Students Know Best: Modeling the Influence of Self-Reported Proficiency, TOEIC Scores, Gender, and Study Experience on Foreign Language Anxiety

Understanding how second language proficiency and other key learner characteristics relate to language learning anxiety subcomponents can help inform the type and quantity of content instructors approach foreign language (FL) students with when teaching English. This study seeks to understand how FL proficiency (i.e., TOEIC reading, TOEIC listening, and self-reported), FL study experience, and gender influence levels of foreign language classroom anxiety and second language writing anxiety. To accomplish the research goals, 245 students completed a questionnaire measuring foreign language classroom anxiety and second language writing anxiety. Three hierarchical linear regressions were used to test the study’s models, and structural equation modeling was used to confirm and illustrate the observed relationships. Overall, self-reported FL proficiency was a stronger predictor of anxiety than actual FL proficiency. Findings further indicate that FL listening and self-reported proficiency significantly negatively correlate with FL speaking anxiety but not writing anxiety. Only FL reading proficiency significantly influenced FL writing anxiety; however, this relationship disappeared when self-reported FL proficiency was added to the model. Further, differences in how FL proficiency influences FL anxiety when controlling for years of experience studying EFL and gender are discussed. Recommendations on how instructors can approach language instruction to assist highly anxious students in the EFL context are also discussed.


Introduction
Addressing anxiety is essential in second language acquisition as it is a central emotional construct that is directly connected to learning outcomes in second language classrooms (Teimouri et al., 2019).For a student to learn, they must receive, understand, and apply knowledge; however, anxiety limits one's ability to pay attention, and this lack of focus impairs cognitive processing, which results in poor organization and assimilation of new information.Foreign language (FL) anxiety and FL proficiency are posited here to have a recursive relationship in which one's current proficiency partly influences anxiety, and this anxiety consequently influences proficiency gains when learning.
There is a growing body of literature that acknowledges the role L2 proficiency has in explaining foreign language anxiety (Alamer & Lee, 2021;Thompson & Lee, 2014).For the proposed model in the current study, justification for modeling the effects of L2 proficiency on foreign language anxiety stems from Alamer and Lee's (2021) longitudinal study that evidenced L2 proficiency's significant influence on foreign language anxiety.Alamer and Lee (2021) drew attention to the lack of clarity on the relationship between L2 proficiency and FL anxiety.Their longitudinal study identified a causal relationship between L2 achievement to FL anxiety and not the other way around.Their conclusion was that language achievement precedes anxiety and their findings provide evidence for more research to explore the influence L2 proficiency subskills have on FL anxiety subcomponents.To this end, the proposed model in the current study places L2 proficiency as an explanatory variable on the levels of foreign language anxiety.
The influencing effect of living experience abroad has been found to also influence FL anxiety and therefore also worth investigating alongside L2 proficiency (Thompson & Lee, 2014).Thompson and Lee's (2014) investigation on the effect experience living abroad and L2 proficiency have on foreign language anxiety found that both experiences abroad and L2 proficiency were significant predictors of language learning anxiety.
Foreign language anxiety has severe negative effects on the attitude and consequential behavior students exhibit in the classroom.Foreign language anxiety in the class can arise when students look down upon those whose language they cannot clearly understand.This anxiety can make students underestimate their capabilities and hinder their social connectivity (Boudreau et al., 2018).Further, some students feel that their teachers favor high English language proficient learners who participate and contribute in the classroom.Consequently, low English language proficient students may feel anxiety when speaking in front of a group (Dewaele & Li, 2020) or sharing writing in an online class forum (Bailey et al., 2017).
When preparing lessons, FL instructors should consider factors that influence FL anxiety, like L2 proficiency, if they want to scaffold apprehensive students through learning goals (Pham, 2012).Classes containing students with mixed FL proficiency levels can be problematic because the instructor must prepare flexible lessons that correspond to individual learning needs (van Geel et al., 2019).Foreign language classroom instruction strategies that take into account different L2 proficiency levels are recommended for instructors in countries like South Korea where classes often contain students with mixed FL proficiency levels.Class activities can be connected to the content, procedure, performance, and learning context.In other words, instruction in conversational English classes attempt to provide multiple ways to ''receive the facts, concepts, generalizations or principles, attitudes, and skills related to the subject matterforeign language system in an FL classroom -as well as the materials that represent those elements'' (Tomlinson & Strickland, 2005, p. 7).
One such factor that is consistently influenced by L2 proficiency perceptions, self-efficacy, and performance is foreign language anxiety (V. Park & Datnow, 2017).Learners with high levels of second language apprehension will avoid engaging in the target language (e.g., English), and this is especially true if the language exercise is beyond their proficiency level, or in other words, outside their learner's zone of proximal development (ZPD; Roy et al., 2013).Gender and experience studying EFL are two further variables of concern here.Regarding gender, some studies (Namaziandost et al., 2019) show that females are generally more capable of learning a second language and consequently experience less language apprehension.However, these findings are not ubiquitous throughout the literature (Rianto, 2021), therefore, need further exploration.Next, years of experience is another variable of interest.In South Korea, where the current study occurred, students typically begin studying English in kindergarten and this continues throughout their primary and secondary (i.e., elementary, middle, and high school) education.However, variation in when students begin FL study does exist so experience studying English is considered, here, to be a more accurate measure than age.
The Test of English for International Communication (TOEIC) Reading and Listening scores were included alongside the other explanatory variables (i.e., selfreported proficiency, gender, and FL study experience).The TOEIC Reading and Listening test includes multiple choice questions which ask students to respond correctly and complete sentences using correct grammar or vocabulary.Since test takers are assessed on their ability to choose which word or phrase correctly completes a sentence, the TOEIC reading is considered a reliable and valid measure of writing skills (Imsa-ard, 2019).In a similar vein, TOEIC listening partly assesses one's speaking skills since accurately comprehending speech is necessary for successful discourse.Listening is crucial for FL classroom communication and speaking since one must comprehend speech to respond appropriately.
Overall, this study aims to investigate the magnitude of and causal relationships among self-reported proficiency, TOEIC Reading and Listening test scores, gender, FL study experience, foreign language classroom anxiety (i.e., evaluation and communication anxiety), and second language writing anxiety.To this end, the following questions were asked: RQ1: What are the relationships among and magnitude of self-reported proficiency, TOEIC Reading and Listening test scores, study experience, gender, and foreign language anxiety?RQ2: How does self-reported proficiency, TOEIC Reading and Listening test scores, study experience, and gender influence second language writing anxiety and foreign language classroom anxiety?

Literature Review
The theoretical framework of this study pivots on the foreign language learning anxiety theory (Horwitz et al., 1986), which recognizes that the more students manage their levels of learning anxiety, the more likely they will show improvement (Canaran et al., 2020;Liu & Ni, 2015;Otair & Aziz, 2017).One topic of interest that has seized researchers' attention is the relationship between foreign language anxiety (FLA) and FL proficiency, especially in EFL classrooms (Brunfaut & Revesz, 2015).The effects of learner characteristics on FL anxiety have been discussed at length.Foreign language anxiety, here, refers to ''the subjective feeling of tension, apprehension, nervousness, and worry associated with an arousal of the autonomic nervous system'' (Horwitz et al., 1986, p, 125).Moreover, foreign language anxiety is regarded as situation-specific anxiety (Horwitz, 2010), and domainspecific according to the English subskill being practiced, either speaking, listening, reading, or writing.In a similar vein to FL anxiety theory, language learning strategy theory (Oxford, 2011) posits that language acquisition occurs when students practice a wider array of learning strategies and practice them more accurately.Students who are apprehensive to learning a FL are less likely to engage in learning strategies (e.g., cognitive, metacognitive, and social strategies).Contrarily, students with less FL anxiety are more self-efficacious and therefore more likely to reap rewards from successfully engaging in language learning strategies.

Proficiency and Foreign Language Classroom Anxiety
In Horwitz et al.'s (1986) theoretical framework, the foreign language classroom anxiety scale (FLCAS) are used as ''an instrument to measure anxiety levels as evidenced by negative performance expectancies and social comparisons, psychophysiological symptoms, and avoidance behaviours'' (Tran, 2012, p. 71).According to Horwitz et al. (1986), FL anxiety refers to ''a distinct complex of self-perceptions, beliefs, feelings, and behaviors related to classroom language learning arising from the uniqueness of the language learning process'' (p.128).It is connected principally with speaking and listening since these skills are most influenced by anxiety (Kra´lova´, 2016).Tran (2012) argues that despite critiques of the FLCAS, it still proves to be a reliable instrument to solve conundrums in FL anxiety research.Foreign language classroom anxiety, including fear of missing important details and fear of not understanding the FL content, were found to be negatively affected by the students' self-efficacy beliefs (Canaran et al., 2020), one's inclination to communicate (Kalsoom et al., 2020), and the learners' FL proficiency (Lee, 2018).
An abundance of research has reached a consensus indicating a negative correlation between FL learning and FL anxiety (i.e., Hewitt & Stephenson, 2012;Kalsoom et al., 2020;Soleimani et al., 2020).Serraj and Noordin (2013) carried out a bivariate correlation analysis to investigate the relationships among listening comprehension, foreign language speaking anxiety, and foreign language listening anxiety among a group of 210 Iranian EFL students.Results evidenced a statistically significant relationship between FL listening anxiety and listening comprehension skills and recommended EFL instructors be aware of the impending effect of low FL proficiency, be it actual or self-reported, on apprehension to use a second language.Assuming anxiety predicted L2 proficiency scores, Zhang (2013) modeled the relationship between FL listening anxiety and listening performance with 300 first-year majors in China and results indicated that FL anxiety had a significant negative effect on FL listening performance.In contrast to Zhang's (2013) tested hypotheses, the current study suggests proficiency level influences anxiety which consequently influences performance (e.g., proficiency gains).
In short, despite some nuances among most studies that evaluate FL anxiety and improved learning outcomes, foreign language anxiety plays a pivotal role, and it is usually not a positive one.Tran (2012) argues that foreign language anxiety can be both a cause for negative consequences on FL learning and an effect of a failure in learning the foreign language (Ganschow & Sparks, 1996;Horwitz et al., 1986;Lee, 2018).It should be taken into consideration that, nevertheless, the specifics of their studies differ in which Soleimani et al. (2020) focus more on the relationship between writing motivation and FL writing anxiety.Alemi et al. (2011) and Kalsoom et al. (2020) assessed the relationship between FL communication anxiety and the willingness of the student to communicate in a foreign language.Few studies have looked at anxiety and subskills of a second language like reading or listening (Graham, 2011).Previously reviewed studies have shared one consensus, which is the idea that FL anxiety proves to be a primary setback to improving FL skills and learning success throughout the foreign language learning journey.

Proficiency and Writing Anxiety
Writing anxiety also shows a negative relationship with FL proficiency (Abbas, 2016;Al Asmari, 2013;Ismail et al., 2010;Rezaei & Jafari, 2014).Second language writing has temporal and spatial differences from FL speaking since second language writing often occurs in asynchronous conditions.The instructor can assign a writing task for the student to complete at a time and place of their choosing.The presentation of writing composition is also starkly different than speaking which may influence levels and sources of FL anxiety.Cheng (2004) identified three subcomponents of writing anxiety including cognitive, somatic, and behavioral anxiety.The cognitive component entails fear of failure or humiliation and is most similar to some of Horwitz et al.'s (1986) FLCA items pertaining to evaluation and communication.Cheng, however, includes a somatic component to measure physiological responses the student experiences when writing in a second language (e.g., I feel my heart pounding.).The behavioral component refers to the student taking action to avoid writing in English (e.g., I usually do my best to avoid writing English composition).Cheng's second language writing anxiety instrument (SLWAI) has been used extensively in recent decades (Woodrow, 2011;Xiao & Wong, 2014).The three FL writing anxiety sub-components have shown distinct relationship patterns with learner characteristics (Bailey et al., 2017).When comparing a blended and fully offline writing course, Bailey et al. (2017) asked students to complete a series of process writing tasks over the course of an academic semester.Through pre-post survey analysis, it was found that blended learning students reported higher increases in behavioral anxiety compared to students in offline conditions.Bailey and Cassidy's (2017) students reported higher levels of cognitive anxiety and low levels of somatic and behavioral anxiety, indicating students have distinct makeup when it comes to their FL anxiety.Studies like Bailey and Cassidy (2017) provide evidence that explanatory variables like one's current FL proficiency level, gender, and study experience uniquely influence different types of FL anxiety subcomponents.

Gender, FL Study Experience, and Anxiety
A relatively limited number of studies on the relationship between gender and anxiety exist (Aydin et al., 2006;Namaziandost et al., 2019), with mixed results showing that males have higher anxiety, females have higher anxiety or no difference at all.Several studies reveal that female students exhibit lower levels of foreign language apprehension (Lian & Budin, 2014;McAllister, 2014;Shang, 2013;Zorbaz, 2011).The reasoning for the difference in anxiety levels between genders has been attributed to more positive feedback from their teachers (Daly, 1985) or that females are more emotionally stable than males in stressful circumstances (Spielberger, 1983).In other studies, the opposite was observed with females have reported higher levels of FL writing anxiety than males (Abdul-Fattah, 1995;Cheng, 2002;Cocuk et al., 2016;Rodriguez et al., 2009) due to reasons including fear of making mistakes, being criticized by others, and failure to meet their instructor's expectations.Regarding foreign language anxiety, Mohtasham and Farnia (2017) found that female students (n = 125) reported higher levels of FL classroom anxiety than males (n = 50) in impromptu speaking activities.In the same line of research, previous studies showed that females are more anxious than males (Naghadeh et al., 2014;G. P. Park & French, 2013).With Korean students particularly, K. J. Kim (2006) found that female students were more apprehensive writers than male students, including overevaluating themselves and overcorrection.In other studies, no statistically significant differences in second language writing anxiety (Abu Shawish & Abdelraheem, 2010;Al Asmari, 2013;Alsowat, 2016;Bell & McCallum, 2012;Kostic´-Bobanovic´, 2016;Marshall & Varnon, 2009).Kabigting et al. (2020) described how FL writing anxiety, using the SLWAI, affected ESL learners' ability to write.Among 33 tenth-grade ESL students, they found that students perceived themselves as having high levels of writing apprehension overall, while no differences were noticed between males and females.The greater anxiety students reported, the worse writing performance they exhibited (Kabigting et al., 2020).
Regarding FL learning experience, some studies find that older students with more FL study experience are more apprehensive (Al Asmari, 2013), while other studies show no difference at all (Abu Shawish & Abdelraheem, 2010;Cheng, 2002;Shang, 2013).Similarly, Bae et al. (2020) found that young females, in fourth grade, revealed higher FL proficiency than their male counterparts but these differences were not apparent with more experience fifth and sixth graders.Moreover, experience learning EFL/ESL, on occasion, proves to be an indicator of more advanced writing strategy use, with more advanced writing strategies used by older students with more experience with English study.Other studies (Qadir et al., 2021) found that gender, academic level, and age showed no difference in FLA among a group of 33 postgraduate students learning EFL.With the contrasting nature of gender and anxiety, and experience studying and anxiety, a continuous finger on the pulse of these relationships is warranted.Drawing on the studies that have illustrated the causal relationship between anxiety and proficiency, this study investigates the level of influence of actual and self-reported FL proficiency on FL classroom anxiety and FL writing anxiety.

TOEIC Reading and Listening Test
The multiple-choice format with the TOEIC Reading and Listening test make it a more affordable and accessible choice for FL assessment than the TOEIC Speaking and Writing test.The four language skills (i.e., listening, speaking, writing, and reading) share fundamental processes including the conceptualization of words, knowledge of sounds, and grammar.The TOEIC Reading and Listening directly measure how language correctly used and this assessment significantly predicts how students correctly speaks or writes (D.E.Powers & A. Powers, 2015).Likewise, J. Kim and Chon (2013) contend that exposure via reading leads to productive use of new and expanded lexical knowledge which can be applied to L2 output (i.e., speaking and writing).This was also previously observed by Ponniah (2011), who found that students could acquire and use new words, but also were able to use correct grammar after being exposed to it via reading materials.
Drawing on the studies that have illustrated the causal relationship between FL proficiency and anxiety, this study investigates the level of influence of self-reported FL proficiency, TOEIC scores, gender, and study experience on FL communication, negative evaluation, and writing anxiety.

Methods
This study followed a hierarchical regression analysis research design to learn the extent self-reported proficiency, TOEIC scores, gender, and study experience influence FL anxiety subcomponents.In all, 245 freshman students (Male = 121, Female = 124, gender-nonbinary = 0) were recruited from the Colleges of Humanities (n = 142) and Science and Engineering (n = 103) from a university in the central province of South Korea.Data were collected during the spring semester of 2017 and students were attending test preparation classes for the TOEIC.Table 1 displays demographic data for the participants.Students were required to complete the TOEIC test at the end of their first year.Pertaining to actual FL proficiency as measured by the TOEIC, students had slightly lower levels of reading proficiency than listening proficiency.Regarding selfreported FL proficiency, students reported having a moderate level with normal distribution around the median value of 5. Self-reported proficiency is a reliable assessment for FL proficiency (Lee, 2018).

Materials
Data Collection Instruments.In addition to demographic data, the questionnaire included two scales (i.e., negative evaluation and communication anxiety) from the foreign language classroom anxiety survey (FLCAS; Horwitz et al. 1986) and second language writing anxiety instrument (SLWAI; Cheng, 2004).Specifically, the cognitive and somatic anxiety scales from Pae's (2007) modified version of Cheng's (2004) SLWAI was employed and included 16 items (see Table 3).For the FLCAS (Horwitz et al., 1986), communication and negative evaluation anxiety subcomponents were employed.All survey items are displayed in Tables 3 and 5. Items were scored on a seven-point Likert scale, from 1 (never true of me) to 7 (always true of me).For FL writing anxiety, communication anxiety, and evaluation anxiety, Cronbach's alpha scores were .951,.891,and .888,respectively, indicating good internal reliability.
Data were collected through an online survey created using Google Forms.Participants were informed about the purpose of the study through an informative paragraph and instructor explanation.Students were also informed that participation was fully voluntary.Then, consent was obtained.The survey was completed during class time under instructor supervision.Data Analysis.Descriptive analyses were carried out with SPSS (version 25.0).Specifically, mean score and  (George & Mallery, 2010).Multicollinearity refers to a situation in which more than two independent variables are highly related.No correlation coefficients among the explanatory variables (i.e., demographic variables and L2 proficiency scores) were above .70,providing evidence of no multicollinearity issues in the data (Tabachnick & Fidell, 2001).Next, the study measured variable inflation factors (VIF) and found no values greater than two which is below the recommended threshold of 10.Lastly, the Kaiser-Meyer-Olkin (KMO) was 0.960 and Bartlett's Test of Sphericity was 6,369.40(df = 496, p \ .001),both with acceptable ranges (George & Mallery, 2010).Given the fulfilled statistical prerequisites, the data set was ready for analysis.

Results
Overall, students reported high levels of second language writing anxiety (M = 4.33, SD = 1.15) and moderate levels of both communication (M = 3.87, SD = 1.33) and negative evaluation (M = 3.73, SD = 1.20) anxiety.For the mean score values of the study variables, students reported studying English on average for 7.27 years (SD = 2.86).Table 2 shows mean scores and Pearson correlation analysis for the variables of interest.Several points of statistically significant correlations were found.The measured anxiety types show strong positive correlations with one another.Likewise, the three FL proficiency types also showed a strong positive correlation with one another.Further, the three anxiety types and three FL proficiency measures shared strong negative correlations.Regarding demographic variables, females reported being more apprehensive about each of the three observed anxiety subtypes than males.Further, years of studying English showed a moderate negative correlation with each of the anxiety types and a medium to high correlation with each of the three types of FL proficiency.
Findings from answering research question one are in line with previous studies, which show anxiety and FL proficiency are negatively correlated.Table 3 provides descriptive statistics at the item-by-item level for writing and foreign language classroom anxiety scales.Items were ranked in order from highest to lowest.Overall, students reported higher levels of writing apprehension than classroom anxiety (i.e., communication or negative evaluation).Seven of the 16 SLWAI items were in the 4.50 to 7.0 range, indicating generally high levels of writing anxiety.The two highest mean scores pertained to fear of making grammar mistakes (rank 1) and lack of confidence in fluency (rank 2).Contrary to writing anxiety, students reported more moderate levels of FLCA.Students were less anxious about communicating in class with one another and receiving negative evaluations from others.Regarding communication apprehension, students reported feeling anxious about ''speaking English in front of others'' (Communication Apprehension, ranked 1; M = 4.08, SD = 1.70), ''panic if not prepared'' (Communication Apprehension, ranked 2; M = 4.03, SD = 1.61), and ''think other students are better at English than I am'' (Negative Evaluation, rank 1; M = 4.23, SD = 1.63).
According to the Pearson correlation analysis in Table 2, anxiety varied in relation to proficiency type (i.e., TOEIC reading, listening, and self-reported), gender, and experience studying English, and these variations ranged from medium to large effect size.Such high variation suggests these factors help explain the types of student characteristics that contribute to different types of language learning anxiety.
For research question two, the next step in the study entailed carrying out regression analysis to understand how learner characteristics influence distinct FL anxiety subtypes.There were three sets of hierarchical regression for the observed anxiety types.Each model revealed statistically significant F values, R 2 values, and changes in R 2 values.
Model one contained demographic data, including gender and the years of English study.In each analysis, the model was significant.In each model, females showed a positive path coefficient with anxiety, while study experience showed a negative correlation, supporting initial findings from the correlation analysis in research question one.For model two, actual FL proficiency scores were added.Once added, gender continued to have a statistically significant influence on each of the three anxiety types; however, years studying English no longer accounted for any effect, regardless of anxiety type, indicating study experience (in years) was not an influential factor when considering standardized test scores.
In model 2, the reading and listening scores showed distinct levels of influence on each of the anxiety types.Reading proficiency had a statistically significant influence on FL writing anxiety, while listening proficiency had no influence.Contrarily, listening proficiency had a statistically positive influence on both of the FLCA types.While showing a positive influence, reading proficiency approached significant levels of predicting fear of negative evaluation.
In model three, self-reported FL proficiency was added to the hierarchical regression.In this instance, gender continues to reveal an influence on each of the anxiety types, indicating the females were consistently more anxious about writing in a second language, communicating, and receiving negative feedback.Self-reported proficiency accounted for the largest influence on each of the anxiety types.In fact, once added to model 3, reading proficiency along with listening proficiency showed no influence on FL writing anxiety.However, listening proficiency, along with self-reported proficiency influenced levels of communication and evaluation anxiety (Table 4).Hierarchical regression evidenced that study experience (in years) was a poor predictor of FL anxiety, followed by TOEIC Reading scores.Contrarily, selfreported FL proficiency was the strongest predictor for each anxiety subcomponent, followed by TOEIC Listening scores for both evaluation and negative evaluation anxiety.

SEM Analysis
Structural equation modeling (SEM) was carried out to provide a more powerful analysis than linear regression of the final model (i.e., model 3).A three-factor correlated model was used to validate the three anxiety subcomponents.Principal component analysis with Varimax rotation was used to extract factors for the three anxiety subcomponents.Unidimensionality was confirmed as evidenced by acceptable average variance extracted (AVE) values above .50and composite reliability (CR) scores above .60(Fornell & Larcker, 1981).Initially, a poor model fit was observed when running the SEM, x 2 (df = 402) = 1,057.75,NFI = 0.834, IFI = 0.892, CFI = 0.891, RMSEA = 0.082, PCLOSE = 0.000.Upon following modifications according to the modification indices (MI), moderate model fit was achieved.x 2 (df = 393) = 868.669,IFI = 0.922, TLI = 0.913, CFI = 0.921, RMSEA = 0.070, PCLOSE \ 0.001.Regarding modifications, five instances of covariance were recommended for the SLWA scale (i.e., items 1 and 2,4 and 11,5 and 6,7 and 11,and 8 and 11;MI change = 156.42),three instances of recommended covariance within the negative evaluation scale (i.e., items 2 and 9, 1 and 9, and 4 and 7, MI change = 40.52),and two within the FL communication scale (i.e., items 3 and 5, and 4 and 5, MI change = 29.58).All instances of covariance were within the items' respective scales.The CFI, NFI, and TLI values are recommended to be above .95(Hair et al., 2006;Steiger, 2007) and RMSEA values between .50 and .70 are considered moderate fit (Steiger, 2007).The mediocre fit observed in the study's model was partly due to high levels of collinearity among the anxiety subcomponents and considered acceptable here due to the established construct validity evidenced in past research (Cheng, 2004;Horwitz et al., 1986).Next, self-reported proficiency, TOEIC Reading and Listening scores, gender, and EFL study experience were added, resulting in sufficient model fit, x 2 (df = 528) = 1,010.97,IFI = 0.928, TLI = 0.918, CFI = 0.927, RMSEA = 0.061, PCLOSE = 0.001.Path coefficients illustrated in Figure 1 produced through SEM coincided with the stepwise regression analysis presented in model 3 (see Table 5).Only significant path coefficients are illustrated in Figure 1.
Answering research question two identified distinct paths among the explanatory variables.SEM confirmed significance levels identified through hierarchical linear regression.Overall, the model predicted FL communication anxiety the most (R 2 = 48 ** ), followed by FL negative evaluation (R 2 = 39 ** ) and FL writing anxiety (R 2 = 32 ** ).In the next chapter, explanation for why these emerging models appeared among the current population of students is provided, along with implications for how these findings can apply to instruction within EFL classrooms that have students with varying levels of L2 proficiency.

Discussion
The current section summarizes the findings and contributions made related to the influences FL proficiency (i.e., TOEIC reading, TOEIC listening, and selfreported), gender, and FL study experience have on FL negative evaluation, communication, and writing anxiety.Findings from research question one revealed that actual and self-reported FL proficiency negatively correlates with anxiety types, which supports results from extant literature (Lee, 2018).
Findings from answering research question 1 suggest that students with higher levels of actual and selfreported FL proficiency tend to have lower FL anxiety.This finding presents evidence indicating that educators should recognize the significance of FL anxiety as a factor contributing to learners' reluctance to engage in FL communication (Alemi et al., 2011).Another point worth noting is that levels of anxiety decrease while levels of FL proficiency, actual or self-reported, increase with the number of years students have studied English, which is in line with past literature (Lee, 2018).
Mean score comparison shows that students reported higher levels of FL writing anxiety compared to FL classroom anxiety, be it fear of communicating in English or fear of receiving a negative evaluation.Thus, providing new evidence that supports the claim that students experience higher levels of second language anxiety when they are asked to write than when asked to speak.Gender has shown mixed results in past foreign language anxiety studies.In some cases, females reported more apprehension when using a second language (i.e., Abdul-Fattah, 1995;Bae et al., 2020;Cheng, 2002;Cocuk et al., 2016;Rodriguez et al., 2009), and this was the case here.In the current study, correlation analysis found that females reported higher levels of writing and classroom anxiety, and this path stayed consistent in each of three regression models tested in research question two.
Research question 2 used hierarchical linear regression to identify the influence key demographic variables, actual FL proficiency (i.e., reading and listening) and self-reported FL proficiency have on three types of FL anxiety.Initially, only gender and years of English study were modeled and both student characteristics resulted in significant path coefficients for all three types of anxiety.This implies that when considering the factors influencing anxiety, independent of FL proficiency, gender and years of experience studying English have an impact on FL writing and FL speaking anxiety, specifically in relation to classroom communication and negative evaluation.Model 2 added actual FL proficiency to the regression and, once added, the effect of years studying English on anxiety disappeared, indicating mediation.To this end, the relationship between time spent studying English and anxiety is explained through the students' competency with English.Therefore, regardless of the time spent studying a second language, anxiety is most likely to decrease with a corresponding increase in second language acquisition.These basic findings are consistent with research showing that external factors causing FL listening anxiety were found to have a strong negative effect on self-efficacy (Canaran et al., 2020) and a statistically significant negative correlation between foreign language anxiety and speaking (Canaran et al., 2020;Hewitt & Stephenson, 2012;Otair & Aziz, 2017).
Another point worth discussing from model 2 pertains to the type of FL proficiency and its influence on foreign language anxiety.Reading proficiency predicted FL writing anxiety yet listening proficiency did not.Reading, as a language skill, is more closely related to writing since both involve internal processing of language without a partner (Ghaith, 2020;Imsa-ard, 2019).Contrarily, listening proficiency was a significant predictor of both classroom anxieties measured while reading proficiency was not.Again, the explanation provided here posits that listening and speaking entail a communication partner, be it a classmate or teacher, and therefore more closely related to one another.Likewise, FL proficiency in reading was more closely related to writing anxiety.
The third and final model added self-reported FL proficiency to the regression.Once added, the effect actual reading proficiency had on FL writing anxiety disappeared, indicating that students' perceptions of their capabilities in a second language have a more powerful effect on their writing anxiety than their actual FL proficiency.Contrarily, no mediating effect was noticed by self-reported proficiency on the relationship between listening proficiency (actual) and classroom anxiety, be it fear of communication or fear of negative evaluation.Lastly, with the inclusion of self-reported proficiency to model 3, gender continued to influence each of the three measured anxiety types.Females, from our findings from this questionnaire study, clearly report higher levels of anxiety than males, and this is in lieu of years studying, actual FL proficiency, and self-reported proficiency, echoing past studies (Abdul-Fattah, 1995;Cheng, 2002;Cocuk et al., 2016;Naghadeh et al., 2014;G. P. Park & French, 2013;Rodriguez et al., 2009), and this may be attributed to reasons including fear of making mistakes, being criticized by others, and failure to meet their instructor's expectations.
Through structural equation modeling, a comprehensive analysis of model 3 from regression analysis was provided.In the SEM analysis, self-rated proficiency emerged as the strongest predictor of the outcome variable when compared to TOEIC reading scores, TOEIC writing scores, EFL study experience, and gender.The path coefficients clearly indicated that self-rated proficiency had the most substantial impact on the anxiety scales measured here.Additionally, gender demonstrated significant associations with the anxiety subcomponents.These SEM findings align with the results obtained from the regression analysis, indicating that TOEIC reading scores and EFL study experience do not significantly influence the anxiety subcomponents.Recognizing the importance of students' self-evaluation of their L2 proficiency level is crucial in managing anxiety, as it directly impacts their self-confidence and ability to effectively cope with communication anxiety.The inclusion of items in the communication anxiety scale that address confidence-related fears further strengthens the relationship between self-reported proficiency and EFL communication anxiety levels.

Pedagogical Implications
Explicitly informing students on their FL anxiety levels may not be as beneficial as ensuring students acknowledge and accept their L2 proficiency level.Therefore, instructors should encourage students to be aware of their L2 proficiency level, even more so than their FL anxiety level.Students may have anxiety when realizing that their low proficiency levels are not sufficient for communicating.Consequently, instructors should help students accept their current L2 proficiency, and prepare learning programs that match the student's ability.Students can further manage their anxiety by engaging in activities within their L2 proficiency level.Gains in their L2 will contribute to better managing FL anxiety, and possibly help anxiety decrease over time.To help desensitize anxiety among EFL students, instructors with anxious students are recommended to use a wide range of teaching and learning platforms and channels (e.g., social media, blog, and smartphone apps).Exercises that instructors use should be calibrated at or below the student's proficiency level at some times and at or above the students' proficiency level at other times.Lesson planning entails balancing between easy learning activities that evoke feelings of boredom and difficult activities that evoke feelings of anxiety.
With the quick onset of educational technology in the foreign language classroom, classes are flipped, blended, or fully online with video conference tools.Writing and FL classroom anxiety should be considered as impeding forces on the students learning potential.While, often, students will generally be proficient at writing, speaking, listening and reading, this is not always true.In certain cases, students will be distinctly weak writers or weak speakers, and this reflects in their levels of FL writing and classroom anxiety.When considering EFL classrooms, the instructor must consider FL proficiency as a landscape with peaks and valleys across the different linguistic skills one must master during their SLA journey.

Conclusions
The paper concludes by arguing that self-reported FL proficiency was a stronger predictor of anxiety than actual FL proficiency as measured by the TOEIC Reading and Listening test.The analysis leads to the following conclusions FL listening and self-reported proficiency have a significant negative correlation with FL classroom anxiety, especially pertaining to communication with peers.Only FL reading proficiency significantly influenced FL writing anxiety; however, this relationship disappeared when self-reported FL proficiency was added to the model.Further differences in how FL proficiency influences FL anxiety and gender showed that female students reported higher levels of apprehension about acquiring a second language than males.
Future research investigations are necessary to validate the kinds of conclusions that can be drawn from this study.A mixed-method design that includes both individual interviews and focused group ones would shed light on the validity of the current study.Furthermore, the study was undertaken to investigate specific variables (i.e., actual, and self-reported FL proficiency and FL classroom anxiety and writing anxiety) within a limited population (i.e., South Korean university students), adding further variables (e.g., studying abroad) and increasing the diversity of participants could prove quite beneficial to the literature.The FL study experience factor entailed only a single item (i.e., How long have you studied English) which did not account for frequency (e.g., hours per day and days per week) or biological age.Future study should consider these metrics and more to increase the validity and reliability of study experience as an explanatory variable.Future research should also consider the potential effects of developing lesson plans that includes content, development, the performance of ongoing diverse assessment, and dynamic instruction (Khan, & Asif, 2018).Within such an environment, all students are granted the opportunity to learn the equal curriculum; nevertheless, following verified teaching methods to encounter the students' individual learning requirements, including learners with anxiety issues.EFL instruction, at its finest, should reflect a new pedagogy that encourages practical integration and knowledge transformation (Pham, 2012).This is due to the fact that EFL education, if properly implemented, can demonstrate institutional performance and provide students with a variety of learning experiences that will enable them to effectively adapt to rising obstacles in the global community.

Authors Contributions
Dr. Daniel Bailey (theoretical framework, data collection and analysis, writing original draft, review and editing).Dr. Norah Almusharraf (theoretical framework, writing original draft, review and editing).

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The researchers thank Prince Sultan University for funding this research project under grant (Education Research Lab-ERL-CH-2023/2).In the context of promoting an academic-rich working environment, this paper also acknowledges Konkuk University in 2023 for academic support.

Table 1 .
Demographic Information of Study Participants.Data Cleaning.Data cleaning entailed several steps.First, Mahalanobis and Cook's distance values identified outliers of which eleven were identified and consequently removed, leaving 245 completed surveys.Kurtosis values for the study's anxiety scales were within the 22 and + 2 range and skewness was within the 2 1.96 and + 1.96 range, exhibiting appropriate levels of normal univariate distribution

Table 2 .
Correlations and Mean Score Analysis.

Table 4 .
Results of Hierarchical Regression Analysis for Study Variables (n = 245).

Table 5 .
Foreign Language Classroom Anxiety Scale Items.Note.Items adopted fromHorwitz et al.'s (1986)FL classroom anxiety scale.