Professional Documents
Culture Documents
People often believe that some language varieties are more prestigious than others, which can trigger
speech-centered biases and inform social judgments of the speaker. However, it is largely unknown what
types of language experience and exposure might mitigate language biases, especially for second lan-
guage (L2) learners. The goal of this study was to investigate this issue by focusing on L2 French learners’
attitudes toward European and Quebec varieties of French. L2 French learners in Montreal (N = 106)
rated 2 audios recorded by native speakers from France in a listening comprehension task, with 1 of
the 2 speakers introduced as a speaker of Quebec French. The learners described their language learn-
ing experience, filled out a French social network questionnaire, and completed a French proficiency
test. Results revealed some evidence of reverse linguistic stereotyping, with learners preferring to speak
like one speaker significantly more than the other, based on the speaker’s assumed identity, not actual
speech. Four of the 6 speaker ratings were also associated with participants’ oral proficiency scores, social
network density, and positive experiences in Quebec. Findings have implications for the use of speech
models in L2 teaching and for the mitigation of language-centered biases in L2 classrooms.
Keywords: Quebec French; European French; reverse linguistic stereotyping; language attitudes; L2
French
Consent Participants were introduced to the study and affirmed consent 3 minutes
Practice The researcher administered a practice exercise with sample 3 minutes
map and scale
Speaker The first speaker was introduced as a French teacher from 1 minute
introduction Quebec or from France, depending on counterbalanced order
assignment
Map Task 1 Participants followed the speaker’s directions by drawing a line 1 minute
on their map accordingly
Rating scales Participants used scales to rate the speaker and optionally left 7 minutes
comments regarding their ratings
Map Task 2 Steps 3–5 were repeated with the second speaker, who was 9 minutes
introduced as a French teacher from France or from Quebec,
depending on counterbalanced order assignment
Questionnaires Participants filled out the background questionnaire and social 10 minutes
network survey
Listening Participants completed the TEFAQ listening comprehension test 20 minutes
comprehension (used to derive a listening comprehension measure)
Speaking activity Participants recorded responses to two open-ended questions in 5 minutes
French with a partner (used to derive a measure of L2
speaking)
Debrief Participants were informed of the true purpose of the study and 3 minutes
were told that both speakers were from France
Component
Questionnaire item 1 2 3 4
compared to European French, was labeled Per- specific measure of French listening compre-
sonal relevance of Quebec French. Factor 3, labeled hension. To obtain a measure of participants’
Positive experience in Quebec, captured how welcome L2 speaking, their audio recordings were evalu-
the participants felt in Quebec and the extent to ated by two native French judges (one European
which they experienced positive interactions with French speaker, one Quebec French speaker)
Quebec French speakers. Factor 4, labeled Quebec who both had 10 years of experience teaching
French use, dealt with participants’ amount of ex- French and were employed as evaluators for the
posure to Quebec French through media or na- TEFAQ exam (with 1 and 6 years of experience,
tive speakers, and with how important they per- respectively). To stay consistent with the grading
ceived the use of French to be in Quebec. Four of the TEFAQ, the standardized evaluation crite-
factor scores (one per component) were derived ria for the Common European Framework of Ref-
for further analyses using the Anderson–Rubin erence for Languages (CEFR) were used to rate
method for obtaining noncorrelated scores fol- participants’ range in linguistic forms and vocabu-
lowing PCA. lary, their grammatical accuracy, speech flow, and
coherence. The levels consisted of A0, A1, A2, B1,
B2, C1, and C2 (where A0 corresponds to very lim-
Social Network Survey
ited knowledge of French, and C2 implies native-
Following Doucerain et al. (2015), four mea- like performance). For statistical analysis, each
surements were calculated per participant: L2 net- level was converted to a numerical value (where
work size (total number of French speakers in C2 was worth 10 points, as the highest level at-
one’s social network), L2 intimacy (average close- tainable, and A0 was worth 4 points). Therefore,
ness rating across all people listed), L2 inclusive- a score corresponding to a level was assigned to
ness (number of nonisolated people divided by each speaking component (range, accuracy, flu-
the L2 network size), and L2 density (number of ency, coherence), for a total score of 40 possible
links between the people divided by the number points. The two raters demonstrated high consis-
of possible links). Because we focused on attitudes tency in their ratings (.90), so their overall scores
toward Quebec French, further analyses only in- for each participant were averaged to derive a
cluded the scores for Quebec French speakers in single measure. The two tests thus yielded two
participant networks. separate measures: a L2 listening comprehension
score, which was relevant to the target listening
task, and a L2 speaking score, which provided a
Measures of L2 French Speaking and Listening
study-specific estimate of participants’ speaking
Comprehension
ability (given that our participants came from vari-
The TEFAQ comprehension test was graded ous language programs and course levels). These
out of 20 points, with the score used as a study- two measures were explored separately for their
Rachael Lindberg and Pavel Trofimovich 831
FIGURE 1
Histogram of Participants’ Ratings of Speakers’ Regional Identity
50 50
40 40
Number of parcipants
Number of parcipants
30 30
20 20
10 10
0 0
0 20 40 60 80 100 0 20 40 60 80 100
European guise rang Quebec guise rang
Note. 0 = Quebec French, 100 = European French.
potential relationships with participants’ audio plain) the variation in participants’ tendency to
ratings. go along with the manipulation and to explore its
possible impact on speaker ratings, participants’
rating of the Quebec guise (reflecting their de-
RESULTS
gree of sensitivity to the manipulation) was in-
Preliminary Analysis cluded as a variable in subsequent quantitative
analyses.
Before proceeding with the main analysis, we
first checked the extent to which participants
L2 Learners’ Attitudes Toward Quebec French
believed that the Quebec guise indeed involved
a Quebec French speaker. This check focused Speech Ratings. To answer the first RQ, which
on the scale asking participants to identify the asked whether L2 learners demonstrate any at-
speaker’s French variety (0 = Quebec French, titudinal biases toward Quebec French, we first
100 = European French). The speaker in the Eu- compared participants’ ratings of the speakers in
ropean guise was recognized consistently as Eu- the two guises for the six target dimensions: ac-
ropean French, as illustrated in the negatively centedness, comprehensibility, intelligence, com-
skewed histogram in Figure 1 (left panel). In con- petence as a teacher, desire to have the speaker
trast, the speaker in the Quebec guise elicited as a teacher, and desire to sound like her. If par-
a wide range of ratings, sometimes identified as ticipants held any attitudinal biases toward Que-
Quebec French and sometimes (correctly) as Eu- bec French, they should rate the two speakers dif-
ropean French (Figure 1, right panel). Thus, ferently depending on their presented linguistic
there was a range of sensitivity to the RLS ma- identity as a speaker of either European or Que-
nipulation among participants, such that some bec French. As summarized in Table 3, planned
participants believed the false presentation of the pairwise comparisons yielded only one significant
speaker’s identity while others rejected the false difference—namely, participants were less willing
identity by labeling the speaker correctly as Euro- to sound like the speaker in the Quebec guise
pean French. We assumed that those who believed (M = 69.39) than the speaker in the European
our false presentation of the speaker’s identity guise (M = 76.20), with a small effect size (Plon-
would be those whose preconceived ideas about sky & Oswald, 2014). This result should be inter-
Quebec French might be activated. We expected preted cautiously given that a conservative Bonfer-
these participants to reveal attitudinal biases to- roni correction (α = .008) would have rendered
ward Quebec French through more negative rat- the obtained difference nonsignificant.
ings, compared to the ratings for the speaker in Although participants generally showed a pref-
the European guise (e.g., Kang & Rubin, 2009). erence to sound like the speaker in the Euro-
Therefore, to account for (and ultimately ex- pean guise, this tendency was amplified by their
832 The Modern Language Journal 104 (2020)
TABLE 3
Summary Statistics for Rated Dimensions and Between-Guise Comparisons
Note. Degrees of freedom (df) vary because of several missing responses per rated dimension.
FIGURE 2
Relationship Between Participants’ Sensitivity to the Guise Manipulation and Desire to Emulate Guise
100
Desire to speak like Quebec guise
80
60
40
20
0 20 40 60 80 100
Quebec guise rang
Note. The regression line shows the best linear fit to the data.
sensitivity to the guise manipulation (illustrated tential attitudinal reactions that might underlie
in Figure 2 through regression). As shown by a their ratings. Regarding the question after each
significant β coefficient of .22, where a linear re- audio asking participants to choose descriptors
lationship proved to provide the best fit, t = 2.31, characterizing each speaker’s speech, they tended
p = .02, 95% CI = [.03, .40], those who tended to to choose positive and negative adjectives at simi-
believe the Quebec guise manipulation (low rat- lar rates in response to the Quebec French guise
ings on the x-axis) were less likely to prefer the (81% positive, 19% negative) and the European
presumed Quebec French speaker as a desirable French guise (86.2% positive, 13.8% negative).
speech model (by an average of −22 points on The descriptors most often chosen for Quebec
a 100-point scale) than those who were not sus- French were normal (63 out of 106 participants),
ceptible to the manipulation (high ratings on the clear (57), smooth (56), standard (49), and ac-
x-axis). cented (38), and these rates were similar to those
shown for the European French guise (see On-
Qualitative Responses. We next examined par-
line Supporting Information C for full frequency
ticipants’ qualitative responses to capture any po-
Rachael Lindberg and Pavel Trofimovich 833
TABLE 4
Sample Reasons for Participants’ Preferred Speaker
Preference for European French guise Preference for Quebec French guise
“[She] speak as native France French. She didn’t “I’m more used to her pronunciation and the rest
use the accent, her voice was pure and of the people at Quebec pronounces like her.”
international.” “If you are living in Quebec, you should get in use
“Preferred that accent and more universal to this accent.”
(standard).” “Since I live in Quebec, I would choose this
“I felt that it’s more useful to learn standard teacher at the moment.”
French as it is better understood worldwide.” “Since I personally need to get more used to a
“Because French originated in France and I prefer Quebecois accent.”
French teachers.” “Easier to understand, more accented speaking.”
“More sophisticated.” “Because she has the accent Quebecois and I want
“More elegant and was easy to understand and to learn like that.”
clear, which is required.” “I liked the more accented voice.”
“More used to France French accent, so find it
easier to follow.”
data). However, participants appeared to have pants who justified their preference for the Que-
attributed certain positive descriptors more fre- bec French guise, most reasons were also related
quently to the European French guise than to to the speaker’s speech being easier to understand
the Quebec French guise: international (26 vs. 8), (52% of the reasons given), more pleasant to lis-
proper (50 vs. 36), the original (24 vs. 10), sophisti- ten to (20%), or more familiar (8%)—but also
cated (27 vs. 11), elegant (33 vs. 20), and pure (26 because they preferred to learn Quebec French
vs. 17). In addition, when given the same word (20%). As seen in Table 4, at least some of these
bank on the background questionnaire and di- reasons seem to have been based on the assump-
rectly asked to circle those describing Quebec tion that the two speakers had different linguistic
French, twice as many participants overall chose backgrounds, revealing attitudes toward the two
accented (76 out of 106 participants), making it French varieties. Taken together, the results from
the most frequent descriptor, followed by difficult speech ratings and questionnaire responses sug-
(47), unclear (41), nonstandard (33), bizarre (27), gest that participants did not generally differ in
and normal (27). Negative descriptors were cho- their evaluations of the two speech guises (shown
sen over three times more frequently to describe through ratings) but that they nevertheless held
Quebec French on the questionnaire (32.9% pos- some bias against Quebec French (shown through
itive, 67.1% negative) than after listening to the their qualitative responses).
speaker (81% positive, 19% negative).
At the end of the session, participants selected
Proficiency, Exposure, and Experience and Learners’
the speaker they preferred for the activity, allow-
Attitudes Toward Quebec French
ing them to express their overall impressions af-
ter being able to compare both speakers. Based To answer the second RQ, which investigated
on 33 participants who explained their prefer- possible associations between L2 learners’ atti-
ence for the European French guise, their pref- tudes toward Quebec French and their profi-
erence was largely based on the speaker’s speech ciency and exposure profiles, the next analysis in-
being clearer and easier to understand (46.5% volved Pearson correlations (two-tailed) between
of the reasons), even though they did not distin- participants’ ratings of the speaker in the Quebec
guish the two speakers through comprehensibil- guise and their proficiency, exposure, and expe-
ity ratings. In addition, their choice was explained rience variables specific to Quebec French (see
through reference to the speaker’s French be- Online Supporting Information D for a parallel
ing more standard (14%), more aesthetically set of analyses targeting participants’ attitudes to-
pleasing, purer and more sophisticated (11.6%), ward European French; although these analyses
more pleasant to listen to (9.3%), more famil- fall outside this study’s scope, they might be of in-
iar (9.3%), or because they simply preferred the terest to some readers). As shown in Table 5, four
accent from France (9.3%). Of the 25 partici- of the six ratings of the speaker in the Quebec
834 The Modern Language Journal 104 (2020)
TABLE 5
Correlations Between Background Variables and Speaker Ratings in the Quebec French Guise
French guise (comprehensibility, desire to speak in the Quebec French guise (0 = Quebec French,
like the speaker, her competence as a teacher, and 100 = European French) were entered in Step 1
desire to have her as a teacher) were associated in both models, to account for participants’ sensi-
with several background and exposure variables. tivity to the guise manipulation, and the relevant
The PCA-derived background variable of having background and exposure variables were entered
had positive experiences in Quebec (i.e., feeling in a stepwise procedure in Step 2. For comprehen-
welcome in Quebec, having positive experiences sibility, the predictors entered in the model were
with Quebec French speakers) was the only vari- Variables 1–3, 7, 8, 11, and 12 (as shown in Ta-
able positively linked to the two teacher ratings, ble 5). For desire to speak like the speaker, the
with weak associations (Plonsky & Oswald, 2014). predictors entered in the model were Variables 4–
Those participants who had reported more pos- 6 and 11 (as listed in Table 5). The predictors cho-
itive experiences in Quebec, compared to those sen for each model only included those that had
with fewer positive experiences, provided higher the potential to influence the criterion variable in
ratings for the speaker in the Quebec guise, eval- each model, based on their nontrivial associations
uating her as a more competent teacher and ex- (r > .20) with the measures of comprehensibility
pressing a stronger desire to have her as their and desire to speak like the speaker. The sample
teacher. size of 106 was deemed sufficiently large to detect
To determine the relative weight of multiple a medium-size effect with up to 10 predictors en-
background and exposure variables associated tered into the model (Field, 2009).
with the ratings of comprehensibility and desire The regression model for comprehensibility
to speak like the speaker (see Table 5), we carried (summarized in Table 6) yielded a two-factor solu-
out two regression analyses separately for these tion, accounting for 19% of shared variance. First,
two ratings. In each analysis, the speech rating participants’ sensitivity to the Quebec guise was
was the criterion variable, and the background associated with the rating of how comprehensi-
and exposure variables in Table 5 that were signif- ble they found the speaker (7% of shared vari-
icantly correlated with the relevant speech rating ance), F(1,90) = 6.94, p = .01, such that those
served as predictors. The ratings of speaker origin who were more likely to go along with the guise
Rachael Lindberg and Pavel Trofimovich 835
TABLE 6
Results of Multiple Regression Analyses Using Background and Exposure/Experience Variables as Predictors
of Speaker Ratings in the Quebec French Guise
Comprehensibility
Sensitivity to guise manipulation .27 .07 0.26 [0.07, 0.46] 2.63 .010
Speaking score .44 .12 2.13 [0.95, 3.31] 3.58 .001
Desire to speak like the speaker
Sensitivity to guise manipulation .23 .05 0.22 [0.03, 0.41] 2.31 .023
Social network density .36 .08 −25.64 [−43.46, −7.81] −2.86 .005
Positive experience in Quebec .47 .09 8.01 [3.22, 12.79] 3.32 .001
manipulation (i.e., labeling the speaker as Que- network density (8% of variance explained) and
bec French) tended to provide lower compre- positive experience in Quebec (8% of variance ex-
hensibility ratings. Speaking score was the only plained) predicting participants’ desire to sound
other significant predictor accounting for addi- like the presumed Quebec French speaker.
tional variance (12%) in rated comprehensibil-
ity, F(2,90) = 10.34, p < .001. Thus, regardless of
their sensitivity to the guise manipulation, partic- DISCUSSION
ipants with higher L2 speaking scores, compared
L2 Learners’ Attitudes Toward Target Language
to those with lower scores, tended to assign higher
Varieties
ratings to the speaker in the Quebec guise.
The regression model targeting participants’ The first RQ asked whether L2 French learn-
desire to sound like the speaker (see Table 6) ers demonstrate any attitudinal bias toward Que-
yielded a three-factor solution, with sensitivity to bec French, either through speaker ratings in a
guise manipulation, social network density, and RLS procedure or through their qualitative re-
positive experience in Quebec jointly accounting sponses. With respect to speaker ratings, any dif-
for 22% of shared variance in participants’ de- ference in the ratings given to two speakers of
sire to sound like the speaker (R2 = .22). Partici- European French (one of whom was falsely pre-
pants’ sensitivity to the guise manipulation again sented as a Quebec French speaker) would be
predicted the rating (5%), F(1,95) = 5.35, p = consistent with RLS, due to learners’ judgments
.023. The density of participants’ Quebec French being influenced by the label of the speaker’s
social networks accounted for additional variance identity. Despite our expectations, there was only
(8%), F(2,95) = 6.96, p = .002, with a negative as- weak support for possible bias. Of the six target
sociation. The PCA-derived background variable measures, only the rating of learners’ desire to
of having had positive experiences in Quebec was sound like the speaker (a solidarity trait) was asso-
the final predictor (9%), F(3,95) = 8.83, p < .001, ciated with a between-speaker difference, where
with a positive association. Thus, regardless of learners expressed a marginally stronger prefer-
their sensitivity to the guise manipulation, partici- ence to speak like the speaker in the European
pants with sparser Quebec French social networks French guise than in the Quebec French guise.
and those with more positive experiences in Que- Thus, at least for this measure, learners seemed to
bec tended to express a stronger desire to sound have engaged in RLS, with knowledge of Quebec-
like the speaker in the Quebec guise compared to specific stereotypes contributing to their ratings.
those with denser social networks and fewer posi- Social bias linked to a particular social group
tive experiences in Quebec. Finally, we conducted often affects not only speaker preference but
two post hoc regression analyses (suggested by an also listening comprehension (Hu & Su, 2015;
anonymous reviewer) where participants’ ratings Kang & Rubin, 2009; Rubin, 1992). To the ex-
of speaker origin in the European French guise tent that comprehensibility (a measure of ease of
was entered as a variable in Step 1. Results re- understanding) captures some aspects of speech
mained unchanged, with participants’ speaking comprehension (see Kang, Thomson, & Moran,
scores being the only significant predictor of the 2019), there was no quantitative evidence in this
presumed Quebec French speaker’s comprehen- study of preconceived ideas influencing learn-
sibility (9% of variance explained) and with social ers’ comprehension of the speakers, as both were
836 The Modern Language Journal 104 (2020)
rated equally comprehensible. Similarly, there cult, yet these expectations did not seem to lead
were no between-speaker differences for mea- them to assign harsher ratings to the presumed
sures of speaker status. For example, the ratings of Quebec French speaker, as both guises were rated
speaker intelligence were comparable across both as equally accented and equally easy to under-
guises because the majority gave a neutral rating stand. Of course, it is possible that our guise ma-
and responded that they could not accurately rate nipulation did not produce the intended effect,
either speaker based solely on a voice recording, such that learners did not believe that one speaker
while also suggesting that a speech variety is inde- was speaking Quebec French, which might have
pendent of a speaker’s intelligence. precluded them from expressing their attitudes.
Based on prior results by Hume et al. (1993), In addition, even though participants were as-
where undergraduate L2 French learners in On- sured that their responses would remain anony-
tario rated the European French teacher highest mous, any learners holding negative bias may have
on status traits and professional competence, we been unwilling to express their attitudes (Munro
also expected that the European French speaker et al., 2006). Nevertheless, there were many learn-
would be evaluated more positively regarding her ers who went along with the manipulation, down-
teaching competence and would be rated more fa- grading the presumed Quebec French speaker
vorably as a potential teacher, compared to the re- (see Figures 1 and 2), such that those who be-
actions to the presumed Quebec French speaker. lieved the false presentation of the speaker’s iden-
However, there was no difference in learners’ tity were the ones who expressed more negative
preference for having either speaker as a teacher, attitudes toward wanting to speak like her.
and one was not believed to be a better teacher A more likely explanation for the lack of strong
than the other. Instead, learners generally showed RLS effects in this study is that we targeted two
awareness of the importance of being exposed to native varieties, where ratings could depend on
different varieties, with 86 out of 106 responding listeners’ familiarity and experience with each
that having a combination of both French vari- variety. In contrast, in most prior RLS studies,
eties spoken would be ideal. listeners were led to believe that they were ex-
Whereas speech ratings (collected through a posed to L2 speakers (e.g., Hu & Su, 2015; Ru-
RLS procedure) yielded weak evidence for con- bin, 1992), which triggered their biases against
sistent bias effects, learners’ qualitative responses accented speech. Indeed, when it comes to ac-
suggested that many believed European French to cented speech, listeners can easily distinguish
carry more prestige than Quebec French. Outside native speakers from L2 speakers (e.g., Munro,
the rating task, learners often expressed explicit Derwing, & Burgess, 2010), but they may know
preference for European French or generally less about which speech patterns differentiate
associated it with positive descriptors, choosing two native varieties (e.g., Clark & Schleef, 2010)
negative adjectives for Quebec French over three and, more specifically, have little understanding
times more frequently on the posttask question- of what phonetic features characterize Quebec
naire than during the rating task. L2 learners French (Bergeron, 2019). It remains for future re-
referred to European French as being more search to establish whether and to what degree
standard, universal, familiar, and sophisticated, L2 learners can distinguish different varieties of
implying that they consider it the standard variety. French and to examine whether biases are am-
Similarly, they most frequently chose accented to plified or mitigated when learners evaluate actual
describe Quebec French (76 out of 106 partici- (rather than presumed) speakers of a particular
pants), which is a biased expectation because ac- variety.
centedness was assessed similarly in the two guises.
These attitudes toward Quebec French, expressed Proficiency, Social Networks, and Exposure
outside the rating task, indicated a prestige hi-
erarchy consistent with that held by participants If listeners’ attitudes depend on their familiar-
of Kircher’s (2012) study, who classified Quebec ity and experience with each speech variety, then
French as bizarre, anglicized, and accented. it is important to understand how listener back-
The conclusion that emerges from both sets of ground and experiential profiles contribute to
findings (rating task, qualitative responses) is that their ratings. Our second RQ therefore aimed to
learners held negative attitudes toward Quebec explore what background and experiential vari-
French but that these attitudes did not generally ables were associated with learners’ ratings of
influence their ratings in a RLS procedure. For the Quebec French guise, which represents a
example, the top two descriptors chosen by learn- novel contribution. Learners’ experience in in-
ers for Quebec French were accented and diffi- teractions involving Quebec French speakers was
Rachael Lindberg and Pavel Trofimovich 837
the composite variable associated with four of instead related more to the type and quality of
the six rated dimensions, such that having more the experience, as suggested by Kang & Rubin
positive experiences was linked to greater speaker (2009). The current results reflected this idea,
comprehensibility, stronger preference to sound such that L2 learners who had experienced more
like the speaker, greater perceived teacher com- positive interactions with Quebec French speak-
petence, and stronger desire for learners to have ers and felt more welcome in Quebec were more
the speaker as their teacher (see Table 5). Multi- likely to wish to sound like the presumed Que-
ple regressions further clarified that learners’ per- bec French speaker, rate her as comprehensible,
ception of the presumed Quebec French speaker perceive her as a good teacher, and want her as
as being more difficult to understand and as a their French teacher. In fact, positive experience
less desirable speech model could be explained in Quebec emerged as one of two factors uniquely
through background variables, where lower oral predicting L2 learners’ desire to sound like the
proficiency (for comprehensibility) and fewer speaker in the Quebec French guise.
positive experiences in Quebec and greater so- Besides positive experience in Quebec, learn-
cial network density (for desire to sound like ers’ social network density involving Quebec
the speaker) predicted negative ratings. These French speakers was also predictive of their desire
relationships held even after learners’ sensitivity to sound like the speaker in the Quebec French
to the guise manipulation had been accounted guise, such that denser social networks were as-
for. sociated with lower ratings. L2 speakers with so-
Considering that “attitudes specifically associ- cial network density scores in the top quartile pro-
ated with the group or the language are quite vided a rating of 62 (on a 100-point scale) for
probably dormant until the student is confronted wanting to speak like the Quebec guise, while
with learning the language” (Gardner, 1985, those with density scores in the bottom quar-
p. 8), we expected that lower level learners (e.g., tile assigned a rating of 77. Because denser net-
presumably those who had started learning the works allow for greater exposure to and famil-
language recently) would display more neutral iarity with Quebec French, this linguistic expe-
attitudes toward the Quebec French guise com- rience may have enabled learners to develop a
pared to higher level learners (e.g., those with more accurate perception of how Quebec French
a lengthier experience with language learning). differs from European French (e.g., Baker &
This expectation was also consistent with Lam- Smith, 2010; Bergeron, 2019), resulting in learn-
bert’s (1967) hypothesis that L2 learners’ atti- ers’ preference to not consider Quebec French
tudes can be influenced by language learning, as a desirable speech model. Consistent with this
as they begin to identify with the language and idea, learners with network density greater than 0
the speech community during this process. How- rated their familiarity with Quebec French higher
ever, there were no major associations between (52) than learners with low network density (35).
learners’ proficiency measures and their ratings. It is also possible that Quebec French speak-
In fact, as shown in Table 5, listening compre- ers themselves—known to hold negative views of
hension and speaking scores were linked to one their own variety (e.g., Kircher, 2012)—also ex-
of the six measures—speaker comprehensibility posed learners within dense social networks to
(with small-to-medium associations). Learners in negative attitudes toward Quebec French (e.g.,
the top quartile of the proficiency range (B1 and Baker, 2008), leading learners to solidify their
above) rated the speaker’s comprehensibility at stereotypes.
74 (on a 100-point scale) while those in the bot- Taken together, these findings begin to sup-
tom quartile (A1) rated her comprehensibility port the idea that developing quality contacts
at 46. More advanced L2 proficiency likely en- within a language community can have a measur-
abled learners to assess speaker comprehensibil- able social impact, which can determine how atti-
ity more accurately, in a manner that was not tudes are adopted (Cortes–Colomé et al., 2016).
colored by attitudes. Alternatively, learners’ own In some cases, quality contacts can foster posi-
ease with language (presumed at higher L2 lev- tive feelings, leading to positive attitudinal con-
els) helped them to assess the speaker as more sequences. However, other quality contacts—such
comprehensible. as those involving dense social networks within
Because these results generally do not align the target language variety—may result in some
with previous findings indicating more pro- attitudes becoming potentially entrenched, for
nounced attitudes for learners who gain greater example, through more intensive exposure ex-
amounts of experience with and linguistic knowl- periences. These findings also crucially support
edge of the target language (Giles et al., 1974, the idea that it is often the quality, not the quan-
1979), perhaps the development of attitudes is tity, of linguistic experiences that matters for L2
838 The Modern Language Journal 104 (2020)
development (e.g., Moyer, 2011), including the munity, it could be that their integrative motiva-
development and maintenance of attitudes. tion toward learning French played a role in how
they viewed Quebec French speakers. This was ev-
LIMITATIONS AND FUTURE WORK ident in our participants’ desire to hear different
French varieties in their language input. Conduct-
Although these findings are suggestive of the ing similar research in a less linguistically diverse
complex interplay of learners’ language attitudes, city, or with learners who have no intention of im-
their L2 proficiency, and their social experiences migrating to the local community, may reveal a
with the target language variety, these findings different attitudinal profile.
must be revisited in future research. For exam- Similarly, our lack of strong attitudinal differ-
ple, researchers could examine in more detail ences in response to two French varieties may be
the root causes of any negative feelings that attributed to issues of methodology. Unlike prior
learners adopt toward Quebec French during research, where images have often been used to
their language instruction and even prior to their establish a speaker’s identity (e.g., Rubin, 1992),
exposure to it, to see if language teachers could our procedure involved only providing partic-
address these issues in class. Perhaps learners ipants with a verbal description of a speaker’s
encounter difficulties and frustrations when com- origin, because one cannot unambiguously differ-
municating with Quebec French speakers, and entiate a French speaker from France from a Que-
such occurrences of communication breakdowns bec French speaker based on physical features
could be lessened through more practice and in- alone. Thus, a verbal label of a speaker’s origin
teraction with the local variety in a safe classroom may not have been sufficiently powerful to acti-
environment. Therefore, one possible implica- vate any potential attitudinal biases. Last but not
tion for L2 French instructors could be to expose least, because we tested most of this study’s partic-
French learners to diverse pedagogical models, ipants during their regular classroom instruction,
which would also familiarize learners with the we could not include many experimental features
linguistic reality of la Francophonie and potentially typical of the standard matched-guise procedure
contribute to learners having balanced attitudes (e.g., presentation of multiple files with distractor
toward different French varieties, speakers, and audios). Therefore, our findings must be revisited
cultures. Alternatively, perhaps learners are sen- in research employing stricter experimental con-
sitive to Quebec speakers’ own negative attitudes trols among other more focused methodological
toward their home variety (e.g., Kircher, 2012; tools, such as including an obligatory comment
Lambert et al., 1960), which would call for a section and conducting stimulated recall inter-
change in attitudes in Quebec French speakers views to assess participants’ sensitivity to the guise
themselves, before they can project positive feel- manipulation and its impact on their ratings.
ings about their language to others. Similarly, this
study has begun to identify L2 learners’ existing CONCLUSION
attitudes toward French varieties and to uncover
their possible experiential origins. However, what In summary, our findings shed light on the
remains to be investigated is how these attitudes attitudes that L2 French learners in Montreal
can affect learners’ L2 motivation and their learn- hold toward Quebec and European French. L2
ing outcomes. This work would be important to learners generally held some negative attitudes
establishing the role, if any, that attitudes toward toward Quebec French. However, apart from
language varieties play in L2 development. rating the presumed Quebec French speaker
The results of this study are also inextricably as the less preferred model to follow, they did
linked to the research context. For instance, it not seem to attribute their negative attitudes in
is possible that our participants’ attitudinal re- rating native French speakers while completing
actions are less pronounced than those docu- a realistic listening-comprehension activity. The
mented in prior work (Hu & Su, 2015, Rubin, most important finding was that feeling welcome
1992) due to the specific characteristics of our in Quebec and having positive interactions with
participants and our research setting. Perhaps Quebec French speakers, along with higher L2
Montreal’s culturally diverse environment lends speaking scores and greater social engagement
itself to a more linguistically aware population of with Quebec French speakers within dense social
learners who may not hold as strong language bi- networks, was the strongest predictor of learners’
ases as those in a more monolingual setting. Ad- attitudes toward what was presumed to be Quebec
ditionally, because many of our participants were French speech, revealing a strong experiential
immigrants or international students who arrived component to language attitudes. We call for
in Montreal with intentions of settling in the com- more research, particularly in multilingual and
Rachael Lindberg and Pavel Trofimovich 839
multicultural contexts, targeting the role of social parer au TEF. Accessed 16 October 2020
and experiential variables in the development at https://www.lefrancaisdesaffaires.fr/tests-
and maintenance of L2 learners’ attitudes toward diplomes/se-preparer/tutoriels-tef
native varieties of the target language, to gain a Chiba, R., Matsuura, H., & Yamamoto, A. (1995).
Japanese attitudes toward English accents. World
better understanding of how language attitudes
Englishes, 14, 77–86.
shape L2 learning.
Choy, S., & Dodd, D. (1976). Standard-English speaking
and non-standard Hawaiian-English-speaking chil-
dren: Comprehension of both dialects and teach-
Open Research Badges ers’ evaluations. Journal of Educational Psychology,
68, 184–193.
Clark, L., & Schleef, E. (2010). The acquisition of so-
ciolinguistic evaluations from Polish-born adoles-
This article has earned an Open Materials badge. cents learning English: Evidence from perception.
Open Materials are available at https://www.iris- Language Awareness, 19, 299–322.
database.org. Cortes–Colomé, M., Barrieras, M., & Comellas, P.
(2016). Changes in immigrant individuals’ lan-
guage attitudes through contact with Catalan: The
mirror effect. Language Awareness, 25, 272–289.
REFERENCES Dalton–Puffer, C., Kaltenboeck, G., & Smit, U. (1997).
Learner attitudes and L2 pronunciation in Aus-
Ahn, S. Y., & Kang, H. S. (2017). South Korean university tria. World Englishes, 16, 115–128.
students’ perceptions of different English varieties D’Anglejan, A., & Tucker, R. (1973). Sociolinguistics
and their contribution to the learning of English correlates of speech style in Quebec. In R. Shuy &
as a foreign language. Journal of Multilingual and R. Fasold (Eds.), Language attitudes, current trends
Multicultural Development, 38, 712–725. and prospects (pp. 1–27). Washington, DC: George-
Baker, W. (2008). Social, experiential, and psychologi- town University Press.
cal factors affecting L2 dialect acquisition. In M. Davila, A., Bohara, A., & Saenz, R. (1993). Accent penal-
Bowles, R. Foote, S. Perpiñán, & R. Bhatt (Eds.), ties and the earnings of Mexican Americans. Social
Selected Proceedings of the 2007 Second Language Re- Science Quarterly, 74, 902–916.
search Forum (pp. 187–198). Somerville, MA: Cas- Derwing, T. (2003). What do ESL students say about
cadilla Proceedings Project. their accents? The Canadian Modern Language Re-
Baker, W., & Smith, L. C. (2010). The impact of L2 view, 59, 547–567.
dialect on learning French vowels: Native En- Doucerain, M., Varnaamkhaasti, R., Segalowitz, N., & Ry-
glish speakers learning Québécois and European der, A. (2015). Second language social networks
French. The Canadian Modern Language Review, 66, and communication-related acculturative stress:
711–738. The role of interconnectedness. Frontiers in Psychol-
Bergeron, A. (2019). L2 French learners’ sociolinguis- ogy, 6, 1–12.
tic awareness of Quebec French (Unpublished doc- Dragojevic, M., & Giles, H. (2016). I don’t like you
toral dissertation). Montreal, Canada: Concordia because you’re hard to understand: The role
University. of processing fluency in the language attitudes
Bestelmeyer, P. E. G., Belin, P., & Ladd, D. R. (2015). process. Human Communication Research, 42, 396–
A neural marker for social bias toward in- group 420.
accents. Cerebral Cortex, 25, 3953–3961. Edwards, J. (1999). Refining our understanding of lan-
Boulé, J. J. (2002). Attitudes of young Quebecers towards En- guage attitudes. Journal of Language and Social Psy-
glish and French (Unpublished MA thesis). Mon- chology, 18, 101–110.
treal, Canada: Concordia University. Evans, B. E. (2002). Attitudes of Montreal students to-
Bourhis, R. Y. (1997). Language policies and language wards varieties of French. In D. Long & D. R.
attitudes: Le monde de la francophonie. In N. Preston (Eds.), Handbook of perceptual dialectology
Coupland & A. Jaworski (Eds.), Sociolinguistics: A (pp. 71–93). Philadelphia/Amsterdam: John Ben-
reader and coursebook (pp. 306–322). New York: St. jamins.
Martin’s Press. Field, A. (2009). Discovering statistics using SPSS (3th ed.).
Brown, B., Giles, H., & Thakerar, J. (1985). Speaker eval- Thousand Oaks, CA: SAGE.
uations as a function of speech rate, accent and Gardner, R. (1985). Social psychology and second language
context. Language and Communication, 5, 207–220. learning. London: Edward Arnold.
Carlson, H. K., & McHenry, M. A. (2006). Effect of ac- Giles, H., Bourhis, R., & Davies, A. (1979). Prestige
cent and dialect on employability. Journal of Em- speech styles: The imposed norms and inherent
ployment Counseling, 43, 70–83. value hypotheses. In W. McCormack & S. Wurm
Chambre de commerce et d’industrie Paris Ile- (Eds.), Language and society: Anthropological issues
de-France. (2018). Les tutoriels pour se pré- (pp. 589–596). New York: Mouton.
840 The Modern Language Journal 104 (2020)
Giles, H., Bourhis, R., Lewis, A., & Trudgill, P. (1974). national teaching assistants. TESOL Quarterly, 49,
The imposed norm hypothesis: A validation. Quar- 681–706.
terly Journal of Speech, 60, 405–410. Kang, O., Thomson, R., & Moran, M. (2019). The ef-
Giles, H., Wilson, P., & Conway, A. (1981). Accent and fects of international accents and shared first lan-
lexical diversity as determinants of impression for- guage on listening comprehension tests. TESOL
mation and employment selection. Language Sci- Quarterly, 53, 56–81.
ences, 3, 92–103. Kinzler, K. D., Dupoux, E., & Spelke, E. S. (2007). The
Gill, M. (1994). Accent and stereotypes: Their effect on native language of social cognition. PNAS: Proceed-
perceptions of teachers and lecture comprehen- ings of the National Academy of Sciences of the United
sion. Journal of Applied Communication Research, 22, States of America, 104, 12577–12580.
348–361. Kircher, R. (2012). How pluricentric is the French
Gluszek, A., & Dovidio, J. (2010). Speaking with a non- language? An investigation of attitudes towards
native accent: Perceptions of bias, communication Quebec French compared to European French.
difficulties, and belonging in the United States. Journal of French Language Studies, 22, 345–
Journal of Language and Social Psychology, 29, 224– 370.
234. Kircher, R. (2014). Thirty years after Bill 101: A contem-
Hay, J., Warren, P., & Drager, K. (2006). Factors in- porary perspective on attitudes towards English
fluencing speech perception in the context of a and French in Montreal. The Canadian Journal of
merger-in-progress. Journal of Phonetics, 34, 458– Applied Linguistics, 17, 20–50.
484. Lambert, W. E. (1967). A social psychology of bilingual-
Hopper, R., & Williams, F. (1973). Speech characteris- ism. Journal of Social Issues, 23, 91–109.
tics and employability. Speech Monographs, 46, 296– Lambert, W. E., Hodgson, R. C., Gardner, R. C., & Fillen-
302. baum, S. (1960). Evaluational reactions to spoken
Hu, G., & Lindemann, S. (2009). Stereotypes of Can- languages. The Journal of Abnormal and Social Psy-
tonese English, apparent native/non-native status, chology, 60, 44–51.
and their effect on non-native English speakers’ Laur, E. (2008). Contribution à l’étude des perceptions
perception. Journal of Multilingual and Multicul- linguistiques. La méthodologie des faux-couples
tural Development, 30, 253–269. revisitée. [Contribution to the study of lin-
Hu, G., & Su, J. (2015). The effect of native/non-native guistic perceptions. Methodology of the
information on non-native listeners’ comprehen- matched guise technique revisited]. Gou-
sion. Language Awareness, 24, 273–281. vernement du Québec. Accessed 16 October
Hume, E., Lepicq, D., & Bourhis, R. (1993). Attitudes 2020 at https://www.oqlf.gouv.qc.ca/ressources/
des étudiants Canadiens anglais face aux accents sociolinguistique/2008/notemetholaur.pdf
des professeurs de français en Ontario. [English Lev–Ari, S., & Keysar, B. (2010). Why don’t we believe
Canadian students’ attitudes towards the accents non-native speakers? The influence of accent on
of French professors in Ontario]. The Canadian credibility. Journal of Experimental Social Psychology,
Modern Language Review, 49, 209–235. 46, 1093–1096.
Hutcheson, G. D., & Sofroniou, N. (1999). The multivari- Lindemann, S. (2002). Listening with an attitude: A
ate social scientist: Introductory statistics using general- model of native-speaker comprehension of non-
ized linear models. Thousand Oaks, CA: SAGE. native speakers in the United States. Language in
Hyman, H. K. (2002). Foreign accented adult ESL learn- Society, 31, 419–441.
ers: Perceptions of their accent changes and employabil- McGowan, K. B. (2015). Social expectation improves
ity qualifications (Doctoral dissertation). Accessed speech perception in noise. Language and Speech,
5 October 2020 at ProQuest Dissertations & The- 58, 502–521.
ses (Accession number 200210856). McKenzie, R. (2008). Social factors and non-native at-
Jarvella, R. J., Bang, E., Jakobsen, A. L., & Mees, I. M. titudes towards varieties of spoken English: A
(2001). Of mouths and men: Nonnative listeners’ Japanese case study. International Journal of Applied
identification and evaluation of varieties of En- Linguistics, 18, 63–88.
glish. International Journal of Applied Linguistics, 11, Moyer, A. (2011). An investigation of experience in L2
37–56. phonology: Does quality matter more than quan-
Kang, O., & Rubin, D. L. (2009). Reverse linguistic tity? The Canadian Modern Language Review, 67,
stereotyping: Measuring the effect of listener ex- 191–216.
pectations on speech evaluation. Journal of Lan- Munro, M. J. (2003). A primer on accent discrimination
guage and Social Psychology, 28, 441–456. in the Canadian context. TESL Canada Journal, 20,
Kang, O., & Rubin, D. L. (2012). Intergroup contact 38–51.
exercises as a tool for mitigating undergradu- Munro, M. J., Derwing, T. M., & Burgess, C. S.
ates’ attitudes toward nonnative English-speaking (2010). Detection of nonnative speaker status
teaching assistants. Journal on Excellence in College from content-masked speech. Speech Communica-
Teaching, 23, 159–166. tion, 52, 626–637.
Kang, O., Rubin, D., & Lindemann, S. (2015). Mitigat- Munro, M. J., Derwing, T. M., & Sato, K. (2006). Salient
ing U.S. undergraduates’ attitudes toward inter- accents, covert attitudes: Consciousness-raising for
Rachael Lindberg and Pavel Trofimovich 841
pre-service second language teachers. Prospect, 21, at https://www.statcan.gc.ca/eng/dai/smr08/
67–79. 2014/smr08_190_2014#a4
Niedzielski, N. (1999). The effect of social information Strand, E. (1999). Uncovering the role of gender stereo-
on the perception of sociolinguistic variables. Jour- types in speech perception. Journal of Language and
nal of Language and Social Psychology, 18, 62–85. Social Psychology, 18, 86–100.
Piechowiak, A. (2009). What is “good” quality oral Weener, P. D. (1967). The influence of dialect differences on
French? Language attitudes towards “differently” ac- the immediate recall of verbal messages (Unpublished
cented French in Quebec (Unpublished doctoral dis- doctoral dissertation). Ann Arbor, MI: University
sertation). Montreal, Canada: McGill University. of Michigan. Accessed 5 October 2020 at the Edu-
Plonsky, L., & Oswald, F. L. (2014). How big is “big”? cation Resources Information Center (ERIC Num-
Interpreting effect sizes in L2 research. Language ber ED017901).
Learning, 64, 878–912. Williams, F. (1976). Explorations of the linguistic attitudes of
Rubin, D. L. (1992). Nonlanguage factors affecting un- teachers. New York: Newbury House.
dergraduates’ judgments of nonnative English- Zhang, Q. (2013). The attitudes of Hong Kong stu-
speaking teaching assistants. Research in Higher Ed- dents towards Hong Kong English and Mandarin-
ucation, 33, 511–531. accented English. English Today, 29, 9–16.
Ryan, E. B., & Sebastian, R. J. (1980). The effects of Zhang, W., & Hu, G. (2008). Second language learn-
speech style and social class background on social ers’ attitudes towards English varieties. Language
judgments of speakers. British Journal of Social and Awareness, 17, 342–347.
Clinical Psychology, 19, 229–233.
Seligman, C., Tucker, G. R., & Lambert, W. E. (1972).
The effects of speech style and other attributes on SUPPORTING INFORMATION
teachers’ attitudes towards pupils. Language in So-
ciety, 1, 131–142. Additional supporting information may be found
Statistics Canada. (2017, September 15). Back to online in the Supporting Information section at
school… by the numbers. Accessed 2 November 2018 the end of the article.