You are on page 1of 24

883903

research-article2019
JLSXXX10.1177/0261927X19883903Journal of Language and Social PsychologyHansen

Article
Journal of Language and Social Psychology
2020, Vol. 39(1) 148­–171
Accent Beliefs Scale (ABS): © The Author(s) 2019
Article reuse guidelines:
Scale Development and sagepub.com/journals-permissions
DOI: 10.1177/0261927X19883903
https://doi.org/10.1177/0261927X19883903
Validation journals.sagepub.com/home/jls

Karolina Hansen1

Abstract
People’s accents in speech strongly influence how they are perceived by others. The
current Accent Beliefs Scale was inspired by work on stigmatization, implicit theories
of intelligence, and essentialism. The scale has two dimensions: accent diagnosticity
and accent stability. The scale was developed, validated, and applied using a mixed
methods approach with a QUAN–qual sequential design. Pretest and Study 1
developed the items, the subscales, and showed that diagnosticity and stability beliefs
are independent of each other. Study 2 confirmed the scale’s two-factor structure on
a large sample and proved its divergent and convergent validity. Study 3 addressed
predictive validity and showed that the more perceivers viewed accents as diagnostic
of other traits and the more they believed accents can be changed, the worse they
evaluated a nonnative speaker with a strong accent. The developed scale can help
understanding and predicting negative reactions to nonnative speakers.

Keywords
nonnative accent, language attitudes, implicit theories, stigma, essentialism

Social categories of race and ethnicity have been extensively studied by social scien-
tists. For instance, it has been shown many times that a foreign appearance in terms of
skin color or ethnic physical features can lead to a negative social evaluation (Allport,
1954; Dovidio & Gaertner, 2010). Nevertheless, recent research has shown that in
numerous social contexts appearance might be less informative as a social cue than
accent in speech (Hansen, Rakić, & Steffens, 2017; Kinzler, Shutts, Dejesus, & Spelke,
2009; Pietraszewski & Schwartz, 2014). In general, people’s accents can strongly
influence how they are perceived by others, but also these others can differ in the way

1University of Warsaw, Warsaw, Poland

Corresponding Author:
Karolina Hansen, Faculty of Psychology, University of Warsaw, ul. Stawki 5/7, Warszawa 00-183, Poland.
Email: karolina.hansen@psych.uw.edu.pl
Hansen 149

they interpret accents and what they infer from them. These interpretations and infer-
ences can, in turn, influence whether a nonnative speaker is employed, finds an apart-
ment or how she or he is evaluated in an educational setting (Gluszek & Hansen,
2013). For example, many people believe that accent is something one can change. It
is true that people’s accents often get weaker over years or decades spent in a new
country, but there is extensive research showing that accents cannot be eliminated after
childhood no matter how hard the speaker tries (e.g., Moyer, 2004; Scovel, 2000).
Perceivers’ erroneous beliefs about accent can do a lot of harm to nonnative speakers.
In the current article, two such beliefs about accents are presented: beliefs about accent
stability over time and about accent diagnosticity. The current article reports on the
development and validation of a scale that measures these beliefs: Accent Beliefs
Scale (ABS). The scale is important to understand the, until now overlooked, role of
these beliefs in perceptions of nonnative speakers.

Accent Beliefs: Need for Research


Although several researchers have suggested that beliefs about accents can be impor-
tant for people’s impressions of accented speakers, there is hardly any empirical
research on the topic. Already in the 80s Ryan (1983), at the end of her theoretical
article, mentioned that “causal attributions research could be useful to determine how
much control listeners attribute to second language speakers over their speech stan-
dardness” (p. 159). At the same time, Wertheimer (1983) mentioned similar topics in
an article about ethics of employers’ choices in job recruitment process. In his exam-
ples he wrote about, what psychologists would call, accent controllability or stability
(p. 103) and about accent diagnosticity—how much one can infer about a person
based on their accent (p. 104). Gluszek and Dovidio (2010b) in a more recent article
reasoned that “a person with a nonnative accent may be perceived in control over
progress of an accent and ( . . . ) might suffer from negative consequences associated
with being perceived as willingly holding onto a stigmatized identity” (p. 222). In an
even newer article, showing surprisingly positive evaluation of a nonnative Mexican
speaker in the United States, Hopkins (2015) wrote that “language is often perceived
as a choice” (p. 22) and he suggested measuring perceptions of control over accent as
an avenue for future research.
Despite the aforementioned curiosity and theorizing about accent beliefs, there is
hardly any research directly testing these claims. Recent studies addressed this issue
from the perspective of nonnative speakers and their concept of language learning (not
specifically accent) showing that language learners who believe that language skills
can be trained more strongly endorse learning goals and deal better with failure (Lou
& Noels, 2016, 2017). However, to the best of my knowledge, there is no research on
this topic from the perspective of the listeners. To understand accent attitudes and
potential discrimination of accented speakers, it is crucial to empirically test and
understand how listeners may react to nonnative speakers depending on the listeners’
beliefs about accents. The goal of the current research is to provide tools to study such
beliefs. The current scale is inspired by the theorizing on accent stability (e.g., Gluszek
150 Journal of Language and Social Psychology 39(1)

& Dovidio, 2010b) and more generally by the research on stigmatization (Crandall,
1994), by implicit theories of intelligence and other traits (Dweck, Chiu, & Hong,
1995), as well as by essentialism theory (Haslam, Rothschild, & Ernst, 2000). Based
on these lines of research, it was expected that two types of beliefs would be crucial
for accents perception: accent stability and accent diagnosticity/informativeness.
Accent stability, as the name suggests, refers to how stable and unchangeable is a per-
son’s accent over their lifetime. Accent diagnosticity refers to how diagnostic is accent
for other traits of a person, namely how much we can infer about a person based on
their accent.

Accent Stability
There seems to be no published research on perception of accent stability, but one
piece of preliminary research is mentioned by Gluszek and Dovidio (2010b) in their
review about accents. They cite their conference poster and write that

Although empirical evidence about listeners’ beliefs about controllability over progress
is limited, in recent research, we found that listeners who were more likely to think that
people can readily eliminate nonnative accents felt more uncomfortable talking with
people with nonnative accents and tried to avoid these encounters more. (Gluszek &
Dovidio, 2010b, p. 222)

Hansen and Dovidio (2018), also in an unpublished work, found that the more partici-
pants agreed with statements like “Immigrants are capable of eliminating their accent,”
the worse they evaluated Chinese-accented speakers who spoke with a strong nonna-
tive accent. This was especially the case when participants believed that the speakers
were living in the United States for many years.
Research on stigmatization also seems relevant for accent stability beliefs. This
research has addressed, among others, questions of how people’s beliefs influence
perception of stigmatized individuals. It was postulated that when a source of stigma
is perceived to be something evoked by external causes, people are less prejudiced
toward individuals with such a stigma than when the causes are perceived as internal
(Weiner, Perry, & Magnusson, 1988). It was shown, for example, that people show
less prejudice toward obese individuals when they believe that obesity is genetically
based rather than when they believe that it depends on ones choices or will (Crandall
& Moriarty, 1995; Crocker, Cornwell, & Major, 1993). Along these lines, beliefs
about accent stability could coincide with evaluating nonnative speakers with less
accountability for the way they speak and for that reason with less stigmatization.
Partly similar concepts, but opposite predictions, are described by Dweck’s (1999)
implicit theories of intelligence. Dweck says that people can be categorized as two
types of naïve theorists: entity theorists and incremental theorists. Entity theorists
believe that intelligence is something fixed, given once for good. To some extent,
they are like essentialists of biological basis, believing that some traits are inborn and
we do not have any influence over them (Haslam et al., 2000). In contrast,
Hansen 151

incremental theorists believe that intelligence is something flexible that can be trained
like a muscle. Those who believe that intelligence is fixed tend to put less effort in
learning, especially when confronted with failure, than those who believe that intel-
ligence can be developed with own effort (Molden & Dweck, 2006). Similar results
emerged for “language intelligence” (Lou & Noels, 2017). If we equate entity theo-
rists with believers of trait (e.g., intelligence or accent) stability, then we see that the
entity reasoning goes in the opposite direction to stigmatization research: Stability
does not take the fault away of the person, but rather makes the perceiver hopeless
that the person will ever change. Entity versus incremental beliefs were also applied
to other domains than intelligence. It was shown that prejudice entity theorists (i.e.,
people who believe that one’s prejudice level is fixed) are less interested in activities
reducing prejudice or in interracial interactions, and are more anxious and unfriendly
in such interactions (Carr, Dweck, & Pauker, 2012). Crucially, the role of these theo-
ries has been also shown in the perception of others, not only in self-perception. In
the Israeli–Palestinian conflict, when Israelis and Palestinians believed in a fixed
view of groups, they had more negative attitudes toward the out-group and were less
willing to compromise than when they had a flexible view (Halperin, Russell,
Trzesniewski, Gross, & Dweck, 2011).
Both the stigmatization and the implicit theories reasoning are backed up with
extensive evidence, but have conflicting predictions and findings when it comes to the
role of stability beliefs and evaluation of others. On the one hand, general research on
stigma and unpublished research on accents (Gluszek & Dovidio, 2010b; Hansen &
Dovidio, 2018) suggest that people who believe that accent is something stable would
not blame nonnative speakers for having an accent and would evaluate them more
positively than would people who believe that the speakers could have changed their
accent. On the other hand, research on fixed and flexible traits, although focused more
on self-perception than other-perception, would suggest the opposite: people who
believe that accent is flexible would evaluate nonnative speakers better. The reasoning
behind could be that if accent is flexible, then there is hope and chance that the person
will diminish it with time and effort, similarly as one can heighten intelligence or
diminish prejudice (Carr et al., 2012).
Having a scale to measure people’s accent beliefs, one can verify which reasoning
is more probable when it comes to accents. One can also study under which conditions
the effects might resemble the ones shown by stigmatization research or the ones
shown by research on implicit theories of intelligence. Possibly such features as accent
strength or accent type (i.e., accent of a positively or a negatively perceived group)
could influence these evaluations. If one believes that accent can be changed and hears
a person speaking with a strong accent, the perceiver might evaluate the speaker nega-
tively. If the same person, believing that accent can be changed, hears a person with a
very weak accent, he or she might evaluate the speaker positively, as the speaker has
tried hard to eliminate their accent and has almost succeeded in doing so. A scale of
accent stability can help testing these scenarios.
152 Journal of Language and Social Psychology 39(1)

Accent Diagnosticity
As shown above, accent stability seems important for evaluations, but the mechanisms
of its influence can be complex. Accent diagnosticity (i.e., accent’s predictive potential)
might be a more straightforward element of accent beliefs. Research on essentialism is
here of particular relevance. This research showed that essentialist beliefs have two main
dimensions: one showing how much social categories are understood as natural kinds
and the second one showing how coherent these categories are as entities (Haslam et al.,
2000). The natural kind (or biological basis) dimension shows that a social category is
something stable and has a biological basis, like gender or race. This dimension can be
seen as a broader version of the fixed-incremental beliefs described earlier. The entitativ-
ity dimension has two subdimensions: discreteness and informativeness. Most relevant
for accents is the informativeness dimension. It is expressed in statements such as to
know about many aspects of a person once you become familiar with a few of their basic
traits (Bastian & Haslam, 2006, p. 235). People scoring high on essentialist informative-
ness dimension believe that “knowing that someone belongs to the category tells us a lot
about that person” (Haslam et al., 2000, p. 117). In other words, informativeness tells us
that a piece of information about a person is diagnostic and can be used to infer more
about this person. Different aspects of essentialist beliefs, among them the informative-
ness beliefs, were shown to be positively related to stereotyping (Bastian & Haslam,
2006). In the context of accents, it is possible that people who base their judgments of
others on the others’ accents are likely to generally perceive these others in a stereotyped,
simplified manner.
Although I described implicit intelligence/personality theories in the section about
accent stability, in the scales of implicit theories stability was often intermixed with
diagnosticity. It happened in questions such as Your intelligence is something very basic
about you that you can’t change very much or You can do things differently, but the
important parts of who you are can’t really be changed. These and similar items include
parts that reflect stability of a trait such as you can’t change (your intelligence) very
much but they also include a diagnosticity essentialist parts like Your intelligence is
something very basic about you or the important parts of who you are. Translated into
accents, with such questions one would be unable to distinguish between people who
believe that a nonnative accent is something that cannot be changed and people who
believe that it is an important part of a person, a person’s “essence.” One could believe
that accent cannot be changed, but at the same time believe that it does not tell much
about a person. These two aspects might be related to each other, but they should not
need to be. Essentialism research keeps them apart and shows only weak relationship
between the two (r = .25 in Bastian & Haslam, 2006).

Current Research
The present research was designed to create a scale to measure accent beliefs. It was
envisioned that two main dimensions of the scale would be accent diagnosticity and
accent stability. Implicit theories of intelligence and prejudice were a basis for some of
Hansen 153

Figure 1. Graphical representation of the sequential mixed design of the current research.

the scale’s stability items, but were adapted not to mix diagnosticity with stability. The
essentialist natural kind/biological basis dimension enriched the accent stability sub-
scale. The informativeness part of essentialism research inspired accent diagnosticity
scale.
The scale was developed and validated in a pretest (see supplemental material
available online) and three studies. The scale’s creation process was partly theory-
driven, partly data-driven, and was inherently iterative and incremental (Larman &
Basili, 2003). It was a mixed methods approach (Creswell, 2013; Driscoll, Appiah-
Yeboah, Salib, & Rupert, 2007) that combined quantitative methods used traditionally
for scale development such as exploratory factor analysis and confirmatory factor
analysis with analysis of participants’ comments, group discussions, and an expert
consultation of the scale’s items. It was a QUAN–qual (more quantitative than qualita-
tive) sequential mixed design (see Figure 1).
All studies conducted in Poland were carried out in accordance with the Ethical
Code of Psychologists of the Polish Psychological Society and with the guidelines of
American Psychological Association for research on human participants. Research
conducted on an online U.S. sample was in accordance with the above guidelines and
was additionally approved by the Yale University Institutional Review Board.
154 Journal of Language and Social Psychology 39(1)

Items Creation and Preselection


The items were generated based on the concepts and operationalization of Dweck’s
implicit theories of intelligence and prejudice, as well as on Haslam’s essentialism.
Many of the initial accent stability items were adaptations of Dweck’s scale of fixed
versus growth mind-set (Dweck et al., 1995; Lachowicz-Tabaczek, 2004; Levy,
Stroessner, & Dweck, 1998). For example, in an item No matter who you are, you
can significantly change your intelligence level the words intelligence level were
exchanged for accent strength. This simple strategy was earlier successfully used to
adapt the implicit theories intelligence scale into an implicit theories prejudice scale
(Levy et al., 1998). Additionally, a couple of similar items were freely generated.
Accent diagnosticity items were created based on the informativeness subscale of
essentialism (Bastian & Haslam, 2006), but adjusting the items considerably to fit
the accent context. Besides of the core diagnosticity and stability items, a few items
about situational stability and situational control over accents were added (Gluszek,
2010; see supplemental material available online). The initial items pool was dis-
cussed during an expert meeting, reformulated, pretested, and then adjusted again
after the comments of pretest participants (see supplemental material available
online for a full list of items).

Study 1
The goals of Study 1 were to prepare and test a preliminary version of the ABS. When it
comes to preparing, an initial list of nine diagnosticity and nine stability items was pre-
sented to an expert group. After discussion, a few items were excluded and a few rewrit-
ten or created new. Seven diagnosticity and nine stability items were selected for Study
1. The already consistent (after the pretest) three situational control items were also used
in this study, but not in further studies (see supplemental material available online).
Additionally, scales referring to essentialism, general trait immutability/stability,
social dominance orientation (SDO; Pratto, Sidanius, Stallworth, & Malle, 1994), and
right-wing authoritarianism (RWA; Altemeyer, 1981) were added in order to provide a
convergent validity test. Various aspects of essentialist beliefs and trait immutability
beliefs were shown to be related to prejudice and stereotyping (Bastian & Haslam,
2006). In the current study, all aspects of essentialism were expected to correlate with
accent diagnosticity beliefs. As this was the first test of the ABS and accent stability
was a more distant concept from essentialism than accent diagnosticity, it was unclear
whether it would correlate with essentialism, but was tentatively expected to be related.
General trait stability was not expected to correlate with accent diagnosticity, but to
correlate with accent stability. Furthermore, SDO and RWA are major predictors of
prejudice (Duckitt & Sibley, 2007) and these were also expected to correlate with
accent diagnosticity beliefs. In this first test, it was hard to clearly predict relationships
between accent stability, SDO, and RWA. Motivation to respond unprejudiced was
included to test for divergent validity. This motivation was expected to correlate only
weakly with both types of accent beliefs, as it is sometimes claimed that accent or
Hansen 155

language bias is a socially acceptable form of prejudice and as such might be expressed
more freely than other, less acceptable, forms of prejudice (Hansen & Dovidio, 2016;
Ng, 2007).

Participants
Participants were recruited through science-related fanpages and groups on Polish
social media and through an earlier large longitudinal study on other topic. The study
link was clicked by 417 participants, but some quit after the first page of the study. The
final sample consisted of 262 participants (Mage = 22.75, SD = 5.30, age range: 19-50
years, 83% women). All were native Polish speakers.

Method
Participants were asked to think about the way nonnative speakers speak and indicate
to what extent they agree or disagree (1 = strongly disagree to 7 = strongly agree) with
the statements of the ABS such as, for diagnosticity From someone’s strong or weak
accent one can infer many things about the speaker, or for stability People are capable
of eliminating their accent. After the questions about accent, other scales were pre-
sented, all using anchors from 1 = strongly disagree to 7 = strongly agree. One of the
scales was essentialism (Bastian & Haslam, 2006) with subscales of biological basis
(eight items such as The kind of person someone is can be largely attributed to their
genetic inheritance, α = .86), discreteness (eight items, e.g., Everyone is either a cer-
tain type of person or they are not, α = .77), informativeness/diagnosticity (seven
items, e.g., It is possible to know about many aspects of a person once you become
familiar with a few of their basic traits, α = .78), and additionally a scale of general
trait immutability/stability (eight items, e.g., The kind of person someone is something
very basic about them and it can’t be changed very much, α = .89, Levy et al., 1998).
Other scales used were SDO (five items, α = .82, Pratto et al., 1994) with statements
such as Superior groups should dominate inferior groups and RWA (six items, α = .83,
Altemeyer, 1981) with statements like The real keys to the good life are obedience and
discipline. Motivation to respond unprejudiced (six items, α = .72, Devine, Plant,
Amodio, Harmon-Jones, & Vance, 2002; Dunton & Fazio, 1997) was assessed with
statements such as If I have a prejudiced thought or feeling, I keep it to myself. At the
end, participants filled out demographic questions and could leave comments about the
study.

Results and Discussion


Exploratory Factor Analysis. A principal component analysis with a varimax rotation was
performed. The results showed that eigenvalue was greater than 1.0 for four factors, but
for the fourth factor it was only 1.18 and the scree plot clearly suggested three factors
(Cattell, 1966). The three factors explained 51% of the total variance. The first factor
was about accent stability and the second—accent diagnosticity (for loadings, see
156 Journal of Language and Social Psychology 39(1)

Table 1. Correlations of the Accent Subscales With Other Scales in Study 1.

Ess. Ess. Ess. inf./ Trait


Stability biolog. discret. diagn. stability SDO RWA Motivation
Diagnost. .07 .13* .22*** .13* .06 .18** .28** −.20**
Stability .15* .20** .01 .21*** .04 .14* .09

Note. Diagnost. = accent diagnosticity; Stability = accent stability; Ess. biolog. = essentialism biological
basis; Ess. discret. = essentialism discreteness; Ess. inf./diagn. = essentialism informativeness/
diagnosticity; Trait stability = general trait immutability/stability; Motivation = motivation to respond
without prejudice.

supplemental material available online). In the third factor, three items that loaded in the
same direction were about situational control over accent and four that loaded into the
opposite direction were about motivational aspects of accent stability (and are discussed
only in the supplemental material available online). The two main subscales were: accent
stability (six items, α = .83) and accent diagnosticity (seven items, α = .78).

Correlations. As can be seen in Table 1, accent diagnosticity was not significantly cor-
related with accent stability. Although earlier research has often included aspects of
stability of a trait and its diagnosticity in the very same items and scales, this correla-
tion and the results of the factor analysis show that accent diagnosticity and stability
are separate constructs. Furthermore, proving convergent validity, accent diagnosticity
was positively related to different aspects of essentialism, SDO, and RWA. Diagnos-
ticity was weakly but, incongruent with predictions, significantly related to motivation
to respond without prejudice. Accent stability was also related to two (out of three)
aspects of essentialism, general trait stability/immutability, and RWA, but not to SDO.
It was, as expected, not related to motivation to respond without prejudice.

Expert Consultation. The created scale was presented to prof. John Dovidio—an expert
on, among others, perception of nonnative speakers and on mechanisms of accent
discrimination (e.g., Dovidio & Gluszek, 2012; Dovidio, Gluszek, John, Ditlmann, &
Lagunes, 2010; Gluszek & Dovidio, 2010a, 2010b). As a result of these consultations,
diagnosticity subscale was refined. There was a need to reflect separately accent type
and accent strength diagnosticity. Nonnative speakers can be discriminated either
because their accents reveal their provenience and stereotypes associated with a spe-
cific ethnic group are activated or because nonnative speech is (or is perceived to be)
harder to understand than native speech (Dovidio & Gluszek, 2012). The first mecha-
nism relates to stereotyping, the second one to processing fluency (Oppenheimer,
2008). To reflect the first possibility, some of the items were reworded to explicitly
state that accent type can tell us something about the speaker. One of the most impor-
tant correlates of the second (understanding/fluency) mechanism is accent strength.
People who have stronger accents are harder to understand and are evaluated worse
than people with weaker accents (Lev-Ari & Keysar, 2010). Some recent research sug-
gests that accent type might be of a lower importance than accent strength (Roessel,
Hansen 157

Schoel, & Stahlberg, 2018). As many of the scale’s items were already referring to
accent strength, no rewording was necessary. In the effect of the adjustments, there
was same number of diagnosticity items relating to accent type (stereotyping mecha-
nism) and to accent strength (understanding/fluency mechanism). As there was no
similar earlier research, it was hard to clearly tell whether these two aspects would
really be empirically separate. Therefore, the refined scale was expected to have (a)
either three subscales: stability, type diagnosticity, and strength diagnosticity with the
two diagnosticity aspects related to each other or (b) two subscales: stability and diag-
nosticity, with diagnosticity representing its two types but within one subscale. This
was to be tested in the next study.

Study 2
The main goal of Study 2 was to validate the created ABS. Furthermore, the study
tested empirically whether accent strength and accent type diagnosticity are separable
from each other or are rather two aspects of the same construct. This was done in an
exploratory factor analysis on one half of the data and a confirmatory factor analysis on
the second half of the data. Measurement invariance across genders was also assessed.
It was expected that accent diagnosticity would not be related to accent stability. To
verify the scale’s validity, the study explored relationships of the accent subscales to
other measures. For discriminant validity (i.e., what the scale should not or only
weakly correlate with), it was expected that neither of the two subscales would corre-
late with demographics and personality measures. For convergent validity (i.e., what
other similar constructs should the scale correlate with), the diagnosticity subscale was
expected to correlate with general out-group bias measures, such as SDO and RWA.
Stability subscale, based on results of Study 1, was expected not to correlate with these
or do so only weakly. For predictive validity (i.e., whether the scale can predict some
outcomes), both subscales were expected to correlate with attitudes toward immi-
grants and refugees and with measures of social distance to immigrants.

Participants
The study was a part of a larger prescreening for a panel of people interested in taking
part in social science studies in exchange for winning small prizes. In the large pre-
screening study, the prize was an e-book reader. Participants were recruited similarly
as in Study 1 (mostly through science-related fanpages and groups on social media).
Participants of the panel were 1,445 people, but excluding a few nonnative Polish
speakers and concentrating on the variables of interest, these were answered fully by
1,248 participants (Mage = 26.15, SD = 6.98, age range: 18-65 years, 78% women).

Method
After registering in the panel system and giving their informed consent, participants
answered a series of scales. Some were the same as in Study 1, some were new. First, they
158 Journal of Language and Social Psychology 39(1)

answered the same motivation to respond without prejudice scale (six items, α = .69).
Then, they answered a personality measure (Donnellan, Oswald, Baird, & Lucas, 2006)
with four items in each of the three subscales: neuroticism (e.g., Have frequent mood
swings, α = .77), extraversion (e.g., Am the life of the party, α = .87), and openness to
experience/intellect (e.g., Have a vivid imagination, α = .68), all with anchors from 1 =
does not describe me well to 7 = describes me perfectly. A short version of a classic self-
esteem scale (five items, α = .84, Rosenberg, 2015) that included statements such as I like
myself was administered using a scale from 1 = strongly disagree to 4 = strongly agree.
Afterward, the RWA (six items, α = .85) and SDO (five items, α = .83) scales same as in
Study 1 were used.
Then, the accent beliefs questions appeared. The study used a refined version of the
ABS. The goal was to have a short but reliable scale with items formulated in both
directions to avoid response biases. Accent diagnosticity had now two subscales (type
and strength diagnosticity). It was tested both as one and as two subscales. Each of the
subscales had four items, including two reverse-coded ones (see the appendix). All
items were, again, answered on a scale from 1 = strongly disagree to 7 = strongly
agree. Linguistic expectations to immigrants were measured with one item asking,
What level of proficiency in the majority language do you think immigrants should
have? and answers that could range from 1 = they do not need to learn it at all to 7 =
same as Poles, the level of native speakers.
Afterward, we also measured social distance to refugees (α = .92) and the two larg-
est immigrant groups in Poland: Ukrainians (α = .89) and Vietnamese (α = .86). All
social distance scales consisted of three items inspired by classic work of Bogardus
(1926) and included items such as Would you accept a refugee/Ukrainian/Vietnamese
as a neighbor? with possible answers from 1 = I would be strongly against to 4 = I
would definitely accept. For the clarity of presentation, before the analyses, the scale
was reversed so that higher scores would reflect larger social distance. Attitudes to
immigrants were assessed with five items such as Immigrants are taking away our
jobs (α = .85, Zick, 2011), also rated on a scale from 1 = strongly disagree to 7 =
strongly agree. Attitudes to refugees were measured with a 12-item scale with state-
ments such as Refugees who illegally arrived in Poland should be deported, even if it
would mean a restriction of freedom rated on a scale from 1 = strongly disagree to 7
= strongly agree (α = .93, Świderska, Winiewski, & Hansen, 2017). At the end, par-
ticipants filled out demographics (age, gender, size of the place of residence, and edu-
cation), were thanked and could leave their comments about the study.

Results and Discussion


The sample was split into two random halves and an exploratory factor analysis was
conducted on the first half and a confirmatory factor analysis on the second.

Exploratory Factor Analysis. It was expected that either three factors would be present
(accent strength diagnosticity, accent type diagnosticity, and accent stability) or two
factors (accent diagnosticity and accent stability). A principal component analysis
Hansen 159

Table 2. Factor Loadings From an Exploratory Factor Analysis in Study 2.

Factor

Diagnosticity Stability Name in CFA


From someone’s strong or weak accent one .74 Str. Diagn. 2
can infer many things about the speaker.
(r) The type of accent one has does not tell .72 Type Diagn. 3re
much about the speaker.
The strength of an accent in one’s speech is .70 Str. Diagn. 1
a sign of their personality.
It is possible to tell how someone will act by .69 Type Diagn. 1
hearing their accent.
(r) The strength of a person’s accent has .62 Str. Diagn. 3re
nothing to do with what kind of person
they are.
The type of accent in a person’s speech is an .62 Type Diagn. 2
important trait.
(r) One cannot learn much about someone .57 Type Diagn. 4re
on the basis of their accent.
(r) One cannot draw conclusions about .49 Str. Diagn. 4re
someone’s intelligence based on the
strength of their accent.
People are capable of eliminating their .81 Stability 1
accent.
(r) Everyone has an accent, and they cannot .79 Stability 4re
change it, even if they tried.
(r) If someone was raised in one language, .77 Stability 3re
they will have its accent their whole life.
An accent is something that is learned, so .74 Stability 2
one can change it if necessary.

Note. CFA = confirmatory factor analysis. Method was a principal component analysis with varimax
rotation and Kaiser normalization. Values smaller than ±.25 are not presented.

with a varimax rotation suggested either a solution with three or with two factors, but
in the three factors solution the two diagnosticity factors were accent diagnosticity
standard-coded items and reverse-coded items. This suggested that the way the items
were formulated was more important than whether they mentioned type or strength
diagnosticity. As this was not theoretically based, an analysis with two factors was
selected. It provided a solution with 49% explained variance and a clear eight-item
diagnosticity factor and a four-item stability factor (Table 2).

Confirmatory Factor Analyses. A confirmatory factor analysis was conducted for two
factors (accent diagnosticity and accent stability). The model had no discriminative
issues and an acceptable model fit. After the results of the exploratory factor analysis,
it was to be expected that reverse-coded items that shared a question structure would
160 Journal of Language and Social Psychology 39(1)

Figure 2. Results of confirmatory factor analysis with standardized regression coefficients of


the Accents Beliefs Scale in Study 2 (second half of the sample).
Note. Latent variables are shown in ellipses and observed variables in rectangles. All coefficients are
significant at p < .001 level.

have their error terms correlated. Indeed, modification indices suggested that some
error terms were correlated (Figure 2) and these were the reverse-coded items (see
supplemental material available online for the results before correlating the errors).
Correlating errors between factors can lead to misinterpretations of the results, but
correlating errors within one factor is a common practice that reflects a realistic view
that similar wording provokes the errors to be correlated (Bollen & Lennox, 1991;
Brown, 2014). After correlating the errors, the chi-square was significant, χ2 = 175.31,
p < .001, but the sample was large (n = 640). Other indices showed a good model fit:
comparative fit index (CFI) = .95, root mean square error of approximation (RMSEA)
= .06, goodness-of-fit index (GFI) = .96. All regression coefficients were significant
at p < .001 level (Figure 2). Additionally, the scale showed gender measurement
invariance (see supplemental material available online).

Reliability and Validity. For the whole sample, the reliability of the diagnosticity scale
was α = .79 and of the accent stability scale α = .81. The results clearly showed that
Hansen 161

Table 3. Correlations Between Accent Beliefs and Other Constructs in Study 2.

Accent Diagnosticity Accent Stability


Discriminant validity
Age −.10** .05
Gender (male) .06* −.03
Place of residence −.13*** .00
Education −.05 −.01
Motivation to respond unprejudiced −.17*** .03
Neuroticism −.05 .08*
Extraversion .06* −.03
Openness/Intellect −.09** −.12***
Self-esteem .05 −.06*
Convergent validity
SDO .28*** −.01
RWA .33*** .09**
Predictive validity
Prejudice immigrants .30*** .05
Prejudice refugees .35*** .06*
Distance refugees .33*** .05
Distance Ukrainians .27*** .07*
Distance Vietnamese .28*** .07**
Linguistic expectations to immigrants .25*** .05

*p < .05. **p < .01. ***p < .001.

diagnosticity and stability were not related to each other (r = .002), suggesting again
that these are independent constructs.
The study also explored relationships between accent beliefs and other variables
verifying different types of validity of the scale. Regarding the scales’ discriminant
validity, both types of diagnosticity were only weakly related to demographics, moti-
vation to respond without prejudice, personality measures, and self-esteem (Table 3).
The strongest (but still weak) correlations suggested that people who lived in bigger
cities perceived accents as less diagnostic than people living in towns or villages. Also,
people who were motivated to respond in an unprejudiced way perceived accents as
less diagnostic. Accent stability was even weaker related to demographics and person-
ality measures. The only relationship that was a bit stronger suggested that people who
were more open to experience (some personality theorists call this dimension intellect)
perceived accents as less stable.
When it comes to convergent validity, diagnosticity subscale was positively related to
RWA and SDO. Stability subscale was not related to SDO and its correlation with RWA
was very weak. It seems that accent beliefs, and especially diagnosticity beliefs, have
something in common with very general measures of prejudice and intergroup bias.
Regarding predictive validity,1 accent beliefs were related to various prejudice
measures. Especially the diagnosticity was related to general prejudice to immigrants
162 Journal of Language and Social Psychology 39(1)

and refugees, to social distance to refugees and two immigrant groups (Ukrainians and
Vietnamese), as well as to linguistic expectations to immigrants. Accent stability was
less strongly related to the aforementioned measures. It could be tested whether per-
ceiving accent as diagnostic for other traits might be related to prejudice and emotions,
while accent stability to more intellectual interpretations and explanations of nonna-
tive speakers’ accentst. Furthermore, accent stability might be less related to declara-
tive measures, but more to real evaluations of accented speakers.

Study 3
Study 3 was designed to further assess predictive validity of the ABS, this time on an
auditory material. The first goal of the study was to verify whether the two accent
subscales were related to evaluations of a nonnative speaker in terms of hirability and
perceived assimilation to the host society. The second goal was to demonstrate that the
scale adds to the prediction of evaluations of nonnative speakers beyond existing pre-
dictors of prejudice: SDO and RWA. The study was conducted as a part of a bigger
online survey. Participants listened to a recording of a Mandarin Chinese accented
speaker talking in English about why she would be a good candidate for a lower man-
ager position. As strong or weak accent can be interpreted differently, it is important to
note that the speaker’s accent was strong.
It was expected that people believing in accent diagnosticity would evaluate the
speaker as less hirable and as less assimilated. As noted in the introduction, there is
conflicting reasoning and evidence regarding the role of stability beliefs on evalua-
tions of others. From naïve theories of intelligence and other human attributes, one
would predict that stability of an accent when listening to a strongly accented speaker
does not give hope that this person will reduce it in the future (Dweck, 1999). However,
there seems to be an implicit essentialist part of this claim—that an accent is not only
stable, but is also an important trait of a person that tells us something about them. As
the current scale divorced diagnosticity from stability, other explanation might be
more probable. The other explanation comes from research on stigma and from unpub-
lished research on accents (Gluszek & Dovidio, 2010b; Hansen & Dovidio, 2018),
which both suggest that people who believe that accent is something stable would not
blame nonnative speakers for having an accent and would evaluate them more posi-
tively than people who believe that the speakers could have changed their accent. The
current scale allows testing such alternative hypotheses. In the present study, it was
expected that perceived accent stability would be related to higher hirability and higher
perceived assimilation of the speaker.
When it comes to SDO and RWA, both are general predictors of prejudice, but some
authors show that attitudes toward dangerous groups are more strongly related to RWA
and attitudes toward derogated groups to SDO (Duckitt & Sibley, 2007). It can depend
on the studied group, but a Chinese nonnative job candidate for a lower-manager posi-
tion was likely to be perceived as derogated rather than dangerous. Thus, SDO was
expected to predict evaluations more strongly than RWA. Accent beliefs were expected
to be even better predictors and to add explained variance in the evaluations even when
controlling for SDO and RWA.
Hansen 163

Participants
Participants were adults living in the United States recruited through Amazon
Mechanical Turk to complete an online survey in exchange for a small monetary com-
pensation. After excluding nonnative English speakers, people who did not pass audio
check, and those who filled out nonsense (random letters and text clearly copied from
the Internet) in an open-ended question, the sample consisted of 159 participants (Mage
= 36.37, SD = 11.13, age range: 19-69 years, 45% women, 72% White, 14% Black,
4% Latino/a, 5% Asian, 5% other).

Method
In the beginning, participants answered similar short SDO (four items, α = .87) and
RWA (six items, α = .83) scales as in earlier studies. Then, after an audio check, they
were asked to imagine that a company is looking to fill a lower management position.
They were asked to listen to a brief audio clip of an applicant during a job interview
and form an impression of the person. The audio clip was recorded by a Chinese
woman in her late 20s. An earlier test of the recording (n = 32) showed that the accent
was perceived as strong (M = 5.97, SD = 1.33, on a 7-point scale) and as clearly Asian
(M = 6.13, SD = 1.43, on a 7-point scale). The clip lasted about 1 minute and the
applicant was saying, why she would be a good fit for the job. Participants rated the
applicant’s hirability (three items, α = .92, e.g., I would hire the applicant for this
position) and amount of assimilation (three items, α = .90, e.g., The candidate tries to
integrate into American society) on a scale from 1 = strongly disagree/definitely not
to 7 = strongly agree/definitely yes. As the study was a part of a larger survey, other
questions followed, and after those participants answered the items of the ABS (see the
appendix, α = .83 for diagnosticity, α = .63 for stability). The study ended with a few
demographic questions, debriefing, and a space to leave comments.

Results and Discussion


Correlations. A correlational analysis showed that, again, accent diagnosticity was not
related to accent stability (Table 4). The results showed that the more participants
believed in accent diagnosticity, the less they recommended hiring the applicant and
the less they perceived her as assimilated. In opposition to this, the more participants
believed that accent is something stable and difficult to change, the more hirable and
assimilated they perceived the immigrant who spoke with a strong accent. The two
subscales seemed to be correlated to a different extent with these outcomes. Diagnos-
ticity was descriptively stronger related to a behavioral intention of hiring the speaker
and stability was stronger related to the perception of the person as well-assimilated,
but the differences were not large enough to reach significance: diagnosticity, z =
−0.97, p = .17; stability, z = −1.11, p = .14 (Lenhard & Lenhard, 2014).
Furthermore, the higher the participants’ SDO and RWA, the more they perceived
accents as diagnostic. Surprisingly, accent stability beliefs were not related to participants’
164 Journal of Language and Social Psychology 39(1)

Table 4. Correlations Between Accent Beliefs and Dependent Variables in Study 3.

Diagnosticity Stability Hirability Assimilation SDO


Stability −.04
Hirability −.25** .16*
Assimilation −.18* .24** .57**
SDO .50*** .00 −.15† −.27**
RWA .39*** .23** .00 .00 .40***
†p < .10. *p < .05. **p < .01. ***p < .001.

Table 5. Results of a Multiple Stepwise Regression With SDO and RWA (First Step), as well
as Accent Diagnosticity and Accent Stability (Second Step) Predicting Hirability and Perceived
Assimilation of a Nonnative Speaker.

Hirability Perceived Assimilation

b p 95% CI for b β b p 95% CI for b β


Intercept 5.07 <.001 [4.50, 5.63] — 5.31 <.001 [4.82, 5.81] —
SDO −0.18 .04 [−0.36, −0.01] −.18 −0.29 <.001 [−0.44, −0.14] −.32
RWA 0.08 .34 [−0.09, 0.25] .08 0.13 .09 [−0.02, 0.27] .14
Intercept 5.08 <.001 [4.17, 6.00] — 4.72 <.001 [3.92, 5.53] —
SDO −0.07 .48 [−0.25, 0.12] −.07 −0.24 .01 [−0.40, −0.07] −.26
RWA 0.10 .26 [−0.07, 0.28] .10 0.09 .26 [−0.07, 0.24] .10
Accent −0.30 .01 [−0.51, −0.08] −.25 −.09 .37 [−0.28, 0.10] −.08
diagnosticity
Accent stability 0.17 .10 [−0.03, 0.37] .13 0.25 .01 [0.07, 0.43] .22

Note. n = 156. CI = confidence interval.

SDO, but were in a positive manner related to RWA—more authoritarian/traditional par-


ticipants perceived accents as more stable and difficult to change. As could be expected,
RWA was related to SDO, but was not related to perceived hirability or assimilation of the
nonnative speaker (who was rather from a derogated than a dangerous social group).

Regressions. In order to further test predictive validity of the ABS, two hierarchical
multiple regressions were used, one with hirability as dependent variable and one with
perceived assimilation. In each of the regressions, SDO and RWA were entered in the
first step and in the second step accent diagnosticity and accent stability were added.
Tests of collinearity indicated that multicollinearity was not a concern, tolerance = .69
to .93, variance inflation factor (VIF) = 1.08 to 1.45.
For hirability, the final model was significant, R2 = .10, F(4, 152) = 3.99, p = .004,
and it improved when adding accent beliefs, ΔR2 = .07, F(2, 152) = 5.61, p = .004.
As indicated in Table 5, in the first step higher SDO was a significant predictor of
Hansen 165

lower hirability, but the RWA was not. In the second step, SDO became insignificant,
but in turn accent diagnosticity was now an important predictor of hirability. Accent
stability was not a significant predictor (p = .10) with this sample size, but the stan-
dardized coefficient was not negligible (β = .13).
For perceived assimilation, the final model was significant, R2 = .14, F(4, 152) =
6.02, p < .001, and it improved when adding accent beliefs, ΔR2 = .05, F(2, 152) =
4.59, p = .01. In the first step, similarly as for hirability, SDO was a significant predic-
tor of hirability, but RWA was not. In the second step, coefficient for SDO was reduced,
but still significant. In this second step, accent stability was an important predictor of
hirability, but accent diagnosticity was not.
In sum, SDO itself was a good predictor of hirability and perceived assimilation
and RWA was not, which is in line with previous research and reasoning that a nonna-
tive job candidate would represent rather a derogated than a dangerous social group
(Duckitt & Sibley, 2007). Crucially, accent beliefs predicted the evaluations beyond
the general out-group bias measures: SDO and RWA. Accent diagnosticity was the
only significant predictor of hirability. Confirming the expectations about the scale, if
a person perceives accent as diagnostic and encounters an immigrant who speaks with
a strong foreign accent—she or he will not want to hire the candidate. Accent stability,
in turn, was a better predictor than accent diagnosticity in the case of perceived assimi-
lation of the speaker. The more people believed that accent is something stable that
cannot be changed with effort, the more they believed that the speaker (despite the
accent) is well assimilated. This resembles results of Study 2 and suggests that accent
diagnosticity may serve well as a predictor of judgments or behavioral intentions that
can be prejudice-driven and emotionally loaded. Accent stability may better serve as a
predictor when it comes to more cognitively loaded outcomes such as explaining and
interpreting behavior or traits of nonnative speakers.

General Discussion
The developed ABS is a measure of how people perceive accent diagnosticity (includ-
ing accent strength and accent type diagnosticity) and accent stability. Thanks to its
short length, the scale can be used in various studies, also in large-scale surveys. It
could be also used as a manipulation check or an intervention effectiveness measure in
studies trying to change people’s beliefs about accents. The results of the conducted
studies show good (in Study 3 for stability—acceptable) reliabilities of the subscales,
their discriminatory as well as convergent and predictive validity. The studies show
that the scale is related to, but at the same time distinct from, more general measures
of prejudice toward out-groups. It is also related to perceptions of immigrants and
refugees in general, as well as to evaluations of specific accented speakers whom one
can hear.
Some earlier research studying people’s beliefs about the world has included in the
same scales, and even in the same items, an aspect of diagnosticity and an aspect of
stability (Dweck et al., 1995; Levy et al., 1998). The current studies show that, at least
for the perception of accent in speech, it is important to separate one from the other. In
166 Journal of Language and Social Psychology 39(1)

the current research, diagnosticity and stability items formed separate factors and the
created subscales were not related to each other in none of the three studies. Also, the
two subscales were related to evaluations of a nonnative speaker in the opposite direc-
tions and predictive of different inferences about the speaker.
In general, the results showed that accent diagnosticity was related to prejudice and
speaker evaluations, as would be predicted from essentialism research that the scale
was inspired by (Bastian & Haslam, 2006). In the current studies, people believing that
accent is diagnostic of (i.e., is a good proxy for) other traits of a person, evaluated a
strongly accented speaker more negatively—as less hirable. This resembles findings
on visual and auditory gaydar showing not only that visual and vocal cues can evoke
stereotyping but also that the perceiver’s beliefs influence these effects (Fasoli,
Hegarty, Maass, & Antonio, 2018; Stern, West, Jost, & Rule, 2013). The diagnosticity
of accents can be seen as a possible moderator of the negative evaluations of accented
speakers found in the literature: the accent bias might be most prevalent among those
who believe that accent is diagnostic for other traits. The relationships between the
diagnosticity subscale and other variables or constructs were strong and replicated in
multiple studies.
The subscale of stability showed high reliability in all studies but Study 3, and was
related to other variables. However, these relationships were often weaker than in the
case of diagnosticity. The results showed that people who perceived accent to be
something stable that cannot be changed with an effort, perceived a nonnative speaker
better, especially in terms of the speaker’s assimilation to the host society. This is in
line with research on stigma showing that when a stigmatizing illness or trait is per-
ceived as caused by external factors, people with the stigma are evaluated more posi-
tively, as it is not their fault that they have the illness or trait (Crocker et al., 1993;
Gluszek & Dovidio, 2010b).

Conclusion and Future Directions


The ABS can be used to better understand perceptions and evaluations of nonnative
speakers. It could also be applied in different cultural contexts to study potential dif-
ferences in the strength of the relationships between, for example, perception of
accent as stable over lifetime and perceived hirability of a nonnative speaker. It
could be also assessed whether beliefs about accents influence evaluations differ-
ently for high-status (e.g., British, German) accents and low-status (e.g., Arabic,
Russian) accents.
Accents can have a higher or lower status, but also accent strength could shift peo-
ple’s evaluations and the role accent beliefs play in these evaluations. If someone
speaks with a relatively strong accent (as in Study 3) and the perceiver believes that
accents are diagnostic or that accents can be changed, then the speaker would be eval-
uated negatively. However, when the listener believes the same in general, but the
speaker speaks with a very mild accent, then it can rather mean that the hardly notice-
able accent is a cue for the speaker’s positive traits (willing to assimilate, hardworking,
and patient). The ABS can be used to study all these fascinating and socially relevant
Hansen 167

processes. It can help understand them and, as a consequence, prevent discrimination


and promote appreciation of others’ language skills.

Appendix
Items of the Accent Beliefs Scale.

English version Polish version


Diagnosticity
Accent strength
1. The strength of an accent in one’s 1. Siła akcentu w mowie świadczy o
speech is a sign of their personality. cechach charakteru tej osoby.
2. From someone’s strong or weak 2. Słysząc czyjś silny lub słaby akcent,
accent one can infer many things można wiele na temat tej osoby
about the speaker. wywnioskować.
3. (r) The strength of a person’s accent 3. (r) To z jak silnym akcentem ktoś mówi
has nothing to do with what kind of nie ma związku z tym, jaka ta osoba jest.
person they are. 4. (r) Nie można wnioskować o inteligencji
4. (r) One cannot draw conclusions danej osoby na podstawie siły akcentu, z
about someone’s intelligence based jakim mówi.
on the strength of their accent.
Accent type
5. It is possible to tell how someone will 5. Słysząc, jaki ktoś ma akcent, można się
act by hearing their accent. domyślić, jak ta osoba się zachowa.
6. The type of accent in a person’s 6. To, z jakim akcentem ktoś mówi, jest
speech is an important trait. ważną cechą tej osoby.
7. (r) The type of accent one has does 7. (r) Rodzaj akcentu w mowie niewiele
not tell much about the speaker. mówi o danej osobie.
8. (r) One cannot learn much about 8. (r) Na podstawie tego, z jakim akcentem
someone on the basis of their accent. ktoś mówi, nie można się wiele o tej
osobie dowiedzieć.
Stability
9. (r) People are capable of eliminating 9. (r) Ludzie są w stanie wyeliminować swój
their accent. akcent.
10. (r) An accent is something that 10. (r) Akcent to tylko coś nabytego i można
is learned, so one can change it if go zmienić jeśli potrzeba.
necessary. 11. Jeśli ktoś się wychował w jednym języku,
11. If someone was raised in one to będzie miał jego akcent przez całe
language, they will have its accent życie.
their whole life. 12. Każdy człowiek ma swój akcent i choćby
12. Everyone has an accent, and they się starał, nie będzie mógł go zmienić.
cannot change it, even if they tried.

Acknowledgments
I wish to thank Karolina Żelechowska, Olga Kuzawińska, and Klaudia Pączkowska for their
help in preparing and conducting the studies. Thanks to Monika Prusik for statistical consulta-
tion and to the Center for Research on Prejudice as well as the editors and anonymous reviewers
for their feedback on my work.
168 Journal of Language and Social Psychology 39(1)

Declaration of Conflicting Interests


The author declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

Funding
The author disclosed receipt of the following financial support for the research, authorship, and/
or publication of this article: This work was supported by the European Commission (MSCA-
RISE CoLing, grant agreement ID: 778384) and by the Polish National Science Centre (Fuga
DEC-2013/08/S/HS6/00573).

ORCID iD
Karolina Hansen https://orcid.org/0000-0002-1556-4058

Supplemental Material
Supplemental material for this article is available online.

Note
1. It should be noted that it is difficult to draw a very clear line between convergent and
predictive validity. Due to the study being a part of a longer panel survey, only declarative
measures were used and some of them that are presented under predictive validity might be
also seen as belonging to convergent validity. In Study 3, there was a possibility of present-
ing real recordings of accented speakers and measuring predictive validity more directly.

References
Allport, G. W. (1954). The nature of prejudice. Oxford, England: Addison-Wesley.
Altemeyer, B. (1981). Right-wing authoritarianism. Winnipeg, Manitoba, Canada: University
of Manitoba Press.
Bastian, B., & Haslam, N. (2006). Psychological essentialism and stereotype endorsement.
Journal of Experimental Social Psychology, 42, 228-235. doi:10.1016/j.jesp.2005.03.003
Bogardus, E. S. (1926). Social distance in the city. Proceedings and Publications of the
American Sociological Society, 20, 40-46.
Bollen, K. A., & Lennox, R. (1991). Conventional wisdom on measurement: A structural equa-
tion perspective. Psychological Bulletin, 110, 305-314.
Brown, T. A. (2014). Confirmatory factor analysis for applied research. New York, NY:
Guilford Press.
Carr, P. B., Dweck, C. S., & Pauker, K. (2012). “Prejudiced” behavior without prejudice? Beliefs
about the malleability of prejudice affect interracial interactions. Journal of Personality and
Social Psychology, 103, 452-471. doi:10.1037/a0028849
Cattell, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research,
1, 245-276.
Crandall, C. S. (1994). Prejudice against fat people: Ideology and self-interest. Journal of
Personality and Social Psychology, 66, 882-894. doi:10.1037/0022-3514.66.5.882
Crandall, C. S., & Moriarty, D. (1995). Physical illness stigma and social rejection. British
Journal of Social Psychology, 34, 67-83. doi:10.1111/j.2044-8309.1995.tb01049.x
Hansen 169

Creswell, J. W. (2013). Research design: Qualitative, quantitative, and mixed methods


approaches. London, England: Sage.
Crocker, J., Cornwell, B., & Major, B. (1993). The stigma of overweight: Affective conse-
quences of attributional ambiguity. Journal of Personality and Social Psychology, 64, 60-
70. doi:10.1037/0022-3514.64.1.60
Devine, P. G., Plant, E. A., Amodio, D. M., Harmon-Jones, E., & Vance, S. L. (2002). The
regulation of explicit and implicit race bias: The role of motivations to respond without
prejudice. Journal of Personality and Social Psychology, 82, 835-848. doi:10.1037//0022-
3514.82.5.835
Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The Mini-IPIP
scales: Tiny-yet-effective measures of the Big Five factors of personality. Psychological
Assessment, 18, 192-203. doi:10.1037/1040-3590.18.2.192
Dovidio, J. F., & Gaertner, S. L. (2010). Intergroup bias. In S. T. Fiske, D. T. Gilbert, & G.
Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 1084-1121). Hoboken,
NJ: John Wiley.
Dovidio, J. F., & Gluszek, A. (2012). Accents, nonverbal behavior, and intergroup bias. In
H. Giles (Ed.), The handbook of intergroup communication (pp. 87-99). New York, NY:
Routledge.
Dovidio, J. F., Gluszek, A., John, M.-S., Ditlmann, R., & Lagunes, P. (2010). Understanding
bias toward Latinos: Discrimination, dimensions of difference, and experience of exclu-
sion. Journal of Social Issues, 66, 59-78. doi:10.1111/j.1540-4560.2009.01633.x
Driscoll, D. L., Appiah-Yeboah, A., Salib, P., & Rupert, D. J. (2007). Merging qualitative
and quantitative data in mixed methods research: How to and why not. Ecological and
Environmental Anthropology, 3(1), 19-28.
Duckitt, J., & Sibley, C. G. (2007). Right wing authoritarianism, social dominance orientation
and the dimensions of generalized prejudice. European Journal of Personality, 21, 113-
130. doi:10.1002/per.614
Dunton, B. C., & Fazio, R. H. (1997). An individual difference measure of motivation to
control prejudiced reactions. Personality and Social Psychology Bulletin, 23, 316-326.
doi:10.1177/0146167297233009
Dweck, C. S. (1999). Self-theories: Their role in motivation, personality and development.
Philadelphia, PA: Psychology Press.
Dweck, C. S., Chiu, C., & Hong, Y. (1995). Implicit theories and their role in judgments
and reactions: A world from two perspectives. Psychological Inquiry, 6, 267-285.
doi:10.2307/1448940
Fasoli, F., Hegarty, P., Maass, A., & Antonio, R. (2018). Who wants to sound straight? Sexual
majority and minority stereotypes, beliefs and desires about auditory gaydar. Personality
and Individual Differences, 130, 59-64. doi:10.1016/j.paid.2018.03.046
Gluszek, A. (2010). A word worth a thousand pictures: Non-native accents and their strength in
perceptions of stigmatization, communication challenges, and social belonging (Doctoral
dissertation), Yale University, New Haven, CT.
Gluszek, A., & Dovidio, J. F. (2010a). Speaking with a nonnative accent: Perceptions of bias,
communication difficulties, and belonging in the United States. Journal of Language and
Social Psychology, 29, 224-234. doi:10.1177/0261927X09359590
Gluszek, A., & Dovidio, J. F. (2010b). The way they speak: A social psychological perspective
on the stigma of nonnative accents in communication. Personality and Social Psychology
Review, 14, 214-237. doi:10.1177/1088868309359288
170 Journal of Language and Social Psychology 39(1)

Gluszek, A., & Hansen, K. (2013). Language attitudes in the Americas. In H. Giles & B. Watson
(Eds.), The social meanings of languages, dialects, and accents: An international perspective
(pp. 26-44). New York, NY: Peter Lang.
Halperin, E., Russell, A. G., Trzesniewski, K. H., Gross, J. J., & Dweck, C. S. (2011). Promoting
the Middle East peace process by changing beliefs about group malleability. Science, 333,
1767-1769. doi:10.1126/science.1202925
Hansen, K., & Dovidio, J. F. (2016). Social dominance orientation, nonnative accents, and
hiring recommendations. Cultural Diversity & Ethnic Minority Psychology, 22, 544-551.
doi:10.1037/cdp0000101
Hansen, K., & Dovidio, J. F. (2018). Beliefs in eliminating accent. Retrieved from https://
www.researchgate.net/publication/329984341_Beliefs_in_eliminating_accent/
link/5c275e9692851c22a34d2175/download
Hansen, K., Rakić, T., & Steffens, M. C. (2017). Competent and warm? How mismatching
appearance and accent influence first impressions. Experimental Psychology, 64, 27-36.
doi:10.1027/1618-3169/a000348
Haslam, N., Rothschild, L., & Ernst, D. (2000). Essentialist beliefs about social categories.
British Journal of Social Psychology, 39, 113-127. doi:10.1348/014466600164363
Hopkins, D. J. (2015). The upside of accents: Language, inter-group difference, and attitudes
toward immigration. British Journal of Political Science, 45, 531-557. Retrieved from
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1879965
Kinzler, K. D., Shutts, K., Dejesus, J., & Spelke, E. S. (2009). Accent trumps race in guiding chil-
dren’s social preferences. Social Cognition, 27, 623-634. doi:10.1521/soco.2009.27.4.623
Lachowicz-Tabaczek, K. (2004). Potoczne koncepcje świata i natury ludzkiej [Lay theories of
the world and human nature]. Gdańsk, Poland: Gdańskie Wydawnictwo Psychologiczne.
Larman, C., & Basili, V. R. (2003). Iterative and incremental developments: A brief history.
Computer, 36(6), 47-56. doi:10.1109/MC.2003.1204375
Lenhard, W., & Lenhard, A. (2014). Hypothesis tests for comparing correlations. Retrieved
from https://www.psychometrica.de/correlation.html
Lev-Ari, S., & Keysar, B. (2010). Why don’t we believe non-native speakers? The influence
of accent on credibility. Journal of Experimental Social Psychology, 46, 1093-1096.
doi:10.1016/j.jesp.2010.05.025
Levy, S. R., Stroessner, S. J., & Dweck, C. S. (1998). Stereotype formation and endorsement:
The role of implicit theories. Journal of Personality and Social Psychology, 74, 1421-1436.
doi:10.1037/0022-3514.74.6.1421
Lou, N. M., & Noels, K. A. (2016). Changing language mindsets: Implications for goal orienta-
tions and responses to failure in and outside the second language classroom. Contemporary
Educational Psychology, 46, 22-33. doi:10.1016/j.cedpsych.2016.03.004
Lou, N. M., & Noels, K. A. (2017). Measuring language mindsets and modeling their relations
with goal orientations and emotional and behavioral responses in failure situations. Modern
Language Journal, 101, 214-243. doi:10.1111/modl.12380
Molden, D. C., & Dweck, C. S. (2006). Finding “meaning” in psychology: A lay theo-
ries approach to self-regulation, social perception, and social development. American
Psychologist, 61, 192-203. doi:10.1037/0003-066x.61.3.192
Moyer, A. (2004). Age, accent and experience in second language acquisition. Clevedon,
England: Multilingual Matters.
Ng, S. H. (2007). Language-based discrimination: Blatant and subtle forms. Journal of Language
and Social Psychology, 26, 106-122. doi:10.1177/0261927X07300074
Hansen 171

Oppenheimer, D. M. (2008). The secret life of fluency. Trends in Cognitive Sciences, 12, 237-
241. doi:10.1016/j.tics.2008.02.014
Pietraszewski, D., & Schwartz, A. (2014). Evidence that accent is a dimension of social cat-
egorization, not a byproduct of perceptual salience, familiarity, or ease-of-processing.
Evolution & Human Behavior, 35, 43-50. doi:10.1016/j.evolhumbehav.2013.09.006
Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social dominance orientation:
A personality variable predicting social and political attitudes. Journal of Personality and
Social Psychology, 67, 741-763. doi:10.1037/0022-3514.67.4.741
Roessel, J., Schoel, C., & Stahlberg, D. (2018). What’s in an accent? General spontaneous
biases against nonnative accents: An investigation with conceptual and auditory IATs.
European Journal of Social Psychology, 48, 535-550. doi:10.1002/ejsp.2339
Rosenberg, M. (2015). Society and the adolescent self-image. Princeton, NJ: Princeton
University Press.
Ryan, E. B. (1983). Social psychological mechanisms underlying native speaker evaluations
of non-native speech. Studies in Second Language Acquisition, 5, 148-159. doi:10.1017/
S0272263100004824
Scovel, T. (2000). A critical review of the critical period research. Annual Review of Applied
Linguistics, 20, 213-223. doi:10.1017/S0267190500200135
Stern, C., West, T., Jost, J., & Rule, N. (2013). The politics of gaydar. Journal of Personality
and Social Psychology, 104, 520-541. doi:10.1037/a0031187
Świderska, A., Winiewski, M., & Hansen, K. (2017). Przemoc jako rozwiązanie? Napływ
uchodźców w opiniach Polaków [Violence as a solution? Poles opinions about the ref-
ugee influx]. Retrieved from http://cbu.psychologia.pl/uploads/images/foto/Raport_
WinSwiHan2016_clean.pdf
Weiner, B., Perry, R. P., & Magnusson, J. (1988). An attributional analysis of reactions to
stigmas. Journal of Personality and Social Psychology, 55, 738-748. doi:10.1037/0022-
3514.55.5.738
Wertheimer, A. (1983). Jobs, qualifications, and preferences. Ethics, 94, 99-112.
doi:10.1086/292512
Zick, A. (2011). Intolerance, prejudice and discrimination: A European report. Berlin,
Germany: Friedrich-Ebert-Stiftung.

Author Biography
Karolina Hansen is an assistant professor at the Psychology Faculty of the University of
Warsaw. Her research interests are in the fields of social psychology. sociolinguistics, and
cross-cultural psychology, and include topics such as language and accent attitudes, gender-fair
language, linguistic purism, and cross-cultural differences in social cognition.

You might also like