Hostname: page-component-74656f9884-kv2lp Total loading time: 0 Render date: 2026-11-05T09:34:35.794Z Has data issue: false hasContentIssue false
  • English
  • Français

A network analysis of the relationship between foreign language anxiety and listening fatigue in Arabic and Chinese learners of English

Published online by Cambridge University Press:  28 July 2026

Faisal Aljasser ,
Yusen Wang  and
Michael Vitevitch Open the ORCID record for Michael Vitevitch [Opens in a new window]
Faisal Aljasser
Affiliation:
Qassim University, Saudi Arabia
Yusen Wang
Affiliation:
Kunming University of Science and Technology, China
Michael Vitevitch*
Affiliation:
Speech-Language-Hearing: Sciences & Disorders, University of Kansas, USA
*
Corresponding author: Michael Vitevitch; Email: mvitevit@ku.edu
Rights & Permissions [Opens in a new window]

Abstract

Network analysis assessed the relationship between listening fatigue (Vanderbilt Listening Fatigue Scale; VFS) and foreign language learning anxiety (Foreign Language Classroom Anxiety Scale; FLCAS). Nodes represented the questions on each scale, and edges between nodes represented the strength and direction of the correlation between those questions/nodes. A simple correlation of the sum scores revealed a negative correlation: students reporting high fatigue reported low anxiety. This result may be due to the way in which cognitive resources are allocated (i.e., “ego depletion”). Analysis of the micro- and meso-level structure of the network revealed two distinct subsets of nodes (i.e., communities), with one community containing many of the FLCAS items (and the 5 nodes highest in strength) and the other community containing the VFS items and several items from the FLCAS that assessed confidence in using L2 rather than anxiety in learning L2. Additional ways that network science can be used to increase our understanding of language learning are discussed.

Keywords

Foreign language anxietylanguage learninglistening fatiguenetwork analysisnetwork science

Information

Type
Original Article
Information
Applied Psycholinguistics , Volume 47 , 2026 , e43
Check for updates
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2026. Published by Cambridge University Press

Introduction

Acquisition of one’s first language appears to occur effortlessly and without explicit instruction (Kuhl, Reference Kuhl2000). However, the acquisition of a second language—often occurring in the context of a classroom—typically requires significant cognitive effort, which could lead to feelings of mental fatigue from use of the second language for an extended period of time (Brännström et al., Reference Brännström, Rudner, Carlie, Sahlén, Gulz, Andersson and Johansson2021). Learners of a second language (L2) may also experience anxiety related to learning and speaking in the classroom (Horwitz et al., Reference Horwitz, Horwitz and Cope1986). In the present study, we used computational tools from network science to examine how listening-related fatigue and foreign language anxiety might be related to each other in Arabic and Chinese learners of English.

Studies of listening-related fatigue among L2 learners indicate that L2 learners may exert high levels of cognitive effort, leading to fatigue and reduced performance (Borghini & Hazan, Reference Borghini and Hazan2018; Sarem et al., Reference Sarem, Marashi and Siyyari2018). One way to measure cognitive effort is with the physiological measure of pupillometry, which measures the extent to which the pupils of the eye dilate in response to task-related demands, such as the presence of background noise when native English speakers listen to English speech (Koelewijn et al., Reference Koelewijn, Zekveld, Festen and Kramer2012). Borghini and Hazan (Reference Borghini and Hazan2018) used pupillometry to demonstrate in native English speakers and in Italian speakers of English that non-native listeners exerted greater listening effort (as measured by the pupil response) when listening to English (i.e., the second language) than the native speakers of English, providing physiological evidence of the cognitive effort and fatigue associated with processing L2.

Further demonstrating that listening fatigue influences performance in L2, Sarem et al. (Reference Sarem, Marashi and Siyyari2018) used a semi-structured interview, several questions pulled from various questionnaires about mental and physical fatigue and scores on the test of listening from the TOEFL. They observed a significant negative correlation between scores on the fatigue questions and the L2 listening comprehension score from the TOEFL. That is, English as a foreign language learners who expressed low levels of fatigue performed better on the test of listening from the TOEFL, whereas language learners who expressed high levels of fatigue performed more poorly on the test of listening from the TOEFL. Responses in the semi-structured interview further corroborated the relationship between listening fatigue and performance in L2.

Listening to an L2 may not only be effortful and induce fatigue, but may also induce anxiety in the learner, which may also contribute to poor learning outcomes. Numerous studies have demonstrated that foreign language classroom anxiety by itself or in conjunction with other factors adversely influences language learning (Chen & Guo, Reference Chen and Guo2026; Saito et al., Reference Saito, Garza and Horwitz1999; Shen, Reference Shen2021). The adverse influence of foreign language classroom anxiety has been observed whether the learning environment is in-person or via distance learning (Bosmans & Hurd, Reference Bosmans and Hurd2016). In the present study, we explored the potential relationship between listening fatigue and foreign language classroom anxiety. Although we are only assessing the relationship between listening fatigue and foreign language classroom anxiety—and not measuring classroom performance—observing a relationship between listening fatigue and foreign language classroom anxiety would increase our understanding of the (inter-related) factors that may influence successful language learning, create opportunities for future researchers to further explore how listening fatigue and foreign language classroom anxiety might affect language learning performance, and could provide foreign language instructors with guidance on how to reduce anxiety and fatigue to facilitate language learning performance.

The current study

In the present study, we used the tools of network science to explore the possible relationship between listening fatigue and foreign language classroom anxiety. First, we describe the scales we used to assess listening fatigue and foreign language classroom anxiety. Second, we describe the network science approach (i.e., the mathematical tools from network science that we used), and the potential insight that this approach can provide compared to contemporary approaches. Finally, we discuss three possible outcomes (i.e., tentative or exploratory predictions) that could be observed in the present study.

Recent work by Sarem et al. (Reference Sarem, Marashi and Siyyari2018) used their expert opinions to compile questions from various questionnaires about mental and physical fatigue to examine the influence of (listening) fatigue on language learning performance. Rather than use “…a self-made questionnaire constructed by the researcher…” (Sarem et al., Reference Sarem, Marashi and Siyyari2018, p. 106) in the present study we elected to use the Vanderbilt Listening Fatigue Scale (VFS) to assess fatigue related specifically to listening. The VFS is a set of self-report scales designed to assess listening-related (rather than general mental and physical) fatigue (Davis et al., Reference Davis, Schlundt, Bonnet, Camarata, Bess and Hornsby2021a) among individuals with hearing loss and other communication difficulties. The adult version (VFS-A; Hornsby et al., Reference Hornsby, Camarata, Cho, Davis, McGarrigle and Bess2021) is a 40-item questionnaire that specifically evaluates fatigue experienced by adults during listening tasks, measuring aspects such as concentration, mental effort, and emotional exhaustion. Other versions have been developed to assess listening-related fatigue in children, with versions completed by the child (VFS-C), or by parents (VFS-P) or teachers (VFS-T; Davis et al., Reference Davis, Schlundt, Bonnet, Camarata, Hornsby and Bess2021b). The VSF has been translated and validated in several other languages including Spanish (Benítez-Barrera et al., Reference Benítez-Barrera, Jones, Hornsby and de Diego-Lázaro2025), Italian (Piazzalunga et al., Reference Piazzalunga, Hornsby, Parravicini, Dossi, Negri, Schindler and Cristofari2025), and Turkish (Taşdemir et al., Reference Taşdemir, Türkyılmaz and Hornsby2024). And a brief version of the adult VFS containing 10 instead of 40 items has also been developed (VFS-A-10; Hornsby et al., Reference Hornsby, Camarata, Cho, Davis, McGarrigle and Bess2023).

Although most research utilizing different versions of the VFS has focused on individuals with hearing impairments, it has been suggested that the scale may also apply to second language (L2) learners (Hornsby et al., Reference Hornsby, Camarata, Cho, Davis, McGarrigle and Bess2021). Given the rigorous psychometric evaluation of the VFS scales, its translation into several other languages, the development of an abbreviated form (VFS-A-10), and the possibility for use in L2 learners, we used the VFS-A-10 in the present study.

To measure anxiety specifically related to foreign language learning, Horwitz et al. (Reference Horwitz, Horwitz and Cope1986) developed the Foreign Language Classroom Anxiety Scale (FLCAS). The FLCAS is a 33-item self-report scale that captures various components of language anxiety, including communication apprehension, test anxiety, and fear of negative evaluation by others. With high reliability and validity, it has become a widely used instrument in research on language learning anxiety (cited 14,118 times according to Google Scholar as of 29-June-2026). The scale helps identify students experiencing high levels of anxiety in the classroom, which can significantly hinder their language acquisition (Horwitz, Reference Horwitz2001).

The network science approach

Conventional statistics, such as the simple correlation between the sum score on the VFS-A-10 and the sum score on the FLCAS, provide coarse-grained information about the relationship between listening fatigue and anxiety related to language learning. Also, although sum scores are easy to compute, they may discard useful information. To illustrate this point, imagine a scale with five questions and responses using a 5-point rating scale. Group A has two respondents: Person A1: 1,2,3,4,5; Person A2: 1,3,5,2,4. Group B also has two respondents: Person B1: 1,2,3,4,5; Person B2: 1,2,3,4,5. The sum scores are the same in all four cases (Person A1 = 15; Person A2 = 15; Person B1 = 15; Person B2 = 15). However, in contrast to the people in Group A, the people in Group B responded to the questions in a consistent manner. By looking just at sum scores, as is the convention for many scales, an important pattern or relationship among the test items might be missed. However, if one used a network approach to look at the relationship among test items, that (potentially important) pattern might be revealed. To examine the relationship between listening fatigue and anxiety related to language learning in a more fine-grained way, we used computational tools from network science (Barabási, Reference Barabási2009; Cramer et al., Reference Cramer, Waldorp, van der Maas and Borsboom2010).

In the network science approach, nodes are used to represent the individual questions on each scale (i.e., the VFS-A-10 and the FLCAS) and edges are used to connect the questions/nodes based on the strength of the correlation that exists between the responses to those two questions/nodes. By assessing the correlations that exist among all of the pairs of responses, a web-like network can be created to represent the concepts of listening fatigue (as assessed by the VFS-A-10) and of anxiety related to language learning (as assessed by the FLCAS), as well as the relationship (if any) that may exist between these two concepts.

The tools of network analysis also allow researchers to measure various structural characteristics of the resulting network. Understanding the structure of a given network (i.e., the ways in which the nodes are connected or tend to group together) is important because a central tenet of network science suggests that the structure of a network influences how efficiently processes operate in that network. For example, Kleinberg (Reference Kleinberg2000; see also Latora & Marchiori, Reference Latora and Marchiori2001) found that navigation through a network with a certain structure (known as a small-world network) was more efficient compared to the same search algorithm in a network with the same number of nodes and the same number of edges, but a different overall structural pattern.

The structure of a network can be measured at different levels of analysis. At the micro-level, metrics such as node strength quantify the importance of individual nodes. Node strength measures the number and weight of the connections that connect to a node (Barrat et al., Reference Barrat, Barthélemy, Pastor-Satorras and Vespignani2004). Nodes that are low in strength have few connections to other nodes and/or connections with small weights. Nodes that are high in strength have many connections to other nodes and/or connections with larger weights. If such nodes are activated (i.e., that symptom is triggered), then large amounts of activation may spread rapidly to many other nodes in the network, resulting in those other symptoms being experienced, and an overall increase in classroom anxiety and/or fatigue. Identifying items in scales like the FLCAS or VFS-A-10 that are high in node strength may help instructors prioritize which anxiety or fatigue symptoms/behaviors to address with some type of intervention to potentially reduce anxiety or fatigue in the classroom.

At the meso-level, community detection algorithms identify subsets of nodes that are more closely interconnected with each other than with the rest of the network. The identification of such communities may reveal whether foreign language anxiety is better understood as a single cluster of interrelated symptoms or as a multidimensional phenomenon encompassing distinct but related facets, such as the facets originally identified by Horwitz et al. (Reference Horwitz, Horwitz and Cope1986): communication apprehension, fear of negative evaluation, and physiological arousal. Meso-level findings can inform whether a single-pronged (addressing just anxiety or just fatigue) or multi-pronged intervention (addressing both anxiety and fatigue) might be more appropriate.

In the present network analysis, we measured node strength at the micro-level to determine if specific anxiety or fatigue symptoms were especially “important” in the network. We also used a community detection algorithm to measure the meso-level to determine if several nodes tend to be related to each other in some way, forming smaller, tightly connected subsets (known as a community) in the network instead of the network simply forming one big collection of nodes. Because a network analysis (but not a conventional analysis) represents the relationship between listening fatigue and anxiety related to language learning at a finer-grained level (compared to the coarse-grained measure of a simple correlation between the sum score of the two scales), the network analyses described here may also provide clues (that could be further examined in future studies) about the mechanism that may underlie the possible relationship between listening fatigue and anxiety related to language learning.

Possible outcomes

Because the present analysis is exploratory in nature, we describe the three outcomes that are logically possible when considering the relationship between two constructs, such as listening fatigue and foreign language classroom anxiety. If listening fatigue and foreign language classroom anxiety are simply two independent phenomena that just happen to both occur in the L2/EFL classroom, then one would expect no correlation (i.e., Pearson’s r ∼ 0) between scores of listening fatigue (as assessed by the VFS-A-10) and scores of foreign language learning anxiety (as assessed by the FLCAS). Further, a visualization of the network might show all of the nodes representing the items on the VFS-A-10 and all of the nodes representing the items on the FLCAS being completely disconnected from each other (or with very low but statistically nonsignificant weights on the connections between items on the VFS-A-10 and items on the FLCAS).

Alternatively, there may exist a positive correlation between scores of listening fatigue and scores of foreign language learning anxiety. This possibility is supported by the diagnostic criteria found in the Diagnostic and statistical manual of mental disorders (American Psychiatric Association, 2022) which shows that “being easily fatigued” is a symptom of Generalized Anxiety Disorder, and “uncontrolled anxiety” is a symptom of Chronic Fatigue Syndrome. Given the positive relationship between fatigue and anxiety observed in these two clinical conditions, perhaps the more specific types of fatigue (i.e., listening fatigue) and anxiety (i.e., foreign language anxiety) are also positively correlated in the language classroom. Further, a visualization of the network might show that nodes representing the items on the VFS-A-10 have many, highly weighted and positive connections to the nodes representing the items on the FLCAS.

Finally, a negative correlation may exist between listening fatigue and anxiety related to language learning. That is, individuals high in listening fatigue experience less anxiety related to language learning than those low in listening fatigue. To understand how this relationship might emerge considering the concept of “ego depletion” (Baumeister et al., Reference Baumeister, Bratslavsky, Muraven and Tice1998), which suggests there is a limited amount of cognitive resources available to an individual. Engaging in one cognitive task diminishes the resources available for subsequent or concurrent cognitive tasks (Baumeister et al., Reference Baumeister, Bratslavsky, Muraven and Tice1998). Work by Fan et al. (Reference Fan, Cheng, Tang, Huang and Yu2024) indicates that ego depletion influences the relationship between anxiety and academic procrastination in university students, suggesting that ego depletion could also play a role in the foreign language learning classroom.

To understand how ego depletion may influence the foreign language learning classroom, consider that processing one’s first language (L1) requires some cognitive effort (Peelle, Reference Peelle2018), and that processing in L2 is significantly more effortful (Borghini & Hazan, Reference Borghini and Hazan2018; Scharenborg & van Os, Reference Scharenborg and van Os2019; Song & Iverson, Reference Song and Iverson2018). From this perspective, allocating cognitive resources to manage the anxiety related to foreign language learning may not only lower the level of anxiety one feels but may also deplete cognitive resources, leaving fewer resources for the demanding tasks related to the processing and comprehension of L2. As a result, learners who report low levels of anxiety may experience greater feelings of fatigue when engaging in language tasks (e.g., reading, writing, speaking, and listening) in L2. In this case, one might observe a negative correlation between scores of listening fatigue (as assessed by the (VFS-A-10) and scores of foreign language learning anxiety (as assessed by the FLCAS), such that those who are low in anxiety also tend to experience greater amounts of fatigue. Further, a visualization of the network might show that nodes representing the items on the VFS-A-10 have many, highly weighted and negative connections to the nodes representing the items on the FLCAS.

Methods

Participants: This project was approved by the Institutional Review Boards at Qassim University, Kunming University of Science and Technology, and University of Kansas. All participants provided informed consent to participate in this study. Learners of English were recruited from Qassim University in Suadi Arabia (who were native speakers of Arabic; n = 76) and from Kunming University of Science and Technology in China (who were native speakers of Mandarin; n = 174). Data on gender/sex were not collected. Like most statistical analysis techniques, a larger sample size is preferred for network analysis to provide higher precision, increased statistical power, greater generalizability, etc. Therefore, in the analyses that follow, the data from the Arabic and Mandarin learners were combined. See Appendix A for additional details about the learners and analyses that further justify combining the two groups.

Procedure: This study was conducted on-line using Qualtrics. After providing consent, participants answered several questions to provide demographic information, then completed the 33 questions on the FLCAS (Horwitz et al., Reference Horwitz, Horwitz and Cope1986). The questions of the FLCAS (listed in Appendix B) were modified to specify “English” class/teacher/etc. Responses to each item were made with a 5-point Likert scale (Strongly Agree, Agree, Neither agree nor disagree, Disagree, Strongly Disagree).

Participants then completed the 10 questions (also listed in Appendix B) on the abbreviated version of the adult VFS (VFS-A-10). Responses to each item were made with a 5-point Likert scale (Never/Almost Never, Rarely, Sometimes, Often, Almost Always/Always). The questions on the VFS-A-10 were modified to specify “listening to English.”

The R packages igraph (Csardi & Nepusz, Reference Csardi and Nepusz2006) and qgraph (Epskamp et al., Reference Epskamp, Cramer, Waldorp, Schmittmann and Borsboom2012), Gephi (Bastian et al., Reference Bastian, Heymann and Jacomy2009), and JASP (JASP Team, 2024) were used to perform (or confirm) the network analyses, to create the network visualization, and perform conventional analyses. In the network analyses, a node in the network represented a question on the FLCAS or the VLF-A-10, and the connection between nodes represented the strength and direction of the correlation between the responses to those two items.

Responses from 250 participants (i.e., the data from the two groups of learners were combined; see Appendix A for additional analyses that justify combining the two groups) to the 43 questions (33 from the FLCAS and 10 from the VLF-A-10) were used to create a Gaussian graphical model with 43 nodes (i.e., each node represented a question on the scale), and with edges between nodes representing the correlation between pairs of nodes. Given the large number of correlations being performed, the “least absolute shrinkage and selection operator” (LASSO; Tibshirani, Reference Tibshirani1996) and the Extended Bayesian Information Criterion (EBIC; Chen and Chen, Reference Chen and Chen2008) were used to reduce small correlations (i.e., edge weights) to zero to create a sparser (in this case sparsity = .72) and more interpretable network.

Node strength is a micro-level metric that quantifies how important an individual node is in a network. It measures the number and weight of the connections that connect to a node (Barrat et al., Reference Barrat, Barthélemy, Pastor-Satorras and Vespignani2004). Nodes that are low in strength have few connections to other nodes and/or connections with small weights. Nodes that are high in strength have many connections to other nodes and/or connections with larger weights.

A community detection algorithm, known as the Louvain optimization method (Girvan & Newman, Reference Girvan and Newman2002), was used to examine the meso-level structure of the network to determine if distinct sub-sets of nodes could be observed. In a network, a community refers to a set of nodes that tend to be more interconnected with each other than to nodes in other communities in the network. There are several community detection algorithms that differ (among other ways) in whether a node can belong to only one community or be a part of more than one community (Fortunato & Darko, Reference Fortunato and Darko2016). In the Louvain optimization method used here, a node can belong to only one community.

The number and size of the communities that might be detected is influenced by the resolution parameter. Values greater than 1 lead to a greater number of communities, with extremely large resolution values resulting in each node in the network being a community unto itself. Resolution values that are smaller than 1 result in a smaller number of communities, with extremely small resolution values resulting in a single community that is isomorphic to the entire network. The modularity metric, Q, measures the density of links inside communities as compared to links between communities (Newman, Reference Newman2004; Reference Newman2006) and is often used to assess the quality of the partitioning of nodes into groups (Fortunato, Reference Fortunato2010). Values of Q tend to range from −1 to +1 (although sometimes do exceed +1.0), with values close to 0 suggesting that the groupings that are obtained are not better than random groupings, and positive values indicating the presence of community structure (i.e., potentially meaningful groupings of nodes).

Results

Conventional analysis of FLCAS and VFS-A-10 sum scores

Pearson’s r was used to provide a coarse-grained measure of the relationship between the sum score on the FLCAS (mean FLCAS sum score = 94.58, sd = 13.47) and the sum score on the VFS-A-10 (mean VFS-A-10 sum score = 26.24, sd = 6.31). In computing the sum score of the FLCAS, we used the 5-point scale as described in Horwitz et al. (Reference Horwitz, Horwitz and Cope1986) with the labels: strongly agree; agree; neither agree nor disagree; disagree; strongly disagree. There is no instruction in Horwitz et al. (Reference Horwitz, Horwitz and Cope1986; or in Saito et al. (Reference Saito, Garza and Horwitz1999), which is an adaptation of FLCAS to focus on just reading in a foreign language) to reverse score any items on the FLCAS. However, some subsequent adaptations of the FLCAS for other languages (e.g. Korean; Park & French, Reference Park and French2013) specify that certain items (#2, 5, 8, 11, 14, 18, 22, 28, 32) should be reverse scored. In the present analyses, we followed the instructions as in Horwitz et al. (Reference Horwitz, Horwitz and Cope1986), meaning that no items were reverse scored. We found a significant negative relationship between the sum score on the FLCAS and the sum score on the VFS-A-10 (r (250) = −.38, p < .001; r 2 = .14, a medium effect), suggesting that individuals with greater levels of anxiety reported lower levels of listening fatigue. Alternatively, individuals with lower levels of anxiety reported greater levels of listening fatigue.

Network analysis of FLCAS and VFS-A-10

Figure 1 shows the network of the FLCAS and VLF-A-10 items. Nodes in the network represent the items from the two questionnaires (FLCAS items are red, VLF-A-10 items are green). A list of which FLCAS and VLF-A-10 items correspond to the keywords used to label the nodes in Figure 1 can be found in Appendix B. The edges (i.e., connections) between nodes represent the correlation in the response to those items. The strength of the correlation is represented by the thickness of the edge, with thinner lines corresponding to weaker correlations and thicker lines corresponding to stronger correlations. The color of the edge indicates whether the correlation is positive (green edges) or negative (red edges).

Figure 1.

Network of items from the FLCAS (red nodes) and VLF-A-10 (green nodes). The keywords used to label the nodes can be found in Appendix B. The edges (i.e., connections) between nodes represent the correlation in the response to those items (positive correlation = green edges; negative correlation = red edges). The strength of the correlation is represented by the thickness of the edge, with thinner lines corresponding to weaker correlations and thicker lines corresponding to stronger correlations. Two communities were detected: the nodes inside the yellow circle constitute one community, and the nodes outside the yellow circle constitute the other community.

A network diagram showing the correlation between items from the FLCAS and VLF-A-10, with nodes representing different items and edges indicating correlations.
Figure 1. Long description

A network diagram showing the correlation between items from the FLCAS and VLF-A-10. The diagram consists of red and green nodes connected by edges of varying thickness. The red nodes represent items from the FLCAS, and the green nodes represent items from the VLF-A-10. The edges between nodes indicate the correlation in the response to those items, with green edges representing positive correlations and red edges representing negative correlations. The thickness of the edges represents the strength of the correlation, with thinner lines corresponding to weaker correlations and thicker lines corresponding to stronger correlations. Two communities are detected: the nodes inside the yellow circle constitute one community, and the nodes outside the yellow circle constitute the other community.

Node strength

Node strength is a micro-level metric that quantifies how important an individual node is in a network by summing the number and weight of the connections that connect to a node (Barrat et al., Reference Barrat, Barthélemy, Pastor-Satorras and Vespignani2004). Nodes that are high in strength have many connections to other nodes and/or connections with larger weights. Normalized values (in parentheses) are reported for the 5 nodes that were highest in node strength: 27.[speak] (2.001) I get nervous and confused when I am speaking in my English class; 20.[heart] (1.574) I can feel my heart pounding when I’m going to be called on in English class; 26.[tense] (1.361) I feel more tense and nervous in my English class than in my other classes; 31.[laugh] (1.286) I am afraid that the other students will laugh at me when I speak English; 33.[questions] (1.168) I get nervous when the English teacher asks questions which I haven’t prepared in advance. Note that the 5 nodes that were highest in node strength are all items on the FLCAS. An item from the VLF-A-10 does not appear among the nodes that were highest in node strength until 8th place (F_tired .823).

Community detection

Using the Louvain community detection algorithm, two subsets of nodes (i.e., communities) were identified, as indicated in Figure 1 by the yellow circle (the nodes inside the circle form one community, and the nodes outside the yellow circle form the other community). The solution depicted in Figure 1 (obtained with a resolution value = 1.0) had a Q = 1.3, indicating that the community structure is of high quality. That is, the nodes in each subset are more tightly interconnected with each other than they are to the nodes in the other community. Note that the 5 nodes that were highest in node strength are all found in the community contained by the yellow circle.

Discussion

Conventional analysis of FLCAS and VFS-A-10 sum scores

In the present study, we used conventional statistical analyses and computational techniques from network science to examine the relationship between foreign language anxiety (as measured by the FLCAS) and listening fatigue (as measured by the VFS-A-10). Using a coarse-grained conventional statistical analysis (i.e., Pearson’s r), we found a significant negative correlation between the sum score on the FLCAS (a measure of foreign language anxiety) and the sum score on the VFS-A-10 (a measure of listening fatigue). That is, individuals with greater levels of anxiety reported lower levels of listening fatigue, or said another way, individuals with greater levels of listening fatigue reported lower levels of anxiety.

Because the present study was exploratory in nature, we considered all the possible relationships that could exist between foreign language anxiety and listening fatigue. First, foreign language anxiety and listening fatigue may be two independent phenomena that happen to co-exist in the second language classroom. In that case, there would be no significant correlation between the sum scores for the FLCAS and VFS-A-10. Given the significant negative correlation between the sum scores that was observed, this possibility can be rejected.

Second, we considered the possibility that foreign language anxiety and listening fatigue may be positively correlated, just as anxiety and fatigue are correlated in the clinical conditions of Generalized Anxiety Disorder (which lists fatigue as a symptom) and Chronic Fatigue Syndrome (which lists anxiety as a symptom). Given the significant negative correlation between the sum scores that was observed, this possibility can also be rejected.

However, Horwitz et al. (Reference Horwitz, Horwitz and Cope1986) argued that foreign language anxiety is a distinct form of anxiety related specifically to the language learning classroom. It may be the case that the more specific forms of fatigue (i.e., listening fatigue rather than physical or mental fatigue) and of anxiety that exist in the language classroom (i.e., foreign language anxiety rather than Generalized Anxiety Disorder) are simply less extreme instances of the clinically relevant forms of fatigue (i.e., Chronic Fatigue Syndrome) and anxiety (i.e., Generalized Anxiety Disorder). Although Horwitz et al. (Reference Horwitz, Horwitz and Cope1986) did not specify a certain score or value that might categorize someone as having clinically relevant anxiety, we considered the range of possible FLCAS scores: a low [33 items x rating of 1] of 33 to a high [33 items x rating of 5] of 165. The levels of anxiety reported in the learners of English in the present study (mean FLCAS sum score = 94.58, sd = 13.47) could be described as moderate, suggesting that the students in the present study may not have been experiencing extreme or “clinically relevant” levels of anxiety in the language classroom. We acknowledge that a different relationship between listening fatigue and classroom anxiety could exist at more extreme levels of classroom anxiety.

The third possibility of a negative correlation between listening fatigue and anxiety related to language learning was indeed observed in the present study. That is, individuals high in listening fatigue reported less anxiety related to language learning. Alternatively, individuals with greater levels of anxiety reported lower levels of listening fatigue. Drawing on the ego depletion concept (Baumeister et al., Reference Baumeister, Bratslavsky, Muraven and Tice1998), the negative correlation we observed may arise if second language learners allocate their cognitive resources to varying degrees to manage anxiety and to engage in L2 processing. Previous work by Fan et al. (Reference Fan, Cheng, Tang, Huang and Yu2024) suggests that ego depletion influences the relationship between anxiety and academic procrastination in university students. In the present study, learners who allocate their cognitive resources to reduce their anxiety (resulting in low levels of anxiety) may have fewer resources to engage in L2 processing, resulting in higher levels of listening fatigue. Other learners may allocate their cognitive resources to L2 processing (resulting in low levels of fatigue), but at the expense of experiencing higher levels of language related anxiety.

Although we did not measure classroom performance in the present study, one might argue that experiencing anxiety may hinder performance, making it counter-intuitive that listeners might allocate their cognitive resources to L2 processing at the expense of experiencing higher levels of language related anxiety. It is important to note, however, that it is excessive levels of anxiety that may hinder performance in the classroom. It has been observed that a moderate level of statistical anxiety (i.e., not too high and not too low) leads to optimal performance in the statistics classroom (Keeley et al., Reference Keeley, Zayac and Correia2008). Low levels of statistical anxiety may fail to motivate a student to engage with the material leading to poor performance, whereas very high levels of statistical anxiety may lead to unproductive behaviors (e.g., avoiding class, etc.) and poor performance. However, a moderate level of statistical anxiety may be enough to motivate students to engage in appropriate behaviors (e.g., taking notes, reviewing class slides, forming study groups, etc.), but not be too overwhelming or debilitating, leading to good performance in the classroom. Said another way, anxiety and classroom performance may be better described as an inverse-U shape function rather than a straight line.

Perhaps the same is true in the language classroom. Although we did not assess performance in the foreign language classroom in the present study, the levels of anxiety reported in the learners of English in the present study could arguably be described as moderate. Future studies might explore further how the level of anxiety contributes to negative or positive outcomes in the language learning classroom.

Network analysis of FLCAS and VFS-A-10

We grant that a coarse-grained conventional statistical analysis (i.e., Pearson’s r), allowed us to observe an interesting result (i.e., a negative correlation between classroom anxiety and listening fatigue), and perhaps pointed to a new research direction (i.e., to assess if anxiety level is related to performance with an inverse-U shape function rather than a straight line; Keeley et al., Reference Keeley, Zayac and Correia2008). However, the more fine-grained analysis of the FLCAS and VLF-A-10 scales using the tools of network analysis provided additional insight into the relationship between foreign language learning anxiety and listening fatigue that cannot be observed using conventional measures.

Node strength

Consider first our micro-level analysis of the network using node strength, a measure that identifies “important” nodes. Using this metric, we found that the 5 nodes that were highest in node strength (27.speak, 20.heart, 26.tense, 31.laugh, and 33.questions) were all items from the FLCAS and were all found in the same community (i.e., the subset of nodes enclosed by the yellow circle in Figure 1). An item from the VLF-A-10 did not appear on an order list of node strength until 8th place. Although the present (coarse-grained conventional statistical) analysis showed that foreign language anxiety and listening fatigue are related, the network analysis of node strength suggests that a foreign language instructor might have better outcomes implementing an intervention that is targeted at certain anxiety-related symptoms than at fatigue-related symptoms. For example, the 5 items highest in node strength relate to the anxiety a student might experience when having to “perform” in front of the teacher or their peers in class. Instead of randomly calling on students in class to answer a question, an instructor may reduce anxiety by having students engage in think-pair-share exercises, where students individually think about the question, pair up with a partner to discuss their ideas, and finally share the pair’s thoughts with the whole class (Guenther & Abbott Reference Guenther and Abbott2024). A proper experiment would be required to test this assertion, but the tools of network science provide important guidance for a starting point for a potential intervention instead of randomly selecting from the 43 behaviors in the present study (33 items on FLCAS + 10 items on VLF-A-10).

Community detection

Using the Louvain community detection algorithm, two subsets of nodes (i.e., communities) were identified (as indicated in Figure 1 by the yellow circle). One might naïvely assume that the items from the FLCAS would be grouped together in one community, and the items from the VLF-A-10 would be grouped together in the other community. However, this was not the case. One community (i.e., the FLCAS items in the yellow circle) did contain most of the items from the FLCAS. The FLCAS items found in that community were positively correlated with each other (as indicated by the green edges in Figure 1).

However, the other community contained all the items from the VLF-A-10 and 9 items from the FLCAS. For the convenience of the reader, the 9 items from the FLCAS are listed here: 2.[worry] I do not worry about making mistakes in English class; 5.[no_bother] It wouldn’t bother me at all to take more English classes; 8.[ease] I am usually at ease doing my assignments in my English class; 11.[u_stand] I don’t understand why some people get so upset over English class; 14.[native] I would not be nervous speaking English with native speakers; 18.[confident] I feel confident when I speak in English class; 22.[no_press] I don’t feel pressure to prepare very well for English class; 28.[sure_rlx] When I’m on my way to English class, I feel very sure and relaxed; 32.[ok_native] I would probably feel comfortable around native speakers of English. (A similar community structure was also observed in Rossomondo and Vitevitch (Reference Rossomondo and Vitevitch2026) for English speakers learning Spanish.)

Note that the 9 items from the FLCAS found in the community with all the VLF-A-10 items don’t appear to assess “anxiety” related to learning a second language. Rather they appear to assess “confidence” in using the second language. Further note that those 9 items from the FLCAS in that community are positively correlated with each other (as indicated by the green edges in Figure 1), positively correlated with the VLF-A-10 items in that community (as indicated by the green edges in Figure 1), and negatively correlated (as indicated by the red edges in Figure 1) with the FLCAS items found in the other community (i.e., the FLCAS items in the yellow circle). (Note that reverse scoring these items would not change the strength of the relationship between these nodes and the other FLCAS nodes in the other community; only the color of the connection would change.) Furthermore, all the VLF-A-10 items are negatively correlated (as indicated by the red edges in Figure 1) with the FLCAS items found in the other community (i.e., the FLCAS items in the yellow circle). Putting these correlations into words: language learners with high levels of confidence in their abilities report low levels of anxiety in the language classroom, but high levels of fatigue. Alternatively, language learners with low levels of confidence in their abilities report high levels of anxiety in the language classroom, but low levels of fatigue.

Observing more than one community in the community detection analysis at the meso-level of the network suggests that foreign language anxiety may be better understood as a multidimensional phenomenon instead of a monolithic construct. This is perhaps not surprising given that Horwitz et al. (Reference Horwitz, Horwitz and Cope1986) originally identified several factors that contribute to foreign language anxiety and that guided their development of the FLCAS: communication apprehension, fear of negative evaluation, and physiological arousal (among others). Additional evidence of foreign language anxiety being multidimensional in nature is found in subsequent analyses of the FLCAS using principal components analysis (in several different languages/countries) where 2 components (Cheng et al., Reference Cheng, Horwitz and Schallert1999; Park, Reference Park2014), 3 components (Liu & Jackson, Reference Liu and Jackson2008), or 7 components (Matsuda & Gobel, Reference Matsuda and Gobel2004) have been observed. In the network approach, adjustments to the resolution parameter may lead to the detection of more/fewer communities (enabling one to possibly replicate the number of components that have been previously observed). However, the resulting communities may not be as high in quality (as assessed by the modularity metric, Q) as those presently reported.

Although the results of the present community detection analysis support the idea that foreign language anxiety is multidimensional rather than monolithic in nature, the observed community structure may suggest that the dimensions of foreign language anxiety that language learners experience may differ from the dimensions that language experts, instructors, and educational researchers perceive and use to develop various scales, measures, questionnaires, etc. Future studies using the network approach to map out the semantic space of novices (i.e., learners) and experts (i.e., instructors or researchers) in the domain of foreign language learning may prove useful in determining which dimensions of foreign language anxiety are “important” to whom (Siew & Guru, Reference Siew and Guru2023).

Finally, the results of the community detection analysis in conjunction with the node strength analysis provide some guidance on whether a single-pronged approach or a multi-pronged approach might be more useful if attempting an intervention in the foreign language classroom. Given that 2 communities were observed, one might surmise that a multipronged approach targeting anxiety, confidence, and fatigue might be beneficial. However, recall that one of the observed communities contained most of the items on the FLCAS and contained all 5 of the nodes that were highest in strength. Further, the items from the VLF-A-10 occupied a different community, and an item from the VLF-A-10 appeared in 8th place in an ordered list of nodes that were highest in strength. When considering the results of the community detection analysis and the node strength analysis together, an intervention targeting certain symptoms/behaviors related to anxiety (rather than fatigue) may instead appear more productive. We caution that the results of the present analyses do not confirm this assertion. Rather, the results of the present analyses simply provide instructors/ researchers with guidance on which of the 43 symptoms/behaviors might be a good place to start when conducting a proper experiment to assess the success of an intervention.

Limitations and future directions

The present study of foreign language anxiety and listening fatigue demonstrates how the finer-grained analysis techniques from network science can reveal a more complicated or nuanced relationship between concepts than coarser-grained conventional statistical techniques. Although the present study demonstrated one way in which network science might be useful to increasing our understanding of language learning, the present study does have some limitations. To our knowledge, this is the first use of the VLF-A-10 with L2 learners. We recognize that the various versions of the VFSs have undergone rigorous psychometric evaluation (which was one of the reasons why we used it here). However, those evaluations have been predominantly in individuals with varying degrees of hearing loss. It may be the case that a different scale better assesses the fatigue experienced by second language learners, especially given that learning a language also requires reading, writing, and speaking in addition to listening.

Another limitation of the present study is that only listening fatigue and foreign language anxiety were examined; we did not include a measure of classroom performance, or of cognitive effort in learning a foreign language, such as the pupillometry measure used in Brännström et al. (Reference Brännström, Rudner, Carlie, Sahlén, Gulz, Andersson and Johansson2021). Future studies might also consider including a measure of motivation in foreign language learning. It may be the case that students who are less motivated to learn a second language may feel low or no anxiety (because they don’t care about learning the language) and may not feel fatigue (because they do not put in any effort to learn something they don’t care about). We note that the students in the present study likely had moderate to high levels of motivation to learn English given that they need to demonstrate a certain level of English proficiency to be admitted to the university and to graduate.

Despite the limitations of the present study, we believe the results of our conventional and network analyses have provided important insight into the relationship between listening fatigue and foreign language anxiety. Further, the present results have raised new and interesting questions for future research to pursue. Furthermore, the present study demonstrates one way in which network science may increase our understanding of (second) language learning. Networks have also been used to represent the words that language learners acquire over time to examine how the structure of the mental lexicon influences how quickly students learn new words that are phonologically (Luef, Reference Luef2022) or semantically (Hills et al., Reference Hills, Maouene, Maouene, Sheya and Smith2009) related to already known words. One could also apply network analysis to model the people a student associates with (i.e., nodes in the network) to examine how peer interactions (i.e., edges in the network) in a “social network” influences the cognitive process of learning a second language during an intensive study abroad experience (Paradowski et al., Reference Paradowski, Jarynowski, Jelińska and Czopek2021). As demonstrated by the present study as well as these additional examples, there are many ways that the network approach can increase our understanding of second language learning.

Competing interests

The authors declare none.

Appendix A

Information and analyses of the two groups of language learners

All learners were currently enrolled in an English language class at their respective institutions. As shown in the table below, the minimum and maximum values on the FLCAS indicate that the distributions for the Arabic and Chinese listeners overlapped. Also shown in the table below, the minimum and maximum values on the VLF indicate that the distributions for the Arabic and Chinese listeners overlapped. Given the overlap in the distributions for the Arabic and Chinese learners, their scores on the FLCAS were combined, and their scores on the VLF were combined to create a larger sample size for the network analyses.

A table comparing FLCAS and VLF values for Arabic and Chinese.
Long description

A table comparing FLCAS and VLF values for Arabic and Chinese. The table has four columns and three rows. The columns are labeled FLCAS Arabic, FLCAS Chinese, VLF Arabic, and VLF Chinese. The rows are labeled n, Minimum, and Maximum. Row 1: n, 76, 174, 76, 174. Row 2: Minimum, 69, 66, 10, 12. Row 3: Maximum, 134, 127, 36, 39.

Some previous studies (e.g., Xu & Xie, Reference Xu and Xie2024) have reported negative correlations between foreign language anxiety (as measured by the FLCAS or similar scales) and foreign language proficiency (as measured by GPA, grade level, performance on a standardized test, etc.; but see Rossomondo & Vitevitch, Reference Rossomondo and Vitevitch2026). For the Arabic learners, proficiency in English was determined by the semester of English classes that they had enrolled in at Qassim University (min. = 5; max. = 10). This was not significantly correlated with their score on the FLCAS (r = −.08, p = .52, n.s.). For the Mandarin speakers, proficiency in English was determined by their score on the College English Test (CET-Band 4), a standardized test (using a 710-point scoring system) created and administered by The National Advisory Committee on Teaching English Language to Non-English Majors in Higher Education under The Ministry of Education (National College English Testing Committee, 2006). For the Chinese learners who reported their CET-Band 4 score (n = 138; min. = 308; max. = 564), proficiency in English was not significantly correlated with their score on the FLCAS (r = −.02, p = .86, n.s.). The lack of a correlation between (two different measures of) proficiency in English and anxiety levels (as measured by the FLCAS) in both groups of English learners further justified combining the two groups into a larger sample for the network analyses.

Appendix B

Items from the Foreign Language Classroom Anxiety Scale (FLCAS; Horwitz, et al., Reference Horwitz, Horwitz and Cope1986). Words in [square brackets] show the abbreviation for that item used in the network shown in Figure 1. Items in the FLCAS were modified to specify English class/teacher/etc.

  1. 1. [sure_spk] I never feel quite sure of myself when I am speaking in my English class.

  2. 2. [worry] I do not worry about making mistakes in English class.

  3. 3. [tremble] I tremble when I know that I’m going to be called on in English class.

  4. 4. [frighten] It frightens me when I do not understand what the teacher is saying in English.

  5. 5. [no_bother] It wouldn’t bother me at all to take more English classes.

  6. 6. [no_think] During English class, I find myself thinking about things that have nothing to do with the course.

  7. 7. [better] I keep thinking that the other students are better at English than I am.

  8. 8. [ease] I am usually at ease doing my assignments in my English class.

  9. 9. [panic] I start to panic when I have to speak without preparation in English class.

  10. 10. [failing] I worry about the consequences of failing my English class.

  11. 11. [u_stand] I don’t understand why some people get so upset over English class.

  12. 12. [forget] In English class, I can get so nervous I forget things I know.

  13. 13. [embarrass] It embarrasses me to volunteer answers in my English class.

  14. 14. [native] I would not be nervous speaking English with native speakers.

  15. 15. [upset] I get upset when I don’t understand what the English teacher is correcting.

  16. 16. [anxious] Even if I am well prepared for English class, I feel anxious about it.

  17. 17. [no_go] I often feel like not going to my English class.

  18. 18. [confident] I feel confident when I speak in English class.

  19. 19. [afraid] I am afraid that my English teacher is ready to correct every mistake I make.

  20. 20. [heart] I can feel my heart pounding when I’m going to be called on in English class.

  21. 21. [confused] The more I prepare for a English assignment, the more confused I get.

  22. 22. [no_press] I don’t feel pressure to prepare very well for English class.

  23. 23. [other] I always feel that the other students speak English better than I do.

  24. 24. [self_con] I feel very self-conscious about speaking English in front of other students.

  25. 25. [quickly] English class moves so quickly I worry about getting left behind.

  26. 26. [tense] I feel more tense and nervous in my English class than in my other classes.

  27. 27. [speak] I get nervous and confused when I am speaking in my English class.

  28. 28. [sure_rlx] When I’m on my way to English class, I feel very sure and relaxed.

  29. 29. [t_says] I get nervous when I don’t understand every word the English teacher says.

  30. 30. [rules] I feel overwhelmed by the number of rules you have to learn to speak English.

  31. 31. [laugh] I am afraid that the other students will laugh at me when I speak English.

  32. 32. [ok_native] I would probably feel comfortable around native speakers of English.

  33. 33. [questions] I get nervous when the English teacher asks questions which I haven’t prepared in advance.

Items from the abbreviated adult VFS (VLF-A-10; Hornsby, et al., Reference Hornsby, Camarata, Cho, Davis, McGarrigle and Bess2023). Words in [square brackets] show the abbreviation for that item used in the network shown in Figure 1; the prefix “F_” was added to the VLF-A-10 labels to assist in identifying them in the figure. Items in the VLF-A-10 were modified to specify listening to English.

  1. 1. [F_worn_out] I feel worn out from listening to English.

  2. 2. [F_tired] Struggling to listen and understand English makes me feel tired.

  3. 3. [F_no_enjoy] I get so exhausted from listening to English that I cannot do the things I enjoy.

  4. 4. [F_schedule] I schedule my day to avoid getting tired from listening to English.

  5. 5. [F_details] I get so tired from listening to English that I start to miss details in a conversation.

  6. 6. [F_bed] I get so exhausted from listening to English that I go to bed early.

  7. 7. [F_withdraw] I withdraw when I am unable to follow English conversations in noisy places.

  8. 8. [F_strain] Feeling tired from listening to English causes strain on my relationships.

  9. 9. [F_emo_drain] I feel emotionally drained when it is hard for me to listen and understand English.

  10. 10. [F_energy] It takes a lot of energy to listen and understand English.

References

American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed., text rev.). American Psychiatric Association. https://doi.org/10.1176/appi.books.9780890425787.Google Scholar
Barabási, A. L. (2009). Scale-free networks: A decade and beyond. Science, 325, 412–413. https://doi.org/10.1126/science.1173299.CrossRefGoogle ScholarPubMed
Barrat, A., Barthélemy, M., Pastor-Satorras, R., & Vespignani, A. (2004). The architecture of complex weighted networks. Proceedings of the National Academy of Sciences of the United States of America, 101(11), 3747–3752, https://doi.org/10.1073/pnas.0400087101.CrossRefGoogle ScholarPubMed
Bastian, M., Heymann, S., & Jacomy, M. (2009). Gephi: An open source software for exploring and manipulating networks. International AAAI Conference on Weblogs and Social Media.10.1609/icwsm.v3i1.13937CrossRefGoogle Scholar
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998). Ego depletion: Is the active self a limited resource? Journal of Personality and Social Psychology, 74(5), 1252–1265. https://doi.org/10.1037/0022-3514.74.5.1252.CrossRefGoogle Scholar
Benítez-Barrera, C. R., Jones, A., Hornsby, B. W. Y., & de Diego-Lázaro, B. (2025). Spanish versions of the Vanderbilt fatigue scale for adults: Translation and cultural adaptation. International Journal of Audiology, 64(3), 280–282. https://doi.org/10.1080/14992027.2024.2371886.CrossRefGoogle ScholarPubMed
Borghini, G., & Hazan, V. (2018). Listening effort during sentence processing is increased for non-native listeners: A pupillometry study. Frontiers in Neuroscience, 12(MAR), 1–13. https://doi.org/10.3389/fnins.2018.00152.CrossRefGoogle ScholarPubMed
Bosmans, D., & Hurd, S. (2016). Phonological attainment and foreign language anxiety in distance language learning: A quantitative approach. Distance Education, 37(3), 287–301. https://doi.org/10.1080/01587919.2016.1233049.CrossRefGoogle Scholar
Brännström, K. J., Rudner, M., Carlie, J., Sahlén, B., Gulz, A., Andersson, K., & Johansson, R. (2021). Listening effort and fatigue in native and non-native primary school children. Journal of Experimental Child Psychology, 210, 105203. https://doi.org/10.1016/j.jecp.2021.105203.CrossRefGoogle ScholarPubMed
Chen, C., & Guo, Z. (2026). From foreign language classroom anxiety to English learning engagement: The roles of cognitive reappraisal and expressive suppression. Frontiers in Psychology, 17, 1754113. https://doi.org/10.3389/fpsyg.2026.1754113.CrossRefGoogle ScholarPubMed
Chen, J., & Chen, Z. (2008). Extended Bayesian information criteria for model selection with large model spaces. Biometrika, 95(3), 759–771.10.1093/biomet/asn034CrossRefGoogle Scholar
Cheng, Y. S., Horwitz, E. K., & Schallert, D. L. (1999). Language anxiety: Differentiating writing and speaking components. Language Learning, 49(3), 417–446.10.1111/0023-8333.00095CrossRefGoogle Scholar
Cramer, A. O. J., Waldorp, L. J., van der Maas, H. L. J., & Borsboom, D. (2010). Comorbidity: A network perspective. The Behavioral and Brain Sciences, 33(2–3), 137–150. https://doi.org/10.1017/S0140525X09991567.CrossRefGoogle ScholarPubMed
Csardi, G., & Nepusz, T. (2006). The igraph software package for complex network research. InterJournal Complex Systems 1695, 1–9. Available online at: https://www.necsi.edu/events/iccs6/papers/c1602a3c126ba822d0bc4293371c.pdf.Google Scholar
Davis, H., Schlundt, D., Bonnet, K., Camarata, S., Bess, F. H., & Hornsby, B. (2021a). Understanding listening-related fatigue: Perspectives of adults with hearing loss. International Journal of Audiology, 60(6), 458–468. https://doi.org/10.1080/14992027.2020.1834631 CrossRefGoogle ScholarPubMed
Davis, H., Schlundt, D., Bonnet, K., Camarata, S., Hornsby, B., & Bess, F. H. (2021b). Listening-related fatigue in children with hearing loss: Perspectives of children, parents, and school professionals. American Journal of Audiology, 30(4), 929–940. https://doi.org/10.1044/2021_AJA-20-00216.CrossRefGoogle ScholarPubMed
Epskamp, S., Cramer, A. O. J., Waldorp, L. J., Schmittmann, V. D., & Borsboom, D. (2012). qgraph: Network visualizations of relationships in psychometric data. Journal of Statistical Software, 48, 1–18. https://doi.org/10.18637/jss.v048.i04.CrossRefGoogle Scholar
Fan, J., Cheng, Y., Tang, M., Huang, Y., & Yu, J. (2024). The mediating role of ego depletion in the relationship between state anxiety and academic procrastination among University students. Scientific Reports, 14, 15568. https://doi.org/10.1038/s41598-024-66293-6.CrossRefGoogle ScholarPubMed
Fortunato, S. (2010). Community detection in graphs. Physics Reports, 486, 75–174. https://doi.org/10.1016/j.physrep.2009.11.002.CrossRefGoogle Scholar
Fortunato, S., & Darko, H. (2016). Community detection in networks: A user guide. Physics Reports, 659, 1–44. https://doi.org/10.1016/j.physrep.2016.09.002.CrossRefGoogle Scholar
Girvan, M., & Newman, M. E. J. (2002). Community structure in social and biological networks. Proceedings of the National Academy of Science, U.S.A., 99, 7821–7826. https://doi.org/10.1073/pnas.122653799.CrossRefGoogle ScholarPubMed
Guenther, A. R., & Abbott, C. M. (2024). Think-pair-share: Promoting equitable participation and in-depth discussion. PRiMER, 8, 7. https://doi.org/10.22454/PRiMER.2024.444143.CrossRefGoogle ScholarPubMed
Hills, T, Maouene, M, Maouene, J, Sheya, A, & Smith, L. (2009). Longitudinal analysis of early semantic networks: Preferential attachment or preferential acquisition? Psychological Science, 20, 729–739.10.1111/j.1467-9280.2009.02365.xCrossRefGoogle ScholarPubMed
Hornsby, B., Camarata, S., Cho, S., Davis, H., McGarrigle, R., & Bess, F. (2021). Development and validation of the Vanderbilt fatigue scale for adults (VFS-A). Psychological Assessment, 33(8), 777–788. https://doi.org/10.1037/pas0001021.CrossRefGoogle ScholarPubMed
Hornsby, B. W. Y., Camarata, S., Cho, S.-J., Davis, H., McGarrigle, R., & Bess, F. H. (2023). Development and validation of a brief version of the Vanderbilt fatigue scale for adults: The VFS-A-10. Ear and Hearing, 44(5), 1251–1261. https://doi.org/10.1097/AUD.0000000000001369.CrossRefGoogle ScholarPubMed
Horwitz, E. K. (2001). Language anxiety and achievement. Annual Review of Applied Linguistics, 21, 112–126.10.1017/S0267190501000071CrossRefGoogle Scholar
Horwitz, E. K., Horwitz, M. B., & Cope, J. (1986). Foreign language classroom anxiety. The Modern Language Journal, 70(2), 125–132. https://www.jstor.org/stable/327317.10.1111/j.1540-4781.1986.tb05256.xCrossRefGoogle Scholar
JASP Team (2024). JASP (Version 0.19.3) [Computer software].Google Scholar
Keeley, J., Zayac, R., & Correia, C. (2008). Curvilinear relationships between statistics anxiety and performance among undergraduate students: Evidence for optimal anxiety. Statistics Education Research Journal, 7, 4–15.10.52041/serj.v7i1.477CrossRefGoogle Scholar
Kleinberg, J. M. (2000). Navigation in a small world. Nature, 406, 845.10.1038/35022643CrossRefGoogle Scholar
Koelewijn, T., Zekveld, A. A., Festen, J. M., & Kramer, S. E. (2012). Pupil dilation uncovers extra listening effort in the presence of a single-talker masker. Ear Hear. 33, 291–300. https://doi.org/10.1097/AUD.0b013e3182310019.CrossRefGoogle ScholarPubMed
Kuhl, P. K. (2000). A new view of language acquisition. Proceedings of the National Academy of Sciences U.S.A., 97(22), 11850–11857. https://doi.org/10.1073/pnas.97.22.11850.CrossRefGoogle ScholarPubMed
Latora, V., & Marchiori, M. (2001). Efficient behavior of small-world networks. Physical Review Letters, 87, 198701.10.1103/PhysRevLett.87.198701CrossRefGoogle ScholarPubMed
Liu, M., & Jackson, J. (2008). An exploration of Chinese EFL learners’ unwillingness to communicate and foreign language anxiety. Modern Language Journal, 92(1), 71–86. https://doi.org/10.1111/j.1540-4781.2008.00687.x CrossRefGoogle Scholar
Luef, E. M. (2022). Growth algorithms in the phonological networks of second language learners: A replication of Siew and Vitevitch (2020a). Journal of Experimental Psychology: General, 151(12), e26–e44.10.1037/xge0001248CrossRefGoogle ScholarPubMed
Matsuda, S., & Gobel, P. (2004). Anxiety and predictors of performance in the foreign language classroom. System, 32(1), 21–36.10.1016/j.system.2003.08.002CrossRefGoogle Scholar
National College English Testing Committee. (2006). Syllabus for College English Test Band 4. Shanghai: Shanghai Foreign Language Education Press.Google Scholar
Newman, M. E. J. (2004). Detecting community structure in networks. European Physics Journal B, 38, 321–330. https://doi.org/10.1140/epjb/e2004-00124-y.CrossRefGoogle Scholar
Newman, M. E. J. (2006). Modularity and community structure in networks. Proceedings of the National Academy of Science U.S.A., 103(23), 8577–8582, https://doi.org/10.1073/pnas.0601602103.CrossRefGoogle ScholarPubMed
Paradowski, M. B., Jarynowski, A., Jelińska, M., & Czopek, K. (2021). Out-of-class peer interactions matter for second language acquisition during short-term overseas sojourns: The contributions of social network analysis. Language Teaching, 54(1), 139–143. https://doi.org/10.1017/S0261444820000580.CrossRefGoogle Scholar
Park, G. P. (2014). Factor analysis of the foreign language classroom anxiety scale in Korean learners of English as a foreign language. Psychological Reports, 115(1), 261–275.10.2466/28.11.PR0.115c10z2CrossRefGoogle ScholarPubMed
Park, G. P., & French, B. F. (2013). Foreign Language Classroom Anxiety Scale--Korean Version [Database record]. https://doi.org/10.1037/t61517-000.CrossRefGoogle Scholar
Peelle, J. E. (2018). Listening effort: How the cognitive consequences of acoustic challenge are reflected in brain and behavior. Ear and Hearing, 39(2), 204–214. https://journals.lww.com/ear-hearing/fulltext/2018/03000/listening_effort__how_the_cognitive_consequences.4.aspx.10.1097/AUD.0000000000000494CrossRefGoogle ScholarPubMed
Piazzalunga, S., Hornsby, B. W. Y., Parravicini, G., Dossi, M. M., Negri, L., Schindler, A., & Cristofari, E. (2025). Listening-related fatigue in adults: Cross-cultural adaptation and validation of the Italian version of the Vanderbilt fatigue scale. Folia Phoniatrica et Logopaedica, 77(6), 542–558. https://doi.org/10.1159/000545301.CrossRefGoogle Scholar
Rossomondo, A., & Vitevitch, M. S. (2026). Network analysis of the foreign language classroom anxiety scale. Frontiers in Education, 11, 1866770. https://doi.org/10.3389/feduc.2026.1866770.CrossRefGoogle Scholar
Saito, Y., Garza, T. J., & Horwitz, E. K. (1999). Foreign language reading anxiety. Modern Language Journal, 83(2), 202–218. https://doi.org/10.1111/0026-7902.00016.CrossRefGoogle Scholar
Sarem, S. N., Marashi, H., & Siyyari, M. (2018). The relationship between L2 listening comprehension and listening fatigue among Iranian intermediate EFL learners. The Journal of English Language Pedagogy and Practice, 111(23), 103–124.Google Scholar
Scharenborg, O., & van Os, M. (2019). Why listening in background noise is harder in a non-native language than in a native language: A review. Speech Communication, 108(March), 53–64. https://doi.org/10.1016/j.specom.2019.03.001.CrossRefGoogle Scholar
Shen, J. (2021) A review of the effectiveness of foreign language enjoyment and foreign language classroom anxiety on learners’ engagement and attainment. Frontiers in Psychology, 12, 749284. https://doi.org/10.3389/fpsyg.2021.749284.CrossRefGoogle ScholarPubMed
Siew, C. S. Q., & Guru, A. (2023). Investigating the network structure of domain-specific knowledge using the semantic fluency task. Memory & Cognition, 51(3), 623–646. https://doi.org/10.3758/s13421-022-01314-1 CrossRefGoogle Scholar
Song, J., & Iverson, P. (2018). Listening effort during speech perception enhances auditory and lexical processing for non-native listeners and accents. Cognition, 179(June), 163–170. https://doi.org/10.1016/j.cognition.2018.06.001 CrossRefGoogle ScholarPubMed
Taşdemir, I., Türkyılmaz, M. D., & Hornsby, B. (2024). Reliability and validity of the Turkish version of the Vanderbilt fatigue scale for adults. American Journal of Audiology, 33(2), 532–542. https://doi.org/10.1044/2024_AJA-23-00209.CrossRefGoogle ScholarPubMed
Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological), 58, 267–288.10.1111/j.2517-6161.1996.tb02080.xCrossRefGoogle Scholar
Xu, Y., & Xie, Z. (2024). Exploring the predictors of foreign language anxiety: The roles of language proficiency, language exposure, and cognitive control. Frontiers in Psychiatry, 15, 1492701. https://doi.org/10.3389/fpsyt.2024.1492701.CrossRefGoogle ScholarPubMed
Figure 0

Figure 1. Figure 1 long description.Network of items from the FLCAS (red nodes) and VLF-A-10 (green nodes). The keywords used to label the nodes can be found in Appendix B. The edges (i.e., connections) between nodes represent the correlation in the response to those items (positive correlation = green edges; negative correlation = red edges). The strength of the correlation is represented by the thickness of the edge, with thinner lines corresponding to weaker correlations and thicker lines corresponding to stronger correlations. Two communities were detected: the nodes inside the yellow circle constitute one community, and the nodes outside the yellow circle constitute the other community.