Abstract
Extramural English (EE) has been linked to foreign/second language (L2) English proficiency, but most research studies rely on broad questionnaires, limiting the scope of robust analyses. Although validated EE scales exist in several contexts, none of them have been developed for Türkiye, where EE research remains scarce. This article addresses these gaps through two studies: (1) the development and validation of a Turkish EE scale, and (2) an examination of its predictive value for L2 English proficiency. Study 1 involved 718 university students and followed established procedures for scale development and validation, including exploratory and confirmatory factor analyses, test-retest reliability and known-groups validity. Study 2 applied the scale to a separate sample of 59 university students, whose L2 English proficiency was measured by an institutional exam assessing speaking, writing and reading/listening. Regression and correlation analyses were conducted to explore the links between EE engagement and proficiency outcomes. The final 47-item Turkish EE scale comprises nine factors: Digital Creativity, Gaming, Music, Social Interaction, Reading and Listening, Internalised, Writing, Social Reading and Listening, and Googling and Viewing. Findings showed that EE engagement, especially Internalised EE, significantly predicted L2 English proficiency, particularly speaking. These findings demonstrate that EE is a meaningful predictor of language outcomes in the Turkish context and that specific EE activities contribute more strongly than others. The validated scale provides a reliable tool for future EE research in Türkiye. Based on the results of Study 2, we discuss pedagogical implications for English teaching in Türkiye that explicitly address EE engagement.
Introduction
Technological advancements over the last three decades have prompted a reconceptualisation of foreign/second language (L2) learning environments, now encompassing all settings where individuals are exposed to the target language (see Reinhardt, 2022, for a historical overview). One concept that captures this expanded view of L2 learning is Extramural English (EE; Sundqvist, 2009), which refers to any activities involving the use of English outside formal education that are likely to contribute, either intentionally or incidentally, to an individual’s development of L2 competence.
Several empirical studies have demonstrated the benefits of EE for the development of various L2 skills across different contexts (see Kusyk et al., 2023; Zhang et al., 2021, for reviews). Research has shown that learners can improve their English proficiency by engaging in activities outside the classroom, such as gaming, social media use or watching audiovisual content (e.g., Cole & Vanderplank, 2016; De Wilde et al., 2020; Puimège & Peters, 2019). This evidence underscores EE as an important individual difference in L2 learning (Sundqvist, 2024). However, despite these insights, most studies have relied solely on questionnaires (Lee, 2022). Questionnaires not only restrict analyses to basic statistical procedures like descriptive statistics and correlations, but they have often lacked rigorous statistical validation (Kusyk et al., 2025). Without evidence for their reliability and construct validity, it is difficult to claim that these questionnaires truly capture the multifaceted nature of EE engagement. As noted by Dörnyei and Taguchi (2009), the development of validated and reliable scales is essential to advance this line of research. In recent years, this need has been addressed through the development of EE scales in various countries, including Denmark, Norway, and Sweden (Sundqvist & Uztosun, 2024), Germany (Arndt, 2023) and Indonesia (Lee & Drajati, 2020). However, no statistically validated scale has yet been developed in the Turkish context using rigorous statistical procedures. Although a previous study developed a relevant scale that contributed to EE research in Türkiye (Coşkun & Mutlu, 2017), it focused on high school students and did not employ confirmatory factor analysis to validate the factor structure, which underscores the need for a validated scale suitable for the Turkish context. The present study addresses this gap by introducing a scale to measure EE engagement among Turkish learners, thereby enabling more robust statistical analyses of its impact on L2 English proficiency and other potential variables that influence L2 learning.
Türkiye offers a distinctive context for EE research. Unlike countries where EE research is well established – such as Finland and Sweden (Kusyk et al., 2023) – English instruction in Türkiye generally emphasises grammar and accuracy, with limited focus on communicative competence and several factors further hindering the teaching of English communicatively (Akcan, 2016; Gürsoy et al., 2013; Haznedar, 2012; Valizadeh, 2021). Consequently, many Turkish learners struggle to develop communication skills in English despite receiving years of formal instruction. In such a context, EE engagement may play a particularly important compensatory role by offering opportunities for authentic L2 exposure and use beyond the classroom. Nevertheless, research on EE in Türkiye remains limited, and, to the best of our knowledge, only one study has investigated the relationship between EE engagement and L2 English proficiency (Uztosun & Sundqvist, 2025). While their study demonstrated a correlation between EE and proficiency, it did not provide evidence of causality, which requires regression analysis. The present study is therefore the first in the Turkish context to examine the causal relationship between EE engagement and L2 proficiency.
By offering the first findings based on a scale tailored to the Turkish context, this article presents two empirical studies. Study 1 aimed to develop and validate a scale to measure EE engagement among Turkish university students, while Study 2 employed this scale to explore the impact of EE engagement on L2 English proficiency. The studies address the following research questions:
How reliable and valid is the Turkish EE scale for measuring learners’ engagement in EE activities based on frequency? (Study 1)
What is the frequency of learners’ engagement in EE activities in the Turkish context? (Study 2)
Are there statistically significant relationships between nine measures of EE engagement and L2 English proficiency? (Study 2)
How well does EE engagement predict L2 English proficiency? (Study 2)
Theoretical Background
Several psychological concepts play crucial roles in L2 learning, with engagement being recognised as a key mediator between motivation and learning (Hiver et al., 2024). Engagement is commonly defined as effortful involvement in learning (Henrie et al., 2015), which translates into concrete actions and participation, ultimately facilitating L2 learning (Mercer & Dörnyei, 2020). It is typically conceptualised in three dimensions: behavioural, cognitive, and emotional engagement (Fredricks & McColskey, 2012). From the perspective of EE, behavioural engagement refers to observable actions during EE activities (Nakamura et al., 2021), such as the time spent streaming English videos. Cognitive engagement involves the mental effort dedicated to these activities (Dao et al., 2021), for instance, processing and understanding subtitles. Emotional engagement captures learners’ emotional responses to EE activities (Dao et al., 2021), such as feelings of enjoyment and interest while watching videos. Frequency of participation in EE activities reflects more than just behavioural involvement: voluntary participation indicates emotional engagement, while the use of English signals cognitive effort. Therefore, similar to L2 learning more broadly, EE engagement can be conceptualised as the combined result of behavioural, cognitive, and emotional engagement.
Because EE encompasses a wide range of activities, it can be situated within several influential theories of second language acquisition (SLA; see Sundqvist & Sylvén, 2016; Toffoli & Sockett, 2010). From a Krashenian perspective, EE provides abundant opportunities for comprehensible input while simultaneously lowering the affective filter, given that such activities are typically pursued for pleasure and outside of the constraints of formal instruction (Krashen, 1982; Toffoli & Sockett, 2010; Uhing et al., 2025). In addition, Swain’s (1995) output hypothesis is also relevant, as many EE activities, such as texting, gaming or informal conversations, require learners to produce both spoken and written language, thereby pushing them to use and expand their linguistic repertoire. Interactive EE activities, particularly those involving collaboration or real-time communication, further resonate with Long’s (1981) interaction hypothesis, which underscores how negotiation of meaning facilitates comprehension and language development. Finally, EE activities that involve social participation can be interpreted through socio-cognitivist and sociocultural lenses, which emphasise co-construction of knowledge through interaction and mediation (Lantolf & Thorne, 2006; Vygotsky, 1978). These theoretical linkages show that EE is not a monolithic concept but one that maps onto several theoretical traditions in SLA, with the relevance of each framework depending on the type of activities in question.
Furthermore, EE engagement aligns with key theories in learner psychology. It is well established that negative affective states, such as anxiety, can impede L2 learning, whereas positive states, such as enjoyment, can facilitate it (Dewaele & MacIntyre, 2014). Unlike many classroom environments that may increase anxiety and fail to capture learners’ interest, EE activities inherently support the principles of positive psychology by providing an intrinsically motivated, enjoyable and typically anxiety-free environment.
EE engagement also aligns well with Papi and Hiver’s (2025) proactive language learning theory, which views L2 learning as a strategic endeavour where learners ‘identify a gap in their language abilities, set goals to bridge that gap, and plan actions to achieve their goals’ (p. 8). The theory proposes four behavioural dimensions: input-seeking behaviour, interaction-seeking behaviour, information-seeking behaviour, and feedback-seeking behaviour. These dimensions are closely connected with EE activities that learners consciously engage in with the intention of improving their L2 skills. For example, a learner who identifies a gap in their writing skills and has an interest in manga may exhibit input- and interaction-seeking behaviour by joining a fanfiction community to write fictional stories about manga characters. This illustrates how intentional EE engagement serves as an example of proactive language learning theory.
Research Evidence on the Benefits of EE Engagement
Several studies have investigated the benefits of EE and related forms of informal L2 learning. For example, Cole and Vanderplank (2016) examined classroom-trained and self-instructed adult learners (aged 18–24) in Brazil and concluded that formal instruction is not necessarily required to achieve high levels of English proficiency. Similar results have been reported in several studies conducted in Belgium, where learners who had not received any formal instruction still demonstrated good English skills. For example, studies show that sixth-graders (age 12) knew over 3,000 English word families (Puimège & Peters, 2019), many scored at an A2 level for listening comprehension (De Wilde et al., 2020; Wouters et al., 2024) and for writing and speaking competence (De Wilde et al., 2020). In the Netherlands, Leona et al. (2021) conducted a quantitative study, presenting a statistically verified model where EE engagement predicted vocabulary knowledge among young learners (aged 10) with no formal L2 education. Similar patterns have been observed for even younger learners. For example, in Norway, Gyllstad et al. (2025) found that total EE time and time spent watching TV correlated strongly with vocabulary scores among learners aged 5 to 6. In Hong Kong, among junior-secondary school students, Tsang and Lam (2025) found that EE engagement correlated significantly with reading and listening proficiency. Research on adult L2 learners of English also supports these findings. For instance, Busby (2020) showed that EE engagement was a stronger predictor of vocabulary knowledge than formal English instruction among Norwegian university students. In a Swedish university context, Neagu et al. (2025) reported positive correlations between EE activities and centrally administered English reading comprehension test scores. Collectively, these studies highlight that English proficiency can develop beyond the classroom, emphasising the significant role of EE engagement in L2 learning.
Although EE research consistently provides empirical evidence of its benefits on L2 learning, the extant research on informal L2 learning to date has predominantly been carried out in three ‘clusters’, namely Sweden and Finland, German and France and Mainland China and Hong Kong (Kusyk et al., 2025). This underscores the need for research in countries where the status of English and educational policies and practices differ from these contexts, such as Türkiye, where EE research remains limited. A first study, however, was conducted by Coşkun and Mutlu (2017), who gathered quantitative data from Turkish L2 English students (aged 15–17) on their frequency of EE engagement. Their findings revealed that students reported only occasional engagement in reading, writing and speaking-related EE activities, while listening-related EE activities occurred rarely. A follow-up study by Ipek and Mutlu (2022) in a university setting confirmed that Turkish students engage infrequently in reading, writing and speaking-related activities. Additionally, Uztosun and Kök (2024), conducted a study investigating the relationship between EE engagement, L2 anxiety, and communication apprehension (CA). The results revealed that EE engagement negatively predicted listening and speaking anxiety, as well as CA. More recently, Uztosun and Sundqvist (2025) examined the relationship between EE engagement and L2 English proficiency at a university level. They found that several EE activities (e.g., EE Listening, EE Spoken Interaction, EE Writing and EE Watching) correlated significantly and positively with L2 proficiency. To the best of our knowledge, these are the only EE studies conducted in Türkiye, a country with thousands of learners of English. Thus, we consider the present study both valuable and timely.
Scale Development in EE Research
A few studies have been conducted to develop reliable and valid scales for measuring EE engagement. Arndt’s (2023) study developed a scale to measure affective, cognitive and linguistic engagement with informal L2 practices. The data, gathered from German secondary school students (aged 15–16) led to a scale, consisting of 8 items, that was designed to assess learners’ engagement immediately after participating in an EE activity. Another contribution comes from Sundqvist and Uztosun (2024), who developed the first EE scale in the Scandinavian context (Denmark, Norway and Sweden), with participants of the same age as in Arndt (2023). Their study resulted in a 32 7-point Likert-type items EE scale based on frequency. A third study is the one by Coşkun and Mutlu (2017), who developed an EE scale based on data from 292 Turkish high school students. Their scale comprised 34 5-point Likert-type items of frequency grouped into four factors corresponding to the four skills reading, writing, speaking and listening. However, this scale was not subjected to confirmatory factor analysis, which is an essential step to confirm the structure of underlying factors. To address this gap, the present study follows rigorous procedures for scale development and validation, including the examination of key reliability and validity indicators, to ensure the robustness of the EE engagement measure.
Study 1: EE Scale Development and Validation
Methodology
Study 1 was conducted in English Language Teaching (ELT) departments at seven universities across Türkiye. These universities were selected using convenience and snowball sampling: we contacted colleagues at institutions willing to participate in the study. Including seven universities allowed us to achieve a balanced number of participants at each stage of data collection. ELT departments offer a 4-year teacher education programme, with some institutions also providing a 1-year preparatory course to improve students’ English proficiency. Upon graduation, students are qualified to teach English at all educational levels, from pre-school to university. These departments were included in this study because, compared to other departments, their students are likely to be more actively engaged in EE and possess a broader EE repertoire than students from other departments – an important consideration for ensuring the validity of the scale under development.
Ethical Issues
We collected data after securing ethical approval from the participating institutions. Participants gave their written consent at the beginning of the questionnaire, where they were informed about the research objectives. They were assured that the collected data would be used solely for research purposes. Additionally, except for a test-retest analysis, all data were collected anonymously. No sensitive questions were asked in the questionnaire.
Participants
In total, 718 students participated in the study (for demographics, see Table 1). The scale was developed in two stages: exploratory factor analysis (EFA) and confirmatory factor analysis (CFA), followed by tests for test-retest reliability and known-groups validity.
Participant Demographics.
ELT Department students whose responses were used both for test-retest reliability and known-groups validity.
Elementary Teaching Department students who participated in known-groups validity.
Data Collection Procedures
The data were collected using paper-based questionnaires. The instructions were similar to those used in our earlier publication (Sundqvist & Uztosun, 2024). Specifically, we asked participants to consider a typical week (excluding weekends and holidays) and rate each item referring to specific EE activities on a 7-point Likert scale, ranging from ‘never’ to ‘always’. The 7-point Likert scale was used to provide a clear mid-point and a sufficient range of response options, allowing participants to indicate varying degrees of frequency in their engagement with each EE activity in the scale.
An EFA was conducted to consolidate variables and identify underlying factors of EE engagement (Tabachnick et al., 2013) and a CFA was conducted to validate the factor structure and latent variables identified during EFA (Kline, 2016). To assess the reliability of the scale over time, we used test-retest reliability, administering the scale to the same participants at two different times with a 4-week interval (Kline, 2016). To verify the construct validity, we tested the scale’s known-groups validity, which evaluates whether a scale can distinguish between groups known to differ on relevant variables (Davidson, 2014). In our case, we compared students from ELT and Elementary Education Departments. ELT students receive extensive English education and will become English teachers, whereas Elementary Education students are not trained to teach English and have limited English education. It is therefore reasonable to assume group differences in terms of EE engagement.
Data Analyses
EFA was conducted using IBM SPSS Statistics (version 29). We assessed sample adequacy with the Kaiser-Meyer-Olkin (KMO) test and examined item correlations using Bartlett’s test of sphericity. A KMO value greater than 0.5 and a significant Bartlett’s test are required to proceed with EFA (Field, 2013). The KMO value of 0.92 and a significant Bartlett’s test result (Chi-square = 9670.246, df = 1275, p < .001) confirmed that the data met these assumptions.
We employed principal axis factoring to identify the underlying factors by estimating communalities (Field, 2013). Since the factors in an EE scale are likely to be correlated, we applied oblique rotation (Tabachnick et al., 2013). Items with a factor loading of at least 0.3 were retained, while cross-loaded items were removed based on a factor-loading difference criterion of 0.1 (Field, 2013).
CFA was conducted using JASP (version 0.14.1). CFA is used to test theoretical models by analysing factors, correlations and error values within a data matrix (Kline, 2016). This method helps develop a condensed scale and confirm the factor structure (Mueller & Hancock, 2008). Several fit indices were employed: the chi-square to degrees of freedom (df) ratio, root mean square error of approximation (RMSEA), standardised root mean square residual (SRMR), Tucker-Lewis index (TLI), and comparative fit index (CFI). A chi-square/degrees of freedom ratio less than 2 indicates an acceptable fit. RMSEA values of less than 0.05 suggest a close fit, while values between 0.05 and 0.08 indicate a reasonable fit (Browne & Cudeck, 1993). SRMR values near 0.08 denote a good model fit (Hu & Bentler, 1995). TLI and CFI cut-off values of more than 0.90 are considered acceptable (Heubeck & Neill, 2000).
For tests for test-retest reliability and known-groups validity, normality tests were conducted. The Kolmogorov-Smirnov test and z-scores of skewness and kurtosis showed that data were not normally distributed, with significant Kolmogorov-Smirnov results (p < .001) and z-scores exceeding the 2.58 cut-off (Mayers, 2013). Therefore, Spearman’s correlation was used to assess test-retest reliability. However, for known-groups validity, where data were normally distributed, independent samples t-tests were used to compare EE engagement between two groups.
Results
The questionnaire items were adapted from our earlier publication (Sundqvist & Uztosun, 2024), which was developed based on feedback from international experts to ensure construct validity. For the current study, these items were reviewed and refined to fit the specific context of EE engagement in Turkish university settings. A total of 71 items were subjected to EFA, resulting in the exclusion of 20 items that did not meet the inclusion criteria outlined above. As a result, 51 items were loaded onto 9 factors, which explained 57.51% of the variance (see Supplemental Appendix A).
CFA revealed that four items did not meet the acceptable factor loading criteria and were subsequently removed, resulting in a final 47-item scale with nine factors. Considering the cut-off values mentioned earlier, the CFA results demonstrated an acceptable model fit: Chi-square (933, N = 310) = 1,537.907, p < .001; chi-square/degrees of freedom = 1.64; CFI = 0.93; TLI = 0.92; RMSEA = 0.064; SRMR = 0.046. These results confirmed that the Turkish EE scale, along with its factor structure, provides a good model fit. The English and Turkish versions of the scale are available in Appendices B and C, respectively.
Internal Reliability
Considering the Cronbach’s alpha cut-off values proposed by George and Mallery (2016), the internal reliability of the scale was excellent (.94). The scores for each factor demonstrated acceptable (>.7), good (>.8), or excellent (>.9) reliability. The results indicate that the scale and its factors have strong internal consistency across different dimensions. The reliability scores and number of items for each factor are provided in Table 2.
Internal Reliability of the Factors.
Test-Retest Reliability
To assess the test-retest reliability of the scale, total EE scores were calculated for each participant based on their responses collected at 2 time points (4-week interval). The analysis revealed a strong positive correlation between the two sets of scores (rs = .82, p < .001, N = 55), indicating high test-retest reliability of the scale.
Known-Groups Validity
An independent samples t-test revealed that the ELT department students (M = 4.16, SD = 0.09) reported engaging in EE activities more frequently than Elementary Education students (M = 2.42, SD = 1.22). This difference was statistically significant (p < .001), confirming that the scale effectively distinguished between groups with distinct levels of EE engagement.
Study 2: Extramural English Engagement and L2 English Proficiency
Methodology
Study 2 was conducted at a university in Istanbul that did not participate in Study 1. Similar ethical procedures were followed, with ethical clearance obtained from the university and participants providing informed consent through the online questionnaire. Participation was entirely voluntary, and all data were collected anonymously.
Participants
All students who took the English proficiency exam were invited to participate in Study 2. A total of 59 students volunteered, comprising 40 females, 16 males and 3 individuals identifying as another gender. The mean age was 19.74 years (SD = 4.83), with a median of 19, a mode of 18 and an age range from 17 to 46 years. All participants were newly enrolled students in either ELT or English literature programmes and had taken an English proficiency exam to determine whether they could be exempted from a 1-year preparatory English course.
Data Collection Tools
An online questionnaire was administered, comprising three sections: (a) the Turkish EE scale developed in Study 1, (b) results from an English proficiency exam and (c) demographic information.
We used the scores of an exam that was developed and administered at a university. According to the institution’s guidelines, the exam corresponded to the B2 level of the Common European Framework of Reference for Languages (Council of Europe, 2020). Language proficiency was assessed in three sessions: (i) reading/listening, (ii) writing, and (iii) speaking. Each skill was weighted equally (25 points), and the maximum possible score was 100. The researchers were not involved in any part of the exam process; it was entirely managed by the university’s exam office and course instructors.
In the first session, students engaged with two listening tracks followed by reading texts and responding to multiple-choice questions. The second session required students to write an essay based on provided prompts. The final session evaluated speaking skills in a three-part format test, where students were assessed by an interlocutor and an assessor. Each session was graded by two raters, and any discrepancies in scores were collaboratively resolved.
In the questionnaire, participants were instructed to report their scores. While overall exam scores served as indicators of L2 English proficiency, the scores for each section were used as indicators of proficiency in specific language skills (i.e., reading/listening, writing and speaking).
Data Analysis
Data analyses were conducted using IBM SPSS Statistics (version 29). To assess the normality of the data, we employed several criteria, including the Kolmogorov-Smirnov test results and the calculation of z-scores for skewness and kurtosis (Field, 2013). The results indicated that the exam score data were not normally distributed, as evidenced by significant Kolmogorov-Smirnov p-values (p < .001) and z-scores for skewness and kurtosis exceeding the cut-off value of 2.58 (Mayers, 2013). To establish normality, we excluded 14 outliers from the reading and listening tests, 3 outliers from the speaking exam, 4 outliers from the writing exam, and 5 outliers from the overall exam scores. As a result, the number of participants was reduced to 45 for the reading and listening exam, 56 for the speaking exam, 55 for the writing exam, and 54 for the overall exam. Each data set was processed separately. The removal of these outliers ensured that the data met the criteria for normality and made it possible to perform linear regression analysis.
Results
RQ2: The Frequency of EE Activities
Descriptive analyses revealed that the average engagement with EE activities was 3.65 (SD = 1.21) out of 7. The three factors most frequently engaged with were EE Music (M = 5.85, SD = 1.27), EE Googling and Viewing (M = 5.36, SD = 1.55), and EE Internalised (M = 4.63, SD = 1.75). Activities involving EE Social Interaction (M = 3.60, SD = 1.69), EE Reading and Listening (M = 3.23, SD = 1.46), and EE Gaming (M = 3.18, SD = 1.77) showed moderate engagement, while EE Social Reading and Listening (M = 2.91, SD = 1.72), EE Digital Creativity (M = 2.21, SD = 1.65) and EE Writing (M = 2.06, SD = 1.35) were the least popular. These results indicate that learners prefer more receptive and entertainment-based EE activities, whereas productive or creative activities are less common. The results are presented in Table 3.
The Mean Scores of Scale Factors.
RQ3: Relationship Between L2 English Proficiency and Factors in the Turkish EE Scale
Pearson correlations were computed to explore the relationships between L2 English proficiency and EE engagement. The results were interpreted using Pearson’s r effect size guidelines presented by Mayers (2013, p. 105), where r values between .1 and .3 indicate a small correlation, between .3 and .5 indicate a medium correlation, and above .5 indicate a large correlation. The findings are summarised in Table 4.
The Relationships Among L2 English Proficiency and the Factors of the Turkish EE Scale.
Note. 1 = EE Digital Creativity, 2 = EE Gaming, 3 = EE Music, 4 = EE Social Interaction, 5 = EE Reading and Listening, 6 = EE Internalised, 7 = EE Writing, 8 = EE Social Reading and Listening, 9 = EE Googling and Viewing.
Correlation is significant at the .05 level (two-tailed).
Correlation is significant at the .01 level (two-tailed).
According to the findings, except for EE Digital Creativity and EE Gaming, all types of EE activities correlated significantly with overall L2 proficiency and speaking proficiency. Large correlations were found between EE Internalised and overall L2 proficiency (r = .54); and between EE Internalised and speaking proficiency (r = .57). Additionally, EE Music and EE Internalised were moderately correlated with reading, listening and writing proficiency. EE Reading and Listening also showed a moderate correlation with reading and listening proficiency, while EE Writing exhibited a small correlation with writing proficiency.
RQ4: The Predictive Ability of EE Engagement for L2 English Proficiency
To investigate whether EE engagement predicts L2 English proficiency, four linear regression analyses were conducted. The models included EE engagement (mean scores of the Turkish EE scale) as the independent variable and each exam score as the dependent variable.
For the model with L2 English proficiency as the dependent variable, the regression was statistically significant (F (1, 52) = 9.10; p = .004), with EE engagement as a significant predictor (β = .386; 95% CI [0.95, 4.73]). The model explained 15% of the variance (R2 = .15), indicating a moderate effect. Similarly, for the model with speaking proficiency as the dependent variable, the regression was statistically significant (F (1, 51) = 12.63; p = .00) with EE engagement as a significant predictor (β = .446; 95% CI [0.378, 1.35]). The model explained 19% of the variance (R2 = .19), indicating a moderate effect. These findings suggest that EE engagement may be a significant predictor of both overall L2 English proficiency and speaking proficiency.
In contrast, for the model with reading and listening proficiency as the dependent variable, the model was not statistically significant (F (1, 40) = 4.00; p = .051) (but close), with EE engagement as a predictor [(β = .303; 95% CI [−0.003, 1.78]). The model explained 9% of the variance (R2 = .09), indicating a small-to-moderate effect. Likewise, the regression predicting writing proficiency was not statistically significant (F (1, 50) = 0.155; p = .696), with EE engagement as a predictor [(β = .056; 95% CI [−0.472, 0.702]). The model explained less than 0.1% of the variance (R2 = .001), indicating a none effect. These findings are presented in Tables 5–8 below.
The Predictive Ability of EE Engagement to Overall L2 Proficiency.
The Predictive Ability of EE Engagement to Reading and Listening Proficiency.
The Predictive Ability of EE Engagement to Speaking Proficiency.
The Predictive Ability of EE Engagement to Writing Proficiency.
Discussion
Factor Structure of the Turkish EE Scale
This research demonstrated that EE engagement can be reliably and validly measured using scales that assess the frequency of various EE activities included in each item. The study revealed that, in the Turkish university context, EE engagement consists of nine factors encompassing 47 items: (a) EE Digital Creativity, (b) EE Gaming, (c) EE Music, (d) EE Social Interaction, (e) EE Reading and Listening, (f) EE Internalised, (g) EE Writing, (h) EE Social Reading and Listening, and (i) EE Googling and Viewing.
Factor 1 is labelled EE Digital Creativity because its items pertain to creating digital materials. The distinction between the items lies in whether the materials are shared with others or not. Activities may encompass creating videos, podcasts and music. As noted by Sundqvist and Uztosun (2024), this factor aligns with the ‘action level’ of human behaviour proposed by Lantolf and Thorne (2006). Creating and sharing digital materials may be driven by goal-directed behaviour, such as establishing a community.
Factor 2 relates to EE Gaming and consists of items that refer to specific games, which can be further classified into three categories: (a) games that focus on specific language skills required (e.g., ‘I play games that require speaking’), (b) games that focus on specific genre (e.g., ‘I play adventure games’), and c) games that involve/do not involve other people (e.g., ‘I play games online with others’). This factor also involves items about watching gaming videos and playing non-digital games, such as board games. Gaming is one of the most extensively studied areas of EE research. Several studies have revealed the benefits of gaming for language learning. For instance, Sundqvist (2019) found positive correlations between time spent on gameplay and both receptive and productive vocabulary knowledge, and significant differences between gamers and non-gamers in productive vocabulary levels test. In a longitudinal study on the speaking performances of low-intermediate and high-intermediate gamers, Jabbari and Peterson (2023) reported improvements in complexity, accuracy and fluency. Calafato and Clausen (2024) suggested that the benefits of video gaming for vocabulary knowledge may stem from gamers using vocabulary learning strategies, such as making inferences, using language learning references and note-taking while playing. According to Ulfat (2025), L2 English gains from gaming may occur because of escapism: learners who report being immersed in the fictional worlds created by games report how their proficiency has developed that way. Given that EE Gaming comprises 13 items, this suggests the presence of potential subdimensions within the gaming factor, which would be fruitful avenue for future research focusing specifically on EE Gaming.
Factor 3 is labelled EE Music as it encompasses items related to listening to music and watching video clips. Research has consistently shown that music is the most popular EE activity (e.g., Schwarz, 2020; Sundqvist, 2009; Sundqvist & Uztosun, 2024). However, as Wouters et al. (2024) reported, there are conflicting findings regarding the benefits of music for language learning. For example, Lindgren and Muñoz (2013) found a positive relationship between listening to songs and reading and listening comprehension, while Peters (2018) reported a negative relationship with vocabulary knowledge. These mixed results suggest that, despite its popularity, the specific benefits of EE Music for language learning require further investigation.
Factor 4, labelled EE Social Interaction, comprises items related to establishing written (e.g., texting, writing) and spoken (e.g., talking to others) communication in English. These items reflect individuals’ efforts to seek opportunities for communicative use of English, aligning both with Long’s (1981) interaction hypotheses, which emphasises interaction as central to language learning, and Papi and Hiver’s (2025) feedback-seeking behaviour.
Factors 5 and 8 pertain to reading and listening skills. Factor 5 (EE Reading and Listening) involves individual engagement in reading and listening activities, while Factor 8 (EE Social Reading and Listening) refers to performing these activities with others. Research has reported positive relationships between EE reading and vocabulary knowledge (e.g., Busby, 2020; Peters, 2018). Beyond individual engagement, Factor 8 also relates to the creation of affinity spaces, as described by Gee (2004). Affinity spaces are social environments where individuals interact, and form communities based on shared interests. These spaces foster language learning through social participation and collaboration, providing opportunities for individuals to engage in meaningful communication, exchange ideas and build relationships that enhance their language development.
Factor 6, EE Internalised, includes personal, internal activities such as thinking in English. This construct reflects learners’ use of inner speech or verbalised thought in the L2, which aligns with the principles of sociocultural theory (Lantolf & Thorne, 2006). According to this theory, inner speech plays a key role in cognitive development, allowing individuals to internalise language and use it for self-regulation and problem-solving.
Factor 7, labelled EE Writing, consists of items related to writing in various genres in English, such as blogs, diaries, fanfiction and poems. It also includes collaborative writing in English, which is often associated with writing fanfiction, defined by Duffett (2013) as ‘fictional writing created by the fans inspired by the objects of their interests’ (p. 170). Fanfiction can encompass a wide range of genres, including romance, mystery and various text types such as short stories and graphic novels (Sauro & Sundmark, 2019). This factor highlights the creative and collaborative aspects of writing in English, which can foster L2 development in diverse contexts.
Factor 9, called EE Googling and Viewing, includes items related to following specific YouTubers or vloggers, watching videos and googling for information. These activities align with the concept of extensive viewing, proposed by Webb (2015), which provides learners with large amounts of authentic spoken L2 input. Research has shown that watching documentary TV programmes can facilitate incidental vocabulary learning (Koolstra & Beentjes, 1999; Teng, 2022), particularly in terms of meaning recall and recognition (Peters & Webb, 2018). These findings suggest that EE viewing can contribute to L2 development by increasing learners’ exposure to meaningful, contextualised, interesting and everyday L2 input.
The Frequency of EE Engagement and Its Relationship With L2 English Proficiency
The descriptive analysis of EE engagement frequency revealed that the three most popular activities were EE Music, EE Googling and Viewing and EE Internalised, while EE Writing and EE Digital Creativity were the least popular. These findings align with those of Bardak (2023), who found that listening to English music, watching English movies and viewing English videos and clips were the top EE activities among high school students in Türkiye, while writing activities ranked among the least popular. Similarly, the findings are consistent with Sundqvist and Uztosun (2024), who reported similar trends in their Asian sample, where EE Music, EE Viewing, and EE Internalised were the most favoured activities, with EE Digital Creativity being the least popular in both the Asian and Scandinavian samples. These parallel findings show that Turkish university students exhibit similar tendencies to students in other context regarding their preferred EE activities.
As for the relationship between EE engagement and L2 English proficiency, this study suggests that EE engagement may be a significant predictor of L2 English proficiency and speaking proficiency. However, no statistically significant predictive relationships were observed between EE engagement, reading/listening, and writing proficiency. These findings appear to align with results from some other contexts, such as South Korea (Lee & Dressman, 2018) and Sweden (Sundqvist, 2009). Conversely, the findings diverge somewhat from research conducted in other contexts, such as Belgium (e.g., Wouters et al., 2024), Sweden (e.g., Sylvén & Sundqvist, 2012), and Germany (e.g., Meyer et al., 2024), where EE engagement has been linked to reading and listening proficiency. Nevertheless, these findings of the present research should be interpreted with caution. After the removal of outliers, the sample size for most models was just above 50 participants, which is generally considered the minimum recommended for regression analysis (Tabachnick et al., 2013). For the reading and listening proficiency model, the final sample dropped below 50, reducing the statistical power and potentially limiting the detection of small effects. As a result, some relationships, such as EE engagement predicting reading and listening proficiency, did not reach statistical significance, but it should be noted that it was close (p = .051) and the effect size suggests small-to-moderate effects. That said, the sample size is a limitation that calls for further research with larger samples of Turkish learners to more robustly examine how EE engagement relates to specific language skills in that context.
Interestingly, the only strong correlations were between EE Internalised and both L2 English proficiency and speaking proficiency. This suggests that personal, internal EE activities – such as thinking, daydreaming, and talking to oneself in English – are linked to L2 English proficiency and speaking proficiency. These activities conform inner speech, defined as ‘subjective experience of language in the absence of overt and audible articulation’ (Alderson-Day & Fernyhough, 2015, p. 931). Inner speech, a key concept in Vygotsky’s sociocultural theory, is thought to support cognitive functions such as planning what to say and selecting appropriate word forms and sentence structures (de Guerrero, 2018). Previous research has also shown that L2 inner speech can help learners recognise their thinking and perceive the word through new linguistic lenses and perspectives (see de Guerrero, 2018, for a review). The present study demonstrates that inner speech can be considered as a form of EE engagement and is associated with enhanced L2 English and speaking proficiency.
Furthermore, the study revealed relationships between skill-specific EE activities and reading, listening, and writing proficiency. Specifically, EE Reading and Listening correlated with reading and listening proficiency, while EE Writing was linked to writing proficiency. These findings align with previous research indicating that certain EE activities positively impact lexical diversity in L2 English writing (Kaatari et al., 2023) and lead to longer texts with richer vocabulary (Olsson, 2012). This suggests that skill-specific EE activities may be directly tied to proficiency in the corresponding skill, emphasising the importance of considering the varied impact of different types of EE activities on L2 development.
Conclusions, Limitations and Further Research
This article presents two studies: the first developed and validated a Turkish EE scale, and the second examined the frequency of EE engagement and its impact on L2 English proficiency among Turkish university students. The scale offers a reliable instrument for data collection in Türkiye and, hence, can support more robust statistical analyses that may shed light on the impact of EE engagement on proficiency as well as other key variables in L2 learning.
Study 2 suggested that EE engagement may be a significant predictor of L2 English proficiency, particularly speaking. EE Internalised and EE Music were the only factors that correlated with all four proficiency measures and showed a notably strong correlation with both L2 English proficiency and speaking proficiency. However, the findings should be interpreted with caution as the limited sample size reduced the statistical power of the regression analysis, which in turn constrains the generalisability of the results. As such, we regard the findings as preliminary insights from an underexplored context, highlighting the need for further research to confirm and extend the findings.
Despite these limitations, the findings carry important implications for both English teachers and learners. Given that EE engagement may predict English proficiency – particularly speaking – learners could benefit from actively engaging in EE activities as a complement to formal instruction. Moreover, some skill-specific EE activities appear to support development in corresponding language skills, although this was not the case for all skills. Therefore, it is important to foster students’ awareness of the potential role of EE and encourage them to broaden their EE repertoires.
The findings also emphasise the importance of conducting research in underexplored contexts to identify unique characteristics and contextual factors that influence the impact of EE engagement on proficiency. Continued research is needed to test the reliability and validity of the EE scale within the Turkish context and across different age groups and educational levels. Future studies should also aim to include larger samples to enhance the statistical power and enable the use of more advanced statistics, such as structural equation modelling and path analysis. This will allow for gaining deeper insights into the role of EE engagement in L2 learning.
Supplemental Material
sj-docx-1-sgo-10.1177_21582440251396680 – Supplemental material for Extramural English in Türkiye: Scale Development, Learner Engagement and L2 English Proficiency
Supplemental material, sj-docx-1-sgo-10.1177_21582440251396680 for Extramural English in Türkiye: Scale Development, Learner Engagement and L2 English Proficiency by Mehmet Sercan Uztosun and Pia Sundqvist in SAGE Open
Footnotes
Acknowledgements
We are grateful to the students who agreed to participate in this research, and to the administrators and colleagues who assisted us in collecting the data.
Ethical Considerations
This research study was conducted with the Research Ethics Committee approvals of Çanakkale Onsekiz Mart University dated 25.11.2021 and numbered 021-YÖNP-0830; and Istanbul Medeniyet University, dated 11.09.2023 and numbered 2023/06-05.
Consent for Publication
Informed consent was obtained from all participants involved in the study.
Author Contributions
M. Sercan Uztosun: Conceptualisation; Data curation; Investigation; Methodology; Resources; Writing – original draft; Writing – review & editing. Pia Sundqvist: Conceptualisation; Methodology; Resources; Writing – original draft: Writing – review & editing.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The data of this study are available from the corresponding author upon reasonable request.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
