjudgments nonnative of and deviancein segmentals, pronunciation and syllablestructure. 42, prosody, LanguageLearning, 529-555. Bohn, O.-S. (1995). Cross-languagespeech perception in adults: First language transfer doesn't tell it all. In W. Strange (Ed.), Speech and perception linguistic Issues in cross-language research experience: (pp. 279-304). Timonium, MD: York Press. Bradlow,A. R., Pisoni,D. B., Akahane-Yamada, & Tohkura,Y. (1997). Training R., to of Japanese listeners identify English/r/ and /1/:IV. Some effects perceptual 101, learning on speech production.Journalof theAcoustical Society America, of 2299-2310. Breitkreutz, Derwing,T. M., & Rossiter,M. J. (2002). Pronunciationteaching J., 19, practicesin Canada. TESL CanadaJournal, 51-61. A Brodkey,D. (1972). Dictationas a measure of mutual intelligibility: pilot study. 22, LanguageLearning, 203-220. In Brown,A. (1991). Functionalload and the teachingof pronunciation. A. Brown A (Ed.), Teaching English pronunciation: bookofreadings (pp. 211-224). London: Routledge. Burda, A. N., Scherz,J. A., Hageman, C. F., & Edwards,H. T. (2003). Age and of and understanding speakerswithSpanishor Taiwaneseaccents.Perceptual Motor Skills, 11-20. 97, Burgess,J., & Spencer, S. (2000). Phonology and pronunciationin integrated 28, language teachingand teachereducation. System, 191-215.

SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING

393

Catford, C. (1987). Phonetics and the teaching of pronunciation:A systemic J. on description of English phonology. In J. Morley (Ed.), Current perspectives VA: TESOL. pronunciation 87-100). Alexandria, (pp. or Celce-Murcia,M. (Ed.) (2001). Teaching Englishas a second foreign language(3rd ed.). Boston: Heinle & Heinle. A Celce-Murcia,M., Brinton,D., & Goodwin,J. (1996). Teaching pronunciation: to Oxford,England: Oxford reference teachers English speakers other for of of languages. Press. University Couper, G. (2003). The value of an explicit pronunciation syllabus in ESOL 18(3), 53-70. teaching.Prospect, W. Crawford, W. (1987). The pronunciationmonitor:L2 acquisitionconsiderations In and pedagogical priorities. J. Morley(Ed.), Current on perspectivespronunciation: anchored theory 101-121). Alexandria,VA: TESOL. in Practices (pp. A. Cruttenden, (1994). Gimson's (5th ed.). London: Edward pronunciation English of Arnold. B. Dalton, C., & Seidlhofer, (1994). Pronunciation. Oxford,England: OxfordUniverPress. sity in Oxford,England: Davies, P., & Pearse, E. (2000). Success English language teaching. Press. OxfordUniversity Derwing,T. M. (2003). What do ESL studentssay about their accents? Canadian Modern 59, LanguageReview, 547-566. T. and Derwing, M., & Munro,M.J. (1997). Accent,intelligibility, comprehensibility: Evidence fromfourLls. Studies Second in 19, LanguageAcquisition, 1-16. Derwing,T. M., Munro, M. J., & Carbonaro, M. D. (2000). Does popular speech software workwithESL speech? TESOL Quarterly, 592-603. 34, recognition for Derwing,T. M., Munro, M. J., & Wiebe, G. (1997). Pronunciationinstruction "fossilized" learners:Can it help? Applied 8, LanguageLearning, 217-235. Derwing,T. M., Munro, M. J., & Wiebe, G. (1998). Evidence in favorof a broad instruction. for framework pronunciation 48, LanguageLearning, 393-410. Derwing, T. M., & Rossiter,M. J. (2002). ESL learners' perceptions of their 30, System, 155-166. pronunciationneeds and strategies. on of instruction M. T. Derwing, M., & Rossiter, J. (2003). The effects pronunciation of the accuracy,fluencyand complexity L2 accented speech. Applied Language 13, Learning, 1-18. M. Derwing,T. M., Rossiter, J., & Munro,M. J. (2002). Teaching nativespeakersto and listento foreignaccented speech.Journal Multilingual Multicultural Developof ment, 245-259. 23, of Doughty, C. J., & Long, M. H. (Eds.). (2003). The handbook secondlanguage Malden, MA: Blackwell. acquisition. in pronunciation: English Language Services,Inc. (1966). Drillsand exercises English New York: Collier Macmillan. Consonants vowels. and in pronunciation: English Language Services,Inc. (1967). Drillsand exercises English and Stress intonation, 1. New York: Collier Macmillan. part Esling, J., & Wong, R. F. (1983). Voice quality settingsand the teaching of 17, pronunciation.TESOL Quarterly, 89-95. & Firth,S. (1992). Pronunciationsyllabusdesign: A question of focus. In P. Avery American S. Ehrlich (Eds.), Teaching (pp. English pronunciation 173-193). Oxford, Press. England: OxfordUniversity Journal of Flege,J. E. (1984). The detectionof Frenchaccent byAmericanlisteners. the Acoustical 76, Society America, 692-707. of Flege, J. E. (1988). The production and perception of foreignlanguage speech review--1988 a and sounds. In H. Winitz(Ed.), Human communication itsdisorders, (pp. 224-401). Norwood,NJ:Ablex. 394 TESOL QUARTERLY

of strength Flege,J. E., Munro, M. J., & MacKay,I. R. A. (1995). Factorsaffecting of Society of perceived foreignaccent in a second language.Journal theAcoustical America, 3125-3134. 97, affect how Flege,J. E., & Wang, C. (1989). Nativelanguage phonotacticconstraints well Chinese subjectsperceivethewordfinalEnglish/t/- /d/ contrast. Journal of 17, Phonetics, 299-315. of on Gass, S., & Varonis,E. M. (1984). The effect familiarity the comprehensibility of nonnativespeech. LanguageLearning, 65-89. 34, Gilbert, (1980). Prosodic development:Some pilot studies.In R. C. Scarcella & J. Selected in S. D. Krashen (Eds.), Research second of Los language acquisition: papers the Forum(pp. 110-117). Rowley,MA: Research SecondLanguageAcquisition Angeles House. Newbury in Pronunciation listening and Gilbert, B. (1993). Clearspeech: comprehensionNorth J. book Press. American Student's (2nd ed.). NewYork: CambridgeUniversity English. stressand intelligibility: Research to motivatethe Hahn, L. D. (2004). Primary TESOL Quarterly, 201-223. 38, teachingof suprasegmentals. Harmer, (2001). Thepractice (3rd J. language teaching ed.). Harlow,England: ofEnglish Pearson Education. and learning thelanguageclassroom. in Oxford,England: Hedge, T. (2000). Teaching Press. OxfordUniversity Hillenbrand,J. M., & Clark, M. J. (2000). Some effectsof duration on vowel 108, 3013-3022. recognition. Journal the of Acoustical Society America, of Ioup, G., Boustagi,E., El Tigi, M., & Moselle, M. (1994). Reexaminingthe critical A adult SLA in a naturalistic environperiod hypothesis: case studyof successful ment.Studies Second in 16, LanguageAcquisition, 73-98. J. Jenkins, (2000). The phonology Englishas an international of language.Oxford, Press. England: OxfordUniversity Jenkins, (2002). A sociolinguistically-based, J. empirically-researched pronunciation for 23, syllabus Englishas an international language. Applied Linguistics, 83-103. Kachru, B. B. (Ed.). (1992). The other Urbana, IL: tongue: Englishacrosscultures. of University IllinoisPress. and in practice,revisited. TESOL Quarterly, Levis, (1999). Intonationin theory 33, J. 37-54. intonationin AmericanEnglish. Applied Levis,J. (2002). Reconsideringlow-rising 23, Linguistics, 56-82. are P., (Rev. ed.). Oxford, Lightbown, & Spada, N. (1999). How languages learned Press. England: OxfordUniversity R. with accent: an and in Lippi-Green, (1997). English Languageideology discrimination theUnited States. New York:Routledge. in Long, M. H., & Crookes,G. (1993). Unitsof analysis syllabus design:The case for task.In G. Crookes & S. Gass (Eds.), Tasksinpedagogical contexts: Integrating theory and practice Matters. (pp. 9-54). Clevedon,England: Multilingual to Macdonald, D., Yule, G., & Powers,M. (1994). Attempts improveEnglish L2 of pronunciation:The variable effects different typesof instruction. Language 44, Learning, 75-100. MacDonald, S. (2002). Pronunciation-viewsand practicesof reluctantteachers. 17(3), 3-18. Prospect, P. MacIntyre, D., Baker, S. C., Clement,R., & Donovan, L. A. (2003). Talking in orderto learn:Willingness communicate to and intensive The language programs. CanadianModern 59, LanguageReview, 587-605. S. C. Major,R. C., Fitzmaurice, F., Bunta,F., & Balasubramanian, (2002). The effects of nonnativeaccents on listeningcomprehension:Implicationsfor ESL assessment. TESOL Quarterly, 173-190. 36, SECOND LANGUAGE ACCENTAND PRONUNCIATIONTEACHING 395

Morley, (1991). The pronunciation J. componentof teachingEnglishto speakersof otherlanguages. TESOL Quarterly, 481-520. 25, and experience second in Clevedon, Moyer,A. (2004). Age,accent languageacquisition. Matters. England: Multilingual in Munro, M. J. (2003). A primeron accent discrimination the Canadian context. TESL CanadaJournal, 20(2), 38-51. T. Munro,M.J.,& Derwing, M. (1995). Processing time,accent,and comprehensibilityin the perceptionof nativeand foreign-accented speech. Languageand Speech, 38, 289-306. and Munro, M. J., & Derwing,T. M. (1999). Foreign accent, comprehensibility in 49 intelligibility the speech of second language learners.LanguageLearning, (Supp. 1), 285-310. Munro,M. J., & Derwing,T. M. (2001). Modeling perceptionsof the accentedness and comprehensibility L2 speech: The role of speakingrate. Studies Second of in 23, LanguageAcquisition, 451-468. Munro, M. J., Derwing,T. M., & Burgess,C. S. (2003). The detection of foreign accent in backwardsspeech. In M-J.Sole, D. Recasens, & J. Romero (Eds.), Sciences of Proceedings the 15th International of Congress Phonetic (pp. 535-538). Barcelona, Spain: Universitat Aut6nomade Barcelona. and Boston: Heinle & Heinle. Nunan, D. (1999). Second language teaching learning. for Pennington,M. C., & Ellis,N. C. (2000). Cantonese speakers' memory English sentenceswithprosodic clues. TheModern 84, 372-389. Language Journal, Pennington, M. C., & Richards,J. C. (1986). Pronunciation revisited. TESOL 20, Quarterly, 207-225. M. Perlmutter, (1989). Intelligibility ratingof L2 speech pre-and post-intervention. and Motor Skills, 515-521. 68, Perceptual Piske,T., MacKay,I. R. A., & Flege,J. E. (2001). Factorsaffecting degree of foreign accent in an L2: A review. 29, Journal Phonetics, 191-215. of Rubin, D. L. (1992). Nonlanguage factorsaffecting judgments of undergraduates' Research Higher in nonnativeEnglish-speaking 33, Education, teachingassistants. 511-531. Preferred ScanSoft.(1997). Dragon NaturallySpeaking (Version8) [ComputersoftAvailablefrom ware]. Peabody,MA:Author. http://www.dragonsys.com/naturally speaking/preferred/ 24, Schachter, (1974). An errorin erroranalysis. LanguageLearning, 205-214. J. Schmidt,R. (1990). The role of consciousnessin second language learning.Applied 11, Linguistics, 129-158. Scovel, T. (2000). A criticalreviewof the criticalperiod research.AnnualReview of 20, Linguistics, 213-223. Applied of Smith,L. E., & Bisazza,J. A. (1982). The comprehensibility three varietiesof in 32, Englishforcollege students seven countries.LanguageLearning, 259-269. and second language acquisition:A instruction Spada, N. (1997). Form-focussed research.LanguageTeaching, 73-85. reviewof classroomand laboratory 30, M. Stuparyk, (1996,July14). New Canadians eager to masterEnglish:Course helps G6. Canadianize accents. TheEdmonton Journal, of Suenobu, M., Kanzaki,K., & Yamane,S. (1992). An experimental study intelligibilin Reviewof AppliedLinguistics Language ity of Japanese English. International 30, Teaching, 146-156. R. Sustarsic, (2003). Applicationof acoustic analysisin Englishphoneticsteaching. In M-J. of Sold, D. Recasens, & J. Romero (Eds.), Proceedings the15thInternational Aut6noma Sciences 2841-2844). Barcelona,Spain:Universitat (pp. Congress ofPhonetic de Barcelona.
TESOL QUARTERLY

396

of on Tajima, K., Port,R., & Dalby,J.(1997). Effects temporalcorrection intelligibil25, of ityof foreign-accented English.Journal Phonetics, 1-24. Usher,J. (1995, Winter/Spring). Teaching pronunciation:Inhalation/exhalation sounds. Contact, Newsletterthe of of helps students distinguish of Association Teachers as 21, English a Second LanguageofOntario, 14-15. & for Wang,J., Munro,M.J. (2004). Computer-based training learningEnglishvowel contrasts. 32, System, 539-552. Harlow,England: Longman. Willis, (1996). A framework task-based J. for learning. Wingstedt,M., & Schulman, R. (1987). Comprehension of foreign accents. In 1984: Proceedings the5th International W. Dressler (Ed.), Phonologica Phonology of Press. Meeting (pp. 339-344). Cambridge,England: CambridgeUniversity Zielinski,B. (2004, May). Measurement speechintelligibility: are we actually What of measuring? Paper presentedat the annual meetingof theAmericanAssociationof Portland,OR. Applied Linguistics,

2ESOL QARTERL)

in and See how it toaccessevery article, graph, footnote table, publications. easy is TESO's serial a search authors, subjects, titles, you Using powerful engine, canfind more.iseasy navigate thousandshypertext It of links. andmuch to using

TESOL Journal TESOLQuarterly and on CD-ROM

*Fax Tel. 305-238-5673 305-252-0710

Cybertech Enterprises Toll: Free 1488-506-99

E-mail ctechIlix.netcom.com

SECOND LANGUAGE ACCENT AND PRONUNCIATION TEACHING

397