AN INVESTIGATION INTO INDONESIAN EFL UNIVERSITY
STUDENTS’ SPEAKING ANXIETY
Mukhlash Abrar
Universitas Jambi, Indonesia
mukhlash.abrar@unja.ac.id
Abstract: The general purpose of the study was to
investigate EFL learners‟ speaking anxiety and its
associated factors. More specifically, the study was
purposed to examine the constructs of a speaking
anxiety and to explore the model for predicting speaking
anxiety based on gender, proficiency and class type.
Therefore, explanatory factor analysis and multiple
linear regression analysis were used. The result
indicated that there are two factors in learners‟ speaking
anxiety, but they are difficult to name as a result of
overlapping variables in each component. With regards
to a multiple linear regression test, the finding shows
that proficiency variable is the most significant factor for
predicting the variation in speaking anxiety.
Keywords: speaking anxiety, factor analysis, multiple
linear regression
INTRODUCTION
Successful foreign language learning requires that students
accommodate both cognitive and affective domains in their learning.
Anxiety, an affective factor in language learning, is a frequently
observed issue which arguably influences students‟ performance and
their academic achievement (Chastain, 1975; Dordinejad and
Ahmadabad, 2014; Gardner et al, 1997; Horwitz, 2001; Luo, 2013; Saito
and Samimy, 1996; Scott, 1986) and can negatively impact the learning
process (Arnold and Brown, 1999; Gardner and Maclntyre, 1993;
Oxford, 1996). This phenomenon is often identified in speaking
modules of English as a Foreign Language (EFL) (Horwitz et al, 1986;
Palacios, 1998; Price, 1991) because students are required to think and
give a response to the initiation almost simultaneously. Thus, this
221
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
quantitative study investigates and reports EFL learners‟ speaking
anxiety and its associated factors and aims to provide some practical
suggestions for language tutors and meaningful recommendations for
future research.
LITERATURE REVIEW
Foreign Language Anxiety
Anxiety is one of the most highly researched phenomena in
psychology and education. It is defined, in simple terms, as the
feeling of unease or fear and is often identified in many social and
learning contexts. According to Hilgard, Atkinson, and Atkinson
(1971), anxiety is regarded as a psychological construct, commonly
described as a state of apprehension or worry, which is indirectly
associated with an object. Furthermore, according to Chastain (1988),
anxiety is a state of uneasiness and apprehension initiated by the
anticipation of something threatening.
Along
similar
lines,
Rachman (2004:3) asserts anxiety as “tense, unsettling anticipation of
a threatening but vague event; a feeling of uneasy suspense”. Within
the foreign language context specifically, Horwitz, Horwitz, and Cope
(1986: 128) define anxiety as “a distinct complex of self-perceptions,
beliefs, feelings, and behaviours related to classroom language
learning arising from the uniqueness of the (foreign) language
learning process”. Horwitz (2001) further proves that foreign
language learning anxiety has a negative impact on learners‟
performance and academic achievement. Moreover, MacIntyre (1998)
asserts that anxiety is a feeling, a worry and an emotional reaction
which arises while learning or using a second language, and which
negatively impacts learning process. These imply that anxiety is a
psychological construct which causes individuals worry or fear of
something which negatively influences either the process or the
achievement of learning a foreign language.
Anxiety, broadly speaking, can be classified into three types:
trait anxiety, state anxiety and situation-specific anxiety. However,
Horwitz, Horwitz, and Cope (1986) and William and Andrade (2008)
222
JEELS, Volume 4, Number 2, November 2017
argue that not all anxiety types apply specifically to the foreign
language learning context. Ellis (1994) states that trait anxiety refers to
the stable tendency to be nervous in many circumstances. In other
words, it is a part of a person‟s characteristics and hence is a difficult
trait to shed. An individual whose trait is anxious tends to feel
anxious in a number of situations. State anxiety, as the name implies,
on the other hand, relates to an unstable feeling of anxiousness that
arises in specific circumstances as a response to an external stimulus
(MacIntyre and Gardner, 1989; Spielberger, 1983).
A person experiencing state anxiety will feel tension or worry
because they are exposed to a particular situation which causes them
stress, but it will fade when the threat disappears. Finally, situation
specific anxiety, according to McIntyre and Gardner (1991), can be
seen as trait anxiety limited to a given context. It is stable over time,
but inconsistent across varying circumstances. That is, it is prompted
only by a particular setting or situation, such as taking a test, public
speaking or speaking in a foreign language. Given the features of
situation-specific anxiety, MacIntyre and Gardner (1991) suggest that
foreign language anxiety should be studied with situation specific
measures. Horwitz et al. (1986) further support the theory that foreign
language anxiety is classified as situation-specific anxiety.
As previously indicated, anxiety, in principle, is a response
triggered by external threat. Therefore, it can be attributed to several
factors in the context of language learning in general. Young (1991)
argues the possibility that language anxiety may emerge from three
aspects of learning: the teacher, the learner and the instructional
practice. More specifically, he claims six interrelated factors as the
causes of learning anxiety, namely: (1) personal and interpersonal
anxiety; (2) learner beliefs about language learning; (3) instructor
beliefs about language teaching; (4) instructor-learner interactions; (5)
classroom procedures; (6) language tests.
In the foreign language context, Aida (1994), Casado and
Dereshiwsky (2004), Horwitz et al (1986), MacIntyre and Gardner
(1989), and Pappamihiel (2002) conclude that anxiety is the result of
223
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
three possible affecting factors: communicative apprehension (e.g.
difficulty in understanding the teachers‟ instruction and/or peers‟
talk); fear of negative evaluation (e.g. fear of being corrected by
teachers/peers); test anxiety (e.g. fear of failing the module/lesson).
These three affecting factors are well-known sources of anxiety and
are relevant to the discourse of foreign language anxiety. In the
present study, two of the three constructs - communicative
apprehension and fear of negative evaluation - have been explored in
more depth, by using them as subscales within the questionnaire.
The overall discussion on the concept of foreign language
anxiety suggests that anxiety is one of the most examined affective
factors which may negatively influence foreign language learners‟
performance and/or acquisition of the language due to their difficulty
in understanding the lesson/instruction, worry of getting negative
feedback from onlookers, or fear of failing the class. Since this
psychological construct is personal, every foreign language learner
may experience a different level of anxiety. Students with higher
levels of anxiety had a weaker performance or achieved less
compared to less anxious students (Hewitt and Stephenson, 2012;
Kleinmann, 1977; Steinberg and Horwitz, 1986).
Speaking and Foreign Language Anxiety
Speaking is generally recognised as a fundamental language
skill for effective interaction in any language, including for non-native
language speakers. Its nature is exceptionally distinct from other
language skills in that it is a verbal productive skill that requires the
speakers‟ mastery in linguistic and sociolinguistic competence
(Nunan, 2003, 2009; Hinkel, 2005). Of the two competences, much of
the research reported that linguistic competence appeared to be the
one of the main challenges for non-native speakers in learning and/or
practising a foreign language (Abrar and Mukminin, 2016; Al-Hosni,
2014; Al-Jamal & Al-Jamal, 2014; Arju, 2011; Gan, 2012; Keong et al,
2015; Lee, 2009; Priyatno, 2013; Paakki, 2013; Wang and Roopchund,
224
JEELS, Volume 4, Number 2, November 2017
2015) owing to the numerous components in which one must gain
proficiency.
Syakur (1987) details that there are, at least, five components
with which speakers need to engage when speaking, including:
comprehension (ability to understand the meaning and the capacity
to engage in a conversation); grammar (well-structured sentences);
pronunciation
(comprehensible
articulation);
word
choice
(appropriate diction); fluency (the ease and speed of the flow of
expression). Syakur (1987) theory was used as the framework in
constructing the questionnaire for this study.
In the context of foreign language learning, anxiety is often
associated with speaking ability (Lucas, 1984; Phillips, 1992, Price,
1991). Many studies have consistently shown that foreign language
anxiety and speaking ability are, to a certain extent, interwoven.
Horwitz et al (1986), for instance, argued that communication
comprehension is conceptually relevant to foreign language anxiety.
Price (1991), then, revealed that speaking in front of peers in the target
language provoked the most anxiety for learners in her foreign
language class. On a similar thread, Palacios (1998) asserted that
speaking caused the most anxiety among foreign language learners.
Clearly, the correlation between speaking and anxiety is a cause-effect
relationship in which speaking itself leads to anxiety for foreign
language users/learners.
Studies on Foreign Language Anxiety
There has recently been a marked increase in studies of foreign
language anxiety, specifically relating to speaking skills. Much of the
research examined anxiety with its associated affecting
factors/variables, such as gender and proficiency. Regarding gender
proficiency, Çağatay (2015) researched four associated factors in
speaking anxiety, of which gender was one. She administered an 18item questionnaire to 147 students of an English preparatory
programme of a state university in Turkey. Her findings suggested
that there is a statistically significant difference between male and
225
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
female students‟ anxiety in speaking a foreign language (English),
noting that female students tend to be more anxious than male
students.
Many other studies (Park and French, 2013; Ekström, 2013;
Öztürk, and Gürbüz, 2013; Hsu, 2012; Tianjian, 2010; Occhipinti, 2009;
Huang, 2004) supported this conclusion. A study conducted by
Fariadian, Azizifar and Gowhary (2004), however, showed the
opposite. They investigated gender contribution in EFL speaking
anxiety among Iranian learners. With a total number of 80
participants involved in their study, they found that there is a
significant difference between the two gender categories. Male
respondents were reported to experience slightly higher levels of
anxiety than females. Nevertheless, other research (Debreli and
Demirkan, 2015; Şimsek, 2015; Muhaesin and Al-Haq, 2012; Cui, 2011;
Aida, 2004; Voorhees, 1994) reported that there is no significant
difference between male and female respondents in foreign language
speaking anxiety.
With regard to the proficiency variable, the result of most
related studies showed that more language-proficient participants
tended to have a lower anxiety level when learning and speaking
English than less-proficient participants, although the difference was
not always significant (Çağatay, 2015; Tercan and Dikilitas, 2015;
Zhao and Whitchurch, 2011; Tianjian, 2010; Liu, 2006). Tanjian (2010),
for example, investigated Chinese EFL learners‟ speaking anxiety
including gender and proficiency differences. The researcher involved
240 participants and divided them into three categories of proficiency
level.
The result of the research suggests that students from the
lower proficiency group experienced more anxiety than the higher
proficiency group. A study conducted by Debreli and Demirkan
(2015), in contrast, produced a different result. By comparing anxiety
between elementary and pre-intermediate level of EFL students, their
findings suggests that more proficient EFL learners are more anxious
in learning language than less proficient learners. They argue that as
226
JEELS, Volume 4, Number 2, November 2017
the proficiency increases, the learners need to complete more
demanding tasks and have more concerns about their teachers‟
expectation.
In this present study, the researcher also used gender and
proficiency as independent variables (IVs). Male and female
respondents were used in order to satisfy the gender variable, while
proficiency was subjectively measured and classified by looking at
the respondents‟ experiences in attending English courses. In
addition, the researcher added another IV named “class type” which
was divided into regular and regular-mandiri categories. This variable
is unique to the Indonesian university context whereby regular class
students are admitted on the basis of passing a national entrance
exam while regular-mandiri class students are admitted on the basis of
passing a university entrance exam.
The present study attempts to investigate Indonesian EFL
learners‟ speaking anxiety and its associated factors. The study
addressed the following alternative hypotheses to be tested
statistically.
1. From 10 question instruments, two factors are obviously
identifiable, measured by 5 questions of which they are
comprised.
2. There is a significant prediction of student teachers‟ speaking
anxiety by gender, proficiency, and class type.
METHOD
Participants and Setting
The participants in this quantitative study were 72 second-year
EFL students studying an English Teacher Training programme at a
public unversity in Jambi, Indonesia. The questionnaire was initially
distributed to all second-year students (n=102), but some of them
seemed reluctant to return the questionnaire. The reason for selecting
second-year EFL learners as respondents was because they had taken
all speaking modules offered through their programme. Among the
72 respondents, 18 (25%) participants were male, and 54 (75%) were
227
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
female. 34 (47.2%) participants had previously taken English courses,
while 38 (52.8%) had never taken any. 45 (62.5%) participants enrolled
in regular classes, and 27 (37.5%) enrolled in regular-mandiri classes.
Table 1 details a summary of the sample.
Table 1 Sample Distribution
Number of Participants
Gender
Male
Female
Experience in English Course(s)
Yes (at least once)
No (Never)
Class Type
Regular
Regular-Mandiri
18
54
34
38
45
27
Research Instruments
For the purpose of gathering data from the participants, a
close-ended questionnaire was used as a research instrument for this
study. The statements in the questionnaire were constructed by
developing Syakur‟s (1987) theory on five speaking aspects, and
adapted some elements of Horwitz‟s (1986) questionnaire on FLCAS.
The 13-item questionnaire was developed in English (see Appendix;
questionnaire). The first three items related to the participants‟
personal information: gender, English course experience and class
type. The purpose of items 4-13 in the questionnaires was to explore
the participants‟ level of anxiety in the component of speaking: Q4 &
Q5 (comprehension); Q6 & Q7 (grammar); Q8 & Q9 (word choice);
Q10 & Q11 (pronunciation); Q12 & Q13 (fluency). The five point
Likert scale was used in constructing statements‟ options: (1) strongly
disagree, (2) disagree, (3) neither agree nor disagree, (4) agree, and (5)
strongly agree.
228
JEELS, Volume 4, Number 2, November 2017
Once the questionnaire had been constructed, the researcher
submitted it to an expert, the module tutor, to assess the validity of its
content. Carmines and Zeller, cited in Cohen et al (2011) underline
that content validity focuses on how “the instrument must show that
it fairly and comprehensively covers the domain or items that it
purports to cover” (p.188). Content validity is evidently fundamental
in constructing questionnaires, aiming to determine whether the
questionnaire measures all facets of a given construct or not. In this
case, some revision to the questionnaire was required, primarily
relating to sentence structures and wording.
Data Collection and Data Analysis
Following revision, the questionnaire was subsequently
administered to the second year trainee teachers of an English
Teacher Training programme at a public university in Jambi,
Indonesia with the assistance of the researcher‟s colleagues,
specifically speaking module tutors. The researcher electronically sent
a set of questionnaires to his colleagues and provided instructions
around how the questionnaire should be correctly administered. Each
tutor then administered the questionnaire to their class and returned
the completed questionnaire electronically to the researcher.
When data collection had been completed, the researcher then
manually input the data in Statistical Package for Social Sciences 22.0
(SPSS) and started quantitative analysis. The quantitative data was
analysed using descriptive and inferential statistics, including
explanatory factor analysis (EFA) and multiple linear regression
analysis. EFA was applied to determine the construct structure while
multiple linear regression analysis was applied to investigate anxiety
levels and associated factors: gender, proficiency and class type. The
findings were presented in tables and illustrated in figures.
To satisfy research ethical obligations, prior to undertaking the
study, the researcher sent a letter requesting consent to the Director of
the English Teacher Training programme. When permission had been
granted from the research site, the researcher administered a consent
229
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
form, containing information of the nature and purpose of the study,
to the prospective participants along with a copy of the questionnaire.
They could decline the researcher‟s invitation to take part in the study
by not signing the form and/or returning the questionnaire. The
signed consent form was used as proof that participants had
voluntarily agree to take part and is documented for scholarly
purpose only. To further protect the participants‟ identity, anonymity
was used in the final report.
FINDINGS AND DISCUSSIONS
As indicated, two hypotheses had been formulated in this
study. In the first hypothesis, the researcher wanted to explore the
factor structure underlying speaking anxiety responses in the data set.
The second hypothesis was intended to investigate if the IVs - gender,
proficiency, and class type- become the significant predictors towards
learners‟ speaking anxiety as a dependent variable (DV). Two
inferential statistics, namely explanatory factor analysis and multiple
linear regression, were used.
Result of Factor Analysis
The first step in conducting factor analysis is to produce the
correlation matrix intended to determine whether or not the study
variable correlates and, if so, the extent of the correlation. Tabachnick
and Fidell (2001) outline that the use of factor analysis is questionable
if a correlation, as reflected in the correlation matrix, is less than .30.
Therefore, since the correlations yielded in the correlation matrix in
this test exceeded .30, factor analysis was an appropriate data test.
The result of analysis as suggests that all 10 variables are correlated,
which indicates that there is a patterned relationship among the
variables.
KMO and Bartlett‟s test was utilised to confirm if the variables
have a patterned relationship and to measure the sampling adequacy
for the variables (Hair et al, 2014; Tabachnick and Fidell, 2013). The
result of KMO and Bartlett‟s test is detailed in the following table:
230
JEELS, Volume 4, Number 2, November 2017
Table 2 KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling
.881
Adequacy.
Bartlett's Test of
Approx. Chi-Square 282.579
Sphericity
df
45
Sig.
.000
As shown in table 2, p value of Bartlett‟s test of Sphericity
result is < 0.001 which means very significant. The significance of the
result can be determined when p value ≤ 0.05 (Field, 2013; Hair et al,
2014; Tabachnick and Fidell, 2013). The small p value shows that there
is a statistically significant interrelationship between variables. The
KMO measure of sampling adequacy (MSA) result is 0.881. This result
exceeds the minimum cut-off point, above 0.50 (Tabachnick and
Fidell, 2013; Hair et al, 2014), indicating that the data is sufficient for
EFA.
Table 3 Total Variance Explained
Extraction Sums of Squared
Loadings
% of
Cumulati
% of
Cumulati
Total Variance ve %
Total Variance ve %
4.819 48.192
48.192
4.819 48.192
48.192
58.488
1.030 10.296
1.030 10.296
58.488
.890 8.903
67.391
.746 7.461
74.853
.574 5.743
80.595
.505 5.052
85.647
.444 4.443
90.091
.392 3.922
94.013
.346 3.456
97.468
.253 2.532
100.000
Extraction Method: Principal Component Analysis.
Component
1
2
3
4
5
6
7
8
9
10
Initial Eigenvalues
Rotation Sums of Squared
Loadings
% of
Cumulati
Total Variance ve %
3.735 37.354
37.354
2.113 21.134
58.488
Table 3 shows the total variance explained and indicates the
number of significant factors. Kaiser‟s rule, at this stage, was used to
determine the most eligible factors for interpretation by extracting the
231
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
factors from the variables data. Kaiser‟s criterion (Kaiser, 1958)
stipulates that the only components with an eigenvalue greater than
or equal to 1.0 should be retained for further analysis.
Table 3 clearly indicates that there are only two significant
factors from the data. Together they are capable of explaining roughly
58.5% of all the variable variances. In addition, figure 1 below
confirmed the findings of retaining 2 factors.
Figure 1 Screen Plot
The two components with eigenvalue greater than 1.0 were
rotated using Varimax rotation technique (Comrey and Lee, 2009;
Hair et al, 2014; Stevens, 2009; and Tabachnick and Fidell, 2001) to
generate rotated component matrix. Tabachnick and Fidell (2001)
state that a high value of variable loading, reflected in rotation
component matrix, shows the pure measurement of the component.
Hair et al (2014) comment that each individual variable loading value,
reflected in rotated component matrix, should be .50 or greater to
provide interpretive value and to indicate the interrelation of the
variables in the factor.
By looking at the minimum loading value, as proposed by
Hair et al (2014), it is obvious which variables belong to each
component. As shown in table 4, component 1 consists of 7 variables
with .81 as the highest factor weight value and .59 as the lowest value.
Component 2, on the other hand, comprises 3 variables with the
sequence loading values: .82, .72, and .61. These values in each
232
JEELS, Volume 4, Number 2, November 2017
component imply that variables are highly interrelated. Nevertheless,
it seems impossible for the researcher to label the factors as there are
some overlapping variables in each component. In this respect, a
factor analysis yielded an unexpected result showing different items
are loading in different subscales. Of the 7 variables in component 1,
for instance, 4 items are apprehension subscale and the other 3 items
belong to feedback subscale. Such findings suggest that a factor
analysis does not really support the items that are supposed to be in a
particular subscale and the hypothesis is therefore rejected.
Table 4 Rotated Component Matrixa
Component
1
2
Feedback Pronunciation (11)
Feedback Grammar (7)
.811
.136
.801
.154
Word Choice Apprehension (8)
Feedback Word Choice (9)
.757
.252
.705
.327
Grammar Apprehension (6)
.661
Fluency Apprehension (12)
.633
.413
Pronunciation Apprehension (10)
Feedback Comprehension (4)
Feedback Fluency (13)
.596
.129
.408
.816
.365
.716
Comprehension Apprehension (5)
.153
.604
The unexpected result of factor analysis is likely to occur for
several reasons. Firstly, the sample of research might not be sufficient
because only 72 participants were involved in this study. A wide
range of recommendations in relation to sample in factor analysis has
been made. Experts have different opinions regarding the minimum
number of subjects required for analysis: 100 (Gorsuch, 1983; Kline,
1979); 150 (Hutcheson and Sofroniou, 1999); 200 (Guilford, 1954); 250
(Cattell, 1978); 300 (Nunnally, 1994; Tabachnick and Fidell, 2001,
2013). Comrey and Lee (1992) provided the following guidance
regarding the adequacy of sample size: 100= poor, 200 = fair, 300 =
good, 500 = very good, 1,000 or more = excellent.
233
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
They further encourage researchers to obtain a minimum of
500 samples whenever possible. Adequate sample size is therefore
necessary because it produces a more accurate solution (Costello and
Osborne, 2005). Secondly, content validity of the questionnaire may
be another concern. As previously stated, this type of validity mainly
focuses on how the questionnaire addresses all facets of phenomena
(Carmines and Zeller in Cohen et al, 2011). The total number of items
seems inadequate to measure all facets because there were only 10,
that is, 5 items for each subscale with every item representing the
component of speaking category. Additional items in subscale and/or
categories may be best to the meaningful factorial solution.
Table 5 Reliability Statistics
Cronbach's
Alpha
Cronbach's Alpha Based
N of Items
on Standardized Items
.876
.876
10
Although this questionnaire has issues concerning its construct
validity, it is still found to be reliable. This is because the Cronbach
alpha value is .876 which is more than its acceptable value .70 (Hair et
al, 2014; Nunnally and Bernstein, 1978) as shown in the table 5. Thus,
further analysis can be conducted using this questionnaire.
Result of Multiple Linear Regression
Multiple linear regression is generally defined as one of the
modelling techniques that enable a researcher to assess the
relationship between a DV (predicted) and some IVs (predictor). The
final result of this modelling is the development of a regression
equation (line of best fit) between DV and some IVs. Prior to
conducting regression analysis, the researcher initially recoded IVs
(gender, proficiency and class type) into dummy variables as they are
nominal variables. They are dummy coded by assigning „0‟ to one
category and „1‟ to another category and respectively named as:
„female‟ („0‟ male, „1‟ female); „proficient‟ („0‟ not having
234
JEELS, Volume 4, Number 2, November 2017
course experience, „1‟ having course experience); „regular‟ („0‟
non regular class, „1‟ regular class). The researcher then tested
some assumptions, including multicollinearity, data normality,
linearity, and outliers.
Multicollinearity
Multicollinearity is a phenomenon in which more than one of
the IVs are highly correlated in multiple regression (Field, 2013; Hair
et al, 2014; Stevens, 2009). A researcher must check this assumption to
ensure no predictor variables are correlated in order to avoid the
replication of the tendency predictor variable. The result of the
correlation test, in table 6, clearly demonstrates that there is no
existence of multicollinearity case in this study‟s data-set. No
correlation is found between predictor variables as all the
corresponding values are less than .70.
Table 6 Correlations
Pearson
Correlation
Total anxiety
scale
Female
Proficient
Regular
Sig. (1-tailed) Total anxiety
scale
Female
Proficient
Regular
N
Total anxiety
scale
Female
Proficient
Regular
Total anxiety
scale
Female
Proficient
Regular
1.000
-.115
-.356
-.171
-.115
-.356
-.171
1.000
.161
-.116
.161
1.000
.215
-.116
.215
1.000
.
.167
.001
.075
.167
.001
.075
.
.089
.166
.089
.
.035
.166
.035
.
72
72
72
72
72
72
72
72
72
72
72
72
72
72
72
72
235
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
Data Normality, Linearity, and outliers.
The researcher is advised to check the data normality and
linearity before reporting the model of regression analysis. The
normality and linearity of data distribution can be seen from the
normal P-P plot regression. If the dots, reflected in P-P plot, are
perfectly straight-lined, it can be inferred that the data are highly
normally distributed. The finding shows that the data are normally
distributed because the dots are reasonably close to the best fit line as
shown in chart 2. There were some deviations from the perfect line,
but they were not major ones.
Figure 2 Normal P-P Plot
Aside from the normal distribution shown in figure 2 above,
the researcher also provides some evidence of normality data from
other measurements, including skewness, kurtosis, Kolmogorov–
Smirnov test, and boxplot. In an attempt to check the possible outliers
in the data, scatterplot produced by SPSS was further analysed. If the
dots are clustered between the threshold -3 and 3, it indicates that the
data is acceptable and there are no outliers. The findings of this study,
as shown in figure 3, suggest that there are no outliers found in the
data. Although the dots are not clustered closely to each other, they
are still scattered in the specified threshold.
236
JEELS, Volume 4, Number 2, November 2017
Figure 3 Scatterplot
After assumptions are met, the next stage is to evaluate the
model, the aim of which is to determine the model effectiveness,
significance, and its accurate predictor(s). In order to evaluate the
model, the first consideration should be the ANNOVA table, which is
produced by SPSS.
Table 7 ANOVA
Model
1
Sum of Squares
df
Mean Square
F
Sig.
Regression
Residual
493.552
2988.448
3
68
164.517
43.948
3.743
.015b
Total
3482.000
71
Table 7 above reports the results of an ANOVA test that
represents an overall test value of how well the model as a whole fits
the data. Hair et al (2014) and Steven (2009) underscore that the
model is deemed statistically significant to predict the variation in DV
if p-value is ≤0.05. The significant value (.015), as can be seen in the
table 7, shows that there is a strong evidence that the model is
statistically significant and useful to predict the variation in the
predicted variable. Therefore, the alternative hypothesis of this study
is accepted.
In order to see how much of the variability in the outcome is
accounted for by the predictors, the value of adjusted R2 in the table
237
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
of „model summary‟ should be considered. Hair et al (2014) posit that
adjusted R2 is a “modified measure of the coefficient and
determination that takes into account the number of independent
variables included in the regression equation and the sample size”
pp.152) and is regarded as the best estimate of the degree of the
relationship in the basic population. The following table shows the
value of adjusted R2 of this model.
Table 8 Model Summary
Model
R
R Square
Adjusted R
Square
Std. Error of the
Estimate
1
.376a
.142
.104
6.629
The value of adjusted R2 of this study is fairly small, .104,
which means that the model, using 3 predictor variables, only
explains 10.4% of the variance in speaking anxiety. This relatively
small adjusted R2 value also implies that nearly 90% of the variance in
speaking anxiety is explained by other factors. This unexpected result
encouraged the researcher to see its computed size effect. By running
G* power software, it is found that effect size f2 is .116. This indicates
that the computed effect size corresponds to a small to medium effect
size. To sum up, this model is statistically significant (R2= .104, F(3,
68)= 3.743, p< .05, f2cohen= .116).
Table 9 Coefficients
Model
1
(Constant)
Female
Unstandardized
Coefficients
Standardized
Coefficients
B
34.527
Std. Error
1.971
Beta
t
Sig.
17.520
.000
-1.234
1.851
-.077
-.667
.507
Proficient
-4.454
1.633
-.320
-2.728
.008
Regular
-1.597
1.673
-.111
-.955
.343
Table 9 provides information regarding the relative strength of
individual predictor variables. Only the proficient variable, of the three
238
JEELS, Volume 4, Number 2, November 2017
predictors, seems to contribute more to the model as its p-value was
found significant (p<.05), while female and regular variables do not
contribute (pfemale>.05 and pregular>.05). Another way to see the
significant predictor is from the standardized coefficient column
(Beta-value). The proficient variable appears to have a greater effect (.320) to the model than the other variables. Thus, on the basis of the
mentioned values, it is tempting to conclude that the only useful
predictor is the proficient variable. Table 9 further represents the
information needed to construct the actual model. Using the
„unstandardized coefficient‟ from the table, the construct of statistical
model can be formulated as follows:
Predicted ‘Speaking Anxiety’= 34.572 – 1.234 x ‘Female’ – 4.454
x ‘Proficient’ – 1.597 x ‘Regular’.
The regression equation above shows that if „female‟ is
increased by one unit, total speaking anxiety decreases 1.234 units. In
other words, female students are found to experience less anxiety
than male students. This result surprisingly contradicts the majority
of research (Çağatay, 2015; Park and French, 2013; Ekström, 2013;
Öztürk, and Gürbüz, 2013; Hsu, 2012; Tianjian, 2010; Occhipinti, 2009;
Huang, 2004) in finding that female students tended to be more
anxious at speaking foreign language than male students. However,
from a second language perspective, the result of this study seems
reasonable in that both genders, by nature, are different in the process
of acquiring a language. Females are considered to be more motivated
in learning a second language than males (Gardner and Lambert,
1972).
With regard to „proficient‟ predictor, the model suggests that
the total speaking anxiety significantly decreases as much as 4.454
units if „proficient‟ is increased by one unit. It simply shows that
proficient learners are less nervous about speaking than nonproficient learners. This result is congruent to the findings of other
studies (Çağatay, 2015; Tercan and Dikilitas, 2015; Zhao and
Whitchurch, 2011; Tianjian, 2010; Liu, 2006). This study‟s finding is
239
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
rationally accepted as proficient-language learners have not only
learned the language at school, but also at informal institution(s).
They possibly gain more experiences at speaking practice and
therefore their anxiety level gradually decreases. Skehan (1989)
argues that “students at higher levels might enjoy wider repertoire of
behaviours which would help them to deal with anxiety in language
learning contexts more flexibly” (p.116). In addition, the model
indicates that regular-mandiri students feel more anxious than regular
students as if „regular‟ is increased by one unit, speaking anxiety
generally decreases 1.594 units.
CONCLUSION
This study aims to identify the constructs of a speaking anxiety
questionnaire and to explore the model for predicting speaking
anxiety based on gender, proficiency and class type. Firstly, the latent
constructs of speaking anxiety responses in the dataset are
determined through an explanatory factor analysis. The findings
reveal two underlying factors. Nevertheless, these two factors are
difficult to name due to some overlapping variables in each
component. Therefore, additional participants and questionnaire
items may be included for producing more accurate and meaningful
factorial solution. Secondly, this study aims to examine the
relationship between predictors (gender, proficiency, and class type)
and a predicted variable (speaking anxiety) with a multiple linear
regression test. The findings suggest that the resultant regression
model is determined to be statistically significant for predicting the
variation in speaking anxiety. In addition, among the three predictors,
the proficient variable appears to be the most significant predictor in
that it contributes more to the model.
REFERENCES
Abrar, M. and Mukminin, A. (2016). International Graduate
Classroom Discussion Engagement, Challenges, and SolvingStrategies. Asia-Pacific Collaborative education Journal, 12(1), pp.519.
240
JEELS, Volume 4, Number 2, November 2017
Aida, Y. (1994). Examination of Horwitz, Horwitz, and Cope's
Construct of Foreign Language Anxiety: The Case of Students of
Japanese. The Modern Language Journal, 78(2), pp.155-168.
Al Hosni, S. (2014). Speaking Difficulties Encountered by Young EFL
Learners. International Journal on Studies in English Language and
Literature (IJSELL), 2(6), pp.22-30.
Al-Jamal, D.A. and Al-Jamal, G.A. (2014). An Investigation of the
Difficulties Faced by EFL Undergraduates in Speaking English.
English Language Teaching, Volume 7(1), pp.9-27.
Arju, S. (2011). A Study on ESL Vocabulary Acquisition Needs and
Classroom Practice: A Bangladeshi Context. Journal of English, 6,
pp.51-67.
Arnold, J. and Brown, H.D. (1999). A Map of the Terrain. Affect in
Language Learning, 6, pp.1-24.
Azizifar, A., Faryadian, E. and Gowhary, H. (2014). The Effect of
Anxiety on Iranian EFL Learners Speaking Skill. Applied and
Basic Sciences, 8(10), pp.1747-1754.
Çağatay, S. (2015). Examining EFL Students‟ Foreign Language
Speaking Anxiety: The Case at a Turkish State
University. Procedia-Social and Behavioral Sciences, 199, pp.648656.
Casado, M.A. and Dereshiwsky, M.I. (2004). Effect of Educational
Strategies on Anxiety in the Second Language Classroom: An
Exploratory Comparative Study between US and Spanish FirstSemester University Students. College Student Journal, 38(1), p.23
Cattell, R. B. (1978). The Scientific Use of Factor Analysis. New York:
Plenum.
Chastain, K. (1975). Affective and Ability Factors in Second‐Language
Acquisition. Language learning, 25(1), pp.153-161.
241
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
Chastain, K. (1988). Developing Second Language Skills: Theory and
Practice. Harcourt Brace Jovanovich, Inc.
Cohen, L., Manion, L. and Morrison, K. (2013). Research Methods in
Education. Routledge.
Comrey, A.L. and Lee, H.B. (2009). A First Course in Factor Analysis.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Cui, J. (2011). Research on High School Students‟ English Learning
Anxiety. Journal of Language Teaching and Research, 2(4), pp.875880.
Debreli, E. and Demirkan, S. (2015). Sources and Levels of Foreign
Language Speaking Anxiety of English as a Foreign Language
University Students with Regard to Language Proficiency and
Gender. International Journal of English Language Education, 4(1),
pp.49-62.
Dordinejad, F.G. and Ahmadabad, R.M. (2014). Examination of the
Relationship between Foreign Language Classroom Anxiety and
English Achievement among Male and Female Iranian High
School Students. Int. J. Language Learn. Appl. Linguistics
World, 6(4), pp.446-460.
Ekström, A. (2013). Foreign Language Communication Anxiety in
Correlation to the Sociolinguistic Variables Gender, Age, Performance
and Multilingual Competence: A linguistic Pilot Study of Swedish
Students’ Attitudes.
Ellis, R. (1994). The Study of Second Language Acquisition. Oxford:
Oxford University Press.
Fariadian, E., Azizifar, A., & Gowhary, H. (2014). The Effect of
Anxiety of Iranian EFL Learners Speaking Skill. International
Research Journal of Applied and Basic Sciences, 8(10), 1747-1754.
242
JEELS, Volume 4, Number 2, November 2017
Field, A. (2013). Discovering Statistics Using IBM SPSS statistics.
London: Sage.
Gan, Z. (2012). Understanding L2 Speaking Problems: Implications
for ESL Curriculum Development in a Teacher Training
Institution in Hong Kong. Australian Journal of Teacher Education.
37 (1), 43-59.
Gardner, R.C. and Lambert, W.E. (1972). Attitudes and Motivation in
Second-Language Learning. Rowley, MA: Newbury House.
Gardner, R.C. and MacIntyre, P.D. (1993). On the Measurement of
Affective Variables in Second Language Learning. Language
Learning, 43(2), pp.157-194.
Gardner, R.C., Tremblay, P.F. and Masgoret, A. (1997). Towards a Full
Model of Second Language Learning: An Empirical
Investigation. The Modern Language Journal, 81(3), pp.344-362.
Gorsuch, R.L. (1983). Factor Analysis (2nd ed.). Hillside, NJ: Lawrence
Erlbaum Associates
Guilford, J.P. (1954). Psychometric Methods. New York: McGraw-Hill.
Hair,
J.F., Anderson, R.E., Babin, B.J. and Black, W.C.
(2014). Multivariate Data Analysis: Harlow: Pearson Education
Limited.
Hewitt, E. and Stephenson, J. (2012). Foreign Language Anxiety and
Oral Exam Performance: A Replication of Phillips's MLJ
Study. The Modern Language Journal, 96(2), pp.170-189.
Hilgard, E. R., Atkinson, R. C., and Atkinson, R. L. (1971). Introduction
to Psychology (5th ed.). New York: Harcourt.
Hinkel, E. ed. (2011). Handbook of Research in Second Language Teaching
and Learning (Vol. 2). Routledge.
243
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
Horwitz, E. (2001). Language Anxiety and Achievement. Annual
Review of Applied Linguistics, 21, pp.112-126.
Horwitz, E.K., Horwitz, M.B. and Cope, J. (1986). Foreign Language
Classroom Anxiety. The Modern Language Journal, 70(2), pp.125132.
Hsu, T.C. (2012). A Study on the EFL Students‟ Speech Related
Anxiety in Taiwan. International Journal of Research Studies in
Language Learning, 1(2).
Huang, H. (2004). The Relationship between Learning Motivation and
Speaking Anxiety among EFL Non-English Major Freshmen in
Taiwan. Unpublished Master's Thesis, Chaoyang University of
Technology.
Hutcheson, G.D. and Sofroniou, N. (1999). The Multivariate Social
Scientist: Introductory Statistics Using Generalized Linear Models.
Thousand Oaks, CA: Sage.
Keong, Y. C., Ali, A. I., & Hameed, F. W. (2015). Speaking
Competence of Iraqi EFL Undergraduates of Garmiyan
University. International Journal of Education and Research, 3(5),
157-170
Kleinmann, H.H. (1977). Avoidance Behavior in Adult Second
Language Acquisition. Language Learning, 27(1), pp.93-107.
Kline, P. (1979). Psychometrics and Psychology. London: Acaderric
Press.
Lee, G. (2009). Speaking up: Six Korean Students‟ Oral Participation in
Class Discussions in US Graduate Seminars. English for Specific
Purposes, 28(3), 142-156.
Liu, M. (2006). Anxiety in Chinese EFL Students at Different
Proficiency Levels. System, 34(3), pp.301-316.
244
JEELS, Volume 4, Number 2, November 2017
Lucas, J. (1984). Communication Apprehension in the ESL Classroom:
Getting Our Students to Talk. Foreign Language Annals, 17(6),
pp.593-598.
Luo, H. (2013). Chinese Language Learning Anxiety and Its
Associated Factors. Journal of Chinese Language Teachers
Association, 48(2), pp.109-133.
Luo, H. (2014). Foreign Language Speaking Anxiety: A study of
Chinese Language Learners. Journal of the National Council of Less
Commonly Taught Languages, 15, pp.99-117.
MacIntyre, P.D. (1998). Language Anxiety: A Review of the Research for
Language Teachers. In D.J. Young, (Ed), Affect in Foreign
Language and Second Language Learning. Boston: McGrawHill.
MacIntyre, P.D. and Gardner, R.C. (1989). Anxiety and
Second‐Language
Learning:
Toward
a
Theoretical
Clarification. Language Learning, 39(2), pp.251-275.
MacIntyre, P.D. and Gardner, R.C. (1991). Methods and Results in the
Study of Anxiety and Language Learning: A Review of the
Literature. Language Learning, 41(1), pp.85-117.
Muhaisen, M.S.M. and Al-Haq, F.A. (2012). An Investigation of the
Relationship between Anxiety and Foreign Language Learning
among 2nd Secondary Students in Second Amman Directorate
of Education. International Journal of Humanities and Social
Science, 2(6), pp.226-239.
Nunan, D. (1999). Second Language Teaching & Learning. Heinle &
Heinle Publishers, 7625 Empire Dr., Florence, KY 41042-2978.
Nunan, D. (2003). Practical English Language Teaching. McGrawHill/Contemporary.
Nunnally, J.C. and Bernstein, I.H. (1978). Psychometric Theory. New
York: McGraw-Hill Inc.
245
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
Occhipinti, A. (2009). Foreign Language Anxiety in in-Class Speaking
Activities: Two Learning Contexts in Comparison. Unpublished
Master's thesis: The University of Oslo.
Osborne, J.W. and Costello, A.B. (2009). Best Practices in Exploratory
Factor Analysis: Four Recommendations for Getting the Most
from Your Analysis. Pan-Pacific Management Review, 12(2),
pp.131-146.
Oxford, R.L. (1999). Anxiety and the Language Learner: New
Insights. Affect in Language Learning, pp.58-67.
Öztürk, G. and Gürbüz, N. (2013). The Impact of Gender on Foreign
Language Speaking Anxiety and Motivation. Procedia-Social and
Behavioral Sciences, 70, pp.654-665.
Paakki, H. (2013). Difficulties in Speaking English and Perceptions of
Accents: A Comparative Study of Finnish and Japanese Adult
Learners of English. Unpublished Master‟s Thesis, School of
Humanities, University of Eastern Finland.
Palacios, L.M. (1998). Foreign Language Anxiety and Classroom
Environment: A Study of Spanish University Students. Unpublished
Doctorate Thesis: University of Texas, Austin.
Pappamihiel, N.E. (2002). English as a Second Language Students and
English Language Anxiety: Issues in the Mainstream
Classroom. Research in the Teaching of English, pp.327-355.
Park, G.P. and French, B.F. (2013). Gender Differences in the foreign
Language Classroom Anxiety Scale. System, 41(2), pp.462-471.
Price, M.L. (1991). The Subjective Experience of Foreign Language
Anxiety: Interviews with Highly Anxious Students. Language
Anxiety: From Theory and Research to Classroom Implications,
pp.101-108.
246
JEELS, Volume 4, Number 2, November 2017
Priyanto, A. (2013). The Correlation between English Grammar
Competence and Speaking Fluency of Eleventh Grade Students
in SMAN 1 Sidoarjo. EJournal Unesa. 1 (1), 1-6.
Qaddomi, H. (2013). Investigating Al-Quds Open University
Students' EFL Learning Anxiety. An-Najah Humanities, 27(7),
pp.1533-1562.
Rachman, S. (2004). Anxiety: Clinical Psychology: a Modular Course.
Psychology Press.
Saito, Y. and Samimy, K.K. (1996). Foreign Language Anxiety and
Language Performance: A Study of Learner Anxiety in
Beginning, Intermediate, and Advanced‐Level College Students
of Japanese. Foreign Language Annals, 29(2), pp.239-249.
Scott, M.L. (1986). Student Affective Reactions to Oral Language
Tests. Language Testing, 3(1), pp.99-118.
Şimsek. (2015). The Role of Anxiety in Foreign Language Classes: A
Focus on Gender. International Journal of Language Academy, 3(4),
pp.70-83.
Skehan, P. (1989). Individual Differences in Second Language Learning.
London: Edward Arnold.
Speilberger, C.D. (1983). Manual for the State-Trait Anxiety Inventory
(form Y1). Palo Alto, CA: Consulting Psychologists Press.
Steinberg, F.S. and Horwitz, E.K. (1986). The Effect of Induced
Anxiety on the Denotative and Interpretive Content of Second
Language Speech. Tesol Quarterly, 20(1), pp.131-136.
Stevens, J.P. (2009). Applied Multivariate Statistics for the Social Sciences.
London: Routledge.
Syakur. A. (1987). Language Testing and Evaluation. Surakarta: Sebeles
Maret University Press.
247
Abrar, An Investigation into Indonesian EFL University Students’
Speaking Anxiety
Tabachnick, B.G., Fidell, L.S. and Osterlind, S.J. (2001). Using
Multivariate Statistics. Boston, MA: Allyn and Bacon.
Tabachnick, B.G., Fidell, L.S. and Osterlind, S.J. (2013). Using
Multivariate Statistics. Upper Saddle River: Pearson Education
Tercan, G. and Dikilitaş, K. (2015). EFL Students‟ Speaking Anxiety: a
Case from Tertiary Level Students. ELT Research Journal, 4(1),
pp.16-27.
Tianjian, W. (2010). Speaking Anxiety: More of a Function of
Personality than Language Achievement. Chinese Journal of
Applied Linguistics, 33(5), pp.95-109.
Voorhees, M.G. (1994). Foreign Language Anxiety and Gender.
Unpublished Master Thesis: Iowa State University.
Wang, P. and Roopchund, R. (2015). Chinese Students‟ EnglishSpeaking Anxiety in Asking Questions in the MSc TESOL
Classroom. International
Journal
of
English
Language
Teaching, 2(2), pp.1-18.
Williams, K.E. and Andrade, M.R. (2008). Foreign Language Learning
Anxiety in Japanese EFL University Classes: Causes, Coping,
and Locus of Control. Electronic Journal of Foreign Language
Teaching, 5(2), pp.181-191.
Young, D.J. (1991). Creating a Low‐Anxiety Classroom Environment:
What Does Language Anxiety Research Suggest?. The Modern
Language Journal, 75(4), pp.426-437.
Zhao, A. and Whitchurch, A. (2011). Anxiety and Its Associated
Factors in College-Level Chinese Classrooms in the US. Journal
of the Chinese Language Teachers Association, 46, pp.21-47.
248