An Empirical Analysis of the Relationship between Foreign Language Reading Anxiety and Reading Strategy Training

Anxiety is a psychological factor commonly associated with such feelings as fear, apprehension and uneasiness. It is an individual’s affective reaction to a perceived or a real threat (MacIntyre, 1995). Foreign language reading anxiety (FLRA) refers to one’s negative attitudes which may, to some extent, account for the inhibition that s/he suffers from while reading a text in a foreign language (FL). Given that anxiety associated with FL has a strong impact on learners’ overall achievement (Cheng, Horwitz & Schallert, 1999; Sellers, 2000), FLRA may feature as a potent predictor of success in FL reading courses. Although numerous studies have focused on general FL anxiety, research aiming at unraveling FLRA including its various aspects has been far from being satisfactory. Hence, this study investigates the relationship between FLRA and reading strategy training in FL reading courses. Participants divided into experiment and control groups were selected through convenience sampling. Data were elicited through Foreign Language Reading Anxiety Scale (FLRAS) (Saito, Horwitz & Garza, 1999) and semi-structured interviews. The study indicates that though there has been a decrease in the control group’s FLRA levels, results of the experiment group have revealed intriguing findings. Still, analyses of the interviews with members of the experiment group have yielded divergent remarks regarding their perceptions of their FLRA levels.


Introduction
Foreign language learning (FLL) is an intricate process that requires investing considerable resources in terms of time, effort and materials, yet one that does not warrant ultimate achievement.As FLL by nature comprises a complex set of interactive factors, attempts to explain the process from a mere linguistic perspective are likely to prove futile (Schumann, 1987).Instead, a broader stance embracing social and affective variables that have a role to play in the process may yield a more precise depiction of the path to success in FLL.As such, anxiety related to FLL is one of the key factors that may fill in some lacuna in accounting for possible failures in FLL (Brown, 2000).Though foreign language anxiety (FLA), a set of psychological factors specific to foreign language classroom (Horwitz, Horwitz & Cope, 1986), is commonly associated with oral skills (particularly speaking), much research (Cheng, 2002;Elkhafaifi, 2005;Saito, Horwitz & Garza, 1999) has been conducted in the last two decades that indicate relationships between anxiety and specific language skills.For instance, Elkhafaifi (2005) reveals that listening anxiety is positively related to but distinct from general FLA.Similarly, Cheng, Horwitz & Schallert (1999) put forth that FLA is a broader construct as their study denotes that the type of anxiety associated with writing tasks is related to but distinguished from overall FLA.
As to reading, it holds a vital role in acquiring new information and developing one's thinking (Ennis, 2006).In contrast to oral skills which entail a dialogic construction of meaning through shared attempts of two or more parties, Shariati & Bordbar (2009) argue that reading is a more "individual act"(183).Furthermore, the fact that foreign language (FL) learners have a well-established reading experience rooted in their L1 reading activities may be a factor to help them with difficulties in FL reading (Upton & Lee-Thompson, 2001).Still, much research (Sellers, 2000;Wu, 2001) highlights that anxiety stands out as a prominent factor which may considerably affect FL reading process.Indeed, simply the perception of reading in a language different from one's L1 may aggravate his/her anxiety level since reading requires simultaneous activation of several cognitive processes (like attention and memory) to build the meaning.
This study aims at unearthing the relationship between reading strategy instruction and foreign language reading anxiety (FLRA).More specifically, it investigates if reading strategy training makes any difference in FLRA one suffers from.Given the dearth of research particularly addressing reading anxiety in FL classrooms, this study intends to fill in some space in the literature related to the impact of classroom activities on FLRA levels.With no concern about what specific types of strategy are used, this study highlights fluctuations in learners' FLRA levels over a period of reading strategy training.

Literature Review
FLRA refers to feelings of uneasiness, apprehension or stress one suffers from while s/he is reading a text in FL.Following the interest in research addressing anxiety related to language specific skills, it has been well-established that FLRA does exist and is related but distinguishable from FLA.Despite vast reading experiences one may have in his/her L1, reading in an FL is considered much more demanding.Also, several studies (Horwitz et. al., 1986;MacIntyre, 1995) show that anxiety is merely a matter of perception.In other words, one's perception that his/her FL reading ability is poor may augment his/her FLRA levels.In a preliminary study on FLRA, Saito et. al. (1999) posits that learners' FLRA levels are positively correlated with their self-rated difficulty of reading in FL.As to what stimulates FLRA, Kuru-Gönen (2005in Aydın & Kuru-Gönen, 2012) unravels that there are basically three sources of FLRA.These involve factors associated with a) readers (such as inadequate implementation of reading strategies and lack of motivation), b) the text (like complex linguistic structures and unfamiliar topics), and c) the program (e.g.concerns about assessment).Moreover, Saito et. al. (1999) emphasizes that unfamiliar script system and cultural components presented in the text may induce anxiety in FL reading classrooms.
FLRA has been reported to correlate with various factors including overall FLA levels (Saito et. al. 1999;Wu, 2011), reading comprehension (Sellers, 2000;Zhao, Dynia & Guo, 2013), performance (Zin& Rafik-Galea, 2010), proficiency (Tsai & Li, 2012) and gender (Matsuda & Gobel, 2004;Shariati & Bordbar, 2009).Firstly, various studies (Sellers, 2000;Wu, 2011) indicate that one's FLA levels positively correlate with his/her FLRA levels.For instance, Wu (2011) has examined FLRA levels in Taiwanese learners of English and concluded that students with higher levels of FLA tend to have proportionately higher levels of FLRA.Likewise, much research (Sellers, 2000;Zin & Rafik-Galea, 2010) has found that learners with higher levels of FLRA have more difficulty in comprehending texts in FL.MacIntyre (1995) postulates that regardless of causality between FLA and achievement, anxiety blocks learners' mind and diverts their attention to task-irrelevant thoughts, which in turn interfere with their comprehension.Not surprisingly then, several studies (Saito et. al., 1999;Zhao, Dynia & Guo, 2013;Zin & Rafik-Galea, 2010) report inverse relationship between FLRA and performance.Saito et. al. (1999) maintain that based on their final grades, learners with higher levels of FLRA have received grades lower than their less anxious counterparts.Moreover, Tsai & Li (2012) shows that FLRA is negatively correlated with one's reading proficiency i.e. higher proficiency levels one has, lower FLRA levels s/he is supposed to have.Nonetheless, Brantmeier (2005), in a study conducted with 92 advanced level learners of Spanish, reports that with the highly proficient learners, it is the nature of language classroom rather than reading tasks that aggravate learners' anxiety levels.Finally, much research (Kartal, 2011;Matsuda & Gobel, 2004) reveals that gender is not a factor affecting FLRA levels because no significant difference between males and females regarding their FLRA levels has been reported so far.Now that FLRA is aggravated by difficulties in comprehending FL texts (Sellers, 2000), one can assume that reading strategies may make a facilitative difference in learners FL reading abilities.Reading strategies are tools consciously implemented to overcome problems encountered in FL reading (Lien, 2011;Brantmeier, 2002).McNamara (2011) proposes that utilizing adequate strategies ensures deep and proper understanding of reading texts.Though different researchers (Oxford, 1990;McDough, 1995) have introduced different categorizations, reading strategies fundamentally comprise setting objectives, using schemata, skimming, scanning, guessing, paraphrasing, making inferences, and summarizing.Obviously, there is widespread consensus over the use of teaching FL learners reading strategies.Saito et. al. (1999) emphasize that reading strategy instruction may help learners become more efficient readers and eliminate their unrealistic expectations regarding FL reading.Nevertheless, instructors may face some problems like finding appropriate materials, unwillingness on the part of learners and requirements imposed by the curriculum, to name just a few (Hilden & Pressley, 2007).Yet Singhal (2001) clearly illustrates that despite variations, learners benefit from several reading strategies to acquire, process, store and retrieve information.
Given the plethora of research on reading strategy use in FL classes, Brantmeier (2002) asserts that there is no particular strategy superior to others in promoting FL reading comprehension.Interestingly enough, Feng & Mokhtari (1998) surmise that learners employ reading strategies in FL reading more frequently than they do in L1 reading.Also, the type of strategy they use differs depending on the difficulty of the text (Feng & Mokhtari, 1998).However, Sellers (2000) argues that overall learners' strategy use in FL reading is far from being satisfactory as they still utilize low-level strategies like translation.Likewise, Saito et. al. (1999) postulate that translation remains to be the preliminary strategy to which learners resort in FL reading.Accordingly, Al-Nujaidi (2003) reports positive relationship between FL learners' reading proficiency and the complexity of the reading strategies they implement in reading texts.Last but not the least, research indicates a negative relationship between reading strategy use and FLRA levels as Lien (2011) stresses that the higher FLRA levels one has, the less s/he is likely to implement reading strategies.

Participants
There were 39 freshmen students who were studying at English Language Teaching (ELT) department of a public university.A background information questionnaire was distributed to collect data related to the participants' age, gender, mother tongue, high school they graduated from as well as their parents' educational degree.They were enrolled in the "Advanced Reading Course" offered in the first year of the curriculum.The mean age of the participants all of whom were native speakers of Turkish was 20.The participants were selected through convenience sampling (Cohen, Manion & Morrison, 2007) because they were the only group of freshmen students to whom the researchers could have access.Also, the study recruited only freshmen on the basis of MacIntyre's (1999) suggestion that it is the repeated negative experiences FL learners have in the early stages of formal training that determine their subsequent anxiety development.

Instruments
The present study employed Foreign Language Reading Anxiety Scale (FLRAS) (Saito et. al., 1999) and semi-structured interviews in order to gauge data from the participants.The FLRAS is designed to elicit learners' perceptions about possible difficulties pertaining different aspects of FL reading.It comprises 20 items scored on a 5-point likert-type scale.The theoretical score range in the FLRAS is 0-100.Four items (i.e.items 12, 13, 14 and 18) require reverse-scoring.Thereby, a higher score received from the FLRAS stands for higher FLRA levels.Saito et. al. (1999) posits that further studies on FLRA should reinforce their findings with more in-depth tools because anxiety is too complex a construct to be explained through mere administration of scales.Hence, this study utilized semi-structured interviews to get a deeper insight into the participants' experiences in the reading course.However, one potential risk was that the participants' could improvise to please the researchers.For this reason, the interviews were conducted after all assessment and evaluation period was completed.

Procedure
This study was conducted in the fall semester, which was the first time the participants in ELT program would get instructions on reading.The participants were divided into two groups: group A with 20 participants was randomly selected as the experiment group and group B consisting of 19 participants as the control group.The participants in both groups took the FLRAS at the beginning and end of the semester.Each time the FLRAS was administered took about 20 minutes.A couple of tests on the Statistical Package for Social Studies (SPSS) Version 17.0 were employed to analyze the participants' responses to the FLRAS.
Though the prescribed curriculum was followed in the control group, the curriculum applied in the experiment group was organized in a way to create in the participants familiarity with various reading strategies, and inspire them to utilize those strategies in FL reading.Thereby, the course covered a wide range of strategies such as activating schemata, skimming, scanning, keeping reading journals and reflective thinking.However, as mentioned above, no specific strategy has yet been prescribed as the best (Brantmeier, 2002).Thus, this study was not concerned about what strategy the participants used most or what particular strategy was considered the most relevant to the participants' FLRA levels.
Although both groups took FLRAS twice, it was only the participants in the experiment group who were invited to participate in the interview.Of the total of 20, six participants volunteered to partake in the interviews.As a precaution to eliminate researcher bias (mentioned above), the interviews ranging between 8 to 15 minutes in length were conducted one week after the semester was over.Therefore, the data collection was completed in 17 weeks.The interviews were conducted in the participants' mother tongue, for they could better express their feelings and ideas in their L1.As for the analysis of the interviews, the researchers independently translated them into English and then, compared the translated versions of each interview with one another to spot possible differences.After this process was completed for each interview, disagreements were negotiated through discussion until a consensus was achieved.

Findings
This study aimed to track any changes that could occur in the FLRA levels of FL readers on the basis of the reading strategy training they took in the first year of their undergraduate degree.The mean score that the experiment group received in the first application of the FLRAS was 52.2 whereas it was 58.8 for the control group.The post-test application of the FLRAS hinted an increase in the FLRA levels of both groups as the mean scores were identified to be 63.5 and 62 for the experiment and control groups, respectively.A paired samples t-test was conducted to figure out whether the changes in the FLRAS scores were statistically significant.The results indicated that despite the slight increase in their mean scores, the change in the pre-and post-test FLRAS scores of the control group was not statistically significant.However, the change in the mean scores of the experiment group did show a statistically significant difference in favor of the pre-test scores (See Table 1).That is it was hypothesized that the reading strategy training leaded to a statistically significant increase in the FLRA levels of the participants in the experiment group.To probe further into possible participant-related reasons underlying the statistically significant difference, further analyses were conducted between the participants' FLRA levels and personal information gathered at the beginning of the study.A Mann-Whitney U test was conducted on post-test results of the experiment group to see if gender was a factor affecting the change.The results revealed that the FLRAS scores of female (N=11) and male (N=9) participants did not show a statistically significant difference (See Table 2).Therefore, it was inferred that gender was not a variable to account for the increase in the post-test FLRAS scores of the experiment group.In addition, the relationship between the participants' age and post-test FLRAS scores was analyzed through Mann-Whitney U test.The analysis showed that the FLRA levels of the participants whose age ranged between 19-23 were not significantly different from the scores of those who were in the +23 years of age group (See Table 3).Thus, it was realized that the increase in the FLRAS scores of the participants in the experiment group emerged irrespective of the participants' age, since age did not have a statistically significantly influence on the results.A Kruskall-Wallis test was employed to project the relationship between the post-test FLRAS scores and the type of high school from which they graduated.The analysis demonstrated that there was a statistically significant correlation between the FLRAS scores and high school type (See Table 4).In other words, whether the participants graduated from an Anatolian Teacher Training (ATT)/ Anatolian (AH)/ Vocational (VOC)/ or other high schools was significantly related to their post-test FLRAS scores.The participants who graduated from AH were identified with the highest FLRA levels followed by those having graduated from ATT and Other high schools whereas the graduates of vocational high schools were found to be the least anxious in FL reading.Hence, it was hypothesized that the participants' perceptions about FL reading was affected by previous experiences they had in their particular high schools.Finally, the relationship between the participants' FLRA levels and their parents' educational levels was tested through Kruskall-Wallis test.The analysis indicated no statistically significant relationship between the participants' FLRAS scores and their parents' educational levels (See Table 5).Therefore, it was concluded that categorized into five levels as primary, elementary, high school, bachelor and MA/PhD, the parents' educational degree was not relevant to the post-test FLRA levels of the participants in the experiment group.

Discussion
Reading is a cognitively demanding process particularly in FL classrooms.Despite the simplistic conception that it does not necessarily require active participation and thus, is less anxiety-provoking, FL reading proves to be correlated with anxiety.Apparently, FL reading stimulates higher levels of anxiety compared to reading in L1 as several additional factors like language proficiency, cultural unfamiliarity and motivation interfere with the reading process (Kartal, 2011;Sellers, 2000).Based on the bulk of research favoring for teaching reading strategies in FL classrooms, one can readily assume that providing learners with training on reading strategies may facilitate comprehension of reading texts and thus, help learners cope with anxiety associated with FL reading.However, this study intriguingly indicated that reading strategy training resulted in an increase rather than decrease in learners' FLRA levels.Put simply, the FLRA levels of the participants in the experiment group were observed to ascend after they were exposed to training on various reading strategies involving skimming, scanning, keeping reading journals, and reflective thinking.This finding confirmed previous studies on anxiety (Campbell, 1999;Kitano, 2001;Wu, 2011).In an investigation of the relationship between language anxiety and the four major skills with a specific interest in gender, Campbell (1999) found out that there was a 9% increase in the number of males who reported to suffer from anxiety following the treatment they were offered in the study.Similarly, Kitano (2001) examined the impact of overseas experience on learners' anxiety levels, and noticed that learners of Japanese who had visited Japan were identified with higher levels of anxiety than those without overseas experiences.She (2001) surmised that the ascent in the learners' anxiety levels could be ascribed to the increased awareness imposed on learners by their visit to the target country.
The statistical analyses conducted on participant-related factors to investigate the underlying reasons for the perplexing increase in the FLRA levels of the participants in the experiment group yielded interesting results.
The study unveiled that the increase in the FLRA levels was not related to the participants' gender.This finding gave credence to Wu's (2011) study as she found no significant difference in reading anxiety levels of learners based on gender.Moreover, the results of this study demonstrated that the participants FLRA levels did not correlate with their age, which confirmed Jafarigohar & Behrooznia's (2012) hypothesis that age did not contribute to learners' FLRA levels.Also, this study unraveled that the participants' FLRA levels did not differ in line with their parents' educational degree.However, it was found that the type of high school from which the participants graduated significantly related to the increase in their FLRA levels following the reading strategy training they took.One possible explanation based on this finding could be that the type of reading experiences they had in their respective high schools might have affected their approach to the type of training they were exposed to in the reading course in the first semester of their bachelor degree.Furthermore, it was noteworthy that the participants having graduated from vocational schools reported to have the lowest levels of FLRA.This could be explained by the text types they studied in those schools because vocational high schools necessarily specialize on certain professions and thusly, the reading texts covered in those schools are more complicated in terms of content, vocabulary and language structures.
The above-mentioned explanations and assumptions are based solely on the participants' self-reports to a questionnaire and statistical analyses based on those reports, which may be limited in nature and thus, misleading.For this reason, the data elicited from the participants' in the interviews are used to shed further light on the findings.Though the analyses revealed an increase in the participants' FLRA levels after the training they took on reading strategies, the participants' responses in the interviews yielded contradictory statements, which gave invaluable insights into their perceptions about FLRA and reading strategies.In contrast to the statistical findings, the interviews were shaped by positive comments about the use of instructions on reading strategies.One of the interviewees, with the pseudonym Ayşe, remarked that "It was very useful to learn about different strategies.It increased my reading pace and helped me understand reading passages in English more easily.For example, I used to look up every word in the dictionary.But now, I prefer to make guesses about them or infer their meanings from the whole text".Another participant stated that "It was exciting for me to realize that there were some strategies to facilitate reading in English.I mean, we learned how to quickly go through a text to grasp the general meaning, how to use the title and graphics/ tables instead of reading a lengthy paragraph.It was all very informative" (Gül).However, one of the interviewees underlined the role of text types used in FL reading classrooms, which may also be connected to the statistical increase in their post-training FLRA levels.She stressed that "It was the first time I read such academic texts.Actually, they were difficult for me to understand.The paragraphs were very long and there were several words I did not know.There were some times I thought I would fail the class" (Seda).This criticism echoed Saito et. al.'s (1999) proposition that the type of texts and topics covered in reading classes might be a factor affecting FLRA levels.Another interviewee referred to the role of instructor in the development of learners' FLRA levels as he commented that "I was a bit worried about your (i.e.instructor/ interviewer) teaching in the classroom.I sometimes hesitated to ask you questions about a strategy.Indeed, I did not think that some skills like reflective thinking would be easy for us to learn, and I was afraid to tell this to you in the classroom" (Ali).This supported Wu's (2011) claim that the way instructors teach in the classroom may have considerable impact on their FLRA levels.Eventually, one of the interviewees' response reinforced the finding that the high school type from which they graduated might have affected their FLRA levels.She argued that "In high school, we were reading for the exam.You know, we had to pass the university entrance exam.Thus, our teachers would give us some paragraphs and ask us to read them and select the most appropriate choice" (Gül).Apparently, her remarks have some implications on the difference between reading activities done in high school and university level.It can be asserted that the divergence between high schools and higher education in terms of the routines in FL reading courses may trigger higher FLRA levels.Therefore, it becomes essential that FL reading courses in high schools should be improved to equip learners with skills required to cope with difficulties they may encounter in their university career.

Conclusion
This study investigated the relationship between FLRA and reading strategy instruction in relation to various learner-related variables.The results highlighted that FLRA was a construct related to reading strategy training as there was a change in learners' FLRA levels i.e. an increase in their FLRA level.Results of the statistical analyses demonstrated that the type of high school the learners graduated from could be a factor accounting for the increase in FLRA levels whereas other variables such as gender, age and parents' educational degree were not found to be correlated with learners' FLRA levels.Furthermore, the interviews with learners pinpointed the role of text types used in FL reading classes, instructor's teaching style and differences in reading activities between high school and higher education.Henceforth, it can be suggested that a warm, learner-friendly classroom environment in which the teacher adopts intimate manners may help relieve learners' concerns about reading in an FL.As Aydın & Kuru-Gönen (2012) emphasize, making some modifications based on feedback elicited from learners may also help learners feel valued and safer in the classroom.Likewise, reading programs used in high schools should be revised to better cater for learners' needs in subsequent years of their career.
Nevertheless, there are certain limitations on the findings of this study.The sample size was not big enough to yield generalizable results.Another major drawback was associated with the control group as they merely followed the traditional curriculum.Though the researchers were pre-determined to offer them extra courses on reading strategies, the finding that an increase in familiarity with those strategies was paralleled by an increase in FLRA levels discouraged them to do so.To sum up, a possible suggestion for further research may be replicating the study by including learners' views on the texts to be used.

Table 1 .
Paired samples t-test results of pre-and post-test FLRA scores in control and experimental groups

Table 2 .
Mann-whitney u test results of post-test FLRA scores of the experiment group based on gender

Table 3 .
Mann-whitney u test results of post-test FLRA scores of the experiment group (based on age)

Table 4 .
Kruskall-Wallis test results of the relationship between FLRA scores and high-school type in the experiment group

Table 5 .
Kruskall-wallis test results of the relationship between FLRA scores and parents' education level