Professional Documents
Culture Documents
net/publication/327920443
CITATIONS READS
55 1,618
1 author:
Yasser Teimouri
Georgetown University
17 PUBLICATIONS 584 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Yasser Teimouri on 27 September 2018.
This article aims to introduce shame and guilt from social psychology into second language acquisition
(SLA), to validate their psychometric measurement, and to examine their effects on second language
(L2) learners’ motivation and language achievements. In Study 1, the prevalence of shame and guilt
reactions in L2 settings was explored. Moreover, the Second Language Test of Shame and Guilt Affect
(L2-TOSGA) was developed and validated to measure L2 learners’ individual differences in terms of
proneness to shame and guilt during L2 learning. The results of qualitative and quantitative analyses
evidenced the pervasiveness of shame and guilt in an L2 context, and attested to the reliability, stability,
and validity of L2-TOSGA subscales. In Study 2, the effects of learners’ shame and guilt reactions on
their motivation and language achievements were probed. The results showed that shame strongly but
negatively affected L2 learners’ motivation and language achievements, whereas guilt had positive effects
on their motivation and language achievements.
Keywords: shame; guilt; L2 motivation; language achievement
FOR DECADES, RESEARCH INTO SECOND made to provide a theoretical framework describ-
Language (L2) learners’ emotions witnessed a ing the underlying social, cognitive, affective, and
theoretical lull after the introduction of language motivational components of each emotion. This
anxiety in the field of second language acquisition research tradition in SLA, however, is undergoing
(SLA). The disproportionate volume of research a rapid change with a renewed interest directed
on language anxiety (for reviews, see Gkonou, at both extending and deepening our under-
Daubney, & Dewaele, 2017; Horwitz, 2017; Mac- standing of L2 learners’ emotions during L2
Intyre, 2017) stymied research developments in learning.
other emotions that may be experienced during A recent wave of research has emerged stretch-
L2 learning. In addition, the scope of emotion re- ing the emotional spectrum that L2 learners may
search itself has been traditionally conceptualized experience, such as amusement, anger, contempt,
within broad categories of negative and positive desire, disgust, embarrassment, enjoyment, grati-
emotions. That is, a variety of negative emotions, tude, hate, interest, joy, love, pride, sadness, and
such as anxiety, embarrassment, fear, nervous- serenity (e.g., Dewaele, 2015; Dewaele & MacIn-
ness, sadness, and shame, were lumped together tyre, 2014, 2016; Dewaele et al., 2017; MacIntyre
under the rubric of language anxiety (e.g., Gard- & Gregersen, 2012; MacIntyre, Gregersen, & Mer-
ner, 1985; Horwitz, Horwitz, & Cope, 1986; Mac- cer, 2016; MacIntyre & Mercer, 2014; MacIntyre &
Intyre & Gardner, 1994a), whereas a list of positive Vincze, 2017; Motha & Lin, 2014; Ross & Stracke,
affects, such as creativity, enjoyment, fun, interest, 2017; Teimouri, 2017). Inspired by a complex dy-
pride, and support, was assembled under the sin- namic systems approach in SLA (Larsen–Freeman
gle heading of language enjoyment (e.g., Dewaele & Cameron, 2008), a few studies have also inves-
& MacIntyre, 2014). No further attempts were tigated the simultaneous interactions of L2 learn-
ers’ emotions (e.g., Dewaele et al., 2016) as well as
The Modern Language Journal, 0, 0, (2018) their fluctuations over time (e.g., Gregersen, Mac-
DOI: 10.1111/modl.12511 Intyre, & Meza, 2014).
0026-7902/18/1–21 $1.50/0 In addition, the need for new theoretical per-
C National Federation of Modern Language Teachers spectives in emotions research has been high-
Associations lighted (e.g., Dewaele & MacIntyre, 2014; Gkonou
2 The Modern Language Journal 0 (2018)
et al., 2017; MacIntyre & Vincze, 2017; Şimşek & Using a mixed-method design, then, this article
Dörnyei, 2017). For instance, Dewaele and Mac- aims to provide evidence on the occurrence of
Intyre (2014) have cast doubt on the adequacy shame and guilt during L2 learning and to empir-
of the positive versus negative approach underly- ically examine their effects on L2 learners’ moti-
ing emotions research in SLA. Alternatively, they vation and language achievements.
have called for more specific theories targeting Furthermore, a scenario-based questionnaire
the functions of emotions. The multifaceted na- will be developed and validated to specifically
ture of emotions has been emphasized in some measure L2 learners’ shame and guilt experiences
other studies while documenting the emotional in L2 settings. The need to develop L2-specific
experiences of learners in L2 settings (e.g., Imai, measures has been emphasized in past emotion
2010; Motha & Lin, 2014; Ross & Stracke, 2017; research in SLA (e.g., Horwitz, Horwitz, & Cope,
Şimşek & Dörnyei, 2017; Teimouri, 2017). The so- 1986; Horwitz, 2001, 2017; MacIntyre & Gardner,
cial turn in SLA (see Block, 2003) has also led 1991; MacIntyre, 2017). While the use of general
to the emergence of post-structural/discursive ap- instruments from psychology in early anxiety re-
proaches toward exploring emotions as socially search in SLA generated inconsistent findings,
constructed notions (e.g., Benesch, 2012, 2017). the development and application of language-
The present article aims to introduce shame specific anxiety questionnaires—for instance, For-
and guilt from social psychology into the realm of eign Language Classroom Anxiety (FLCA) (Hor-
SLA and to examine their effects on students’ mo- witz et al., 1986)—led to more consistent findings
tivation and language achievement. Among the regarding the role of anxiety in language achieve-
emotions that L2 learners are likely to feel dur- ments of students (MacIntyre, 2017; Teimouri,
ing language learning are the painful emotions of Goetze, & Plonsky, 2019). Although a plethora of
shame and guilt. For instance, when experiencing L2 anxiety measures have been developed since
a negative event, such as receiving a low grade on the conceptualization of L2 anxiety as a situation-
an exam, receiving negative feedback in the pres- specific emotion, there is a scarcity of measures
ence of others, or failing to communicate one’s in- for assessing other emotions of learners in L2 sit-
tended message during a social interaction, an L2 uations. The need to fill this measurement gap
learner may feel ashamed or guilty, especially if he gains even more significance given the rapid ex-
or she believes others are passing negative judg- tension of emotion research in SLA.
ments about his or her personal characteristics Although scenario-based questionnaires have
or behaviors. Of critical note, L2 learners’ lack of been previously used as data collection tools in
proficiency in the target language and its under- SLA research (e.g., Borg, 2009; Oxford, 2017),
lying sociocultural norms and standards may lead the present article is the first to develop and use
to their occupation of a low social status among a scenario-based questionnaire to measure spe-
the members of the new target community. Such cific emotions in L2 settings. Compared to pre-
an undesirable low social status poses a chronic vious emotion questionnaires used in SLA (e.g.,
threat to L2 learners’ social self, which may even- Dewaele & MacIntyre, 2014; Gardner, 1985; Hor-
tually generate feelings of shame and guilt. witz et al., 1986; MacIntyre & Gardner, 1994a;
While research on shame and guilt is flourish- MacIntyre & Vincze, 2017; Spielberger, 1983),
ing in social and educational psychology for their a scenario-based questionnaire provides a more
profound effects on one’s motivation, interper- trustworthy instrument for assessing L2 learners’
sonal behaviors, and academic performance (e.g., emotions. Each scenario presents a unique con-
Tracy, Robins, & Tangney, 2007; Turner, Husman, text. A context, though not real, is perceived as
& Schallert, 2002), these two emotions have not real by L2 learners through the power of visualiza-
received proper attention in SLA. The need to fill tion. Not only can L2 learners imagine themselves
this research gap becomes even more important in the situations, but they can also feel their pres-
given the fact that few studies have provided suf- ence in them (e.g., Dörnyei, 2014). As a result,
ficient evidence on the pervasiveness of shame in they may evaluate and express their emotional
L2 context and its negative effects on L2 learn- reactions in those situations more accurately. A
ers (e.g., Cook, 2006; Galmiche, 2017; Teimouri, scenario-based questionnaire allows for the mea-
2017; Wang, 2016). Although a few studies have suring of various emotions simultaneously; it also
addressed the negative role of shame, no research allows for the exploration of the dynamic interac-
report to date has closely examined the role of tions between different emotions within a partic-
guilt in SLA. As we will see later, research into ular context and examination of their combined
guilt offers significant pedagogical implications effects. The use of scenario-based questionnaires,
because of its positive role in students’ motivation. thus, advances research in line with the new
Yasser Teimouri 3
complex, dynamic turn in SLA, which focuses on 1995). Upon receiving a low grade on an L2 writ-
the interaction of learner factors within a partic- ing assignment, for instance, a shamed learner
ular context (e.g., Larsen–Freeman & Cameron, might think “If and only if I was not such a poor
2008). L2 writer.” The shamed learner may feel exposed;
that is, her perceived defects or shortcomings
L2 LEARNERS’ FEELINGS OF SHAME AND have been exposed before an audience. The audi-
GUILT ence, of particular note, does not need to be real,
or physically present during the shame-inducing
Shame and guilt belong to the family event (e.g., Leary, 2007; Tangney et al., 1996); its
of self-conscious emotions (shame, guilt, presence, in fact, is often imagined by L2 learners.
embarrassment,1 and pride) as opposed to the Not only does a learner envision the presence of
family of so-called basic emotions (anger, fear, others while reflecting on her characteristic self
sadness, enjoyment, disgust, and surprise) (Ek- and behaviors but she also envisions their reac-
man, 2003; Izard, 1991; Tracy et al., 2007). From tions, thoughts, evaluations, and judgments. Such
a functional perspective, self-conscious emotions negative thoughts, eventually, will trigger feelings
are thought to have evolved through natural selec- of being small, inferior, powerless, and hopeless
tion to facilitate the attainment of complex social (Tangney & Dearing, 2003; Tangney et al., 1996).
goals (or identity goals), such as maintenance or Shame motivates avoidance or withdrawal re-
enhancement of one’s social status, gaining social sponses (Lewis, 1971; Nathanson, 1992; Tangney
acceptance, or avoiding group rejection (Leary, et al., 1996). As noted, since shame involves a feel-
2007; Tracy & Robins, 2004). On the other hand, ing of exposure of one’s defective, unattractive
basic emotions have evolved to facilitate one’s self before a real/imagined audience, a shamed
survival and reproductive needs (Lazarus, 1991).2 learner is more likely to report a desire to flee
Self-conscious emotions are often referred to the shame-inducing situation. In other words, the
as social emotions or other-oriented emotions shamed learner seeks a way to hide from others
(Tracy et al., 2007). This is because a person’s in order to escape the overwhelming feelings of
active appraisal of other people’s thoughts, judg- shame (Tangney & Dearing, 2003). Avoidance or
ments, and feelings plays a substantive role in his withdrawal responses associated with feelings of
or her feelings of shame, guilt, embarrassment, shame may reveal themselves either during actual
and pride (e.g., Gruenewald, Dickerson, & Ke- use of the L2, for instance, when L2 learners move
meny, 2007). According to Leary (2007), “funda- away from a conversation or a group activity, or at
mentally, self-conscious emotions evolved not to L2 learners’ intentional levels, for instance, when
respond to people’s private evaluations of them- L2 learners show less desire to participate in L2
selves but rather to regulate their interactions and class activities or to use the L2 in social interac-
relationships with other people” (p. 46). In the tions. Feelings of shame, hence, will negatively af-
following sections, social, cognitive, affective, and fect L2 learning by reducing L2 learners’ amount
behavioral underpinnings of shame and guilt are of L2 contact or hampering their motivation or
described in detail. willingness to use the L2.
Shame may also motivate learners to adopt de-
Shame fense mechanisms (Lewis, 2000, Tangney et al.,
1996). Not only may a shamed L2 learner adopt
In a prototypical shame experience, L2 learn- avoidant behaviors, he or she may also blame oth-
ers’ primary concern is with their global self as a ers, real or imagined, for their involvement in the
person and how this self is perceived, evaluated, shame-inducing event. Thus, the learner changes
and judged by others (Lewis, 1971; Tangney, 1990, the direction of acute pain of “humiliated anger”
1995; Tangney & Dearing, 2003). In the midst (Lewis, 1971) from his self toward others. Oth-
of a shame experience, an L2 learner may crit- ers are harshly criticized and blamed because they
ically direct her attention toward the self, nega- are held responsible for the learner’s painful ex-
tively scrutinize every aspect of it, and harshly per- perience of shame. Receiving a low grade on a fi-
ceive herself as an incompetent, flawed L2 learner nal exam, for instance, the shamed student—who
or a defective, undesirable speaker of an L2 in the feels inferior compared to other students—may
minds of others in the target language commu- redirect his anger to the teacher (e.g., “the teacher
nity. The counter-factual thinking processes in- was harsh on me”) or other relevant factors (e.g.,
volved in feelings of shame are typically directed “the exam was too difficult”). This other-oriented
at L2 learners’ selves, such as ‘If and only if I was hostility can be conceived of as a defensive strat-
not a such-and-such L2 learner/user’ (Tangney, egy used by shamed learners to regain a sense of
4 The Modern Language Journal 0 (2018)
FIGURE 1
Differential Features of Shame and Guilt
agency or control impaired during the shame ex- ences between shame and guilt are summarized
perience (Tangney, 1990, 1995). in Figure 1.
FIGURE 2
Scree Plot of the Component Eigenvalues
(M = 36.58; SD = 27.72). As it can be seen, the of each scale. Items with ITCs below .30 were con-
sample represents wide diversity in terms of age, sidered suspicious (Field, 2013). Consequently,
educational backgrounds, and language learning the items of one scenario, five items from the
backgrounds. This heterogeneity of participants externalization scale, and one item from the
will add more support to the validity and relia- detachment scale were excluded. Overall, 13
bility of the L2-TOSGA since the results of the scenarios with 46 corresponding items remained
statistical analyses are not affected due to the after the initial item analysis. Cronbach’s alpha
characteristics of a single demographic. coefficients also provided further evidence for
A number of private language institutes were the reliability of the subscales (see Table 1). The
contacted in order to administer the question- average internal consistency of the four scales
naire, and four institutes permitted the study to equals .82, which is excellent considering that
be conducted during regular classes. The volun- subscales were made up of a variety of different
tary and confidential nature of the responses were scenarios.
emphasized while informing students about the
Principal Component Analysis. All 46 items
purpose of the study. Students were further as-
measuring shame-proneness, guilt-proneness,
sured that their performance on the question-
externalization, and detachment were submitted
naire would have no effect on their class grades.
to factor analysis using principal component anal-
The questionnaire, which took about 20 min-
ysis (PCA) with the oblimin rotation method (see
utes to complete, was administered during class
Online Supporting Information, Part B, for an
time.
examination of the assumptions). The scree plot
Item Response Analysis. The initial item analy- was the initial criteria examined to determine the
ses revealed that all items have a good amount of number of components (e.g., Plonsky & Gonulal,
variability, with standard deviations ranging from 2015). As seen in Figure 2, the scree plot is clearly
1.13 to 1.79. The item–total correlation (ITC) was indicative of four components. Next, the eigen-
the next criterion for examining the coherence values of components were examined. Given the
10 The Modern Language Journal 0 (2018)
shortcomings of eigenvalues greater than one for quence or a guilt–shame sequence. On the
determining the number of components (e.g., other hand, the correlation between shame
overestimation) (e.g., Patil et al., 2008), parallel and guilt should be modest or weak so that
analysis (PA) was run to compute the criteria the distinction between them is not vio-
eigenvalues. Presently, PA is considered the most lated. A weak, positive correlation (r = .14,
accurate and objective technique for determin- p = .005, BCa 95% CI [.04, .24]) attested
ing the number of components in PCA (e.g., to the validity of this assumption and war-
Field, 2013). In this procedure, the eigenvalues ranted the theoretical distinction between
from the original data should be compared to shame and guilt.
those calculated from random values of the same 2. Shame should be positively correlated with
dimensionality (i.e., sample size and the number externalization. Although shame and exter-
of variables). PCA eigenvalues greater than PA nalization might seem unrelated and dia-
eigenvalues should be retained as significant metrically opposite, because the former in-
components, and those falling below should be volves blaming one’s self (internal) and the
discarded as spurious (Patil et al., 2008). latter blaming others (external), external-
The results revealed that the raw eigenvalues ization is a defensive maneuver of shame-
of components 1, 2, 3, and 4 are greater than prone L2 learners. That is, L2 learners try to
their corresponding PA eigenvalues, suggesting distance themselves from the overwhelming
the existence of only four significant components pain of shame by simply putting the blame
(Table S4, Online Supporting Information, Part on others.
A). Taken together, the results of the scree plot
In contrast, guilt should be negatively corre-
and PA both confirmed a four-component so-
lated with externalization. Guilt involves a neg-
lution. This is also consistent with the theoreti-
ative evaluation of one’s specific behavior and
cal underpinnings of the constructs included in
partial acceptance of one’s wrongdoing. Further,
the questionnaire. Component 1 (eigenvalue =
guilt is less painful than shame simply because an
6.83), component 2 (eigenvalue = 5.87), com-
L2 learner’s global self is not involved. Hence,
ponent 3 (eigenvalue = 3.76), and component
the L2 learner is more inclined to take repara-
4 (eigenvalue = 2.01) accounted for 15%, 13%,
tive actions to compensate for her shortcomings
8%, and 4% of the variance, respectively. In to-
rather than blaming others. The positive relation-
tal, the four-component solution accounted for
ship (r = .30, p = .000, BCa 95% CI [.19, .41])
40% of variance in the data. A close look at
between shame and externalization of blame, and
the composite items of each component revealed
the negative association between guilt and exter-
that the items representing the students’ shame-
nalization (r = −.15, p = .004, BCa 95% CI [−.26,
proneness, guilt-proneness, externalization, and
−.03]) lent empirical support to the validity of
detachment were neatly loaded onto four distinct
these theoretical hypotheses.
components (Table S5, Online Supporting Infor-
mation, Part A). 3. Shame and guilt should be inversely related
to detachment. Because shame and guilt in-
Correlational Analyses. As noted, externaliza-
volve negative evaluations of one’s global
tion and detachment subscales were included
self or specific behavior, they both have lit-
in the questionnaire to shed more light on
tle in common with a lack of caring mea-
the discriminant validity of the shame and guilt
sured by the detachment scale. The results
scales. In other words, shame and guilt should
showed that detachment was not related to
be differentially related to externalization and
shame (r = −.04, p = .458, BCa 95% CI
detachment in a theoretically consistent manner
[−.13, .06]) and negatively correlated with
(Tangney, 1990, 1995). The following hypotheses
L2 guilt (r = −.24, p = .000, BCa 95% CI
were stated and evaluated based on the results of
[−.37, −.11]).
correlational analyses:
4. Detachment and externalization of blame
should be positively correlated. In external-
1. Shame and guilt should be positively cor- ization, because an L2 learner’s personal
related. Shame and guilt are both negative self is not involved, she is more likely to
self-conscious emotions involving a negative adopt a detached attitude toward the situ-
evaluation of one’s self or one’s behavior. In ation. This hypothesis was confirmed by a
addition, shame and guilt can co-occur in positive correlation (r = .35, p = .000, BCa
the same situation (Lewis, 1971) with stu- 95% CI [.24, .46]) between detachment and
dents experiencing either a shame–guilt se- externalization.
Yasser Teimouri 11
Summary learn English, three qualitatively distinct descrip-
tors of motivation—intended effort, second lan-
The results of Study 1 substantiated the per- guage willingness to communicate (L2 WTC),
vasiveness of shame and guilt reactions in L2 and attention—were measured in this study.
situations. A qualitative analysis of 198 L2 learn- Intended effort assesses students’ intention
ers’ reports of negative experiences at Stage 1 to invest time and effort into learning English
& 2 provided strong evidence that L2 learners (Taguchi, Magid, & Papi, 2009). This construct
often experience shame and guilt during L2 has been used frequently in many L2 motivation
learning, and shame-inducing and guilt-inducing research studies as an important descriptor of
events occur frequently in L2 settings. Based students’ overall motivation (e.g., Dörnyei, 2009;
on L2 learners’ responses, furthermore, the L2- Taguchi et al., 2009). Intended effort has also
TOSGA questionnaire was developed to measure been used as a criterion measure in some stud-
L2 learners’ proneness to shame and to guilt ies examining influence of emotions on students’
during the process of L2 learning. The results of motivation (e.g., MacIntyre & Vincze, 2017). L2
psychometric work—descriptive analyses, relia- WTC refers to students’ “readiness to enter into
bility analyses, factor analyses, and correlational discourse at a particular time with a specific per-
analyses—all provided solid evidence of the son, using a L2” (MacIntyre et al., 1998, p. 547);
reliability and validity of the questionnaire. like intended effort, it represents students’ in-
tention to use the target language voluntarily
STUDY 2 in class. L2 WTC has been used as an impor-
tant criterion variable assessing the influence of
The results of Study 1 on the frequent occur- both negative and positive emotions in past re-
rence of shame and guilt among L2 learners war- search. For instance, anxiety has been found as
rant further investigation of these two neglected a negative predictor and enjoyment as a posi-
emotions in SLA. Similar to previous L2 emo- tive predictor of learners’ L2 WTC in class (e.g.,
tions research (Horwitz et al., 1986; MacIntyre & Khajavy, MacIntyre, & Barabadi, 2017). Of par-
Vincze, 2017; Teimouri et al., 2019), the effects ticular note, both intended effort and L2 WTC
of students’ shame and guilt experiences on their represent L2 learners’ intentions to spend time
motivation and language achievements are em- and energy studying and using the L2. Since in-
pirically investigated in Study 2. Further evidence tentions are the most immediate and important
of the stability, reliability, and validity of the L2- antecedent of one’s behaviors (Sheeran, 2002),
TOSGA questionnaire are also presented. the inclusion of intended effort and L2 WTC in
the present article offers valuable insights into the
Participants effects of shame and guilt on students’ motiva-
tional behaviors. The logical assumption here is
A total of 174 English-major university students that “people do what they intend to do and do
were surveyed in this study. The sample con- not do what they do not intend” (Sheeran, 2002,
sisted of 117 females and 57 males, and their ages p. 1). Past research has shown that intended ef-
ranged from 18 to 58 years old (M = 30; SD = fort is highly related to actual effort (Lake, 2013)
8.23). Most of the students self-reported their lan- and L2 WTC is closely associated with actual L2
guage proficiencies as Intermediate Mid or In- use (e.g., Hashimoto, 2002; MacIntyre & Charos,
termediate High (ACTFL, 2012). The language- 1996).
learning experience of the students ranged from Attention measures students’ motivation in
1 to 240 months (M = 26; SD = 34.23). terms of their actual level of mental attentiveness
in class, such as how much attention they pay
Instrument to their teachers, classmates, and class activities.
According to Kanfer and Ackerman (1989),
A three-part questionnaire was developed in motivation is “the direction of attentional effort,
this study for data collection. The first section the proportion of total attentional effort directed
of the questionnaire included the L2-TOSGA. It to the task (intensity), and the extent to which
consisted of 46 items that measured students’ attentional effort toward the task is maintained
proneness to shame and to guilt in 13 different over time (persistence)” (p. 661). The very close
L2 scenarios. The second section of the question- relation between students’ attention and their
naire consisted of 11 items measuring L2 learn- motivation has also been highlighted by some
ers’ motivation. In order to capture the effects L2 motivation researchers (Crookes & Schmidt,
of shame and guilt on students’ motivation to 1991; Trembley & Gardner, 1995). The results of
12 The Modern Language Journal 0 (2018)
Trembley and Gardner’s (1995) study have evi- of their answers, and their right to withdraw from
denced attention as a major reflector of students’ participating in the study. It took about 25 min-
motivational behaviors. While past research utes for the students to complete the question-
has extensively explored students’ motivation naire. Of note, 53 students in the research sample
in terms of intended effort and L2 WTC, the retook the L2-TOSGA after a 1-month period to
magnitude and direction of students’ attention allow for the test–retest reliability of the question-
during L2 learning as another major indicator naire to be determined. A total of 91 students gave
of motivation has been neglected. In the present consent to the researchers to access their course
study, the inclusion of intended effort and L2 grades.
WTC as immediate predictors of students’ behav-
iors, and attention as a descriptor of students’
Data Analysis
actual behaviors, will allow us to delineate a clear
picture about the effects of shame and guilt on To further assess the reliability of the L2-
students’ motivation in L2 learning.4 TOSGA questionnaire, a test–retest reliability
Three items were adopted from Taguchi et al. analysis was first computed for the part of the sam-
(2009) to measure students’ intended effort (e.g., ple that took the questionnaire twice (n = 53).
I would like to spend lots of time studying English); Next, Cronbach’s alpha for all variables was cal-
four items were adopted from Yashima (2002) to culated and reported based on the whole sample
measure students’ L2 WTC (e.g., If you were free to (N = 174). Finally, a series of multiple regression
choose, how much would you like to speak English in the analyses were conducted to examine the causal re-
class); and four items were newly developed by the lations between L2 learners’ self-conscious emo-
researcher to measure students’ attention in class tions of shame and guilt, and their motivation
(e.g., How much attention do you pay to your teacher and language achievements. The analyses were
when she is speaking in English class). All the item all bootstrapped based on 1,000 samples and
responses were elicited using 6-point Likert scales 95% confidence intervals for bootstrapped es-
anchored at 1 strongly disagree or not at all and 6 timates were calculated using accelerated bias-
strongly agree or very much. correction (BCa) (Field, 2013). Assumptions un-
The third section of the questionnaire elicited derlying each of these statistical analyses were also
background information, such as students’ age, examined (see Online Supporting Information,
gender, language-learning experience, and per- Part B).
ceived language proficiency. To further examine
the effects of shame and guilt reactions on stu-
Results
dents’ language achievements, students’ grades in
three English courses—Grammar Course, Labo- Reliability Analyses. Table 2 shows the results
ratory Course, and Speaking Course—as well as of the test–retest reliability analysis, as well as de-
their GPAs were also collected in this section. The scriptive analyses for L2-TOSGA subscales (n =
aim of the Grammar Course was to enhance stu- 53). As seen, students’ mean scores for each sub-
dents’ knowledge of English grammar and its us- scale at Time 1 and Time 2 are highly similar. The
age. Students were taught English grammar rules dispersion of scores for each scale, however, was
explicitly and were given relevant grammar ex- reduced slightly at Time 2. All the subscales also
ercises and activities. The focus of the Labora- exhibited high reliability coefficients in both ad-
tory Course was on boosting the listening skills ministrations. Test–retest reliability of L2-TOSGA
of students through a range of different listening was calculated by running correlations between
and speaking activities, whereas in the Speaking students’ scores in Time 1 and Time 2. The results
Course students practiced different target tasks in revealed correlation coefficients beyond .95 for
various situations to improve their speaking skills all the subscales of L2-TOSGA, providing further
in English. The final grade for each course was evidence for its stability and precision in measur-
computed based on students’ mid-term and final ing the shame and guilt experiences of students
exams. across time.
Table 3 also shows the results of reliability anal-
Procedure yses and descriptive analyses for all the variables
based on the whole sample. Cronbach’s alpha co-
The questionnaire was administered to the stu- efficients provide evidence of high internal con-
dents during their class time. Before filling out sistency for each scale. As seen, students’ scores
the questionnaire, students were informed of the for guilt reactions are almost twice the scores for
purpose of the questionnaire, the confidentiality shame reactions. Moreover, students reported low
Yasser Teimouri 13
TABLE 2
The Results of Test–Retest Reliability Analyses for L2-TOSGA Sub-scales
TABLE 3
The Results of Reliability Analyses for All the Variables (N = 174)
scores for both externalization and detachment guage achievements—course grades and GPA.
scales. Of the four subscales of L2-TOSGA, shame Cohen’s f2 of .15, .25, .16, and .20 represent
and externalization showed slightly more variance medium effect sizes for each regression model,
compared to guilt and detachment subscales. respectively (Cohen, 1988). Table 5 shows the re-
Considering students’ motivation, intended effort sults of multiple regression analyses. Similar to
scored higher than L2 WTC and attention in class. previous analyses, shame emerged as a strong,
Students also showed more variance in their re- negative predictor of all the language achieve-
sponses to the L2 WTC scale than in those of the ment measures, whereas guilt positively deter-
other two motivational variables. mined Grammar Course grade. Both externaliza-
A set of multiple regression analyses using the tion and detachment had either a negligible, or
Enter method was run to examine the strength of nonexistent effects on the language achievement
the L2-TOSGA sub-scales in predicting students’ outcomes.
motivation. Cohen’s f2 of .37, .56, and .59 repre-
sent quite large effect sizes for each regression DISCUSSION
model, respectively (Cohen, 1988). As the results
show in Table 4, shame strongly but negatively The first research question explored the
predicted L2 learners’ L2 WTC and attention; it frequency of L2 learners’ shame and guilt
had negligible negative effects on intended ef- reactions during L2 learning. A qualitative
fort, though. In sharp contrast, guilt emerged as analysis of 198 L2 learners’ reports of neg-
a positive predictor of all the motivational vari- ative experiences during L2 learning re-
ables. While externalization was found to be only vealed many episodes of shame and guilt
a moderate, negative determiner of intended ef- reactions. More importantly, shame-inducing
fort, students’ lack of emotional involvement, that and guilt-inducing events were found to be fre-
is, detachment, weakly but negatively predicted quent in L2 settings. That is, most of the students
intended effort and attention. have experienced almost half of these situations.
Another set of multiple regression was con- Overall, these results along with the results of
ducted to examine the strength of the L2- other studies in different contexts (Cook, 2006;
TOSGA subscales in predicting students’ lan- Galmiche, 2017; Wang, 2016), provide strong
14
TABLE 4
Regression Analyses of Emotions With Intended Effort, L2 WTC, and Attention as Criterion Measures
TABLE 5
Regression Analyses of Emotions With Course Grades and GPAs as Criterion Measures
Shame −.41 .16 −.025 −.32*** −.73 −.18 −.66 .15 −.014 −.48*** −.98 −.41 −.52 .14 −.011 −.37*** −.80 −.29
Guilt .40 .20 .003 .20* −.02 .78 .03 .19 .014 .01 −.34 .42 .23 .27 −.002 .11 −.27 .77
Externalization .25 .20 .011 .16 −.18 .66 .18 .19 .000 .11 −.24 .56 .19 .17 .000 .11 −.16 .54
Detachment −.05 .79 −.005 −.03 −.49 .36 −.17 .23 .025 −.08 −.70 .36 .08 .21 .008 .04 −.38 .48
R2 / F .13/3.176* .20/5.29** .14/3.47*
Cohen’s f2 .15 .25 .16
GPAs BCa 95% CI
Variables B SE Bias β L U
***
Shame −.45 .13 −.015 −.43 −.74 −.23
Guilt .14 .15 .007 .09 −.16 .46
Externalization .13 .15 .007 .10 −.22 .43
Detachment −.07 .17 .001 −.04 −.41 .25
R2 / F .17/4.41??
Cohen’s f2 .20