When Accent Does Not Match Expectations: A Dynamic Perspective of L2 Speaker Evaluations in a French Interview Context

According to expectation violation theory, job applicants can be upgraded or downgraded during an interview when their accent does not match employers’ speech expectations. Focusing on the employment of second language French job candidates in Québec, this study explored this issue dynamically in terms of how expectations may impact the trajectory of interview evaluations. Participants included 60 Québec French raters and 6 female job candidates applying to a waitress or pizza cook position, presented through their resumes as either first (L1) or second (L2) language French speakers. Each speaker’s interview audios were presented to raters in expectancy-congruent and expectancy-incongruent scenarios. Raters first provided resume-based employability assessments, then two more evaluations throughout a typical sequence of interview questions. The congruent and incongruent scenarios revealed similar evaluation patterns, where the L2 French cook applicant’s employability improved after initially being downgraded. Implications are discussed regarding listeners’ readjustment of their perceptions following first-impression biases.


When Accent Does Not Match Expectations: A Dynamic Perspective of L2 Speaker Evaluations in a French Interview Context
When job applicants participate in an interview, there are many social, linguistic, and affective variables which can influence their perceived competence, subsequently impacting hiring decisions.For example, the presence of a nonstandard, regional, or second language (L2) accent can activate stereotypes, informing judgements of one's personality and credibility (Dixon & Mahoney, 2004;Lambert et al., 1960), leading to lower employability evaluations (Rakić et al., 2011;Spence et al., 2022).However, employers also frequently have access to background information about an applicant before the interview, leading to the formation of certain expectations, and these expectations may have evaluative consequences if the applicant's speech during the interview is incongruent with them.This study focused on the phenomenon of expectancy violations (i.e., when there is a mismatch between stereotype-based social expectations and reality, such as between how a listener imagines a job candidate to look and sound and how the candidate actually looks and sounds) by examining whether cues to L2-speaking applicants' accent provided through their resume elicit expectations that impact subsequent employability evaluations of those candidates in the context of a mock job interview.

Background Literature
As widely studied in research about language attitudes, individuals tend to place others into social categories based on various biases or stereotypes, which impacts their impressions of these people (Dragojevic et al., 2018(Dragojevic et al., , 2020;;Ryan, 1983).For example, a listener might attribute various regional, ethnic, or linguistic variations in a speaker's speech to a social group (e.g., defined through ethnicity, social class, or geography), using this categorization to ascribe stereotypic traits to the speaker, such as a low level of education or an unfriendly personality (Kinzler et al., 2010;Rakić et al., 2011).This can then have far-reaching implications in various social situations, including employment interviews (Kalin, 1982;Matsuda, 1991;Munro, 2003;Purnell et al., 1999).In fact, it is well documented that a speaker's accent-or the degree to which speech is colored by various ethnic, regional, or L2 features-can lead to job applicants being evaluated less favorably (Brennan & Brennan, 1981;Ryan et al., 1977;Spence et al., 2022), with L2 speakers especially vulnerable to accent bias (Carlson & McHenry, 2006;Hopper & Williams, 1973;Rakić et al., 2011).For instance, in a recent study, applicants who spoke English with a strong L2 accent during a mock job interview received the lowest employability ratings, even though their English skills were irrelevant to the position, which was to teach German (Roessel et al., 2019).
Although revealing, existing research focusing on accent bias in employment contexts tends to only elicit applicant evaluations after the interview (Carlson & McHenry, 2006;Rakić et al., 2011;Roessel et al., 2019), which does not consider employers' expectations prior to the interview.Indeed, employers are often presented with background information about the applicants before meeting them, for example, through resumes summarizing their education, language background, or job history (Dipboye et al., 1984), which likely triggers various expectations (Brown & Campion, 1994;Derous et al., 2009Derous et al., , 2014Derous et al., , 2017)).According to expectation violation theory (Burgoon, 1993;Burgoon & Jones, 1976), if the speaker's speech turns out to be more socially desirable than expected, Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 listeners will provide more positive evaluations than if that speaker had matched their expectations.Alternatively, a negative violation occurs when the speaker turns out to be less socially desirable than expected, leading to lower evaluations compared to a congruent situation.This perspective thus predicts evaluative consequences for job applicants if their speech is inconsistent with the expectations created before the interview.
To examine the expectation violation perspective in an employment context, Hansen et al. (2018) used photos of job applicants and short interview greetings, such as "Good morning, nice to meet you!" recorded by first language (L1) speakers of German and by Turkish speakers of L2 German (representing a stigmatized variety).The listeners who first evaluated a candidate's employability based solely on their Western appearance (after inspecting a photo) were found to significantly downgrade their ratings once they subsequently heard the candidate speaking with a Turkish accent.The listeners thus appeared to be negatively "surprised" by the speaker's speech, which they expected (based on visual information alone) to sound nativelike or standard, leading to harsher evaluations for expectation-incongruent individuals compared to those whose speech was consistent with the listeners' expectation (see also Aboud et al., 1974;Jussim et al., 1987).
Although these findings suggest that L2 speakers are likely to be perceived as less employable than L1 speakers or standard-accented speakers (i.e., those perceived to speak a more prestigious, universal, or sophisticated L1 variety), with listener expectations influencing employment assessments, several questions remain unanswered.First, it is unclear whether information other than an applicant's photo might elicit expectations.For instance, applicants' name and background information, such as their place of origin or study, are typically available through their resumes and are normally evaluated before the interview (Cotton et al., 2008;Derous et al., 2014;Dipboye et al., 1984;Oreopoulos & Dechief, 2012).Therefore, to understand how impressions emerge in an interview setting, it is essential to investigate accent bias by capturing expectations triggered during the resumescreening stage.Second, in previous work on listener bias in employment contexts, listeners were typically presented with only a single, brief audio (e.g., 3-20 seconds in duration) per speaker (Dragojevic et al., 2019;Hansen et al., 2018), which makes it unclear whether listener impressions of speakers change over time, as more speech by a candidate is heard.Finally, while nonstandard-accented speakers are often judged more suitable for low-than high-status jobs (Giles et al., 1981;Singer, 1988), it is unclear if they would be evaluated differently for positions where communication skills are more or less critical.For instance, accented applicants might be especially disadvantaged when the position requires communication and customer contact (Derous et al., 2014(Derous et al., , 2017;;Hansen & Dovidio, 2016), given that accented employees tend to receive lower customer satisfaction ratings than L1or standard-accented workers (Wang et al., 2013).Thus, because an L2 applicant's employability might depend on the job to be performed, it is important to consider the type of job when examining employment decisions.

The Current Study
The goal of this study, which focused on French, was therefore to address these issues by examining whether cues to an L2 applicant's accent provided through a resume elicit expectations that might then impact listener judgements of that applicant's employability.First, the study design emulated a typical interview process by including the resume-screening stage where listener expectations were manipulated through resumes.For example, the same L2-speaking candidate was sometimes presented through a resume for evaluation in an expectation-congruent scenario as an L2 French speaker (e.g., as María Ramos Vásquez from Lima, Perú) and sometimes in an expectation-incongruent scenario as an L1 French speaker (e.g., as Laura Morin from Sherbrooke, Québec), such that the identity created through the resume either matched or mismatched the applicant's accent.While theoretically a resume depicting an L2 speaker could be paired with the speech of an L1 French speaker (i.e., in another mismatch scenario), this condition was not included in this study, which focuses specifically on L2 French speakers' oral interview performances.Second, the L2 applicants were evaluated through multiple (two) and longer (40-second) audio interview excerpts, which allowed for a time-sensitive view of how initial expectations (developed through resumes) might evolve throughout the interview.Third, the applicants' employability was assessed in response to interviews for two entry-level positions (pizza cook vs. waitress) to determine the role of expectations for jobs which differ in the amount of communication required.Finally, even though L1 speakers' interview performances were not the target of this investigation, to provide a comparative baseline for interpreting listener reactions to L2 speech, the L2 applicants were broadly compared in their employability ratings with L1 French applicants (speakers of Québec and European French), reflecting the local sociolinguistic context of the study (Montréal, Québec), where French is the only official, majority language.Including both L1 varieties also provided speech samples representative of stereotypically-viewed standard (European French) and nonstandard (Québec French) varieties (e.g., Kircher, 2012;Lindberg & Trofimovich, 2020).
The study addressed the following research question: How do listener-based employability evaluations of L2 French candidates applying for an entry-level position (pizza cook vs. waitress) change from their resume-based impression and throughout the interview when accent expectations are violated compared to when they are confirmed?Consistent with previous work, it was expected that listeners would downgrade the L2 applicant in their employability ratings if this applicant was presented through her resume as an L1 French speaker (i.e., in a negative violation scenario), compared to when the same applicant was presented as an L2 speaker.Because there is no systematic work investigating long-lasting consequences of violated expectations, there was no prediction regarding whether the effects of violated expectations would persist beyond an initial evaluation.Similarly, given the lack of research comparing hiring decisions for jobs with high versus low communicative demands, it was generally predicted that the ratings might be harsher for the position that requires an applicant to communicate in their L2 (waitress) compared to the one requiring little communication (pizza cook).

Speakers as Job Applicants
Six female French speakers were recruited to assume the role of potential job applicants who would engage in a mock job interview.The target job applicants were two L2 French speakers who spoke Spanish as their mother tongue (ages 22 and 30, born in Lima, Perú), as Spanish is one of the most common mother tongues in Montréal (13%) after French and English (Statistics Canada, 2017).The additional four job applicants, recruited to provide L1-speaker baseline data and to serve as distractors for the target Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 audios, were two speakers of Québec French (ages 25 and 28, born in Montréal, Québec) and two speakers of European French (ages 20 and 22, born in Bordeaux and Lille, France).Speakers from both Québec and France were recruited to represent multiple L1 French varieties that employers would likely encounter in interview situations in Montréal.While the Québec French speakers had resided in Québec their entire life, the speakers from France and Perú had been living in Québec for 2.5-5 years.All speakers were enrolled in university-level bachelor (4), masters (1), or doctoral (1) programs.At the time of the study, the L2 French speakers, who had both been studying French for 3 years, were enrolled in an intermediate level 6-7 French course based on the Canadian Language Benchmarks (CLB) and had self-rated their French speaking skills at 57 and 49 on a 100point scale, where 100 corresponded to nativelike.The CLB 6-7 level of the French course attended by the L2 speakers exceeds the minimum CLB level 5 required by the Government of Canada's National Occupation Classification for a cook and a food and beverage server (https://noc.esdc.gc.ca).

Raters as Interviewers
A total of 60 L1 speakers of Québec French (52 females, 8 males) participated as raters, assuming the role of an interviewer.They were on average 24.20 years old (SD = 4.45, and were all born in Québec, having resided in the province most of their life (M = 23.97years, SD = 4.33, range = 19-39).Raters were recruited through announcements posted on various social media groups for universities throughout Québec.They were therefore all students who were studying in either bachelor (36), masters (14), or doctoral (10) programs, and for 53 of them, their university studies were in French.They reported using French for approximately 90.13% of their daily interactions (SD = 12.37%), with 55 using French regularly in the workplace.On a 100-point scale, they reported being moderately familiar with French spoken by L2 speakers (M = 72.23,SD = 25.29,range = 11-100), where 100 corresponded to high familiarity.

Materials
The materials included two job advertisements for entry-level positions, 12 job applicant resumes, rating scales, recordings of two equivalent interviews from the six speakers, and a background questionnaire.

Job Application Materials and Rating Scale
First, two authentic job advertisements, inspired by real job postings, were created for a pizza cook and a waitress position at Montréal restaurants (see Appendix A).Both advertisements included a job description, followed by a list of the main duties and desired skills and qualifications, where communication skills were only emphasized for the waitress position.These positions were chosen to represent jobs which had similar status but differed in the amount of communication required (low vs. high).To ensure that this was indeed the case, both postings were pre-rated by 14 graduate students (8 females, 6 males) in a social science program for job status and importance of communication skills using a 100-point scale, where higher ratings indicated higher status and greater importance of oral communication.Whereas job status was rated similarly low for both the waitress (M Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 = 40.64,SD = 18.15) and the pizza cook (M = 36.18,SD = 12.83), they differed in the importance of communication, where communication skills were deemed very important for the waitress (M = 85.27,SD = 13.98) but less important for the pizza cook (M = 37.81, SD = 22.28).
Second, 12 resumes were created, with six representing candidates applying for the pizza cook job and six corresponding to the waitress position.All resumes were laid out similarly and included one year of relevant job experience (i.e., either one year working as a pizza cook or a waitress), with four bullets describing the tasks performed previously and a completed bachelor's degree in social sciences or the humanities.To manipulate raters' expectations of the speaker's origin, three differences were introduced into the target speakers' resumes: (a) the applicant names, (b) the restaurant and city where they previously worked, and (c) the university and city where they had studied.For example, a resume for an expected L1 Québec French speaker included a common Québécois name, such as Jeanne Tremblay, and described that she had worked and studied in Montréal, Québec, whereas a resume for María Ramos Vásquez, who had worked and studied in Lima, Perú, represented a presumed L2 French applicant (see Appendix B for copies of the target L2 speakers' resumes for both congruent and incongruent scenarios).
Finally, to accompany the resumes and interview responses, raters were presented with a 100-point sliding scale, which was adapted from prior work on language attitudes in employment contexts (e.g., Cargile, 2000).The scale targeted raters' assessment of the applicant's employability, asking if they would hire that person for the advertised position, with anchor points labeled "never" (corresponding to the rating of 0) on the left and "certainly" (corresponding to the rating of 100) on the right.

Mock Job Interview Recordings
Three speakers (one L2 French speaker, one L1 Québec French speaker, one L1 European French speaker) were randomly chosen to participate in mock job interviews for the pizza cook position, while the other three (one L2 French speaker, one L1 Québec French speaker, one L1 European French speaker) were interviewed for the waitress position, with all interviews conducted in French by the first author over Zoom.Put differently, each speaker participated in a mock job interview for only one position, where their responses were specific to that given job.Prior to the interview, the speakers inspected the relevant job ad and their mock resume (with one year of relevant job experience and a bachelor's degree), which they could reference in their responses if desired.During the interviews, each speaker responded spontaneously (i.e., in an unscripted manner) to two sets of matched questions, recording a total of four target audios, which yielded similar yet content-different responses per speaker.One set included responses to the following two matched questions (designated as the first interview response): (a) Tell me about yourself: What makes you the ideal person for this job?And (b) What makes you stand out among other candidates?The other set included answers to these two questions (designated as the second interview response): (a) How do you respond to criticism?And (b) How do you deal with conflicts?
The speakers' responses to the four interview questions were subsequently saved as individual audio files (Mlength = 41.03 seconds, SD = 1.82, range = 38-45), which resulted in 24 recordings (four audios per speaker).To ensure that the audios recorded by the target L2 French speakers were comparable but different from those recorded by the L1 speakers, Applied Linguistics: 26, 2 (2023): 1-31 raters provided three additional evaluations of all speakers outside the main experimental task (i.e., outside the rating of applicants' suitability for the advertised positions) using an additional response recorded by all speakers during their Zoom interviews (What do you think are the qualities required to be successful in this position?).Using a 100-point sliding scale (where 0 indicated beginner French and 100 represented nativelike speech), raters correctly identified all four L1 French speakers as such (M = 96.12,SD = 8.25), and the two L2 speakers were assessed approximately at an intermediate proficiency level, where one was rated at a mean of 52.60 (SD = 20.81)while the other at a mean of 49.65 (SD = 21.21).Regarding the L2 speakers' comprehensibility (where 0 indicated that they were hard to understand and 100 meant that they were easy to understand), one L2 speaker elicited a mean rating of 67.45 (SD = 25.88) while the other obtained a mean rating of 62.62 (SD = 27.18),suggesting that they were similarly comprehensible.In terms of rater perceptions of the L2 speakers' accentedness (where 0 meant no L2 accent and 100 corresponded to a heavy accent), one L2 speaker was rated at a mean of 84.37 (SD = 17.51) while the other was assessed at a mean of 80.24 (SD = 19.25),suggesting that both had a moderately heavy accent.These additional speaker evaluations ensured that the two target L2 speakers were comparable in their L2 performance.

Canadian Journal of
Regarding the interview content, the L2 speaker interviewing for the waitress job produced responses that were on average 73.75 words long (SD = 12.34), which was similar to the mean length of 78.50 words (SD = 9.76) of the interview responses by the L2 speaker applying for the cook position; the average response length for the L1 speakers was 106.13 words (SD = 15.56).To broadly estimate the quality of each candidate's response, we recruited 20 additional external L1 French raters (16 females, 4 males), with a mean age of 23.40 years (SD = 3.01, range = 19-29), for a post hoc analysis (i.e., after the main data collection was completed) so that they could provide judgments of written transcripts of the interview responses (i.e., without being influenced by each speaker's accent).Using a 100point scale, these raters evaluated the quality of the two L2 speakers' linguistic expression similarly at 64.59 (SD = 20.02) and 63.35 (SD = 18.71), which was also comparable to the L1 speakers' average of 63.49 (SD = 22.55),where 100 indicated high quality.Regarding the content of the interview responses, the raters evaluated the ideas and arguments expressed by the two L2 speakers similarly at 68.10 (SD = 21.97) and 62.64 (SD = 19.33),which was also comparable to the content provided by the L1 speakers (M = 66.84,SD = 23.44),where 100 indicated high quality.These post hoc analyses of interview response quality suggested that all speakers were comparable in the sophistication of their responses (see Appendix C for sample transcripts of speakers' interview responses).

Procedure
The entire 45-minute procedure was administered online using French-medium LimeSurvey (https://www.limesurvey.org),which the 60 raters completed independently.After agreeing to the conditions outlined in the consent form (2 minutes), raters were presented with instructions for the main rating task, which informed them that they would be evaluating 12 different applicants for a pizza cook or a waitress position based on their resumes and two interview responses which would play automatically only once.To contextualize the task within a realistic employment situation, raters were told that they were a valued employee invited to an interview panel for the restaurant where they worked to help make hiring decisions.To familiarize raters with the procedure and the rating scales Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 that followed each resume and audio recording, they were asked to complete a practice task (3 minutes) in which they examined a Barista job ad, an example resume, and a sample of an interview response (not used in the main task) shown in writing (to not expose raters to any speaker's speech before the main rating task), accompanied by the employability rating scale.
For the main rating task (30 minutes), raters evaluated L2 job applicants' materials, where each of the six speakers was presented as if they were two different applicants for the same job (with six applicants for the pizza cook job and six applicants for the waitress job), such that one time the speaker was presented in an expectation-congruent scenario while another time the same speaker was presented in an expectation-incongruent scenario.In the expectation-incongruent scenarios for the L1 speakers, the candidates were presented through a resume representing the other L1 variety (e.g., a resume depicting a European French speaker, followed by the Québec French speaker's interview responses).Of key importance, however, is the expectation manipulation for the two L2 French speakers, whereby each was presented, through her resume, as either a presumed L2 French speaker (e.g., María Ramos Vásquez from Lima, Perú) or an L1 French candidate (e.g., Laura Morin from Sherbrooke, Québec).Even though each speaker was presented twice as an applicant for the same position (i.e., in either an L2-speaker guise or an L1-speaker guise), the content of her interview was different, such that one pair of interview responses was presented in an expectation-congruent scenario, while the other pair of matched yet content-different responses was assigned to an expectation-incongruent scenario (or vice versa).Each applicant's employability was rated on three separate occasions using the same employability scale.Raters were first presented with the job ad and evaluated the applicant based solely on her resume, to capture their initial expectations.Raters then evaluated that applicant after hearing her first interview response (i.e., in response to either What makes you the ideal person for this job?or What makes you stand out among other candidates?), and a third time after hearing her second interview response (i.e., responding to either How do you respond to criticism?or How do you deal with conflicts?).
Each rater was randomly assigned to one of four counterbalanced lists, with 15 raters per list, where the applicants for the pizza cook and the waitress jobs were presented together in a random order, with the constraint that there were always at least four other speakers before a given speaker's voice would be heard again in the opposite scenario.It was necessary to present the pizza cook and waitress applicants in the same block (rather than separately) to have a sufficient variety of voices before the same speaker was presented a second time, which was confirmed by a few pilot raters who were unaware that they heard the same speakers twice and had no difficulty evaluating candidates for two separate jobs.Across the four lists, each pair of interview responses (pairs a and b) were presented an equal number of times in both congruent and incongruent scenarios, and the incongruent scenario occurred an equal number of times before and after the relevant congruent scenario per speaker, to minimize potential order effects impacting raters' judgements through comparisons to previous responses.Although raters heard each speaker's voice twice (once in each expectation scenario), the likelihood of raters recognizing the speakers' voices was low, because the speakers were introduced through different resumes and presented in a semi-random order (with sufficient filler audios separating the speaker's recordings), and because the audios recorded by the same speaker featured content-different responses to matched questions.Upon completion of the testing, Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 no rater provided comments about their familiarity with or otherwise any awareness of repeated voices.
At the end of the session, raters engaged in a separate brief rating task (6 minutes), providing speech evaluations of the six speakers, assessing each for French speaking proficiency, comprehensibility, and accentedness (reported above) without seeing resumes and without expectancy manipulation in response to an additional question (What do you think are the qualities required to be successful in this position?).Lastly, raters filled out the background questionnaire (3 minutes).

Data Analysis
Employability ratings (out of 100) were exported into a spreadsheet, and the ratings were first checked for internal consistency using two-way, consistency, average-measure intraclass correlations.For the target L2 speaker evaluations, these values ranged between .73 and .89across the congruent and incongruent scenarios at the three evaluation timepoints (resume, first interview response, second interview response), whereas for the L1-speaker (baseline) evaluations, these values ranged between .70 and .91,corresponding to acceptable internal consistency levels (Larson-Hall, 2016).To examine the role of rater expectations in their evaluations of L2 job applicants for the pizza cook (low status, low communication requirement) or the waitress (low status, high communication requirement) position across different timepoints in the interview process, the employability ratings assigned to the two L2 speakers were compared through two-way repeated-measures ANOVAs, with expectation (congruent, incongruent) and time (resume, first interview response, second interview response) as within-participant variables.Because the target position (pizza cook vs. waitress) was a variable manipulated across different speakers, which precluded direct comparisons between them (i.e., because the speakers were different individuals who engaged in interviews in different, person-specific styles), the analyses of rater evaluations for the pizza cook and the waitress positions were carried out separately.The L1 speakers' evaluations were used only for baseline comparison purposes, so as not to detract from the study's focus on L2-speaking applicants.Even though direct comparisons of rater preferences for L1-versus L2-speaking applicants was outside this study's scope, the data for the two L1 French applicants for each advertised position were pooled together across both incongruent and congruent scenarios to derive a global baseline of the rated employability of L1 French applicants, separately for each position.These L1 speaker data were broadly compared with the L2 applicants' ratings through comparisons of 95% confidence intervals (CIs), which is a preferred, visual way of conducting between-group comparisons (Cumming & Calin-Jageman, 2016).Prior to running parametric analyses, all ANOVA assumptions were checked.In the few cases when the assumption of sphericity (Mauchly's p < .05)was violated, Huynh-Feldt values were reported, as recommended by Field (2018), and all pairwise comparisons were carried out using Bonferroni adjustments, with alpha for omnibus analyses set at .05.Effect sizes for pairwise comparisons were interpreted based on field-specific guidelines (Plonsky & Oswald, 2014), using Cohen's d for repeated-measures designs (0.60, 1.00, 1.40), where each value designates small, medium, and large effects, respectively.

Employability of the L2 French Cook
Raters' evaluations of the cook applicant (summarized in Table 1 and illustrated in Figure 1) were submitted to a two-way (expectation × time) repeated-measures ANOVA.This analysis revealed no statistically significant main effect for expectation, F(1, 59) = 0.50, p = .481,ηp 2 = .008,but a significant main effect for time, F(1.81, 106.77) = 16.52,p < .001,ηp 2 = .219,and a significant two-way interaction, F(1.85, 109.31)= 3.32, p = .043,ηp 2 = .053,which was explored through tests of interaction effects.In terms of the expectation manipulation, as shown in Figure 1, there was an early rating difference between the two conditions, where the applicant whose resume implied that she was an L1 French speaker received higher ratings than the same applicant portrayed as an L2 speaker, Mdiff = -5.25,t(59) = -3.06,p = .003,95% CI = [-8.68,-1.82], d = 0.40 (corresponding to a small effect), with no difference detected between the two conditions after the first and the second interview response (p > .645).As far as the applicant's evaluations across different stages of the interview, the ratings declined after raters heard the first interview response, compared to their initial ratings based off a resume, Mdiff = 10.21,t(59) = 4.70, p < .001,95% CI = [5.86,14.55], d = 0.61 (again with a small effect), but returned to the initial level after the applicant's second response, such that the final ratings did not differ from the initial ones, Mdiff = 1.60, t(59) = 0.82, p = .417,95% CI = [-2.31,5.51], d = 0.11.In fact, as shown visually in Figure 1, the evaluations of the L2 speaker did not differ from the evaluations of the L1 applicants both early (after raters inspected the resumes) and late (after the second interview response) in the interview Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 sequence, with the 95% CI for the ratings of the L1 speakers subsumed within assessments of the L2 speaker.

Figure 1 Mean Employability Ratings (with Brackets Enclosing 95% CIs) for the Cook Candidate
Note.The applicant's resume either implied that she was an L2 speaker (congruent condition) or that she was an L1 speaker (incongruent condition).Baseline evaluations for the L1 French-speaking applicants, with the 95% CI shaded in light gray, are provided for comparison purposes.
To further consider how raters reacted after first hearing the L2 speaker, pairedsamples t tests using difference scores between the resume and first-response ratings were conducted to compare the change in ratings following the resume-based impression between each condition.As before, these analyses were conducted only for the L2-speaking applicant rather than the L1 speakers, in keeping with this study's focus on L2 job candidates.There was a significantly more pronounced drop in ratings in the incongruent scenario, where evaluations dropped by 13.57 points on average (SD = 20.53)after hearing the first interview response, compared to the congruent scenario, where evaluations only dropped by 6.85 points on average (SD = 20.82),t(59) = 2.17, p = .034,95% CI = [0.51,12.92], d = 0.28.Therefore, it appears that raters were more negatively surprised after hearing the L2 cook when she was expected to be an L1 speaker.

Employability of the L2 French Waitress
Raters' evaluations of the waitress applicant (summarized in Table 1 and illustrated in Figure 2) were submitted to a similar two-way (expectation × time) repeated-measures ANOVA.This analysis revealed only a statistically significant main effect for time, F(1.69, 99.80) = 27.22,p < .001,ηp 2 = .316,but no main effect for expectation, F(1, 59) = 0.28, p = .601,ηp 2 = .005,and no interaction effect, F(2, 118) = 2.65, p = .075,ηp 2 = .043.The significant main effect for time was explored further through tests of simple main effects.Regardless of the expectation condition in which the applicant was presented, as shown in Figure 2, raters' evaluations were significantly lower after her first interview response, Mdiff = 11.91,t(59) = 5.79, p < .001,95% CI = [7.79,16.02], d = 0.75, and after her second interview response, Mdiff = 13.52,t(59) = 5.77, p < .001,95% CI = [8.83,18.20], d = 0.74 (both with small-to-medium effects), compared to their initial evaluations based off a resume.As illustrated in Figure 2, raters always evaluated the L2 speaker's interview responses lower than the responses of the L1 French-speaking candidates, such that the 95% CI for the ratings of the L1 speakers did not overlap with the assessments of the L2 speaker outside the evaluations based off the applicant's resume.

Figure 2 Mean Employability Ratings (with Brackets Enclosing 95% CIs) for the Waitress Candidate
Note.The applicant's resume either implied that she was an L2 speaker (congruent condition) or that she was an L1 speaker (incongruent condition).Baseline evaluations for L1 French-speaking applicants, with the 95% CI shaded in light gray, are provided for comparison purposes.To capture if evaluations dropped more severely in one condition over the other, paired-samples t tests comparing difference scores between the resume and first-response ratings for the L2 job applicant showed that ratings dropped by an average of 14.53 points in the incongruent scenario (SD = 22.15) and by 9.28 points on average in the congruent scenario (SD = 17.42), but this difference between conditions was not significant, t(59) = 1.70, p = .095,95% CI = [-0.94, 11.44], d = 0.22.In short, it appeared that listeners were equally negatively surprised by the L2 waitress after her first interview response, whether she was originally expected to be an L1 or L2 speaker.Again, considering this study's focus on L2 speakers, these analyses were conducted only for the L2-speaking applicant.

Discussion
The goal of this study was to examine if the employability of an L2 French speaker would be evaluated differently over the course of a mock job interview if the speaker displayed speech that either matched or mismatched listeners' expectations.The expectations prompted through the applicant's resume generally produced little effect on the employability ratings, where there was only an initial statistically significant rating difference in favor of the presumed L1 speaker and only for the cook position, with a small effect (cf. resume ratings in Figures 1 and 2).Besides this initial difference, the expectation manipulation appeared to have little impact on how the L2 applicants' actual interview responses were evaluated, despite the L2 cook's evaluations decreasing more drastically in the incongruent scenario compared to the congruent scenario.In terms of the time course of the ratings, the L2 applicants were downgraded in their employability, relative to their initial resume-based evaluation, with a small-to-medium effect.However, only the applicant for the cook position, not the waitress position, received a final evaluation that was comparable to the listeners' initial assessment and in fact similar to their rating of L1 French candidates.

No Effect of Violated Expectations
First and foremost, this study's findings generally revealed no appreciable differences in employability ratings as a function of matched versus mismatched expectations, which is contrary to prior work where L2 speakers were evaluated more negatively in situations where listener expectations were violated than when they were upheld (Hansen et al., 2017a(Hansen et al., , 2018(Hansen et al., , 2017b)).While the listeners in this study provided similar ratings to the same L2 candidates when their speech both matched and mismatched how they were introduced through their resume, listeners' "surprised" reaction after hearing the speaker for the first time became apparent through the negative shift in evaluations from raters' initial judgement.Especially for the cook position, the L2 speaker's downgrade appeared to be more pronounced when listeners expected her to be an L1 speaker (negative violation) than an L2 speaker (confirmed expectation).While this may portray the more extreme reactions that are predicted by expectation violation theory (Burgoon, 2009;Burgoon & Burgoon, 2001), the expectations ultimately had no effect on overall employability ratings.
At first glance, it may be that the listeners were impervious to the target manipulation, such that they did not detect or internalize the information in the applicants' Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 resumes.Nevertheless, this is inconsistent with the listeners' initial reaction in the resumescreening stage, where they tended to favor (at least numerically so) the applicant when she was presented as an L1 French speaker (see Figures 1 and 2), thus demonstrating a slight but well-attested bias for L1-or standard-accented applicants (Carlson & McHenry, 2006;Rakić et al., 2011;Roessel et al., 2019).Conceivably, the lack of expectation effects could also have stemmed from potential differences in the quality of the applicant's two matched interview responses at each assessment timepoint, where, for instance, the same speaker produced a more content-relevant response when answering the question of why she is the ideal person for the job than how she stands out among other candidates.However, it is unlikely that any differences in response quality masked expectation-driven effects, because both matched responses from the same speaker occurred in both congruent and incongruent scenarios.An effect of expectancy violation, if it can indeed be detected in a design such as ours, should presumably emerge across the responses that were rotated between the expectancy scenarios.
Instead, the most plausible interpretation for the lack of expectation effects in this study is that violated expectations might not be as detrimental to interview outcomes as previously thought (Aboud et al., 1974, Hansen et al., 2017a, 2018, 2017b,).Indeed, in prior expectancy-focused work, listeners were presented with various accented speakers either saying the same 3-second sentence (Hansen et al., 2017a(Hansen et al., , 2018(Hansen et al., , 2017b) ) or reading the same 20-second story (Dragojevic et al., 2019), whose content was irrelevant to the target job posting, making accent the only diagnostic information about those speakers' employability.However, because stereotypes "operate most powerfully in the abstract, applying primarily to undifferentiated individuals, and they may exert relatively little impact on judgements about concrete, individuated targets" (Nisbett et al., 1981, p. 272), presenting the listeners with extended, individuating content of each applicant's jobspecific responses may have mitigated or altogether eliminated expectation effects.Put differently, any expectation-based bias may have been diluted for the listeners through their focused, extended experience with specific individuals discussing job-relevant issues.Until investigated further, an interim conclusion stemming from this dataset is that job-specific, extended interview experience might mitigate expectation-driven influences that could develop early in a job interview process.

Time Course of Employability Assessment
Besides the absence of expectation effects, a key finding of this study is that L2 applicants' employability fluctuates during a mock interview.For both positions, the applicants' ratings were high in the resume-screening stage, but they were significantly less positive after the first (for the cook position) or even the second (for the waitress position) interview response.This is a novel finding because (to the best of our knowledge) no prior L2 research on speaker employability has elicited multiple ratings of the same speakers over the course of a single interview (but see Louw et al., 2010, for an example of multiple employability ratings in an intervention study).A general U-shaped pattern of the applicants' evaluations may not be altogether surprising.Whereas the resume, which listed a completed bachelor's degree and job-relevant experience and qualifications, showcased the candidate as fairly suitable for each position, her responses to the two questions (each around 40 seconds in duration) provided important interview-relevant content, allowing the listeners to revise and finetune their assessments so they could determine the person's job suitability.
Even though response quality may not have mattered for expectancy violation, response quality may have certainly determined how severely the listeners evaluated each applicant's employability across time.For example, an L2 speaker may have provided a better answer (e.g., in terms of clarity or depth) as their first or second interview response, resulting in an upgrade or a downgrade of that candidate's employability over time.To examine this possibility, a post hoc analysis was conducted to assess the quality of the L2 applicants' responses, independent of their accent, where written transcripts of each speaker's responses were rated through a similar LimeSurvey interface by 20 additional L1 French speakers on a 100-point scale where 100 indicated high quality, with the wording of each question provided before raters saw response transcripts.For the L2 candidate applying for the cook position (Figure 1), there was indeed a significant difference (by Bonferroni-corrected paired-samples t tests) in the content ratings between her first (M = 59.83) and her second (M = 80.33) interview responses, t(19) > -3.40, p < .003,95% CI = [-36.99, -8.81], d < -0.76.However, for the L2 candidate applying for the waitress position (Figure 2), her first (M = 64.35)and second (M = 66.20) interview responses were evaluated similarly, t(19) < -0.68, p > .505,95% CI = [-14.29, 7.29], d < -0.01.Therefore, an upgrade in the employability of the pizza cook applicant at the end of the interview may have been due to the listeners rewarding that speaker for her response quality, relative to her initial performance.In contrast, the waitress candidate had no such advantage at the end of the interview, because the quality of her first and second interview responses was similar.With respect to the time course of employability ratings, then, a tentative conclusion is that these ratings reflected the quality of each candidate's interview content, where better-quality responses elicited higher employability ratings, similar to how pragmatic competence has shown to be vital for L2 speakers' job interview success (Louw et al., 2010).

Employability and Job Communication Requirement
Finally, the present findings speak to the pattern of employability ratings for two entry-level jobs that differ in communication requirement (pizza cook vs. waitress).Because language skills were deemed more important for the waitress position, it was expected that the listeners would be sensitive to these job demands, assigning harsher ratings to the L2 applicants for the waitress than the pizza cook positions.While the waitress applicant, just as the pizza cook applicant, was initially downgraded in her ratings after the first response, the listeners' perception of her employability did not improve by the end of the interview, as it did for the cook.In fact, as shown in Figure 2, the listeners evaluated the L1 French-speaking candidates as more suitable for the waitress position.It is possible that, as expected, the listeners in this study may have categorized the L2 applicants in terms of the functional demands of the advertised job, presumably considering comprehensible L2 speech more important for the waitress than the cook position.However, because both the L2 waitress and L2 cook applicants were rated as relatively difficult to understand (with overall comprehensibility ratings of 62.62 and 67.45 on a 100point scale, respectively), raters may have globally preferred L1-speaking candidates over L2 speakers, especially for what they likely considered to be a customer-facing position requiring substantial communication skills.Put differently, whereas an L2-speaking Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 candidate might be as suitable as an L1 French speaker for the position of a pizza cook, which is a low-status job requiring little communication, an L2 speaker was clearly less acceptable as a waitress, which corresponds to a low-status job requiring considerable French communication skills.This explanation would align with results of a recent metaanalysis of 27 studies which revealed a hiring bias favoring standard-accented speakers only for jobs with high communication demands (Spence et al., 2022).Alternatively, as discussed above, in light of a post hoc analysis of response quality, the L2 French cook may have been upgraded in her final evaluation because, unlike the L2 French waitress, she provided more sophisticated answers in her second interview responses.Even though disentangling these two explanations is impossible in the present dataset-because job position was manipulated across different speakers (and different interview content)-both forces may have at play.The listeners may have factored in the communicative demands of the job, preferring L1-speaking candidates over L2 speakers for the waitress position, and they may have also penalized the particular candidate applying for the waitress position for her (less-than-optimal) interview performance.It remains for future research to examine the precise interplay of these factors in determining L2 applicants' employability for more versus less communication-relevant positions.

Limitations and Future Research
The present findings must be examined in future research to address this study's limitations.First, it would be important to revisit these results in a larger dataset, featuring a more extensive applicant sample, including speakers of different L1 and L2 varieties, focusing on job interviews in languages other than English and French, and targeting a broader set of positions, such as high-status jobs with a strong communication component (e.g., branch manager).It would also be revealing to compare applicants' employability ratings across different contexts, where the target language is a dominant language (e.g., French in Québec) and a minority or local language (e.g., French in English-speaking Canada), assuming that listener expectations are situation-and context-specific.Second, while this study captured how L2 speakers are perceived following expectations of an L1 speaker (negative violation), future research should also include a positive violation scenario to explore how an expected L2 speaker may be upgraded when their speech surmounts expectations (i.e., L2 applicant resume followed by interview audios of an L1 speaker).Third, future interview-focused studies should strive not only to approximate reallife demands of authentic interviews by providing extended, multiple opportunities for listeners to evaluate candidates but also to include measures capturing the depth of applicant interview responses and the extent of listener comprehension, particularly for L2speaking candidates.Fourth, even in the most controlled experiments, it might be difficult to isolate expectation-driven biases from other influences on listeners' decision-making, such as mood, background knowledge, preparedness, or the quality and composition of the applicant pool.Therefore, to better understand listener reactions, future studies should include a qualitative component where listeners could explain their ratings for each candidate, either through open-ended comments or through a stimulated-recall procedure.Finally, future research could also explore the possibility that listeners may rate job applicants differently as a function of inter-individual differences, including their social dominance orientation (Hansen & Dovidio, 2016) and general accent beliefs (Hansen, 2020).

Conclusion
This study's dynamic perspective sheds light on how listeners react toward L2 French speakers during a mock job hiring process.While the expectation condition appeared to have no effect on employment evaluations, the findings portray how resume characteristics irrelevant to job qualifications can impact employability perceptions, which can then fluctuate with the presentation of new information.On the one hand, it is concerning that a resume of an L2 French applicant was downgraded relative to that of an equally-qualified L1 French candidate, even for a job that requires minimal French communication and customer contact.Indeed, this is not an unfamiliar concern, considering that the bias in how people perceive names on resumes has been thoroughly documented (e.g., Cotton et al., 2008;Derous et al., 2014;Oreopoulos, 2011;& Dechief, 2012).Considering this, human resource specialists may wish to blind names during the resume-screening stage to avoid potential bias and any expectations they may generate.On the other hand, it is encouraging to see an improvement in at least one L2-speaking candidate's employability ratings across time, which implies that first-impression biases (however subtle) had limited impact on final interview evaluations for certain applicants.In fact, in the end, L2 candidates' employability evaluations appeared to reflect (as they should) the quality of their interview responses.This is a positive finding for educators teaching a second language, as it demonstrates the value of the ideas expressed and of speakers' pragmatic competence for job interview situations, which is a skill that can be taught to help L2 speakers succeed in interviews despite potentially being less comprehensible than L1 speakers (Louw et al., 2010).At the very least, the results of this study should inspire further investigation into how an employer may readjust their perceptions following resume-based expectations.We therefore call for more research on expectation violations involving individuals who are invested in hiring practices to gain a better understanding of how impression formation of job applicants might change with the addition of new information (e.g., speech, appearance) that may violate or confirm prior expectations.
Correspondence should be addressed to: Rachael Lindberg Email: Rachael.Lindberg@concordia.ca

Acknowledgements:
This study has been approved by members of the University Human Research Ethics Committee (Chair: Dr.Richard DeMont) at Concordia University and was deemed acceptable on ethical grounds for research involving human subjects (certification number: 30010556).This work was supported by the Social Sciences and Humanities Research Council of Canada under Grant # 430-2020-01134 and Grant # 435-2021-0069 awarded to the second author.

Pizza Cook
Restaurant Brigade Pizzeria Napolitaine, Montreal

DESCRIPTION:
We are looking for someone with energy and positivity to join our team.Pizza baking experience is a plus.Must be able to speak French and basic English.The salary will depend on your ability and your level as a pizza chef.We offer training to those who do not have a lot of experience but have a passion to learn the trade.If you're the right fit for our team, we'll train you.

MAIN DUTIES:
• Prepare all foods as directed, hygienically and efficiently • Follow recipes and presentation specifications • Use standard kitchen equipment in a healthy and efficient manner • Clean and maintain the station while maintaining safety and sanitation standards • Maintain inventory of food, utensils and equipment • Comply with all sanitation and food production codes European French Waitress: « J'essaie de trouver un petit job à temps partiel, surtout comme serveuse.J'ai déjà été serveuse dans le passé, alors je sais comment faire.Je sais que j'ai de bons atouts pour faire ce job.Notamment, par exemple, je suis très organisée, j'aime beaucoup parler avec les gens.J'ai des très bons rapports relationnels avec les clients en général.Aussi, je sais bien m'adapter à différentes situations, aussi aux situations où le niveau de stress est très élevé.Ça ne me dérange pas du tout.Je sais très bien gérer des situations compliquées, difficiles.Étant donné que j'ai déjà fait ce travail dans le passé, je pense que je serai la bonne personne pour celui-ci.» "I'm trying to find a small part-time job, especially as a waitress.I've been a waitress before, so I know how to do it.I know that I have good assets to do this job.In particular, for example, I am very organized, I really like talking with people.I have a very good relationship with customers in general.Also, I know how to adapt well to different situations, also to situations where the level of stress is very high.It doesn't bother me at all.I know very well how to handle complicated and difficult situations.Since I've done this job before, I think I will be the right fit for this one."Québec French Waitress : « Je pense que je suis une personne qui est assez sociable, qui a assez de facilité à communiquer avec les gens.Je pense que c'est vraiment important dans la restauration, le lien avec les clients.Aussi le fait que je sois assez rapide et efficace sur le plancher.J'ai déjà travaillé dans un restaurant avant.Donc, j'ai vraiment acquis cette rapidité de service et de mémorisation des commandes des clients et des petites commandes spéciales -'Il faut que je retourne dans telle salle pour telle chose.' Sinon, je pense que je suis quelqu'un de polie aussi, et d'assez organisée.» "I think I am a person who is quite sociable, who is quite at ease with communicating with people.I think it's really important in the restaurant industry, the connection with customers.Also the fact that I am quite fast and efficient on the floor.I have worked in a restaurant before.So, I really acquired this speed of service and memorization of customer orders and special orders -'I have to go back to this room for this thing.'Otherwise, I think I'm a polite person too, and quite organized."L2 French Waitress : « Je suis une personne très organisée.J'aime travailler en équipe.Je suis très, très dynamique.Je sais aussi résoudre les conflits.Et j'aime parler avec des personnes.Je pense aussi que je suis une personne très amicale.J'ai de l'expérience.Je travaillais dans un autre restaurant comme serveuse pendant une année.Je pense que j'ai plus ou moins d'expérience.J'ai appris beaucoup et j'ai vraiment aimé cette expérience.Et c'est pour ça que je pense que je suis une personne idéale pour ce poste.» "I am a very organized person.I like teamwork.I am very, very dynamic.I also Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 know how to resolve conflicts.And I like talking with people.I also think that I am a very friendly person.I have experience.I worked in another restaurant as a waitress for a year.I think I have more or less experience.I learned a lot and really enjoyed this experience.And that's why I think I'm an ideal person for this position." European French Cook : « J'ai déjà eu des expériences dans un restaurant de pizzeria pendant un an où j'ai vraiment adoré faire des pizzas.J'ai même réussi à créer plusieurs recettes que le restaurant n'avait pas avant.Et aussi vous cherchez quelqu'un qui a beaucoup d'énergie, qui est positif et je crois que ces deux adjectifs me décrivent très bien.Et aussi c'est l'esprit d'équipe.J'ai beaucoup travaillé avec des équipes et j'adore ça.Cet esprit d'équipe, de communauté, c'est vraiment ce qui m'enchante dans le travail et c'est vraiment ce que je cherche.» "I've already had experience in a pizzeria restaurant for a year where I really loved making pizzas.I even managed to create several recipes that the restaurant didn't have before.And also you are looking for someone who has a lot of energy, who is positive, and I believe that these two adjectives describe me very well.And also team spirit.I've worked a lot with teams and I love it.This team spirit, spirit of community, is really what delights me in this work and it is really what I am looking for."Québec French Cook : « Je suis quelqu'un qui est ponctuel et qui est fidèle à mon employeur, en qui on peut avoir confiance.J'aime travailler en équipe.Je n'ai pas de problème parce que j'ai travaillé avec des gens avec différentes personnalités.J'aime justement ce travail en cuisine, comme ambiance d'habitude c'est dynamique.J'ai déjà travaillé le même style d'emploi dans le passé pour faire des pizzas comme cuisinière de pizza.Donc, ça serait utile à votre restaurant.Je ne pars pas de zéro, donc c'est pas mal ça.» "I am someone who is punctual and who is loyal to my employer, who can be trusted.I like teamwork.I don't have a problem because I've worked with people with different personalities.I just like this work in the kitchen, typically it's dynamic.I've worked the same style of job in the past making pizza as a pizza cook.So it would be useful to your restaurant.I'm not starting from zero, so that's not bad."L2 French Cook : « Depuis l'année dernière, je travaille dans un restaurant.Le restaurant, il est italien, donc je voudrais continuer de faire ça et gagner un peu plus d'expérience avec la cuisine italienne.Aussi je pense que je serais la personne idéale pour ce travail parce que j'ai déjà de l'expérience à faire de la pizza, préparer les repas.Et je sais utiliser le matériel, le four, et tout ça dans un restaurant.» "Since last year, I have been working in a restaurant.The restaurant, it's Italian, so I would like to continue doing that and gain a little more experience with Italian cuisine.Also I think I would be the perfect person for this job because I already have experience making pizza, preparing meals.And I know how to use the equipment, the oven, and all that in a restaurant."

•
No previous experience required • French (Mandatory) Canadian Journal of Applied Linguistics: 26, 2 (2023): 1-31 Examples of Original (French) and Translated (English) Interview Responses Parlez-moi de vous.Qu'est-ce qui fait de vous la personne idéale pour occuper ce poste?Tell me about yourself.What makes you the ideal person for this job?

Table 1
Means (Standard Deviations) for Employability Ratings by Job, Condition, and Interview Stage