Applied Linguistics 2021: 42/1: 1–23

doi:10.1093/applin/amz056 Advance Access published on 9 December 2019

Role Models in Language Learning: Results

of a Large-Scale International Survey

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

*CHRISTINE
MUIR, ZOLTÁN DÖRNYEI, and
SVENJA ADOLPHS
School of English, University of Nottingham
*
E-mail: [email protected]

Role models can exert considerable influence in shaping individuals’ values, atti-
tudes, and beliefs. A large body of work in the social sciences has investigated the
influence of celebrity role models, and in the context of education, several disci-
plines have a rich research history in this area (e.g. medical education).
However, in the context of second language acquisition, research centred on role
models has largely remained on the periphery. This study presents a large-scale
international survey investigating the role models of English language learners.
With data collected from 8,472 participants, analysis investigated whether these
learners had English language role models, who the role models were and what
characteristics learners valued in them, and investigated systematic variation
among subgroups. Results showed that 68 per cent of respondents reported hav-
ing an English language role model, and four key role model dimensions
emerged: overall command of English, paralinguistic features, personal attrib-
utes and accent/variety of English. We argue that role modelling may be a highly
influential component of the psychological context of second-language acquisi-
tion, and conclude by highlighting several valuable areas for future research.

INTRODUCTION
Role models of all kinds can exert considerable influence in shaping our val-
ues, attitudes, and beliefs. Individuals, both young and old, can develop what
they perceive to be strong and intimate relationships with celebrity role mod-
els (Boon and Lomore, 2001; Cashmore, 2006), and it is well documented that
these virtual bonds can result ‘in powerful forms of personal and social trans-
formation’ (Fraser and Brown, 2002: 200). Although role models are often dis-
cussed in the context or renowned or celebrated personalities, people who we
meet in our daily lives can also function as role models, and role models can
even be fictional or animated characters that we watch or read about.
Role models can affect change through multiple processes. The process
which is arguably most relevant to educational contexts is that of ‘vicarious
learning’ (also referred to as observational learning). The notion of vicarious
learning is a well-established principle in psychology, and, in his seminal book
on social learning theory, Bandura (1977: 12) submits that ‘virtually all learn-
ing phenomena resulting from direct experience occur on a vicarious basis by
C The Author(s) (2019). Published by Oxford University Press. All rights reserved.
V
For permissions, please email: [email protected]
2 ROLE MODELS IN LANGUAGE LEARNING

observing other people’s behaviour and its consequences for them’. People
continually and actively search for models they perceive as representative of
what they wish to achieve, and in doing so, Bandura concludes, this ‘guides
and motivates self-development’ (Bandura 1997: 88).

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

However, owing to the everyday familiarity of the term it has been argued
that the ‘conventional wisdom of role models’ (Solomon, 1997: 396) has cur-
tailed their rigorous empirical investigation. For example, Carrington and
Skelton (2003) submit that government policies with regard to teachers’ roles as
sex- and race-specific models for students have been ‘legitimated by an appeal to
common-sense notions about the salience of ‘role models’ in socialization’ (p.
253), rather than solid empirical results. Criticism has highlighted the fact that
studies are sometimes loosely grounded in theory (Turner and Shepherd, 1999),
and findings are further complicated by the fact that in the social sciences the
term ‘role model’ has been ‘inconsistently used and loosely defined’ (Gibson,
2004: 135). Partly for these reasons, the potentials of role modelling have not yet
been fully exploited across educational domains, and this is even truer of the
field of second-language acquisition (SLA). This article addresses this paucity of
research by presenting the findings of the first large-scale international study of
the role models of learners of English, offering a baseline dataset that researchers
will be able to draw on as an important reference point.
We recruited participants globally, and the primary aims of this exploratory
study were to map out whether, and which, participants reported having English
language role models, and who these role models were. We were further inter-
ested to understand the specific characteristics that participants reported valuing
in their role models, and whether there was systematic variation in responses
among discrete participant subgroups. We begin by offering a brief overview of
relevant literature, before laying out the methodology of the study. We go on to
present the results and discuss their implications both with regard to theory and
pedagogy. We conclude by highlighting the limitations of the study and by high-
lighting what we believe to be fruitful areas for future research.

LITERATURE REVIEW
Theoretical foundations of role modelling
At the heart of this article is the educational significance of role modelling
which, as noted above, has its roots in Bandura’s (1977) social learning the-
ory. This posits that the vicarious experience of observing models involves
four key processes. First, behaviour will only be learned from models to
whom individuals pay attention. Within an individual’s immediate context, the
availability of models may be limited, but this can be expanded through the
mediated frame of the press and the mass media, providing important models
with ‘high status, competence, and power’ (p. 88).
The second and third processes—retention and motor reproduction—highlight
the importance of the way the observed stimulus is processed and stored,
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 3

involving first the strengthening of this information by repeated exposure,

and then various forms of practice through which these ‘symbolic representa-
tions’ can be converted into action. The final aspect relates to accompanying
motivational processes, as people are more likely to enact a modelled action if

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

they observe the action resulting in positive consequences. Indeed, Bandura
(1977: 87) emphasized that ‘Seeing or visualising people similar to oneself per-
form successfully typically raises efficacy beliefs in observers that they them-
selves possess the capabilities to master comparable activities’. We should also
note caution that these processes may not always be successful—that is, not
all observed behaviour will be modelled—for reasons including weaknesses or
gaps in any of the above stages, physical inability or a lack of sufficient
incentives.
More recent investigations of vicarious experience have involved assessing
the potential for learning without direct observation through discussion or
dialogues (Cox et al. 1999; Northedge 2003), through utilizing students’ im-
agination as well as ‘the vicarious experience afforded through good fiction’
(Fox 2003: 99), and through storytelling (Krietemayer and Heiney 1992;
Spouse 2003). As Roberts (2010: 14) explains, exposure to such discourses
‘enables students to develop concepts of themselves in different roles’. Ibarra
(1999) refers to this process as the construction of provisional selves, thereby
allowing individuals to ‘try on’ possible identities before they are required to
act them out.

Role modelling in education and SLA

Given the close links of role modelling with social learning, the notion has
been explored in multiple areas of education. Medical education has a particu-
larly rich history, and the cumulative body of research amassed in this context
lends support to the conclusion that role models have distinct educational
relevance (cf. Wright et al. 1997; Althouse et al. 1999; Paice et al. 2002; Perry
2009).
In SLA, role models have been most directly investigated in the context of
‘near peer role modelling’ (see below), but it is fair to conclude that discussion
of role models has largely been on the periphery of other research objectives.
In research on language learning motivation, role models have been identified
as an important impetus for the creation and refinement of an ideal L2 self
(Dörnyei and Kubanyiova 2014), and as able to help learners conceptualize
‘roadmaps to success’ along with effective strategies to realize these possible
selves (see Thompson and Vásquez 2015). The notion of possible selves
(Markus and Nurius 1986), which underpins the L2 motivational self-system
(Dörnyei 2005, 2009), links these theories indirectly yet inherently with the
notion of role modelling. An intriguing variation on these principles is ‘video
self-modelling’ (see Collier-Meek et al. 2012, for a detailed overview of the
procedure; and Adolphs et al. 2018, for a further innovative approach with re-
gard to visualization and technology). Video self-modelling involves first
4 ROLE MODELS IN LANGUAGE LEARNING

producing a ‘success video montage’ made up of edited clips of a person per-

forming a target behaviour well, and then asking the person to regularly watch
these images of him/herself. In this way, the participant serves as his/her own
role model. Dörnyei and Kubanyiova (2014) have argued that this technique

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

lends itself particularly well to being utilized in language learning contexts.

Near peer role models. Murphey and Arao (2001: 1) define ‘near peer role
models’ as ‘people who might be “near” to us in several ways: age, ethnicity,
sex, interests, past or present experiences, and also in proximity and in fre-
quency of social contact’. Findings in the context of SLA have confirmed that
exposure to near peer role models can result in immediate benefits relating to
student motivation and excitement, risk-taking and the amount of English
used (Murphey and Murakami 1998). Evidence has also suggested that these
positive changes can be long-lasting (Murphey and Arao 2001), that relatively
little class time is required to achieve them (30 minutes in Murphey and
Murakami’s 1998 study) and that a conscious emphasis on near peer role
models in the L2 classroom can lead not only to student change, but also to
positive teacher development (Murphey and Arao 2001). Moreover, when
non-native speaker teachers are positioned as representing desirable language
models for their students—significantly, with knowledge of more than one
language system—they can also represent a powerful form of near peer role
model (Duff and Uchida 1997; Nemtchinova 2005; He and Zhang 2010;
Barkhuizen 2016).

Related research in SLA. We have noted above the lack of research into role
models in the field of SLA. However, we would be remiss not to recognize sev-
eral bodies of research which touch on the same core issues, even if they do
not draw on the same terminology. For example, Bonny Norton and her col-
leagues’ work on the notion of identity, investment, and imagined commun-
ities (Anderson 2006) exemplifies this. To use their words: ‘a learner’s hopes
for the future (or their children’s future) are integral to a language learner’s
identity’ (Norton and Toohey 2011: 415). Drawing also on the notion of pos-
sible selves (Markus and Nurius 1986), Pavlenko and Norton (2007: 670) sub-
mit that as humans, we ‘are capable, through our imagination, of perceiving a
connection with people beyond our immediate social networks’. They go on
to argue that ‘For both Wenger [1998; situated learning theory] and Markus
and Nurius, possible selves, linked to memberships in imagined communities,
shape individuals’ present and future decisions and behaviors and provide an
evaluative and interpretive context for such decisions, behaviors, and their
outcomes’ (Pavlenko and Norton 2007: 670). Social comparison processes
(Festinger 1954), whether upward or downward, therefore also have clear
links with our discussion of English language learner role models (see e.g.
Henry 2015, for a discussion of the dynamic, real-time revisions of learners’
ideal L2 selves).
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 5

Research has also investigated changes in students’ L2 goals, motivations,

and in the perceptions of their target language community during periods of
study abroad (see e.g. Kinginger 2008), and processes of second language so-
cialization are also rooted in the existence of clear L2 cultural and linguistic

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

models (Duff 2007). Even when students do not or are not able to travel,
teaching materials and other resources can help learners develop imagined
transnational networks and identities, through imagining themselves engag-
ing in communication with communities worldwide (Curdt-Christiansen and
Weninger 2015; Duff 2015). These approaches doubtless all have links to the
core principles of role modelling, and we return to expand this discussion in
exploring the implications of our study.

Research questions and aims of this study

The previous overview suggests that role modelling occurs daily in classrooms
around the world, yet research in the field of SLA has not yet examined the
scope and the nature of the process in a systematic manner. Drawing on a
large-scale online questionnaire survey of learners of English, the current ex-
ploratory study strives to begin addressing this gap by seeking answers to the
following five research questions:
1 How common is having a language learning role model?
2 Who are participants’ role models?
3 Are there any salient role model archetypes?
4 Is there any systematic variation in role models described by specific par-
ticipant subgroups?
5 What characteristics do participants value in their English language role
models?

METHODS
Participants
Participants for this study were 8,472 English language learners from 155 L1
backgrounds (see Table 1 for the eight most frequent). The male/female split
was 54.5 per cent /44.2 per cent (with 1.3 per cent of participants opting not
to say), more than 60 per cent of participants were under the age of 30, and
over 80 per cent under the age of 40. Most participants (57.7 per cent) self-
reported having an English language proficiency level of upper-intermediate
and above, and 28.2 per cent an intermediate level. Participants were from a
wide range of teaching and learning contexts (see Table 2), and a special fea-
ture of the dataset was the large number of participants who were English lan-
guage teachers (N ¼ 1,189/14.0 per cent). In order to aid analysis, five broad
geographical groupings were created from the largest participant clusters in
terms of their mother tongue and reflecting their nationality/the place that
they call home: Chinese (all dialects; N ¼ 1,038); Russian (N ¼ 576); languages
6 ROLE MODELS IN LANGUAGE LEARNING

Table 1: Eight most common participant L1s

L1 N (per cent of respondents)

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

Chinese 1042 (12.3)
Spanish 938 (11.1)
Arabic 603 (7.1)
Portuguese 591 (7.0)
Russian 577 (6.8)
Hindi 463 (5.5)
Vietnamese 436 (5.1)
Turkish 326 (3.8)

Table 2: Participants’ learner/teacher status and context

Learner/teacher status and context N (per cent)
Studying at a private language school 794 (9.4)
Studying at school 833 (9.8)
Studying at university 1648 (19.5)
Studying on a year abroad 137 (1.6)
Studying on my own 3871 (45.7)
I am a teacher 1189 (14.0)

spoken in India (N ¼ 898); languages spoken in the European Union (EU)

(N ¼ 1,799); and languages spoken in Central and South America (N ¼ 1,134).

Instruments and procedures

This study was part of a collaboration between the University of Nottingham
and Cambridge University Press. The two institutions have a long-standing re-
lationship, and Cambridge University Press did not have any financial stake in
the results of the study. Ethical approval was sought and gained from both
institutions and fully complied with. Participants were given the opportunity
to participate in a prize draw to win a single £200 Amazon voucher. Contact
details were drawn for this at random after data collection ended, and these
participant emails were then immediately deleted. The link to a bespoke ques-
tionnaire (see Supplementary Appendix A) was placed on the Cambridge
Online Dictionaries website, and data were gathered over the course of
21 days from 15 February to 7 March 2017.
The questionnaire was split into three sections. In the first, participants
were asked if they had an English language learning role model, and those
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 7

who did were invited to share details of up to two such models. Participants
were asked to indicate the sex of their role models (with a separate option e.g.
a fictional character/animation); whether they were native English speakers
and whether they were the same nationality as them; whether the role models

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

were younger, of a similar age or older than them; and whether they worked
in a similar field/profession. Finally, participants were also invited to identify
their role models by name (e.g. ‘Barack Obama – former US president’ or ‘my
grandmother’; note that these examples were only given after participants
had answered these initial questions describing their chosen role model).
This section was introduced with a brief description of a ‘role model’, and
examples of role models in a different context:
A role model is someone that you respect and that you want to be-
come more like – for example, if you are sportsman/sportswoman,
your sporting role models might be Muhammad Ali or Serena
Williams. We would like you to think about your English language
role model: it might be a teacher, a famous actor or singer, a polit-
ician, a friend. . .but it could be anyone who speaks or writes in
English!
In the second section, participants were asked to rate a set of characteristics in
terms of how important they considered them, either regarding their existing
English language role model(s) or for a role model if they were to choose one.
These characteristics were drawn from a previous study that identified a set of
features and attributes highlighted as important by English language learners
(Adolphs et al. 2018). They included, for example, accent, fluency, use of ges-
tures and the perceived ‘naturalness’ of the role models’ English usage.
Participants were asked to indicate their responses on five-point Likert scales,
and were also given the opportunity to share additional characteristics that
they valued in an open-ended question.
The third section of the questionnaire collected background information
concerning participants’ pronunciation goals, the variety of English they
wanted to learn, and various demographic details as well as their current and
intended future patterns of English use (some of which is not analysed in this
article).

Data analysis
In order to develop a detailed picture of the role models described by the par-
ticipants, comprehensive descriptive statistics were collated and chi-square
analyses were conducted to interrogate the data. Drawing on the acknowl-
edged self-organizing capacity of systems—a core tenet of complex dynamic
systems theory (CDST)— we sought to identify key role model archetypes
nested within the vast dataset (Dörnyei 2014). Exploratory factor analysis was
also used to examine the underlying factor structure of the role model charac-
teristics that the participants were asked to rate, first computed with the whole
8 ROLE MODELS IN LANGUAGE LEARNING

dataset and subsequently separated with key participant subgroups (male/fe-

male, teachers/students, and by geographical grouping). Qualitative data
obtained from open-ended questions were interrogated through thematic ana-
lysis (Nowell et al. 2017): once the initial coding had been completed, wider

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

themes were identified with the coding categories reviewed and refined
through an ongoing iterative process. Owing to the size of the dataset it was
not double coded in its entirety. Coding was completed by the first author
(C.M.), coding categories and examples were discussed in detail with the se-
cond and third authors (Z.D. and S.A.), and adjustments and re-coding were
completed as necessary to achieve consensus. This was an iterative process,
and these discussions took place at regular points throughout the process of
analysis, with an aim to collaboratively find the best fit with the data.

RESULTS AND DISCUSSION

Research Question 1: How common is having an English
language learning role model?
A total of 5,767 participants reported having an English language role model,
a total of 68.1 per cent of all respondents. Of these, 2,582 also described a se-
cond English language learning role model, leading to a dataset comprised of
8,349 role model descriptions. In evaluating this proportion, we must take
into consideration that our respondents constituted a self-selected sample,
leading to the likely inflation of the proportion of positive responses: one
could rightly argue that people who completed the questionnaire were inter-
ested because they felt they had something substantial to contribute. We like-
wise note that participants were users of the Cambridge Online Dictionaries
website, possibly a reflection of a high motivation to study. Curiously, the best
evidence of the fact the participant sample was not entirely biased is the large
number of learners (38.1 per cent) who completed the survey even though
they did not have a role model. This suggests that the call for participation was
attractive enough to a wide range of language learners, and further evidence
of the general appeal of the survey is provided by the unexpectedly large sam-
ple that was recruited during the short period the link was active.
Thus, although some systematic bias in favour of those who had role models
is inevitable, the above considerations suggest that the main attraction of the
survey was not restricted to this aspect. This is indicative that the investigation
of role models is indeed a worthy topic for the field of SLA. While the propor-
tion of English language learners with role models may not turn out to match
the figures reported in other educational disciplines (e.g. over 90 per cent as
noted by Wright et al. 1997, in the context of medical education), further re-
search will clarify the pedagogical potential rooted in L2 role models. The
wider relevance of L2 role models is further supported by the international
spread of the respondent sample, with the exact figures exhibiting natural
variation reflecting differences in these learners’ wider cultural contexts (see
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 9

Table 3: Number of participants with role models from different geographical

groupings
Geographical grouping N (per cent)

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

Chinese 612 (59.0)
Russian 396 (68.8)
India 641 (71.4)
EU 1181 (65.5)
Central and South America 831 (73.3)

Table 3 and “Research Question 4” section for further discussion of these

results). Overall, we feel confident in arguing that role modelling may be a
highly influential component of the psychological context of SLA.

Research Question 2: Who are participants’ role models?

Let us start characterizing the role models that our participants described with
some demographic statistics. Of the 8,349 role models in our dataset, 4,980
(59.6 per cent) were male and 3,140 (37.6 per cent) female; 229 role models
belonged to the ‘Other’ category, for example, animated or fictional charac-
ters. Most role models described were English native speakers (64.2 per cent)
and most role models (78 per cent) were older than participants. Of the role
models, 37.8 per cent were personally known by the participants (see
Table 4), and 55 per cent were famous (i.e. not personally known; see
Table 5).
Most participants reported wanting to learn British English (see Table 6),
likely influenced by the fact that data were collected from visitors to the
Cambridge Online Dictionaries website. This was reflected in the fact that the
majority of the most frequently mentioned famous role models were also
British (see Table 7). It is significant that there was a substantial category of
personally known role models made up of teachers and professors (see
Table 4), highlighting the potentially large role modelling impact of educators.
It is also noteworthy that the two most popular role models, Barack Obama
and Emma Watson, are both personalities well known for their social activism
(Obama was awarded the Nobel Peace Prize in 2009 and Watson was
appointed as a UN Women Goodwill Ambassador in 2014). Although we can-
not know whether participants were aware of this, there is an interesting con-
nection here with our qualitative data, which highlighted role models’
character, ethics, and moral stance as important factors contributing to their
selection (see “Research Question 5” section). We should note, however, that
in the context of the overall sample size the frequency numbers attached to
named celebrities are very small, indicating significant variation. Even Barack
Obama only accounted for 3.2 per cent of all role models described, and we
10 ROLE MODELS IN LANGUAGE LEARNING

Table 4: Top five categories of personally known role models

Personally known role N Percentage of Percentage
model categories personally known of all role

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

role models models
All known role models 3,156 100.0 37.8
Teacher/professor 1,848 58.6 22.1
Friend/classmate 588 18.6 7.0
Family member 275 8.7 3.3
Boss/colleague 252 8.0 3.0
Partner/spouse (current or ex) 99 3.1 1.2

Table 5: Top five categories of famous role models

Famous role model categories N Percentage of Percentage of
famous role models all role models
All famous role models 4,590 100.0 55.0
Film/TV industry 1,170 25.3 13.9
Politics 728 15.9 8.7
Author/poet 493 10.7 5.9
Singer/musician 448 9.8 5.4
YouTuber/vlogger (ELT related) 263 5.7 3.2

Table 6: ‘What variety of English would you like to learn?’ (Participants

were able to tick multiple responses)
Variety N
British English 5,471
American/Canadian English 3,691
It is not important to me 2,453
Australian/New Zealander English 1,073
A specific variety of English (e.g. Chinese English/Chinglish) 345
Other 232

acknowledge that the example of ‘Barack Obama – US president’ was given to

participants at the end of Section 1 when participants were invited to write
the full name of their role model (along with e.g. ‘“my grandmother” or “my
first English teacher”’; while this may have influenced participants to
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 11

Table 7: Top eight most frequently reported famous role models

Famous role models N (per cent)

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

Barack Obama 267 (3.2)
Emma Watson 125 (1.5)
Benedict Cumberbatch 113 (1.4)
J. K. Rowling 70 (0.8)
Stephen Fry 58 (0.7)
Queen Elizabeth II 54 (0.6)
Michelle Obama 48 (0.6)
Adele 46 (0.6)

volunteer Obama as a second role model, the vast number of mentions of

Obama—n ¼ 199 of 267 total mentions—can be found as a first role model
described by participants). The overall variation in responses is doubtless
considerable.

Research Question 3: Are there salient role model archetypes?

It appears that just as individual tastes differ, so do individual preferences
regarding role models. This would imply that role models display a virtually
unlimited variety, yet, a principle of CDST is that the self-organization cap-
acity of systems works to reduce variation, almost always resulting in a finite
number of archetypes (Dörnyei 2014). Therefore, we examined our role
model pool to identify any templates for role model archetypes that were par-
ticularly frequent. In order to capture robust tendencies without becoming
lost in the richness of detail that characterizes such vast datasets, we formed
five dichotomies from the basic role model rubrics given to participants in sec-
tion 1 of the questionnaire:
1 Sex: male vs. female
2 L1: native English speaker vs. non-native English speaker
3 Relationship to participant: famous vs. personally known
4 Job: same job vs. different job
5 Nationality: same nationality vs. different nationality
The permutations of these primary categories provided 32 possible combina-
tions (e.g. famous þ male þ native speaker þ different job þ different nation-
ality), and we began the analysis by computing the frequency of role model
occurrences for each combination. Because we were interested in salient
trends, we discounted the combinations which yielded the lowest frequency
numbers (N < 2 per cent). The two tables we enclose in Supplementary
Appendix B include a full summary of the remaining archetypes
(Supplementary Table B1 summarizing famous role models, and
12 ROLE MODELS IN LANGUAGE LEARNING

Supplementary Table B2 summarizing personally known role models), and

the frequency statistics relevant to the current discussion can be found pre-
sented in the first column of these tables.
Although participants described almost 50 per cent more famous than per-

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

sonally known role models, the substantial frequency of the latter (N > 3,100)
underlines their significance. Within the ‘famous’ category, the primary role
model archetype is a male native speaker (NS) in a job different to that of the
respondent. This category represents 34 per cent of all the role models
reported in our study, and the most popular named role model, Barack
Obama, is a prime example of this. The second most endorsed archetype is the
female counterpart of the first, epitomized by the second most popular named
role model, Emma Watson. Interestingly, a third famous archetype also
emerges different from the first two types in that it involves a non-native
speaker (NNS) of English of the respondent’s nationality (i.e. a local celebrity).
The primary archetype in the personally known subgroup was a female
non-native speaker of the same nationality as the respondent, representing
roughly 12 per cent of the total number of role models reported. The second
archetype in this category was its male counterpart, and as we might expect
these two NNS role model archetypes are roughly twice as common as the cor-
responding personally known NS speaker archetypes.
Thus, we can identify a small set of role model archetypes that explain a sig-
nificant proportion of the variance in the overall pool of role models reported.
The next section continues this discussion by examining their interaction with
discrete respondent subgroups.

Research Question 4: Is there any systematic variation in role

models described by specific participant subgroups?
It is clear that the characteristics of a person choosing a role model are in a dy-
namic relationship with the characteristics of the role model chosen. In order
to explore any systematic variation in this respect, we divided the respondents
into basic subgroups according to their sex, status (student/teacher), geo-
graphical grouping, and age group. When we displayed the frequency data on
a comprehensive spreadsheet, some robust patterns became detectable. We
then examined the occurrence of the 32 role model types identified in the pre-
vious section across these participant subgroups, computing the relevant fre-
quencies and adding these into the two tables included in Supplementary
Appendix B in new columns.
In some areas, various subgroups do not represent unique variation, as cer-
tain role model types are equally relevant across the subgroups. However, a
closer look at the figures did reveal some systematic variation, indicating inter-
esting and possibly fruitful areas for further research. After analysing the fre-
quency data in a visual manner and identifying possible trends (an established
approach in qualitative research to processing ‘display data’; see e.g. Miles and
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 13

Huberman 1994: 11), we verified our observations through chi-square statis-

tics. Let us look at findings from the four participant subgroups separately.

Variation stemming from participant sex. Famous NS role models were a key

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

archetype for both male and female participants, with this archetype account-
ing for 49 per cent and 43 per cent of all role models described by men and
women, respectively. A 3  3 chi-square analysis (including participant sex
‘prefer not to say’ and role model ‘other’ e.g. animated character) confirmed a
significant relationship between the sex of participants and their English lan-
guage role models, with male respondents more likely to describe male role
models, and female participants female models, x2 (4, n ¼ 8,349) ¼ 652.6, p <
0.001. Concordant with research in other disciplines, the reporting of other-
sex role models was more marked for female participants, and female partici-
pants in fact described more male famous NS role models than female famous
NS role models (24 per cent versus 19 per cent). Nevertheless, many male
respondents did have female role models: while 494 (45.2 per cent) of these
models were famous, there was also a substantial number of female role mod-
els personally known to male participants (N ¼ 539/49.3 per cent), most of
whom (N ¼ 319/29.2 per cent) were their teachers/professors.
A 2  3 chi-square analysis further confirmed a significant relationship be-
tween the participants’ sex and whether or not they reported NS role models,
with men more likely to describe NSs and women NNSs, x2 (2, n ¼ 8,349) ¼
7.866, p < 0.05. Importantly, a 3  3 chi-square analysis also confirmed a signifi-
cant relationship between the participants’ sex and their relationship with their
role model, with male participants more likely to describe a famous role model
and women a model personally known, x2 (4, n ¼ 8,349) ¼ 79.96, p < 0.001.

Teacher and student role model archetypes. The most common role model
archetypes described by both teachers and students reflected the overall pri-
mary archetypes (male and female NS celebrities), and the third most com-
mon archetype—personally known female NNS—may be explained in light of
the fact that 62 per cent of the teachers in our sample were female. A 2  3
chi-square analyses also indicated significant differences between teachers and
students with regard to whether their role models were famous/personally
known and whether they were in a similar job to them: teachers were more
likely to report personally known role models, x2 (2, n ¼ 8,349) ¼ 31.255, p <
0.001, and role models in a similar profession to themselves, x2 (2, n ¼ 8,349)
¼ 243.61, p < 0.001.

Variation across different geographical groupings. Chi-square analyses also

indicated significant and intriguing differences across the geographical/
mother tongue clusters, with two superordinate clusters emerging: (i)
Central and South America, EU, and Russia and (ii) China and India. The
distinction between them concerned whether the selected role models were
14 ROLE MODELS IN LANGUAGE LEARNING

L1 English speakers, whether they were the same nationality as respondents

and whether they were personally known. A 2  5 chi-square analysis con-
firmed that the second cluster—China and India—were more likely to de-
scribe NNSs than the first, x2 (4, n ¼ 5,310) ¼ 343.49, p < 0.001. This was

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

particularly marked for India (for which 59.1 per cent of all role models
described were NNSs), and was further underscored by the self-reported pro-
nunciation goals which indicated that Chinese and Indian learners valued
communicative ability over native-like pronunciation (see Table 8). A 2  5
chi-square analysis confirmed that China and India were more likely to re-
port on role models of the same nationality as them, x2 (4, n ¼ 5310) ¼
533.61, p < 0.001, as well as on personally known role models, x2 (8,
n ¼ 5310) ¼ 68.35, p < 0.001. These consistent differences indicate the im-
portance the two most populous ethnolinguistic communities of the world
place on ‘local heroes’ over western celebrities. The particularly high figures
in India are also likely to be due to the large number of Indian speakers of
English who can be considered native-like having been brought up and edu-
cated in English. In the context of this exploratory study, we asked partici-
pants themselves to make the judgement as to whether they perceived their
role model to be a native or a non-native speaker of English, and have like-
wise not been able to delve as deeply as we would like in our analysis into
issues surrounding ‘accent’ (both of role models themselves and with regard
to participants’ personal L2 goals). The initial findings we present here there-
fore highlight a fascinating area for future research.

Variation across participant age groups. Multiple chi-square analyses indi-

cated significant differences between the age of participants, the nationality
and L1 of their role models, and their relationship with them. A 5  2 chi-
square analysis confirmed that younger participants were more likely to report
role models that were NNSs, x2 (4, n ¼ 8349) ¼ 93.40, p < 0.000; a 5  3 chi-
square analysis confirmed that younger participants were more likely to report
on role models that were personally known, x2 (8, n ¼ 8349) ¼ 26.07, p <
0.001; and a 5  2 chi-square analysis confirmed that younger participants
were more likely to report on role models of the same nationality, x2 (4,
n ¼ 8349) ¼ 87.17, p < 0.000. Although we might have expected the opposite
(namely, that younger participants would turn to popular celebrity culture for
English language role models), our results suggest the draw of younger learn-
ers towards ‘near peer role models’ (we also note that this may be because
younger learners, more likely to still be in formal education, may be more
likely to be surrounded by near peers than their older counterparts).

Interim summary. There is evidence of systematic variation between role

model choice and characteristics of the various subgroups. However, to some
extent, the overall archetypes interfere with this. For example, even though a
same-sex preference in choosing role models was found for virtually all role
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 15

Table 8: Pronunciation goals reported by different geographical groupings

Geographical grouping ‘Sounding like a ‘Being able to communicate
native speaker is is a more important goal

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

important to me, even to me than to sound
though it is hard’, like a native speaker’,
N (per cent) N (per cent)
Central and South America 500 (44.1) 468 (41.3)
EU 781 (43.4) 652 (36.2)
Russian 297 (51.6) 181 (31.4)
Chinese 436 (42.0) 450 (43.4)
India 278 (31.0) 551 (61.4)

model types, the absolute figure of male celebrity role models chosen by fe-
male respondents was higher than that of female famous role models.
Similarly, while we observed a preference for younger respondents to choose
personally known NNS role models, here, too, the absolute frequency figure
for celebrity NS role models remained higher.

Research Question 5: What characteristics do participants

value in their English language role models?
In order to investigate the basis on which participants selected their role mod-
els, the survey asked respondents to rate 19 potential role model characteris-
tics/features. To assess the underlying factor structure of their evaluations, we
submitted their responses to exploratory factor analysis. The data were ideally
suited to exploratory factor analysis: the ratio of subjects to items is over 400
and the sample size likewise appropriate (1,000 participants or more is consid-
ered excellent by Tabachnick and Fidell 2001). Maximum likelihood extrac-
tion with oblimin rotation produced a four-factor solution (Table 9; see
Supplementary Appendix C for means and standard deviations of each item),
which explained 54.1 per cent of the variance. There was only one item that
did not load onto a factor at a level of at least 0.3 (‘Their rate of speech’), and
the factor matrix was likewise clear in the sense that there were no cross-
loadings. Bartlett’s test of sphericity was significant (<0.000), and the Kaiser–
Meyer–Olkin (KMO) measure of sampling adequacy was well above the ne-
cessary 0.6 threshold (0.856; see Pallant 2005). The clarity of the picture pre-
sented by the factor matrix offers strong support for the claim that all
respondents conceptualized their role models along the following four main
dimensions: overall command of English; paralinguistic features; demographic fea-
tures; and accent/variety of English.
To further investigate the claim that these four dimensions were representa-
tive of the way in which participants conceptualized their role models,
16 ROLE MODELS IN LANGUAGE LEARNING

Table 9: Factor analysis of the role model characteristics in the whole sample
(maximum likelihood extraction; oblimin rotation; loadings under 0.30
deleted)

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

Role model characteristic Factor

1 2 3 4
The size of their vocabulary 0.729
Their ability to explain themselves 0.682
Their spoken fluency 0.682
Their grammatical accuracy 0.677
Their ability to adapt their English for differ- 0.645
ent contexts (at a business meeting, dinner
with friends)
Their written English 0.614
Their confidence when they speak English 0.530
Their reaction when they do not under- 0.363
stand/know a word in English
Their understanding and use of humour 0.344
(telling jokes)
How ‘natural’ they look when they speak 0.305
English
Their rate of speech (how fast they talk)
Their facial expressions (eye contact, smile, 0.852
etc.)
Their gestures (how they use their hands and 0.813
arms when they talk)
Their personality more generally (if they are 0.321
friendly, patient or nice, for example)
Their age 0.762
Their job/profession 0.624
Their nationality 0.609
Their accent 0.730
The type of English they speak (American 0.660
English, British English, etc.)

additional exploratory factor analyses were conducted for key subsamples:

male and female participants, teachers and students, and each of the five cre-
ated geographical/mother tongue clusters. For all of these subgroups, a four-
factor solution produced factor structures similar to the above, lending further
support to this underlying structure. However, quantitative analyses are inev-
itably limited by the restricted number of pre-determined items that are
included in a questionnaire, and with this in mind participants were invited to
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 17

list further characteristics that they felt were important in their English lan-
guage role models. A total of 1,872 participants responded, and after excluding
responses that were unclear, too broad or irrelevant, a pool of over 1,600
responses remained.

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

Thematic analysis identified 29 categories underlying this qualitative data-
set. In Supplementary Appendix D, we include the full mapping of these cate-
gories onto the four dimensions described above (Supplementary Table D1).
The qualitative data offer a rich illustration of the content of these dimensions,
and we include this detailed information here so that it can be drawn upon in
future research. The mapping exercise also helped to identify a key aspect of
the role models concerning their personality and appearance that was not cov-
ered by the questionnaire data, for example, including attributes such as role
models’ charisma, general demeanour, personal appearance, and integrity/
ethics (a full overview of these results and sample data for each category can
likewise be found in Supplementary Appendix D, see Table D2). This can be
viewed as complementary to the demographic factors emerged from the factor
analysis, thereby forming a broader dimension that we have labelled personal
attributes. We can thus answer the fifth research question by concluding that
the role models reported in our study were evaluated by participants accord-
ing to four broad dimensions: overall command of English, paralinguistic features,
personal attributes, and accent/variety of English.

IMPLICATIONS AND LIMITATIONS

Theoretical implications
The findings of this study lend strong support to the claim that role modelling
is thoroughly deserving of more systematic and detailed investigation within
the field of SLA. The variation identified among geographical groupings high-
lights the importance of local context, and the interaction of this with the
international and transnational networks that students directly engage in, or
that they imagine themselves to be a member of. Other theoretical links can
be made to the notion of social capital and interesting questions surround, for
example, the description of other-sex role models, particularly men describing
female role models. There may be little coincidence that over three quarters of
the female role models described by men in this study were in positions with
influence or high social capital, often their teachers/professors (see Bandura
1986; Gibson and Cordova 1999). Investigation of changes in language learn-
ers’ role models over time is a further key area for future research, both
related to learners’ increasing proficiency in the L2 (see e.g. Gibson 2003, for
discussion in the context of investment banking and management consulting
documenting the changing descriptions of individuals’ role models at different
stages in participants’ careers), and changes stemming from experience and
exposure to different networks and communities (Kinginger 2008; Duff
2015).
18 ROLE MODELS IN LANGUAGE LEARNING

Linked to this, research is also needed to investigate variation in learners’

construal of their L2 role models (cf. Trope and Liberman 2010). For example,
in what contexts/situations do learners interact with or imagine their role
models, and in what types of communicative encounters? Investigation of the

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

frequency of any form of contact (real or imagined) of learners with their L2
role models is likewise critical. The implications with regard to L2 develop-
ment (or on intermediary variables such as motivation) are clear: if learners
have a L2 role model but rarely meet/imagine them, their existence is likely to
have little relevance. This might be investigated, for example, in relation to
motivation, engagement (Mercer and Dörnyei, in press), discrete aspects of L2
development (see Ushioda 2016), or learner emotions. The latter is particular-
ly timely considering the recent emphasis in the field of SLA on positive emo-
tions (MacIntyre et al. 2016), and research investigating the types of emotions
triggered by L2 role models (both positive and negative) and their varying im-
pact would be an interesting line of inquiry.

Pedagogical implications
The immediate practical implication of this study implies that teachers might
capitalize on the fact that so many of their students may already have L2 role
models. For example, they may help students to build links with them,
whether in person or—more likely—virtually, via tailored textbook or other
classroom resources, through various mediums of technology or via learners’
imaginations. Darvin and Norton’s (2015, 2017) work highlights the shifting
technological landscape, and ‘the capacity of both learners and teachers to
move fluidly across both time and space in an increasingly digital world’
(2017: 227). In fact, a classroom environment that cannot accommodate
learners’ role models may lead to significant negative consequences, as under-
scored by Norton’s (2001, 2013) related research on student non-
participation. As Pavlenko and Norton (2007: 678) explain, and as may be
equally true with regard to the recognition and inclusion of language learner
role models, ‘If we do not acknowledge the imagined communities of the
learners, we may exacerbate their non-participation and impact their learning
trajectories in negative ways’, leading even to withdrawal from study.
Pavlenko and Norton further highlight Kanno’s (2003) work, emphasizing
the importance of not only the future visions of learners’ themselves, but also
the visions that schools have for their learners, and the impact of this on stu-
dents’ identity and academic development (see also Darvin and Norton 2017,
for a discussion of the importance of the language learning environment in
affording learners opportunities to enact their identity, and to fully invest in
the language learning process). This again underlines the importance of adopt-
ing a person-in-context-relational-view (Ushioda 2009), and of Ushioda’s call to
understand and engage in the classroom all aspects of learners’ transportable
identities (Ushioda 2011): to include and acknowledge their visions for the
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 19

future and the role models they hold dear with regard to their L2 goals and
aspirations.
As has been discussed in the context of the motivational potential of learn-
ers’ ideal L2 selves (Dörnyei and Kubanyiova 2014), so is it likely that there is

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

a specific set of conditions that must be met before any pedagogical value from
L2 role models might be realized. As a starting point, we propose that it is
highly likely that many of the conditions identified in relation to the former
are equally relevant to the latter. For example, conditions relating to the per-
ceived plausibility of reaching competence similar to that of learners’ role mod-
els, or understanding how they might achieve this (i.e. the existence of a
roadmap of relevant plans and strategies). In considering potential pedagogical
applications, research investigating this supposition is needed, and also in rela-
tion to the varying levels of importance of each condition. In the context of
ideal L2 selves there has been little research emphasis on this important ques-
tion, yet one interesting study has found, for example, that the frequency with
which an ideal L2 self is imagined is strongly associated with participants’ mo-
tivation to attain it (Hessel 2015).
Finally, we wish to highlight the pedagogical relevance of role models not
only to students but also to teachers. The large subgroup of data collected from
English teachers in this study has clearly provided an initial demonstration of
relevance, and pedagogical implications with regard to teacher training may
stand as a further line of research that could prove to be particularly fruitful.
This may link, for example, to the importance of L2 teacher role models in the
development of specialized ‘craft knowledge’ that cannot be learned from
textbooks (in the context of medical education see e.g. Perry 2009, for discus-
sion of the importance of this with regard to learning to use silence as an ef-
fective communicative tool).

LIMITATIONS
Our study has several limitations, the most serious arguably being that the
pervasiveness of the term ‘role model’ in everyday parlance may have overrid-
den the definition given to participants: some participants may have described
mentors rather than role models, or have mixed up role models with English
speakers they looked up to but whom they did not necessarily try to imitate.
Moreover, the number of role models was capped at two, and it is not clear
whether participants described current or past role models.
Further limitations pertain to the online questionnaire format. We have not
been able to account for participants completing the questionnaire more than
once, the likelihood of which may have increased due to the prize-draw
offered. This may also have occurred from participants wishing to describe
more than the two role models the questionnaire design allowed for. With re-
gard to the former, we believe it is unlikely that if some participants did com-
plete the questionnaire more than once that this will have impacted
significantly on the results of this exploratory study, particularly because our
20 ROLE MODELS IN LANGUAGE LEARNING

analysis focused on identifying broad trends. With regard to the latter, we feel
participants would have been discouraged from doing this because of the add-
itional information they would have needed to complete in the questionnaire
prior to resubmitting. We further acknowledge that we cannot verify that the

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

background information given by the participants was correct.
A final source of limitation concerns the fact that no variables were included
that allowed for investigation of the impact of these role models on either par-
ticipants’ motivation or other learning behaviours, yet this can be accounted
for in the design of this study as an initial exploratory investigation into these
issues. Existing research on role models in other disciplines has been criticized
for not offering sufficient evidence that the attitudinal influences role models
exert are translated into actual attainment, and research has suggested that
the primary influences of sex- and race-specific role models might be exerted
more broadly and indirectly, for example, in the creation of a more inclusive
school environment (Carrington and Skelton 2003). This again highlights
clear links between role modelling and notions of identity, investment, and
possible selves. In order to usefully utilize the concept of role modelling in
SLA, future research must address this directly by including a broad range of
criterion measures (both behavioural and attitudinal).

CONCLUSION
Our research project was motivated by the initial belief that role models
play an important part in language learning, an assumption supported by
investigations elsewhere in the field of education as well as across other disci-
plines. We conducted an exploratory study to establish the main parameters
of the subject and to address the fundamental questions of how widespread
role models are and what characteristic features people highlight about them.
We aimed to present an extensive baseline dataset that can act as a useful ref-
erence point and springboard for future work in this area. Based on the
results, we have outlined a four-component framework of the underlying
structure of role model appraisal—overall command of English, paralinguistic fea-
tures, personal attributes and accent/variety of English—and in Supplementary
Appendix D have provided rich qualitative illustrations of the content of each
dimension.
We believe that our findings offer sufficient evidence for the importance of
L2 role models within the process of mastering an L2, and that our initial find-
ings warrant further, more focused investigations. An important question to
be answered by future research is whether role modelling is gradable; that is,
are there weaker and stronger role models? Although this issue was not
addressed directly in our study, the information we have gathered suggests
that role models exert variable influence, and if this indeed turns out to be the
case, it may be an interesting research programme to identify correlations be-
tween aspects of role models’ behaviour and the impact they have on others.
Related to this question, Bandura (1977) has suggested that some people are
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 21

more likely to be susceptible to modelling influences than others, and it may

well be the case that there is a dynamic interaction between certain types of
role models and recipients. This received indirect evidence from the fact that
certain participant subgroups in our study displayed marked preferences for

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

certain role model types. Finally, future research might also examine any pos-
sible barriers that stop an individual from considering someone a role model,
such as negative stereotypes or clashes with other possible selves (see e.g.
Buck et al. 2008). In sum, investigations into the intriguing subject of L2 role
modelling are likely to bring forth several, as yet untapped seams to be mined.

SUPPLEMENTARY DATA
Supplementary material is available at Applied Linguistics online.

FUNDING
This study was funded by an Engineering and Physical Sciences Research Council
Impact Acceleration Account grant (EPSRC IAA Reference: RR1402).

REFERENCES
Adolphs, S., L. Clark, Z. Dörnyei, T. Glover, A. adolescent girls and women scientists in identi-
Henry, C. Muir, E. Sánchez-Lozano, and M. fying science role models: A feminist approach,’
Valstar. 2018. ‘Digital innovations in L2 motiv- Science Education 92: 688–707.
ation: Harnessing the power of the Ideal L2 Carrington, B. and C. Skelton. 2003.
Self,’ System 78: 173–85. ‘Re-thinking ‘role models’: Equal opportunities
Althouse, L. A., F. T. Stritter, and B. D. in teacher recruitment in England and Wales,’
Steiner. 1999. ‘Attitudes and approaches of in- Journal of Education Policy 18: 253–65.
fluential role models in clinical education,’ Cashmore, E. 2006. Celebrity/Culture. Routledge.
Advances in Health Sciences Education 4: 111–22. Collier-Meek, M. A., L. M. Fallon, A. H.
Anderson, B. 2006. Imagined Communities: Johnson, L. M. H. Sanetti, and M. A.
Reflections on the Origin and Spread of Nationalism. Delcampo. 2012. ‘Constructing self-modeling
Verso. videos: Procedures and technology,’ Psychology
Bandura, A. 1977. Social Learning Theory. Prentice in Schools 49: 3–14.
Hall. Cox, R., J. Mckendree, R. Tobin, J. Lee, and T.
Bandura, A. 1986. Social Foundations of Thought and Mayes. 1999. ‘Vicarious learning from dialogue
Action: A Social Cognitive Theory. Prentice Hall. and discourse,’ Instructional Science 27: 431–58.
Bandura, A. 1997. Self-Efficacy: The Exercise of Curdt-Christiansen, X. L. and C. Weninger
Control. W.H. Freeman and Company. (eds). 2015. Language, Ideologies and Education:
Barkhuizen, G. 2016. ‘A short story approach to The Politics of Textbooks in Language Education.
analyzing teacher (imagined) identities over Routledge.
time,’ TESOL Quarterly 50: 655–83. Darvin, R. and B. Norton. 2015. ‘Identity
Boon, S. D. and C. D. Lomore. 2001. and a model of investment in language learning,’
‘Admirer-celebrity relationships among young Annual Review of Applied Linguistics 35: 36–56.
adults: Explaining perceptions of celebrity influ- Darvin, R. and B. Norton. 2017. ‘Language,
ence on identity,’ Human Communication identity, and investment in the twenty-first cen-
Research 27: 432–65. tury’ in T. L. McCarty and S. May (eds):
Buck, G. A., V. L. P. Clark, D. Leslie-Pelecky, Language Policy and Political Issues. Encyclopedia of
Y. Lu, and P. Cerda-Lizarraga. 2008. Language and Education, 3rd edn. Springer, pp.
‘Examining the cognitive processes used by 227–40.
22 ROLE MODELS IN LANGUAGE LEARNING

Dörnyei, Z. 2005. The Psychology of the Language Henry, A. 2015. ‘The dynamics of possible selves’
Learner. Individual Differences in Second Language in Z. Dörnyei, P. D. MacIntyre, and A. Henry
Acquisition. Lawrence Erlbaum. (eds): Motivational Dynamics in Language
Dörnyei, Z. 2009. ‘The L2 motivational self sys- Learning. Multilingual Matters, pp. 83–94.

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

tem’ in Z. Dörnyei and E. Ushioda (eds): Hessel, G. 2015. ‘From vision to action: Inquiring
Motivation, Language Identity and the L2 Self. into the conditions for the motivational capacity
Multilingual Matters, pp. 9–42. of ideal second language selves,’ System 52:
Dörnyei, Z. 2014. ‘Researching complex dynamic 103–14.
systems: ‘Retrodictive qualitative modelling’ in Ibarra, H. 1999. ‘Provisional selves:
the language classroom,’ Language Teaching 47: Experimenting with image and identity in pro-
80–91. fessional adaptation,’ Administrative Science
Dörnyei, Z. and M. Kubanyiova. 2014. Quarterly 44: 764–91.
Motivating Learners, Motivating Teachers: Building Kanno, Y. 2003. ‘Imagined communities, school
Vision in the Language Classroom. Cambridge visions, and the education of bilingual students
University Press. in Japan,’ Journal of Language, Identity, and
Duff, P. A. 2007. ‘Second language socialisation Education 2: 241–9.
as sociocultural theory: Insights and issues,’ Kinginger, C. 2008. ‘Language learning in study
Language Teaching 40: 309–19. abroad: Case studies of Americans in France,’
Duff, P. A. 2015. ‘Transnationalism, multilingual- Modern Language Journal 92: 2–124.
ism, and identity,’ Annual Review of Applied Krietemayer, B. C. and S. P. Heiney. 1992.
Linguistics 35: 57–80. ‘Storytelling as a therapeutic technique in a
Duff, P. A. and Y. Uchida. 1997. ‘The negotiation group for school aged oncology patients,’
of teachers’ sociocultural identities and practices Children’s Health Care 21: 14–20.
in postsecondary EFL classrooms,’ TESOL MacIntyre, P., T. Gregersen, and S. Mercer. 2016.
Quarterly 31: 451–86. Positive Psychology in SLA. Multilingual Matters.
Festinger, L. 1954. ‘A theory of social comparison Markus, H. and P. Nurius. 1986. ‘Possible selves,’
processes,’ Human Relations 7: 117–40. American Psychologist 41: 954–69.
Fox, F. F. 2003. ‘Reducing intercultural friction Mercer, S. and Z. Dörnyei. in press. Engaging
through fiction: Virtual cultural learning,’ Students in Contemporary Classrooms. Cambridge
International Journal of Intercultural Relations 27: University press.
99–123. Miles, M. B. and A. M. Huberman. 1994.
Fraser, B. P. and W. J. Brown. 2002. ‘Media, Qualitative Data Analysis: An Expanded Sourcebook,
celebrities, and social influence: Identification 2nd edn. Sage.
with Elvis Presley,’ Mass Communication & Society Murphey, T. and H. Arao. 2001. ‘Reported belief
5: 183–206. changes through near peer role modelling,’
Gibson, D. E. 2003. ‘Developing the professional TESL-EJ 5. http://tesl-ej.org/ej19/a1.html.
self-concept: Role model construals in early, Murphey, T. and K. Murakami. 1998. ‘Teacher
middle and late career stages,’ Organization facilitated near peer role modelling for aware-
Science 14: 591–610. ness raising within the Zone of Proximal
Gibson, D. E. 2004. ‘Role models in career develop- Development,’ Academia 65: 1–29.
ment: New directions for theory and research,’ Nemtchinova, E. 2005. ‘Host teachers’ evalua-
Journal of Vocational Behavior 65: 134–56. tions of nonnative-English-speaking teacher
Gibson, D. E. and D. I. Cordova. 1999. trainees: A perspective from the classroom,’
‘Women’s and men’s role models: The import- TESOL Quarterly 39: 235–61.
ance of exemplars’ in A. J. Murrell, F. J. Crosby, Northedge, A. 2003. ‘Enabling participation in
and R. J. Ely (eds): Mentoring Dilemmas: academic discourse,’ Teaching in Higher Education
Developmental Relationships within Multicultural 8: 169–80.
Organizations. Lawrence Erlbaum, pp. 121–42. Norton, B. 2001. ‘Non-participation, imagined
He, D. and Q. Zhang. 2010. ‘Native speaker communities and the language classroom’ in M.
norms and China English: From the perspective Breen (ed.), Learner Contributions to Language
of learners and teachers in China,’ TESOL Learning: New Directions in Research.
Quarterly 44: 769–89. Longman/Pearson Education, pp. 159–71.
 C. MUIR, Z. DÖRNYEI, AND S. ADOLPHS 23

Norton, B. 2013. Identity and Language Learning: Tabachnick, B. G. and L. S. Fidell. 2001. Using
Extending the Conversation, 2nd edn. Multivariate Statistics, 4th edn. HarperCollins.
Longman/Pearson Education. Thompson, A. S. and C. Vásquez. 2015.
Norton, B. and K. Toohey. 2011. ‘Identity, lan- ‘Exploring motivational profiles through lan-

Downloaded from https://academic.oup.com/applij/article/42/1/1/5670576 by University of Nottingham user on 19 August 2023

guage learning, and social change,’ Language guage learning narratives,’ The Modern Language
Teaching 44: 412–46. Journal 99: 158–74.
Nowell, L. S., J. M. Norris, D. E. White, and N. Trope, Y. and N. Liberman. 2010.
J. Moules. 2017. ‘Thematic analysis: Striving to ‘Construal-level theory of psychological dis-
meet the trustworthiness criteria,’ International tance,’ Psychological Review 117: 440–63.
Journal of Qualitative Methods 16: Turner, G. and J. Shepherd. 1999. ‘A method in
160940691773384–13. search of a theory: Peer education and health
Paice, E., S. Heard, and F. Moss. 2002. ‘How im- promotion,’ Health Education Research 14:
portant are role models in making good doc- 235–47.
tors?,’ BMJ 325: 707–10. Ushioda, E. 2009. ‘A person-in-context relational
Pallant, J. 2005. SPSS Survival Manual, 2nd edn. view of emergent motivation, self and identity’
Open University Press. in Z. Dörnyei and E. Ushioda (eds): Motivation,
Pavlenko, A. and B. Norton. 2007. ‘Imagined Language Identity and the L2 Self. Multilingual
communities, identity, and English language Matters, pp. 215–28.
learning’ in J. Cummins and C. Davison (eds): Ushioda, E. 2011. ‘Motivating learners to speak as
International Handbook of English Language. themselves’ in G. Murray, Q. Gao, and T. Lamb
Springer, pp. 669–80. (eds): Identity, Motivation and Autonomy in
Perry, R. N. B. 2009. ‘Role modeling excellence Language Learning. Multilingual Matters, pp.
in clinical nursing practice,’ Nurse Education in 11–24.
Practice 9: 36–44. Ushioda, E. 2016. ‘Language learning motivation
Roberts, D. 2010. ‘Vicarious learning: A review of through a small lens: A research agenda,’
the literature,’ Nurse Education in Practice 10: Language Teaching 49: 564–77.
13–16. Wenger, E. 1998. Communities of Practice: Learning,
Solomon, R. P. 1997. ‘Race, role modelling, and Meaning, and Identity. Cambridge University
representation in teacher education and teach- Press.
ing,’ Canadian Journal of Education 22: 395–410. Wright, S., A. Wong, and C. Newill. 1997. ‘The
Spouse, J. 2003. Professional Learning in Nursing. impact of role models on medical students,’
Blackwell Science. Journal of General Internal Medicine 12: 53–6.