Second Language Speaking Anxiety Among Malaysian Postgraduate Students at a Faculty of Education

This study examined the level of L2 speaking anxiety among school teachers undertaking a master’s degree at the International Islamic University Malaysia and sought to determine if the anxiety level would differ by gender, school location and teaching subject. A convenience sample of 290 teachers completed a 12-item survey on L2 speaking anxiety adapted from Nazir et al. (2014). The survey data were analyzed using the Rasch measurement modeling for polytomous data, independent samples t-test and one-way ANOVA. Overall, the respondents did not show high levels of speaking anxiety, (Mean -.89 logits). Most were largely worried about being able to express themselves effectively in English (-.66 logits), making mistakes in speaking (-.31 logits), and how lecturers would react to their mistakes (-.38 logits). The sample’s speaking anxiety scores were found to differ significantly by teaching subject, while no statistically significant differences were observed in regard to gender and school location. The results imply that the teachers need to be helped in terms of confronting the factors that cause them to be anxious about speaking in English. In terms of research, in-depth qualitative studies need to be undertaken to further understand the nature of L2 speaking anxiety among schoolteachers, while quantitative studies with larger samples are recommended to uncover underlying factors of speaking anxiety.


INTRODUCTION
a number of factors are attached. In addition, research has identified anxiety as one of the major obstacles which ESL/ EFL learners encounter in the course of their learning (Wu, 2010;Zheng, 2008). This phenomenon has gained the attention of language researchers for the past few decades due to its effects confirmed on one-third of foreign language learners (Ahmad, Sabariah, Nimehchisalem & Ramiza, 2019;Ellis, 2008;Horwitz, 2001;Wang, 2014).
It has been observed that higher education students who speak English as a second language face speaking anxiety which adversely affects their speaking performances both in and out of class activities (Ahmed et al., 2017;Elaldı, 2016;Hasrul et al., 2013;Mohammad et al., 2016;Saranraj & Meenakshi, 2016). Moreover, some studies identify English language anxiety as having a debilitative effect on students' academic performances (Amiri & Ghonsooly, 2015;Kumar, 2018;Saad, 2016). In other words, students who feel highly uneasy in communicating in English would be found to have relatively low academic performance, and vice versa.
English language is the main medium of instruction at the International Islamic university Malaysia as well as many other public universities in Malaysia. Therefore, both undergraduate and postgraduate students must possess a high level of competence in the four main English language skills, namely listening, speaking, reading, and writing, but with an increased proficiency in the productive ones (speaking and writing) since they are engaged in class activities that require the use of English speaking skills, and research tasks which involve writing skills. However, research shows a high level of English language anxiety among Malaysian university students due to certain factors which include fear of making mistakes, fear of negative judgments, and emphasis of culture on the need for positive outlook and saving face (Chin, Hie Ling & Jiin Yih, 2016;Siti Faridah Kamaruddin & Nabilah Abdullah, 2015). Therefore, this study is an attempt to identify the level of English-speaking anxiety and its contributing factors among school teachers enrolled in a master of education program at the International Islamic University Malaysia. It also aims to ascertain the possible differences in the level of Englishspeaking anxiety due to selected variables, namely gender, school location and teaching subject. The study has adapted the scale developed by Nazir et al. (2014), the mini version of FLCAS (Foreign Language Classroom Anxiety Scale) of Horwitz et al. (1986).

English Language Anxiety
Research has shown speaking anxiety as a phenomenon that is peculiar to ESL learners of various levels (Ahmed et al., 2017;Elaldı, 2016;Hasrul et al., 2013;Mohammad et al., 2016;Saranraj & Meenakshi, 2016). Moreover, speaking nervousness and fear of being negatively evaluated do not decrease as these students get elevated into higher levels in their studies. Rather, the more advanced the level reached in their studies, the higher their level of anxiety would be (Elaldı, 2016). A number of studies pinpoint cultural and gender diversity in the classroom, and learner-centered teaching approach handled by authoritative teachers as the prominent factors that unfavorably affect ESL/EFL learners' speaking anxiety (Hashemi & Abbasi, 2013;Young, 1991;Zhang & Zhong, 2012). For instance, in most parts of the Arab world, gender has been identified as an influential factor of speaking anxiety. Female students have been found to record higher levels of anxiety than males in seven Arab countries, but no considerable difference was identified between males and females only in Iraq, Palestine, and Jordan (Ahmed & Alansari, 2004). Nazir et al. (2014) undertook a study to find out if there was a considerable degree of English-speaking anxiety among a group of Pakistani ESL intermediate learners. In efforts to achieve this objective, a questionnaire developed by Horwitz et al. (1986) was adapted to measure the level of speaking anxiety of 50 learners who were chosen from among the whole population of 253 intermediate students at a Pakistani college. The findings indicate that learners suffer seriously from communication nervousness and fear of being negatively evaluated which impede their classroom English speaking skill. A huge number of the respondents attributed their speaking anxiety to inadequate preparation before engaging in classroom speaking activities.
Mohammad Batiha et al. (2016) also conducted a research to identify the circumstances leading to speaking anxiety among EFL learners as well as gender differences pertaining to these circumstances in the Jordanian context. In order to achieve this end, Foreign Language Speaking Anxiety Scale (FLSAS) developed by Horwitz et al. (1989) was used to gather information from a group of 112 freshmen chosen from among the students of Jadara University. The findings show fear of being negatively evaluated, insufficient preparation, fear of speaking in public and timorousness, and general class speaking anxiety as the factors leading respondents to develop ELF speaking anxiety. However, the results indicate no considerable gender differences in the factors that influence speaking anxiety. Elaldi (2016) undertook a study to find out the speaking anxiety level of learners of English language and literature at Cumhuriyet University, Turkey Adopting the Foreign Language Classroom Anxiety Scale (FLCAS) of Horwitz at al. (1989). The researchers elicited data from a group of 98 students both in their preliminary level and fourth year. The findings indicate that students experience a moderate level of foreign language anxiety both in their preparatory and fourth levels even though they were found somewhat more anxious in the fourth year. Moreover, the male students were found to have higher foreign language anxiety than the female students.
Another pertinent study was carried out by Hasrul et al. (2013) to examine the level of English anxiety of Malaysian gifted learners in an ESL context. To actualize this objective, the researchers collected data from a group of 119 gifted learners (aged between 15 and 16 years with a minimum of 9 years of English language learning) chosen from the Malaysian National University. The results point out that gifted learners experience a certain level of anxiety in English language which is relatively lower than that of non-gifted learners who were examined in previous studies. Furthermore, gender difference is considered an insignificant factor of language anxiety even though males were found to be slightly less anxious than females. Furthermore, the results show that English language anxiety is debilitative to their academic achievement.
In their study, Jomaa and Jupri (2014) sought to find out the causes of second/foreign language anxiety through the experiences of students enrolled in a Postgraduate Linguistics program at a Malaysian university. Adopting a qualitative approach, this study conducted a face-to-face interview to obtain information from a group of nine students who were purposively chosen from seven different nationalities, including Malaysia. The findings are categorized into two main causes of second/foreign language anxiety, namely, external and internal factors. The former category (i.e. foreign language anxiety caused by factors other than the anxious student him/herself)includes, teachers' unfriendly attitude towards students in English speaking classes, student-audience asking multiple questions and criticizing their classmates' presentations, classroom anxiety (i.e. differences in the learning situations of the participants), and course assessment. With regards to the latter (i.e. the factors within the student him/herself), fear of committing mistakes and fear of failure were found as sources of foreign language anxiety. Ahmed et al. (2017) also conducted their research to identify factors affecting English language oral communication anxiety among a group of postgraduate students at a university in Pakistan, and to determine whether there was a considerable difference in gender (if any) with regards to speaking anxiety. The researchers adapted the foreign language class anxiety questionnaire developed by Horwitz et al. (1986) with which data were collected from a group of 240 postgraduate students who were enrolled in a course titled, "Functional English Language". The findings reveal that the focused postgraduate students experience English language anxiety in their learning activities owing to their challenging classroom English tasks. Moreover, the majority of these students attributed their English language anxiety to the following: inability to communicate fluently in English language, failure to use error-free sentences in their conversations, and insufficient linguistic competence with which to express themselves in a wide range of communicative situations. The findings reveal no considerable difference between genders regarding the English language anxiety.
Additionally, Zhiping and Paramasivam (2013) undertook a study to determine the causes of English communicative anxiety among international students at a Malaysian university, and the strategies they used to mitigate their anxiety when speaking. It also sought to examine teachers' perspectives and reactions towards learners' speaking anxiety, and students' perspectives of teachers' reactions towards their anxiety. In efforts to achieve these objectives, the researchers purposively chose eight respondents (3 Nigerians, 3 Iranians, and 2 Algerians) from a group of PhD international students who were enrolled for a semester course at the earlier mentioned university. Using semi-structured interview and observations, data were elicited from the respondents. The findings signal that the majority of the Nigerian students do not feel anxious when using English in class, while Algerian and Iranian students experience speaking anxiety, which, according to the findings, is due to fear of effecting conversation in public, fear of being negatively evaluated, and fear of being inaccurate while speaking. As for the strategies used to reduce their speaking anxiety, students were found to use silence with a view to avoiding oral communication in class, avoiding eye-contact with the teacher, mingling with friends with the same first language, and expressive reactions (e.g. smiling or laughing) to mask one's true feelings of anxiety. Concerning teachers' strategies to mitigate students' speaking anxiety, the findings reveal the use of jokes and story narration, acknowledgement of students' responses, and showing positive gestures. According to these findings, students showed their good feelings and perceptions towards their lecturers' strategies to reduce their speaking anxiety.

English Anxiety and Academic Achievement
Much research has reported English language anxiety as a phenomenon with a debilitative effect on students' academic performance (Amiri & Ghonsooly, 2015;Kumar, 2018;Saad, 2016). This seems to imply that students who are highly anxious at English language are usually found to display relatively low academic performance, and vice versa. Kumar (2018) carried out a research to identify the correlation between English language anxiety and academic achievements of students in Malda District, India. In efforts to achieve this end, normative survey method of descriptive research was employed to elicit data from 266 grade six students who were randomly chosen from a number of rural and urban Malta District secondary schools. The findings show that students who are less anxious at English speaking have a higher level of academic achievement compared to their counterparts who are highly anxious. Amiri and Ghonsooly (2015) also undertook a study to examine the relation between tertiary students' English language anxiety and their academic achievements in their examinations. To achieve this objective, the researchers used two instruments (the Persian version of Foreign Language Classroom Anxiety Scale (FLCAS) and students' terminal achievement test to collect data from a sample of 258 freshmen who were conveniently chosen from different scientific departments of the Gonabad University. The findings reveal a statistically significant negative correlation between English language anxiety and students' achievements in their examinations. In other words, the higher the students' level of English language anxiety, the lower their level of achievement in examinations. Females were also found to be more anxious at English language than males, and that negatively affects their achievements in examinations more than it does the males.
In his study, Saad (2016) also investigated the connection between English language anxiety and learners' academic achievements among Moroccan students. In attempts to actualize this aim, the researcher employed Cattle's anxiety questionnaire and achievement test to elicit data from a cohort of English language students selected at the faculty of Arts and Humanities, Moulay Ismail University, Meknes. The findings show a statistically significant inverse relation between English language anxiety and academic achievements of the respondents. That is to say, respondents with a high level of English language anxiety have fewer academic achievements compared to their counterparts with a low level of English language anxiety.

METHODOLOGY
This study used the cross-sectional survey design with an adopted set of questionnaire to measure the participants' level of English-speaking anxiety and identify its factors. This questionnaire is a mini version of FLCAS (Foreign Language Classroom Anxiety Scale) developed by Horwitz et al. (1986) on a five-point Likert scale, (1 = strongly agree, and 5= strongly disagree (see Nazir et al. (2014). However, in this study, the Likert scale is arranged as 1 = strongly disagree and 5= strongly agree for clarity purposes. Through a convenience sampling method, a sample of two hundred and ninety teachers (n= 290) who were first year postgraduate students of a master program at the Kulliyyah of Education, International Islamic University Malaysia, completed the adopted survey on the English language speaking anxiety. The collected responses were keyed into SPSS (version 23) to ensure the accuracy of the data before running the final analysis (Pallant, 2013). Female respondents made up 60% of the sample (n=174), while male respondents 40% (n=116). Among the 290 respondents 63.4 % (n=184) were teaching at schools located in urban area; 35.2% (n=102) teaching at schools located in rural areas, and 1.4 % (n=4) were reported as missing. A majority were teaching non-scientific subjects (53.1%, n=154). The remaining sample were teaching scientific subjects (12.4%, n= 36), other teachers were teaching English subject (24%, n=70); and 10 % (n=30) were reported as missing. The Rasch Measurement Modeling for polytomous data was utilized to analyze the collected data, using Winsteps Rasch measurement software, version 4.1.0. (2018). In Rasch Model, it is possible to display all the items and students on one interval scale with odd units (logits) (Bond & Fox, 2015;Linacre, 2019) and see clearly the position of each item in relation to other items distributed on the scale and the ability of each person in relation to other persons. The most difficult items to be endorsed and the most able persons are positioned at the top of the scale and vice versa. Rasch analysis can also ensure the items are working meaningfully to measure the construct by looking at the item fit statistics. In essence, the Rasch model ensures that the research instrument has met the measurement requirements (i.e. item polarity, fit statistics, unidimensionality and consistency) as explained in section 4. The Rasch model output person data file was converted into SPSS version 23 to conduct the inferential analysis (i.e. the independent samples t-test and the one-way analysis of variance (ANOVA) test). The former was used to determine the significant differences in the levels of English speaking anxiety due to gender and school location (rural and urban), and the latter was used to determine the differences due to teaching subject (Scientific, Non-Scientific and English). It is assumed these selected variables influence the levels of English speaking anxiety. The results were depicted in tables and figures.

RESULTS
Reliability refers to the reducibility of items and persons ordering in Rasch analysis (Bond & Fox, 2015). Table 1 shows that the reliability of item difficulty measures is high (0.94) and the separation index for items is (3.79) > 2. The person reliability is also high (0.90) with person separation index (3.04) > 2. Table 1 also shows that the point-measure correlation coefficients for the 12 items are positive and ranged from 0.45 to 0.82 implying that the items are working or moving together in the same direction to define the construct being investigated (i.e. speaking anxiety). For the fit statistics (infit and outfit Mean-square statistics), Table 1 shows that all items have infit and outfit meansquare values within the recommended range (0.5 -1.5), except for item No. 1 (2.1 logits) > 1.5 logit. However, after deleting a few of the most misinfiting persons > 1.5, the item showed good fit (1.33 logits). This ensures the meaningful contribution of the items to measure the construct as expected by the model (Bond & Fox, 2015). Therefore, all the items were included in the final analysis. Finally, Table 1 shows that the unidimensionality of the scale is not violated. The variance explained by the measures is 56.8%, and the largest factor extracted from the residuals is equivalent to 1.99 units, which has the strength of about 2 items (Linacre, 2019).
Overall, the Rasch analyses reveal that students did not show high levels of English-speaking anxiety as indicated by the person difficulty measure (-.89) which is lower than the item difficulty measure (0.00) (Figure 1). It was not easy for most students to endorse the items on speaking anxiety (Table 1 and Figure 1). The Item-Map (Figure 1) clearly displays the distribution and hierarchy of all items on one interval scale. The highly endorsed items are placed towards the lower part, and the least endorsed ones are placed towards the upper part of the scale. The easiest item to be endorsed by the participants was ANX 2 "I worry about failing to express myself effectively in English language, -0.66 logit"; followed by ANX 12 "I feel afraid that the lecturer will highlight grammatical mistakes while speaking, -0.38 logit"; ANX 10 "I am worried about making mistakes while speaking, -0.31 logit"; ANX 3 "I get nervous when the lecturer asks questions in English, -0.22 logit"; ANX 11 "I still get worried about speaking, even after preparation, -0.17 logit"; and ANX 5 "I feel anxious if someone asks me to explain something in English, -0.4 logit". It seems that the students were mostly worried about speaking or expressing themselves effectively in the classroom. This might be due to their worry about their mistakes, getting lecturers' feedback and being asked to explain or answer orally in English. On the other hand, Figure 1 shows that the most difficult item to be endorsed by the participants was ANX 4 "The more I try to speak English in the classroom, the more confused I get, 0.64 logits", followed by ANX 6 "I get confused when I speak English in my class, 0.47 logit"; ANX 1 "I usually feel easy expressing myself in English language in the classroom (Recoded), 0.37 logit"; ANX 7"I do not feel confident when the lecturer asks me to participate in English in the classroom, 0.16 logit"; ANX 9 "I feel afraid that other students will criticize me when I start speaking English in the classroom, 0.15 logits"; and ANX 8 "It embarrasses me to volunteer the answers in the classroom, 0.4 logit". These items indicate that students in general do not feel confused, embarrassed or worried about being criticized by their friends when speaking in English. They still have a sense of confidence.
The inferential statistics (independent sample t-test and one-way analysis of variance (ANOVA)) were carried out to compare the students' English-speaking anxiety mean scores with the selected demographic variables (gender, school location and teaching subject) ( Table 2). The independent samples t-test shows no significant difference in the speaking anxiety mean scores for both gender and school location (p>.05). The one-way ANOVA shows a significant difference in speaking anxiety scores for teaching subject (scientific,  non-scientific and English) (p<.05). Post-hoc analysis using tukey test indicates that the participants who teach English have the least level of speaking anxiety, followed by those teaching scientific subjects, and those teaching non-scientific subjects, with significant differences between all the categories, p<0.05.

DISCUSSION
From the above analyses, one can reason that the participants experience some degrees of English-speaking anxiety in class activities which involve the presence of their lecturers and professors. The items they endorsed as the causes of their English-speaking anxiety revolve around the three variables identified by Horwitz et al. (1986), i.e. speaking apprehension, fear of negative evaluation, and fear of making mistakes. This anxiety (of various kinds) experienced stems from two factors which can be categorised as linguistic and non-linguistic. Under the linguistic factors, participants were anxious about making grammatical mistakes, and worried about the comments the lecturers would make on their grammatically flawed conversations. On the other hand, the non-linguistic factors represent their failure to speak English in public as a result of having negative perception of the audience, especially their professors. This finding is in Figure 1. Item-person map agreement with Amiri and Puteh (2018) who conducted an English-speaking anxiety research on a group of international doctoral students studying in various Malaysian universities. They found out that those students experience speaking anxiety caused by their inadequate linguistic competency, inadequate knowledge of the topic being presented, negative perception of the examiners, and other factors such as linguistic deficiency of examiners in understanding their presentation. Moreover, the lack of statistically significant difference in the mean scores on the basis of gender suggest that both male and female adult learners in the ESL/EFL context experience an almost equal level of speaking anxiety. This underpins findings of relevant studies which show no gender difference in speaking anxiety (Ahmed et al., 2017;Hasrul et al., 2013;Mohammad Batiha et al., 2016). With regards to the school location which is reported to have no effect on the participants' English-speaking anxiety, this disagrees with the study undertaken by Piechurska-Kuciel (2012) in which she reported rural secondary students as being more anxious at English speaking than their counterparts in urban centers. Finally, the one-way ANOVA results, which indicate that the English teaching participants are the least anxious at speaking, suggest that the linguistic capability of this group has helped them enhance their confidence at speaking, unlike other groups who might not be completely free from inaccurate construction of sentences when expressing their ideas. The ability possessed by Science teaching participants to showcase a lower level of speaking anxiety than non-science teachers is due to the nature of their taught subjects whose terms and expressions do not usually have equivalents in their mother tongues. This phenomenon might have got them acclimatized to the use of English and enhanced their confidence in speaking.

CONCLUSION
The Rasch analyses revealed that the school teachers enrolled in a master of education program at IIUM did not show a high level of English-speaking anxiety. The overall level of anxiety found was caused by speaking apprehension, fear of negative evaluation, and fear of making mistakes as identified by Horwitz et al. (1986). Moreover, the analysis of the independent sample t-test showed that there is no statistically significant difference in English speaking anxiety due to gender and school location, while the One-way ANOVA analysis' results indicate a statistically significant difference in speaking anxiety among the participants due to their teaching subjects. The English teaching participants were shown as the least anxious, followed by those teaching Sciences, whereas the participants teaching non-science subjects were indicated as the most anxious at speaking. In efforts to address this speaking anxiety issue, experts should work further on how to address the factors leading to speaking anxiety. Lecturers should also motivate students to use English in the classrooms and avoid intense criticisms when responding to students' mistakes. Students are also encouraged to increase their use of English language in different situations and build on their confidence as this can lessen their level of speaking anxiety. This research recommends more studies on English speaking anxiety in order that the factors responsible for this problem be researched with more in-depth and their possible solutions could be further identified.