+ All Categories
Home > Documents > Derwing Munro 2013

Derwing Munro 2013

Date post: 04-Jan-2016
Category:
Upload: trenadoraquel
View: 48 times
Download: 2 times
Share this document with a friend
Popular Tags:
23
Language Learning ISSN 0023-8333 The Development of L2 Oral Language Skills in Two L1 Groups: A 7-Year Study Tracey M. Derwing a and Murray J. Munro b a University of Alberta and b Simon Fraser University Researching the longitudinal development of second language (L2) learners is essen- tial to understanding influences on their success. This 7-year study of oral skills in adult immigrant learners of English as a second language evaluated comprehensibility, fluency, and accentedness in first-language (L1) Mandarin and Slavic language speak- ers. The primary data were judgments at three times from two sets of listeners: native monolingual speakers of English and highly proficient English L2 speakers. The Man- darin L1 speakers showed no change over time on any of the dimensions, while the Slavic language L1 speakers improved significantly in comprehensibility and fluency. Improvement in accent was limited to the first 2 years in the Slavic language group. These outcomes appear to be due to the complex interplay of L1, age, the depth and breadth of learners’ conversations in English, and their willingness to communicate. Keywords comprehensibility; fluency; accent; longitudinal; age; pronunciation Introduction Policy makers in immigrant receiving countries such as Australia, Canada, and the United States tend to assume that newcomers require a certain proficiency level in the language of the community in order to integrate suc- cessfully into mainstream society (Burns & De Silva Joyce, 2007; Derwing The findings from this research were presented at the 2010 meeting of the American Association for Applied Linguistics, Atlanta, GA. We thank the participants in this study, who agreed to return for an eighth round of testing and interviews. We are grateful to the administrators and staff at NorQuest College and Metro Continuing Education for allowing us to contact the participants at the outset of the study. We thank Ron Thomson, who was involved with the first seven data collections, and Lori Diepenbroek, Amy Holtby, Jennifer Foote, and Jun Deng, who assisted with the final round of data collection. We are grateful to the Social Sciences and Humanities Research Council of Canada for two grants awarded to us to carry out this research. Correspondence concerning this article should be addressed to Tracey Derwing, Department of Educational Psychology, 6–102 Education North, University of Alberta, Edmonton, AB T6G 2G5 Canada. E-mail: [email protected] Language Learning XX:X, XXXX 2013, pp. 1–23 1 C 2013 Language Learning Research Club, University of Michigan DOI: 10.1111/lang.12000
Transcript
Page 1: Derwing Munro 2013

Language Learning ISSN 0023-8333

The Development of L2 Oral Language

Skills in Two L1 Groups: A 7-Year Study

Tracey M. Derwinga and Murray J. Munrob

aUniversity of Alberta and bSimon Fraser University

Researching the longitudinal development of second language (L2) learners is essen-tial to understanding influences on their success. This 7-year study of oral skills inadult immigrant learners of English as a second language evaluated comprehensibility,fluency, and accentedness in first-language (L1) Mandarin and Slavic language speak-ers. The primary data were judgments at three times from two sets of listeners: nativemonolingual speakers of English and highly proficient English L2 speakers. The Man-darin L1 speakers showed no change over time on any of the dimensions, while theSlavic language L1 speakers improved significantly in comprehensibility and fluency.Improvement in accent was limited to the first 2 years in the Slavic language group.These outcomes appear to be due to the complex interplay of L1, age, the depth andbreadth of learners’ conversations in English, and their willingness to communicate.

Keywords comprehensibility; fluency; accent; longitudinal; age; pronunciation

Introduction

Policy makers in immigrant receiving countries such as Australia, Canada,and the United States tend to assume that newcomers require a certainproficiency level in the language of the community in order to integrate suc-cessfully into mainstream society (Burns & De Silva Joyce, 2007; Derwing

The findings from this research were presented at the 2010 meeting of the American Association

for Applied Linguistics, Atlanta, GA. We thank the participants in this study, who agreed to return

for an eighth round of testing and interviews. We are grateful to the administrators and staff at

NorQuest College and Metro Continuing Education for allowing us to contact the participants

at the outset of the study. We thank Ron Thomson, who was involved with the first seven data

collections, and Lori Diepenbroek, Amy Holtby, Jennifer Foote, and Jun Deng, who assisted with

the final round of data collection. We are grateful to the Social Sciences and Humanities Research

Council of Canada for two grants awarded to us to carry out this research.

Correspondence concerning this article should be addressed to Tracey Derwing, Department of

Educational Psychology, 6–102 Education North, University of Alberta, Edmonton, AB T6G 2G5

Canada. E-mail: [email protected]

Language Learning XX:X, XXXX 2013, pp. 1–23 1

C© 2013 Language Learning Research Club, University of MichiganDOI: 10.1111/lang.12000

Page 2: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

& Thomson, 2005; McHugh, Gelatt, & Fix, 2007). In these and other coun-tries, government-funded language instruction programs for adults provide afoundation on which the individual learners are expected to build. For ex-ample, in Canada, the federal government funds the Language Instructionfor Newcomers to Canada (LINC) program, which is intended to serve as“basic language training to adult permanent residents in one of Canada’s offi-cial languages to facilitate their social, cultural and economic integration intoCanada” (Dempsey, Xue, & Kustec, 2009, p. 1, emphasis added). Support formore advanced language training is more limited, and is inaccessible for manynewcomers. There is an implicit assumption that with basic skills in the ma-jority language, immigrants will continue to develop their second language(L2) at work and in the community. This assumption may have some validity,but it does not apply uniformly to all groups of newcomers. In fact, a recentstudy of oral language in over 4,000 applicants for Canadian citizenship (M =6 years residence) showed that differences in first language (L1) backgroundwere tied to differences in proficiency (Derwing, Munro, Abbott, & Mulder,2010). For example, speakers of Russian-Ukrainian scored significantly higherthan speakers of Mandarin on the Canadian Language Benchmark Assessmenttool (see http://www.language.ca/). Interpretation of these findings, however,is complicated by a lack of information about the applicants’ level of educa-tion and knowledge of English on arrival. The current study aims at a closerexamination of the same two L1 groups, controlling for level of education,immigration class, oral proficiency on arrival, nature of instruction, and lengthof residence.

Background to the Study

In Canada, the term ‘integration’ is not synonymous with ‘assimilation’; rather,it is understood to mean that while retaining aspects of their own cultures andlanguages, immigrants can enjoy full participation in Canadian society. Thisis in contrast to ‘assimilation,’ which is interpreted as the complete adop-tion of the new culture and concomitant abandonment of the traditions of thehome culture. The Canadian definition of integration is not universally accepted(cf. Li, 2003), but it is the foundation of federal government policies of mul-ticulturalism. Immigrant integration is typically measured in economic terms,such as annual total income and occupational status (Hum & Simpson, 2004).However, such measures are influenced by a complex array of factors, such asdiscrimination, difficulties with credential recognition, unavailability of jobscommensurate with immigrants’ occupational training, and language barriers

Language Learning XX:X, XXXX 2013, pp. 1–23 2

Page 3: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

(Krahn, Derwing, Mulder, & Wilkinson, 2000). Moreover, integration entailsmuch more than economic success. Social, cultural, and linguistic factors arealso critical.

To test policy makers’ assumptions about the likelihood that adult immi-grants will pick up the language skills they need once they have the basicunderpinnings of their L2, it is essential to trace their progress over time. Lon-gitudinal studies of the development of oral L2 skills in adults are rare (Ortega& Iberri-Shea, 2005), yet a full understanding of the factors that affect languagelearning requires the type of data obtained from such research. In the presentstudy, we address this need by examining the oral language development of twogroups of adult learners of English as a second language (ESL): Mandarin andSlavic language (Russian and Ukrainian) speakers. Using a picture narrativetask, we evaluated the learners’ comprehensibility, fluency, and accent devel-opment over a 7-year period, starting when they were in Stage 1 LINC classes,but continuing well past their formal language training period. These narrativeswere evaluated by two sets of listeners: native monolingual speakers of Englishand highly proficient English non-native speakers (NNSs).

On the basis of the biographical data collected from these participantsin our previous studies, we could ascertain that, in many respects, the twolearner groups were very similar. At the beginning of the study, all learn-ers were assessed to be beginners in oral language skills, and all had similargovernment-funded language training experiences, comparable levels of ed-ucation in their countries of origin, and a shared goal for paid employmentin their original occupations. Indeed, the two learner groups were judged tohave equal comprehensibility and fluency at the outset of the data collection,when they began Stage 1 LINC classes. We have no reason to suspect anybetween-group difference in aptitude or talent for language learning, which hasbeen shown to have an effect on ultimate attainment in L2 (Abrahamsson &Hyltenstam, 2008; DeKeyser, Alfi-Shabtay, & Ravid, 2010). The participantsdiffered in some respects. They belonged to one of two language backgrounds:Mandarin, which is not related to English, and Slavic languages (Russian andUkrainian), which are Indo-European. They also differed in years of L2 studyin their home country, age of arrival in Canada, and degree of exposure to andpatterns of use of English outside of ESL programming. This variability affordsan opportunity to assess the contributions of L1 background and these factorsto the development of oral language skills.

We are not aware of any studies evaluating a relationship between years ofL2 instruction in the country of origin and general language skills on arrival toan immigrant-receiving country, but adult ESL programs for immigrants exist

3 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 4: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

in most inner circle countries to help newcomers further develop their Englishskills (Derwing, Diepenbroek, & Foote, 2009). Because many skilled immi-grants’ L2 skills on arrival are limited, these programs are intended to developcultural knowledge and linguistic skills to help them integrate into their newsociety (Louw, Derwing, & Abbott, 2010). By contrast, there is ample researchshowing that age of L2 learning is related to ultimate attainment in a numberof areas of language, particularly L2 phonology (Flege, Munro, & Mackay,1995), although most relevant studies have focused on differences betweenchildren and adults, rather than examining age effects after adulthood has beenreached. The few studies that have addressed L2 development in older learnersappear to have mixed results. In a large self-report study, Hakuta, Bialystok, andWiley (2003) found evidence of a broad decline in L2 ultimate attainment acrossadulthood. Derwing et al. (2010) observed a similar age-related decline in adultimmigrants’ speaking and listening test scores at the time of citizenship. Baker(2010) found an age-related decline in phonetic skills between ages 20 and 30.DeKeyser et al. (2010), on the other hand, have argued for a plateau between18 and 40 years of age, based on grammaticality judgments. The present studyaffords us an opportunity to examine the relationship between age and acquisi-tion of oral skills in adults representing a slightly broader age range, as all ourparticipants arrived in Canada between the ages of 19 and 49.

Research Purpose and Expectations

In several previous studies, we have reported on the progress of these sameparticipants, examining their English vowel development (Munro & Derwing,2008), their comprehensibility and fluency (Derwing et al., 2008), their accentand fluency (Derwing, Thomson, & Munro, 2006), and the relationship betweenfluency in their L1 and their L2 (Derwing, Munro, Thomson, & Rossiter, 2009).These and other studies examined the participants’ productions in the first yearor first 2 years of their residence in Canada. In the current study, we focus onthe longer-term performance of the same learners by extending the time frameto 7 years, well after their formal ESL training had ended.

Improvements in second language skills have been documented over timein several studies (e.g., Lightbown, Halter, White, & Horst, 2002; Klapper &Rees, 2003; Mellow, Reeder, & Forster, 1996). In the current paper, we focus onthe ongoing development in three dimensions of oral productions that have beenconsidered in a number of previous investigations: comprehensibility, fluency,and accent (e.g., Derwing, Rossiter, Munro, & Thomson, 2004; Derwing et al.,

Language Learning XX:X, XXXX 2013, pp. 1–23 4

Page 5: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

2008). We define comprehensibility in terms of listeners’ perceived difficultyof understanding. Fluency refers to listeners’ perceptions of the flow of thespeaker’s language output, for example, whether there are frequent pauses,false starts, or other dysfluencies. Accent is the perceived degree of differencefrom the local language variety. All three dimensions are evaluated throughLikert scaling by untrained listeners.

There is no question that the larger group of L2 learners from which thecurrent subset is drawn showed improvement over 2 years on some aspectsof English proficiency, at the lexical, syntactic, and phonetic levels (Munro &Derwing, 2008). However, when we turned our attention to the three dimen-sions of comprehensibility, fluency, and accent, patterns of improvement forthe full sample varied between L1 groups and across the dimensions them-selves (Derwing et al., 2008). While both groups improved in accentedness inthe first year of the study, the change was small (Derwing et al., 2006). At 2years, we also found very little change in the Mandarin L1 speakers’ fluencyand comprehensibility despite their enrollment in a language program and resi-dence in a predominantly English-speaking city. The Slavic language speakers,on the other hand, exhibited improvement in both comprehensibility and flu-ency over a 2-year period. In interpreting these findings, we attributed thisbetween-group effect to differences in both the quality and quantity of the im-migrants’ oral interactions in English. More specifically, we proposed that thewillingness to communicate (WTC) framework (MacIntyre, 2007; MacIntyre,Clement, Dornyei, & Noels, 1998) could account for the motivational, cultural,and interactional differences underlying the performance of the two groupsover time. Based on the interview data we collected at the 2-year landmark,the Slavic language speakers were more inclined to seek out opportunities topractice English. Although some of the Mandarin speakers did attempt to findEnglish-speaking interlocutors, they expressed difficulty in sustaining ongoinginteractions. Our analysis of the interview data suggested that factors withinthe WTC model, such as intergroup climate, social situation, communicativecompetence, L2 self-confidence, and motivation contributed to the differentialoutcomes between the Mandarin and Slavic language speakers (Derwing et al.,2008).

The current research extends our previous investigations 5 years after com-pletion of those earlier studies, with the expectation that, provided there havebeen no dramatic changes in frequency of use, the comprehensibility andfluency of the two groups of learners will continue to diverge. In other words,the Mandarin speakers will show little or no improvement 5 years after the last

5 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 6: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

testing point, but the Slavic language speakers will be perceived as more com-prehensible and more fluent. We would not expect accents to change in eithergroup for several reasons. First, the limited evidence available suggests thataccents tend to stabilize within the first year in the absence of explicit pronun-ciation instruction (Flege, 1988). Second, prior research indicates that accentis of less communicative value than comprehensibility, in that speakers maybe highly comprehensible despite strong foreign accents (Munro & Derwing,1999). Finally, because all our participants showed more limited early improve-ment in accent than in the other dimensions, we hypothesize that there would beless change after 7 years in accent than in either fluency or comprehensibility.

Method

SpeakersThe speakers were 11 Mandarin and 11 Slavic language speakers (7 Russiansand 4 Ukrainians) who were participants in a larger longitudinal study of L2acquisition. Previous research on these speakers’ first 2 years of languagedevelopment was presented in Derwing et al. (2008) and Munro and Derwing(2008). Not all of the original participants were available at the 7-year point, onwhich we are reporting here. Because only 11 Mandarin speakers were testedat each of the 2-month, 2-year, and 7-year points, we restricted the numberof Slavic language speakers to 11 as well, by randomly selecting from the 17who were eligible. There were 6 Mandarin females and 5 males, ranging in agefrom 35 to 47 at Year 7 (M = 42), and 6 Slavic language females and 5 males,ranging in age from 27 to 56 at Year 7 (M = 45.8). We also included threefemale monolingual English speakers, who provided speech samples. Thesewere later used to verify that the listeners remained in step during the listeningtask.

ListenersListeners were recruited through posters in the Education building at the Uni-versity of Alberta. We recruited two sets of listeners: monolingual NSs ofEnglish and highly proficient English NNSs. The criteria for participation forboth sets included undergraduate or graduate student status in a regular (non-ESL) academic program, age under 50 years, and normal hearing. Eventuallythree listeners were dropped (2 NSs and 1 NNS) because of noncompliancewith the task instructions, leaving 34 native listeners (12 males, 22 females)of Canadian English (average age = 22 years, range 18–37) and another 10

Language Learning XX:X, XXXX 2013, pp. 1–23 6

Page 7: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

(1 male, 9 females) who were high proficiency L2 speakers of English (2Mandarin, 2 Cantonese, and 1 each of Portuguese, Russian, Spanish, Tagalog,Ukrainian, and Vietnamese)1 with an average age of 26 years (range 18–42).The proportions of NS and NNS respondents reflect the composition of the stu-dent body in the faculty. All listeners were students at the University of Alberta(and thus met the language requirements for admission to an English-speakinguniversity). All reported normal hearing. All were paid an honorarium of $20for their participation.

StimuliAt each test point (2 months into the study, and the 2-year and 7-year points)all the speakers produced narratives based on an 8-frame cartoon story abouta man and a woman who mistakenly switched suitcases. High-quality digitalrecordings were made at the learners’ schools or in a quiet lab. A 20–25-secondexcerpt was extracted from the beginning of each narrative and normalized toremove variation in volume. The excerpts from all three time periods were thenrandomized for presentation to the listeners.

Rating Procedure

The procedure was the same as that followed in numerous other studies ofthe speech dimensions evaluated here (Munro & Derwing, 1999; Derwing& Munro, 1997; Munro, Derwing, & Morton, 2006; Derwing et al., 2008).Multiple listening sessions were held in a quiet room to accommodate the lis-teners’ schedules. During these sessions, the listeners completed a languagebackground questionnaire and then viewed the cartoon story on which the nar-ratives were based to minimize familiarity effects. They then heard one of threedifferent randomizations of the 70 stimuli and rated each item for comprehensi-bility (on a scale ranging from 1 = easy to understand to 9 = extremely difficultto understand), fluency (on a scale ranging from 1 = extremely fluent to 9 =extremely dysfluent), and accentedness (on a scale ranging from 1 = no accentto 9 = extremely strong accent). Instructions on how to use the scales andwhat the anchor terms meant were provided both in writing and orally (see theAppendix). The listeners heard each complete sample only once before makingtheir judgments and were instructed to use the entire scale. Prior to completingthe task, they rated three practice items (not used in the experiment). Theywere given a 3-minute break partway through the experiment to reduce fatigue.An experimenter controlled the playback of the stimuli to ensure that all raters

7 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 8: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

stayed in step. Total time for the task was approximately 50 minutes. A shortdebriefing of the listeners took place after the experiment.

Biographical DataFinally, we used biographical information about years of English study in thecountry of origin and age on arrival to Canada that we had collected from ourMandarin and Slavic language participants in the earlier studies. We also eliciteddata at the 7-year point, through interviews (to be reported elsewhere) and aquestionnaire about self-reported frequency of conversations in English. Theparticipants indicated how often they engaged in conversations of 10 minutesor more on a 5-point scale ranging from 1 (never) to 5 (several times a day). Wedid not consider shorter conversations because they were likely to be routineand formulaic and thus less likely to contribute to enhanced comprehensibilityand fluency.

Results

The ratings of the three female monolingual NS speech samples were notincluded in any of the statistical analyses that follow. As mentioned above,three listeners were dropped because they did not follow the instructions (e.g.,the judgments were made before the speech sample was completed or responseswere changed long after items had been played). From among the raters retained,all comprehensibility and accent judgments of the NS speech samples werescored at one or two. Furthermore, with the NNS speech samples, 95% or moreof the native listeners assigned at least one score of 9 or 8 at the other end ofeach of the three scales. For the non-native listeners, 80% or more did so. Thedata thus indicated that, as in previous studies, the listeners had recognized theNSs in the samples, had stayed in step throughout the task, and had followedthe instruction to use the full scales as much as possible.

Reliability of Ratings and Comparison of the Two Listener GroupsTo evaluate reliability across raters, we computed the intraclass correlations(Shrout & Fleiss, 1979) for the three measures for each group of listeners. Forthe 34 native English listeners, high reliability was observed on comprehensi-bility (r = .959), fluency (r = .971), and accentedness (r = .950). For the 10non-native listeners, reliability was slightly lower, though still very satisfactory(r = .868, .926, and .896, respectively). Because the intraclass correlation issensitive to the number of judges, the slightly lower scores may be due to thesmaller sample size.

Language Learning XX:X, XXXX 2013, pp. 1–23 8

Page 9: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

We computed Pearson correlations for the mean ratings from the two listenergroups (English NSs and highly proficient NNSs) to determine how similar theirratings of the speech samples were. The correlations for comprehensibility andfluency were very high (r = .94 and r = .95, respectively) and just slightlylower for accent (r = .88), p < .01 in all cases. Nevertheless, data wereanalyzed separately in most of the results presented below, as we wanted todetermine whether native and bilingual listeners responded to aspects of the L2language samples in similar ways and whether they held similar perceptions ofimprovement.

Effects of Time and L1 on JudgmentsMean ratings for each learner on each scale were submitted to a series ofmixed-design analyses of variance (ANOVAs) with L1 (Mandarin or Slaviclanguage) as the between-groups factor, and Test Time (2 months, 2 years, 7years) as the within-groups factor. Table 1 shows the results of the ANOVAsfor both listener groups, including effect sizes (partial eta squared) for theresults that were statistically significant. The first set of ANOVAs evaluatedchanges in comprehensibility ratings over time, while the second and third setsfocused on fluency and accent. A criterion of p < .05 was adopted for statisticalsignificance.

For all three dimensions, the effect of speaker L1 missed significance,while the effect of Time was significant, as was the L1 × Time interaction (seeTable 1). This pattern was observed for both the native English and non-nativelistener ratings. Figures 1a–3b illustrate the ratings of the groups of listenersfor comprehensibility, fluency, and accentedness across the three times. In eachpair of figures the native English listener judgments appear on the top, withthose of the non-native listeners below.

Pairwise Bonferroni adjusted (p < .05) t tests were calculated to explorethe interactions. In general, these revealed significant improvement over timefor the Slavic group but not the Mandarin group (see Table 2). In all butone case, the native English and bilingual listeners showed identical patterns.Neither listener group perceived an improvement in comprehensibility, fluency,or accentedness across any of the times for the Mandarin speakers. By contrast,parallel tests on the Slavic language group revealed significant improvement incomprehensibility from 2 months to 2 years, 2 months to 7 years, and 2 yearsto 7 years. For fluency, both the NS and NNS listener groups judged that theSlavic language group improved between 2 months and 2 years and between 2months and 7 years. However, for the NNS listeners the difference between the2-year and 7-year scores was marginal (p = .035 with a Bonferroni criterion

9 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 10: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

Tab

le1

Res

ults

ofm

ixed

-des

ign

anal

yses

ofva

rian

cew

ith

L1

and

tim

eas

fact

ors

Lis

tene

rG

roup

Sou

rce

NS

(N=

34)

NN

S(N

=10

)

Com

preh

ensi

bili

tyL

1F

(1,20

)=

1.14

p=

.30

F(1

,20

)=

1.92

p=

.182

Tim

eF

(2,40

)=

18.4

4p

<.0

01,η

p2

=.4

8F

(2,40

)=

15.5

7p

<.0

01,η

p2

=.4

4L

1×T

ime

F(2

,40

)=

14.0

9p

<.0

01,η

p2

=.4

13F

(2,40

)=

8.32

p<

.001

p2

=.2

9

Flu

ency

L1

F(1

,20

)=

3.66

p=

.07

F(1

,20

)=

3.40

p=

.08

Tim

eF

(2,40

)=

17.9

2p

<.0

01,η

p2

=.4

7F

(2,40

)=

15.8

2p

<.0

01,η

p2

=.4

4L

1×T

ime

F(2

,40

)=

8.59

p=

.001

p2

=.3

0F

(2,40

)=

5.76

p=

.006

p2

=.2

2

Acc

ente

dnes

sL

1F

(1,20

)=

.05

p=

.82

F(1

,20

)=

1.03

p=

.32

Tim

eF

(2,40

)=

17.7

9p

<.0

01,η

p2

=.4

7F

(2,40

)=

17.1

9p

<.0

01,η

p2

=.4

6L

1×T

ime

F(2

,40

)=

7.64

p=

.002

p2

=.2

8F

(2,40

)=

5.45

p=.

008,

ηp

2=

.21

Not

e.N

S=

nativ

em

onol

ingu

alsp

eake

rsof

Eng

lish

,NN

S=

high

lypr

ofici

entn

on-n

ativ

esp

eake

rsof

Eng

lish

Language Learning XX:X, XXXX 2013, pp. 1–23 10

Page 11: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

1

3

5

7

9 2 month 2 year 7 year

Com

preh

ensi

bilit

y

Mandarin

Slavic

1

3

5

7

9 2 month 2 year 7 year

Com

preh

ensi

bilit

y

Mandarin

Slavic

a b

Figure 1 Mean comprehensibility ratings from native English listeners (a) and non-native listeners (b) for the two speaker groups at 2 months, 2 years, and 7 years.

1 a b

3

5

7

9 2 month 2 year 7 year

Flu

ency

Mandarin

Slavic

1

3

5

7

9 2 month 2 year 7 year

Flu

ency

Mandarin

Slavic

Figure 2 Mean fluency ratings from native English listeners (a) and non-native listeners(b) for the two speaker groups at 2 months, 2 years, and 7 years.

of p < .017), while the NS listeners perceived a significant improvement.For accentedness ratings, both listener groups judged that the Slavic speakersimproved from 2 months to 2 years and from 2 months to 7 years, but not from2 years to 7 years.

In sum, as expected, the Mandarin speakers showed little or no improve-ment 5 years after the last testing point, whereas the Slavic language speakerswere perceived as more comprehensible and more fluent. In other words, thecomprehensibility and fluency scores observed in the two groups of learnersafter 2 years of residence in Canada continued to diverge over 7 years in thecountry. Accent scores, however, did not change from the 2-year to the 7-yearpoint.

To gain deeper insights into the performance of the individual participants,we plotted their scores from 2 years and 7 years for comprehensibility and

11 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 12: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

1 a b

3

5

7

9 2 month 2 year 7 year

Acc

ente

dnes

s

Mandarin

Slavic

1

3

5

7

9 2 month 2 year 7 year

Acc

ente

dnes

s

Mandarin

Slavic

Figure 3 Mean accent ratings from native English listeners (a) and non-native listeners(b) for the two speaker groups at 2 months, 2 years, and 7 years.

Table 2 Perceived improvement over time in the Mandarin (M) and Slavic language (S)groups by two groups of listeners (results of Bonferroni tests)

NS (N = 34) NNS (N = 10)

Scale L1 2 m-2 yr 2 m-7 yr 2 yr-7 yr 2 m-2 yr 2 m-7 yr 2 yr-7 yr

ComprehensibilityM No No No No No NoS Yes Yes Yes Yes Yes Yes

FluencyM No No No No No NoS Yes Yes Yes Yes Yes No

AccentM No No No No No NoS Yes Yes No Yes Yes No

Note. NS = native monolingual speakers of English, NNS = highly proficient non-nativespeakers of English

fluency (see Figures 4a–5b). In each case, results for the Mandarin speakers areshown in the top panel, and those for the Slavic language speakers are in thebottom panel. Because the two groups of listeners provided such similar ratings,we report here only the judgments of the larger native English group. Thedotted lines in the figures indicate numerically worse ratings at 7 years, whilesolid lines indicate an improvement. The most striking differences between thetwo participant groups are in comprehensibility, where 7 Mandarin speakersactually showed worse performance at 7 years, as compared with only 2 from the

Language Learning XX:X, XXXX 2013, pp. 1–23 12

Page 13: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

1

3

5

7

9 2 year 7 year

Com

preh

ensi

bilit

y

1

3

5

7

9 2 year 7 year

Com

preh

ensi

bilit

y

a b

Figure 4 Individual comprehensibility ratings for the Mandarin (a) and the Slaviclanguage (b) speakers at the 2-year and 7-year points.

1

3

5

7

9 2 year 7 year

Flu

ency

1

3

5

7

9 2 year 7 year

Flu

ency

a b

Figure 5 Individual fluency ratings for the Mandarin (a) and the Slavic language (b)speakers at the 2-year and 7-year points.

speakers in the Slavic language group. Furthermore, 5 of the Slavic speakerswere rated better at 7 years than the best Mandarin speaker. Figures 5a and5b show a net fluency improvement in 4 Mandarin speakers but in 8 Slaviclanguage speakers; 7 of those scored better than the best Mandarin speaker. Foraccent (not shown because of the lack of L1-related differences), 6 membersof each group were judged to show a net improvement.

Effects of Years of Prior English Study, Age of Arrival, and Amount

of English Use

Years of English study in the country of origin did not correlate significantlywith comprehensibility, fluency, or accentedness scores at the 7-year point(r(20) = .235, .376, and .11, respectively, p > .05). The participants’ ageson arrival in Canada ranged from 19 to 49 years, all within working age.Age on arrival was significantly correlated with comprehensibility scores forthe combined groups after 7 years, r(20) = .424. When the correlation was

13 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 14: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

1

3

5

7

9 15 25 35 45 55

Acc

ente

dnes

s

Age of Arrival in Canada

Figure 6 Correlation of accent ratings with age of arrival in Canada for all speakers. Aregression line has been added.

computed for Mandarin speakers alone, the finding was r(9) = .372, p > .05,and for Slavic speakers it was r (9) = .774, p = .005. Age on arrival was alsosignificantly correlated with accent scores after 7 years, r(20) = .638, p <. 05,(r (9) = .482, p > .05, for Mandarin speakers alone and r(9) = .831, p = .002, forSlavic speakers alone). The correlation with accent for both groups combinedis illustrated in Figure 6. Later ages of arrival were thus associated with poorerperformance on both these variables. However, age was not correlated withfluency scores for the combined groups, r(20) = .384, p > .05, despite highercorrelations for the individual groups (Mandarin speakers r(9) = .616, p =.044; Slavic language speakers r(9) = .793, p = .004). It is important tointerpret these correlations with caution because of the small sample sizes.When partial correlations for these variables were computed for both groupscombined, controlling for years of English study in the country of origin, age ofarrival was still significantly correlated with both comprehensibility and accentscores, r(20) = .444 and .645, respectively, p < .05. The partial correlationbetween age and fluency was marginal, r(20) = .426, p = .05.

Finally, we considered the participants’ reported frequency of conversa-tions of 10 minutes or more. There was very little shift in the frequency ofconversations in English from the 2-year to the 7-year point. Of the entire setof 22 speakers, only 5 reported increased numbers. We then compared com-prehensibility scores at the 7-year point with participants’ reported frequencyof conversations in English at both the 2-year and the 7-year points. We didso because comprehensibility yielded the most striking 7-year differences be-tween the two participant groups (cf. Figures 4a and 4b). The two weakestperformers were both Mandarin speakers who reported less than one English

Language Learning XX:X, XXXX 2013, pp. 1–23 14

Page 15: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

conversation of 10 minutes or more per day at the 2-year test; neither exhibitedan increase in conversation frequency over the intervening 5 years. The twoMandarin speakers who were rated most comprehensible, on the other hand,reported having extensive interactions in English on a daily basis at both the2-year and 7-year points.

Discussion

This longitudinal study was intended to determine to what extent adult immi-grant L2 learners continued to make progress in the development of oral En-glish skills after their formal ESL training in Canada. In particular, it exploreswhether policy makers are correct in assuming that newcomers will continue toacquire English without further instruction after foundational language skillshave been developed. Following well-established procedures, extemporaneousnarratives produced by two groups of speakers at 2 months, 2 years, and 7 yearswere evaluated by listeners for comprehensibility, fluency, and accentedness ina blind rating task. The 11 Slavic language speakers improved in comprehensi-bility and fluency both during the time they were enrolled in an ESL programand afterward, while accentedness improved only during the first 2 years. The11 Mandarin speakers, on the other hand, showed much less improvement, withmost of them making no perceptible gains across the entire 7-year study inany of the three dimensions of comprehensibility, fluency, or accentedness. Inaddition, their ratings across all three dimensions remained markedly worseover the 7-year span than the ratings given to the Slavic group. The rating taskfindings are also further testament to the fact that accent and comprehensibil-ity are partially independent; an improvement in comprehensibility does notnecessarily require that strength of accent be diminished (Derwing, Munro, &Wiebe, 1998).

In this section, we discuss the findings and propose some explanations. Weargue that the between-group differences appear to be the result of the complexinterplay of first language, the extent of linguistic interactions in the L2, andoverall WTC.

Years of Formal English Language Training in Country of OriginIt might be assumed that a greater amount of English study prior to arrival inCanada would afford an advantage to immigrants. There was no indication ofsuch an advantage for the participants here, however. First, all of them wereplaced at a beginner level in speaking and listening by the central assessmentcentre in their city. Second, our analyses yielded no significant correlations

15 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 16: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

at any of the three times between comprehensibility or fluency and years ofEnglish study in their countries of origin prior to arrival in Canada. While thisfinding may seem unexpected, the participants were all adult immigrants, andeven though many of them had studied English in school, they may not haveanticipated having to use it extensively later on in life. Although one mightexpect an advantage at least at the 2-month stage of the current study (e.g.,faster lexical access), the fact that most of the foreign language courses intheir countries of origin were focused primarily on written language may haveprecluded a benefit in the development of oral skills.

Effect of Age of ArrivalThe finding of a correlation between age of arrival in Canada and comprehensi-bility scores after 7 years contributes to the small but growing body of evidencethat the level of ultimate L2 attainment decreases steadily across the lifespan.Furthermore, the older arrivals in this study showed a marked tendency to havestronger accents than younger arrivals. These findings were observed in learn-ers between the ages of 19 and 49, a time period that is thought to be relativelystable for grammatical learning (DeKeyser et al., 2010). However, Baker (2010)noted a decline in Korean English speakers’ performance on word final stopsdepending on whether they arrived in the United States in their 20s or 30s. Ourfindings, therefore, appear to confirm that pronunciation learning is subject toage effects even during adulthood.

A Comparison of the Mandarin and Slavic Language Learners’PerformanceOur choice of the two L1 groups allows us to examine learning trajectoriesfor learners from very different linguistic and cultural backgrounds. Histori-cally, these two L1 groups are significantly represented in Canada’s immigrantpopulation, and to this day Canada receives large numbers of newcomers whospeak either Mandarin or a Slavic language as their L1. Given the importanceof these L1 groups, we wanted to compare their learning performance for boththeoretical and pedagogical reasons. Of course there is an important L1 dif-ference here, in that Slavic languages are Indo-European and thus show somelexical and structural similarities to English, which undoubtedly contributed tosome of the differences in the outcomes. However, in Derwing et al. (2008),we found that the WTC framework (MacIntyre et al., 1998) was especiallyhelpful in accounting for between group differences for the larger sample fromwhich the current participants are drawn. In that study, the Mandarin and Slaviclanguage speakers were judged to have equal comprehensibility and fluency at

Language Learning XX:X, XXXX 2013, pp. 1–23 16

Page 17: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

the outset, but the Mandarin speakers were less comprehensible and less fluentthan the Slavic language speakers after 10 months and 2 years in Canada. TheWTC framework accounts for communication behavior on the basis of factorssuch as intergroup climate, intergroup attitudes, L2 self-confidence, and socialsituation. A qualitative analysis of the interview data at the 2-year point indi-cated that the participants’ reflections on their language learning and their ownsituations fit the framework very well. The Mandarin speakers overall showedgreater ties to their L1 community, more reluctance to initiate conversations asa result of lower self-confidence in their English abilities, and fewer opportu-nities to interact in English. The outcomes for comprehensibility, fluency, andaccent at 7 years are remarkably similar to those seen at the 2-year point. Inother words, the Mandarin speakers continue to make few or no gains, while theSlavic language speakers continue to improve in both comprehensibility andfluency. Interestingly, neither group showed any improvement in accentednessbetween Year 2 and Year 7, despite the significant improvement in comprehen-sibility evidenced in the Slavic language group. In Derwing et al. (2008), weconcluded that overall experience with an L2, as determined by WTC factors,affected L2 oral development. In this study, at the 7-year point, the findings aresimilar. The two Mandarin speakers rated best on comprehensibility at 7 yearsalso reported high levels of exposure to English on a daily basis at the 7-yearpoint. In contrast, the two with the worst comprehensibility scores were alsothe two who reported the least daily exposure after 7 years in Canada. Whenasked at the 7-year point what advice on language learning the participantswould give to newcomers to Canada, the Mandarin speaker judged to be leastcomprehensible, who also had the least contact with English interlocutors said,“I think first, always talk, talk with people. If you have any time, any chance.”

Native English and L2 English ListenersA unique feature of this study was the elicitation and comparison of ratingsfrom two groups of listeners, 34 NSs of English and 10 high-proficiency NNSsof English. These two groups showed remarkably parallel ratings; overall re-liability within each group was high and the two groups’ scores correlatedwith each other at high levels, suggesting that native and bilingual listenersresponded to aspects of the L2 language samples in similar ways. With respectto perceptions of improvement, there was only one difference between the twogroups: The native English listeners perceived a significant improvement influency for the Slavic language speakers between the 2-year and 7-year points,whereas the non-native listeners’ scores showed only a marginal trend in the

17 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 18: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

same direction. The strong similarities between the ratings of the native andbilingual listeners are important because they indicate that the high proficiencyL2 listeners in this study interpreted oral productions in a manner very similarto the native listeners. That is, a speaker who was perceived as having lowcomprehensibility and low fluency by native English listeners was perceivedthe same way by proficient NNSs.

The fact that the non-native listeners in this study came from a diverse rangeof L1 backgrounds makes this outcome all the more impressive. They appearto share a similar criterion with NSs for what constitutes ‘easy to understand’and ‘fluent’ speech in Canadian English. These findings replicate and extendother research studies. Flege (1988), for instance, found strong similaritiesin accent ratings of Taiwanese, Mandarin, and English native listeners whojudged Mandarin-accented English speech. Mackay, Flege, and Imai (2006)found a close relationship between native English speakers’ and Arabic speak-ers’ judgments of Italian-accented English. Finally, Munro et al. (2006) hadCantonese, Japanese, Spanish, and Polish native listeners evaluate non-nativespeech samples from speakers of those same languages. Scores on intelligibility,comprehensibility, and accentedness were again highly similar across listenergroups. Furthermore, the scores were also similar to those elicited from Englishnative listeners in Derwing and Munro (1997). Thus, even though factors suchas bias (Kang & Rubin, 2009) and familiarity (Gass & Varonis, 1984) can in-fluence responses to L2 speech, listeners are nonetheless powerfully influencedby properties of the speech signal itself.

Implications and Conclusion

Given the central finding of the current study—that the two groups of learnersdiffered dramatically in their progress in oral English over a 7-year period—wereiterate Ortega and Iberri-Shea’s (2005) call for more longitudinal studies inapplied linguistics. A diverse range of L1 groups should be investigated tofully understand factors that contribute to ultimate attainment. It is only withlongitudinal studies that we can observe the outcomes of language instructionand interaction in the target community. This type of study can inform policymakers who are charged with determining the allocation of resources to immi-grant language training. Our research shows that so-called basic language skillsare insufficient for some newcomers to ensure full integration, even after a longperiod of 7 years living and working in the country. The study also suggeststhat ESL programs should put a greater focus on oral language skills in the

Language Learning XX:X, XXXX 2013, pp. 1–23 18

Page 19: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

beginner stages of language acquisition, particularly because some L2 studentsdo not access much oral language outside the classroom.

A limitation here is the somewhat crude measure of frequency and amountof oral language use as measured by self-reported length of interactions inEnglish (frequency of conversations of 10 minutes or more). Of the moresophisticated approaches that have been proposed, experience sampling (Kubey,Larson, & Csikszentmihalyi, 1996) would not be suitable for the lifestyles ofthe speakers in this study. However, an electronic language log, along the linesof that employed by Ranta and Meckelborg (2008), might be feasible. Improvedmeasures of patterns of L2 use and oral interaction with greater detail may pointto ways in which to enhance learners’ WTC, a fundamental goal for pedagogy(MacIntyre et al., 1998). In addition, one aspect of the method used in this ratingstudy was the fixed order of comprehensibility, fluency, and accent ratings fromthe listeners. We opted to assess comprehensibility first because of its primacyin communication. However, additional research could explore whether thereis an order effect in such judgment tasks.

This study featured two sets of listeners, NSs of English and highly pro-ficient bilingual speakers from a wide range of L1 backgrounds. The strikingcongruence of the two groups’ ratings suggests that studies of oral proficiencyneed not restrict listeners to one group or the other. However, the level of L2proficiency required for bilinguals to assess L2 speech reliably has not beenestablished. More research is necessary to determine whether there is a thresh-old of proficiency in an L2 to make such judgments. Although it is possiblefor listeners to detect a foreign accent in a language they do not speak (Major,2007), such an ability does not extend to comprehensibility, which requiressome as yet unknown level of knowledge of the L2.

Finally, correlational analyses carried out with relatively small numbers ofparticipants have to be interpreted with caution. Our finding of a relationshipbetween age of learning and comprehensibility and accent scores for adultsbetween 19 and 49 years at the outset of the study should be tested with alarger number of learners. Although DeKeyser et al.’s (2010) examination ofgrammaticality did not yield such an effect, it may be that pronunciation issubject to different constraints.

Learning a new language and integrating into a new workplace, a newcommunity, and a new society are challenging endeavors. If language programsare to facilitate these processes they should take into account the factors thatinfluence ultimate attainment. This study, in particular, points to the importanceof WTC and its relationship to L1s as a factor in successful acquisition of oral

19 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 20: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

language skills. It is clear that one-size-fits-all programs will not serve theneeds of all learners, as this longitudinal study demonstrates.

Final revised version accepted 3 January 2012

Note

1 We did not attempt to compare the Mandarin, Russian, and Ukrainian listeners’ratings with the others because the numbers did not warrant it.

References

Abrahamsson, N., & Hyltenstam, K. (2008). The robustness of aptitude effects innear-native second language. Studies in Second Language Acquisition, 30, 481–509.

Baker, W. (2010). Effects of age and experience on the production of English word finalstops by Korean speakers. Bilingualism: Language and Cognition, 13, 263–278.

Burns, A., & De Silva Joyce, H. (2007). Adult ESL programs in Australia. Prospect,22(3), 5–17.

DeKeyser, R. Alfi-Shabtay, I., & Ravid, D. (2010). Cross-linguistic evidence for thenature of age effects in second language acquisition. Applied Psycholinguistics, 31,413–438.

Dempsey, C., Xue, L., & Kustec, S. (2009). Language instruction for newcomers toCanada: Performance results by LINC level. Ottawa: Citizenship and ImmigrationCanada.

Derwing, T. M., Diepenbroek, L., & Foote, J. (2009). A literature review of Englishlanguage training in Canada and other English-speaking countries. Unpublishedreport prepared for Citizenship and Immigration Canada.

Derwing, T. M., & Munro, M. J. (1997). Accent, comprehensibility and intelligibility:Evidence from four L1s. Studies in Second Language Acquisition, 19, 1–16.

Derwing, T. M., Munro, M. J., Abbott, M., & Mulder, M. (2010). An examination of theCanadian language benchmark data from the citizenship language survey. Retrievedfrom http://www.cic.gc.ca/english/resources/research/language-benchmark/index.asp

Derwing, T. M., Munro, M. J., & Thomson, R. I. (2008). A longitudinal study of ESLlearners’ fluency and comprehensibility development. Applied Linguistics, 29,359–380.

Derwing, T. M., Munro, M. J., Thomson, R. I., & Rossiter, M. J. (2009). Therelationship between L1 fluency and L2 fluency development. Studies in SecondLanguage Acquisition, 31, 533–557.

Derwing, T. M., Munro, M. J., & Wiebe, G. (1998). Evidence in favor of a broadframework for pronunciation instruction. Language Learning, 48, 393–410.

Language Learning XX:X, XXXX 2013, pp. 1–23 20

Page 21: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

Derwing, T. M., Rossiter, M. J., Munro, M. J., & Thomson, R. I. (2004). L2 fluency:Judgments on different tasks. Language Learning, 54, 655–679.

Derwing, T. M., & Thomson, R. I. (2005). Citizenship concepts in LINC classrooms.TESL Canada Journal, 23(1), 44–62.

Derwing, T. M., Thomson, R. I., & Munro, M. J. (2006). English pronunciation andfluency development in Mandarin and Slavic speakers. System, 34,183–193.

Flege, J. E. (1988). Factors affecting degree of perceived foreign accent in Englishsentences. Journal of the Acoustical Society of America, 84, 70–79.

Flege, J. E., Munro, M. J., & MacKay, I. R. A. (1995). Factors affecting strength ofperceived foreign accent in a second language. Journal of the Acoustical Society ofAmerica, 97, 3125–3134.

Gass, S. M., & Varonis, E. M. (1984). The effect of familiarity on thecomprehensibility of nonnative speech. Language Learning, 34, 65–89.

Hakuta, K., Bialystok, E., & Wiley, E. (2003). Critical evidence: A test of thecritical-period hypothesis for second-language acquisition. Psychological Science,14, 31–38.

Hum, D., & Simpson, W. (2004). Economic integration of immigrants to Canada: Ashort survey. Canadian Journal of Urban Research, 13, 46–61.

Kang, O., & Rubin, D. L. (2009). Reverse linguistic stereotyping: Measuring the effectof listener expectations on speech evaluation. Journal of Language and SocialPsychology, 28, 441–456.

Klapper, J., & Rees, J. (2003). Reviewing the case for explicit grammar instruction inthe university foreign language learning context. Language Teaching Research, 7,285–314.

Krahn, H., Derwing, T. M., Mulder, M., & Wilkinson, L. (2000). Educated andunderemployed: Refugee integration into the Canadian labour market. Journal ofInternational Migration and Integration, 1, 59–84.

Kubey, R., Larson, R., & Csikszentmihalyi, M. (1996). Experience sampling methodapplications to communication research questions. Journal of CommunicationResearch, 46, 99–120.

Li, P. (2003). Deconstructing Canada’s discourse of immigrant integration. Journal ofInternational Migration and Integration, 4, 315–333.

Lightbown, P., Halter, R., White, J., & Horst, M. (2002). Comprehension-basedlearning: The limits of “do it yourself.” The Canadian Modern Language Review,58, 427–464.

Louw, K. J., Derwing, T. M., & Abbott, M. L. (2010). Teaching pragmatics to L2learners for the workplace: The job interview. Canadian Modern Language Review,66, 739–758.

MacIntyre, P. D. (2007). Willingness to communicate in the second language:Understanding the decision to speak as a volitional process. Modern LanguageJournal, 91, 564–576.

21 Language Learning XX:X, XXXX 2013, pp. 1–23

Page 22: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

MacIntyre, P. D., Clement, R., Dornyei, Z., & Noels, K. A. (1998). Conceptualizingwillingness to communicate in a L2: A situational model of L2 confidence andaffiliation. Modern Language Journal, 82, 545–562.

Mackay, I. R. A., Flege, J. E., & Imai, S. (2006). Evaluating the effects ofchronological age and sentence duration on degree of perceived foreign accent.Applied Psycholinguistics, 27, 157–183.

Major, R. (2007). Identifying a foreign accent in an unfamiliar language. Studies inSecond Language Acquisition, 29, 539–556.

McHugh, M., Gelatt, J., & Fix, M. (2007). Adult English language instruction in theUnited States: Determining need and investing wisely. Retrieved fromhttp://www.migrationinformation.org/integration/language.cfm

Mellow, J. D., Reeder, K., & Forster, E. (1996). Using the time-series design toinvestigate the effects of pedagogic intervention on SLA. Studies in SecondLanguage Acquisition, 18, 325–350.

Munro, M. J., & Derwing, T. M. (1999). Foreign accent, comprehensibility andintelligibility in the speech of second language learners. Language Learning, 49,Supplement 1, 285–310.

Munro, M. J., & Derwing, T. M. (2008). Segmental acquisition in adult ESL learners:A longitudinal study of vowel production. Language Learning, 58,479–502.

Munro, M. J., Derwing, T. M., & Morton, S. (2006). The mutual intelligibility of L2speech. Studies in Second Language Acquisition, 28, 11–131.

Ortega, L., & Iberri-Shea, G. (2005). Longitudinal research in second languageacquisition: Recent trends and future directions. Annual Review of AppliedLinguistics, 25, 26–45.

Ranta, L., & Meckelborg, A. (2008, April). Questioning the language exposurequestionnaire. Paper presented at the meeting of the American Association ofApplied Linguistics, Washington, DC.

Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing raterreliability. Psychological Bulletin, 2, 420–428.

Appendix: Oral Instructions Provided to Raters

You will hear both second language learners and NSs of English describing thestory—all of the speech samples are taken from the first 20–25 seconds.

What we would like you to do is make three judgments about each sample.First, we will ask you to say how easy or difficult the sample is to understand,

using a 9-point scale. You might be able to understand everything but it mayrequire a lot of effort on your part—so what we are interested in is the effortyou put in. Can you understand it without even thinking about it, or do youhave to work at it?

Language Learning XX:X, XXXX 2013, pp. 1–23 22

Page 23: Derwing Munro 2013

Derwing and Munro Development of L2 Oral Skills

Second, we will ask you to rate fluency. This is the flow of the language—does the person have problems finding words, using a lot of ums and ahs, orpauses, or do the words come easily? Don’t worry about grammar mistakes—that doesn’t matter. So, someone who is very fluent—that is, the words justflow with no struggle, would be at the left end of the scale, while someone whohas a hard time expressing him or herself would be closer to the right end ofthe scale.

Third, we are interested in accent. We all have accents, but what we areinterested in knowing is how different the speakers’ accents are from a standardCanadian English accent. Accent is different from comprehensibility—youmight be able to understand somebody easily and still hear a heavy accent.Again the scale is 1–9: 1 = no accent and 9 = extremely heavy accent.

We would like you to try to use the whole scale over the course of theexperiment. Please listen to the whole sample before making your decisions.

23 Language Learning XX:X, XXXX 2013, pp. 1–23


Recommended