The Role of the L2 Learning Environment in Shaping Individual Learner Factors and Language Achievement: A Comparison Study of Danish and Spanish Learners of L2 English

Based on the assumption that individual learner factors are context-dependent, this exploratory study examined whether differences in two learning environments (Denmark and Spain) have a differential influence on a set of learner factors that have together been previously identified as important in second language (L2) research, specifically, foreign language classroom anxiety (FLCA), English competence beliefs (ECB), motivation (ideal L2 self) and attitudes towards English language learning. The study also examined whether the L2 learning environment and learner factors had a differential influence on the proficiency of two groups of adolescent learners, as measured by a more instruction-related test (grammaticality judgment test) and a more out-of-school-exposure-related test (listening comprehension test). The results showed that learning environment had a differential influence on FLCA and that gender had an impact on FLCA and ECB. Furthermore, the study showed a differential impact on the L2 English proficiency of the two learner groups of four different factors (i.e., FLCA, ECB, learners’ ideal L2 self, learners’ attitudes towards the presence of English in the academic context, and ECB interacting with FLCA). These findings point to a crucial role of the learning environment in L2 learning and provide empirical support for a context-dependent view of the expression of learners’ individual characteristics in relation to L2 learning.


INTRODUCTION
Research on second language (L2) learners has shown a great deal of variability in L2 achievement.Among the factors that can explain such variability are the following learner factors, namely, foreign language classroom anxiety (FLCA), learners' competence beliefs, and their motivation and attitudes towards the L2, as well as the learning environment.These are factors that in concert have consistently shown to have an influence on L2 achievement (Cadierno et al., 2022;Fenyvesi et al., 2020;Muñoz & Cadierno, 2021) and may impact outcome differentially depending on the learning context.
It is commonly recognized nowadays that learners' affective characteristics, such as anxiety, motivation, and attitudes play an important role in language learning processes (Dörnyei & Ryan, 2015).However, no single factor is expected to explain the observed variability in achievement because "each variable is but a small part of a complex picture" (N.Ellis & Larsen-Freeman, 2006, p. 559; see also Dewaele, 2005).Moreover, several scholars in the field of second language acquisition (SLA) have highlighted the situated nature of individual differences (e.g., Dörnyei, 2005), rejecting the notion that the various traits are context-independent and absolute.In the new dynamic conceptualizations, individual factors "enter into some interaction with the situational parameters rather than cutting across tasks and environments" (Dörnyei, 2009a, p. 218).
Certainly, most SLA researchers would now agree that the role of learner characteristics can best be evaluated with regard to their interaction with specific environmental and temporal factors or conditions (e.g., N. Ellis & Larsen-Freeman, 2006;Mihaljević Djigunović & Nikolov, 2019).However, the inclusion of context in studies focusing on individual differences is not yet common.As a recent metaanalysis on the L2 motivational system (Al-Hoorie, 2018) indicated, there has been a lack of sufficient attention to important learner characteristics, such as the effect of age, gender, or context.
Our study is based on the assumption that learner factors are situated constructs and that their expression may depend on their specific learning environment.It aims to contribute to the literature on individual differences in general, and on affective factors in particular, by focusing on adolescent learners, an under-researched age group, and, particularly, on their learning context.The study has two aims.The first is to explore the role of the learning environment on a set of learner factors (FLCA, competence beliefs, motivation, and attitudes towards English) by comparing two groups of adolescent learners of English in two settings that critically differ with respect to their exposure to English outside the classroom.While Denmark can be characterized as an English-rich environment with large presence of English in the media, Spain can be considered an English-poorer environment given the lower presence of English in society.The second aim is to gauge the role of learning environment and individual factors in the L2 achievement of the two learner groups.

BACKGROUND Foreign Language Classroom Anxiety
FLCA is seen as a unique form of anxiety, that is, a feeling of tension and apprehension that learners experience when learning and/or using a foreign language (FL) (Gregersen & MacIntyre, 2014;Horwitz et al., 1986).FLCA has been described as the strongest predictor of success or failure for L2 students (MacIntyre, 1999), with studies generally showing a negative impact of FLCA on L2 achievement, both in relation to students' final course grades and specific language measures, such as vocabulary, grammar, reading, writing and spelling (see Horwitz, 2010, for a review and Botes et al., 2020, for a recent meta-analysis).
However, FLCA has been mostly associated with oral classroom activities, thus indicating that listening and speaking in the L2 are specific anxiety-provoking activities for L2 learners (e.g., Aida, 1994;Horwitz et al., 1986).Speaking in class is probably the most cited cause of anxiety in the classroom (Mak & White, 1997) because it may threaten learners' self-concept and world-concepts, which are not commonly challenged when communicating in their native language (Horwitz et al., 1986).Research has also shown that FLCA can affect the learners' ability to decode and understand messages in listening activities (e.g., Elkhafaifi, 2005;Kim, 2000).Highly anxious learners have difficulties in discriminating sounds while engaged in listening activities (Horwitz, 1986).The nature of the speech (e.g., fast speech, unfamiliar accents or topics) and the level of vocabulary can also be a source of anxiety for L2 learners (e.g., Kim, 2000).In a recent meta-analysis, Botes et al. (2020) found a significant negative relationship between FLCA and listening achievement.
Moreover, previous research has shown that the learners' immediate learning context and the broader social setting play an important role in learners' anxiety.In relation to the former, a study by Kim (2000) with Korean learners of L2 English has shown that students' level of anxiety was higher in a conversation course than in a reading course as different classroom contexts bring about different instructional methods and procedures that lead to specific types of learning experiences for the students.Additionally, classroom social factors (e.g., the teachers' dress code, age and tone of voice) have also been found to affect learners' FLCA (e.g., Effiong, 2016;Yan & Horwitz, 2008).
In relation to the broader social setting, research has shown that the students' cultural background can affect their level of FLCA.For example, Woodrow (2006) found that European and Vietnamese students of English in Australia tended to be less anxious than Japanese, Korean and Chinese students probably due to the Confucian values of "face" and value being placed on "silence" (Liu, 2002).Additionally, and more directly related to our present study, research carried out in Sweden (Sundqvist, 2009) has shown that even within a specific geographic location, adolescents who spent more time in extramural English activities reported feeling less anxious about speaking English than those who spent less time on such activities.

English Competence Beliefs
Foreign language competence beliefs, understood as learners' evaluations of their own FL competence, is an individual factor that has recently attracted attention in SLA research (Mihaljević Djigunović, 2015).Learners' evaluations of their own FL competence can be considered part of, at least, three constructs, namely self-esteem, selfefficacy, and self-concept.Although these constructs share a common emphasis on an individual's beliefs about their abilities as a person (Valentine & DuBois, 2005), they vary in their degree of specificity and the relative importance of their cognitive and evaluative aspects (Mercer, 2011). 1 In this paper, we use the term English competence beliefs (ECB; see also Fenyvesi et al., 2020) to refer to students' evaluation of their English competence in comparison to other school subjects and their classroom peers.
Crucially, children's achievement-related self-concept has been shown to be closely related to FLCA (Heinzmann, 2013), a link made by Bandura (1986), who views anxiety as both a source and effect of self-efficacy beliefs.For example, MacIntyre et al. (2002) found that highly anxious university students of L2 French tended to underestimate their competence relative to their actual competence, whereas less anxious students exhibited the opposite pattern, that is, they tended to overestimate their competence.In a recent study with first language (L1) Danish learners of L2 English, Cadierno et al. (2022) observed a different pattern, namely, that high ECB scores accurately predicted high receptive vocabulary and grammar scores but only for children with low FLCA.For children with high FLCA, even children with strong beliefs in their own English competence obtained lower scores in both language dimensions.These findings support recent calls in the literature on the need to think of learner factors as dynamic characteristics that interact in multiple ways (Dewaele & Pavelescu, 2021;MacIntyre, 2017).
Finally, links have also been suggested between the learners' L2 learning context and learners' competence beliefs.For example, Clément et al. (1994) suggested that learners' self-esteem would result from frequent and qualitatively pleasant contacts with members of the L2 ethnic group both in naturalistic contexts and through media contact and travelling to the target language country.The interaction between the individual's self-efficacy beliefs and the environment was also stressed in Bandura's (1986) social cognitive theory, which views individuals as active agents who both influence and are influenced by their environment.More recently, Onorato and Turner (2004) claimed that the self-concept would best be viewed as a situated construct which "is conceived as a contextdependent cognitive representation" (p.260).Whereas some studies have found that demographic factors, such as the students' cultural/ethnic background can impact individuals' competence beliefs (e.g., Twenge & Crocker, 2002), other studies have found that the characteristics of different learning environments can have an impact on children's self-concept.For example, Oettingen et al. (1994) found lower levels of self-efficacy in primary school learners attending schools in East Berlin than in West Berlin, probably due to their emphasis on public performance feedback and unidimensional teaching practices which emphasize group-rather than individual-based teaching practices, thus favoring social comparison among classroom peers.

Learner Motivation
Motivation and attitudes are among the individual factors that have attracted most attention in the field of SLA.After the pioneering work by Gardner and Lambert (1959), research into the role of motivation in L2 learning has been heavily influenced by Dörnyei's (2005Dörnyei's ( , 2009a) L2 motivational self system (L2MSS).One of the components of his theoretical framework is the ideal L2 self, which refers to the learner's imagined ideal future self as an L2 speaker.The ideal L2 self is seen as an important component of motivation as learners are likely to be more motivated to learn an L2 if they have vivid, future images of themselves as future speakers of this language.
L2 research conducted on the basis of Dörnyei's (2005Dörnyei's ( , 2009a) ) framework has confirmed the important role of learners' ideal L2 self on their intended learning efforts (e.g., Csizér & Kormos, 2009;Taguchi et al., 2009) whereas its role in learners' actual L2 achievement has produced mixed results, a finding that has been corroborated in a recent meta-analysis by Al-Hoorie (2018).This finding provides evidence that self-reported motivation does not automatically translate into L2 achievement, a view shared by Dörnyei (2001) who argues that motivation is an antecedent of behavior rather than of achievement, that is, motivation can explain why people behave in the way they do rather than how successful their behavior will be.
The discrepancy in the above-mentioned results may be due to methodological issues, that is, how the L2 selfstatements are formulated (Al-Hoorie, 2018;Csizér, 2019).However, a further source of variability may be the influence of specific L2 learning contexts on motivation.In fact, Dörnyei (2009b) argued that individual differences, such as motivation cannot be properly examined unless the characteristics of the learning environment are considered, a claim that has been supported in cross-cultural research on L2 motivation.For example, Taguchi et al. (2009) found that for Chinese and Iranian learners of English, the ideal L2 self tended to develop on the basis of both positive attitudes towards the L2 culture and community and of instrumentality-promotion reasons, that is, usefulness of English in finding a job (i.e., on the basis of both integrative and instrumental motivation).Japanese learners' idealized English self, however, was not strongly linked with the latter.
Furthermore, in the original conceptualization of the L2MSS model, Dörnyei (2009a) outlined several conditions that can enhance or hinder the motivational impact of the ideal L2 self.Three of these conditionsthe availability of an elaborate and vivid future self-image by the learners, its perceived plausibility (i.e., how likely it is for an individual to reduce the discrepancy between one's actual self-state and the ideal self-state; cf.self-discrepancy theory by Higgins, 1987), and the necessary activation/primingseem to be especially context-dependent.It could be argued that learning contexts that facilitate contact with the FL in naturalistic settings (i.e., outside the classroom) will be more likely to activate/prime the development of more elaborate, vivid, and plausible images of oneself as future speaker of the L2.

Attitudes Towards the L2
Language attitudes have been defined in several ways.For example, Wenden (1991) viewed attitudes as comprising three components: (1) a cognitive component that involves perceptions and beliefs about objects and situations related to the attitude; (2) an evaluative component, according to which the objects or situations related to the attitude can provoke like or dislike; and (3) a behavioral component, which may prompt learners to develop specific learning behaviors.Similarly, Montana and Kasprizyk (2008) define attitudes as consisting of persons' beliefs, behaviors, and evaluations of certain outcomes.In our study, we understand attitudes as being composed by these three components and specifically focus on the cognitive dimension exploring students' perceptions and beliefs about the presence of English in school.
L2 research has examined students' attitudes towards different aspects of the English language and the English language learning process.One of the areas that is especially relevant for the present study is learners' attitudes towards the use of English in educational contexts.For example, in a study examining the attitudes of Chinese learners towards English before and after their tertiary studies in Hong Kong, Yang and Lau (2003) found that the students believed that it was important for both secondary school and university teachers to teach in English regardless of the subjects, as this would enhance their knowledge of English.In contrast, in a study examining the attitudes of Turkish 8 th grade primary school students towards the English language and its use in Turkey, Karahan (2007) found that students did not support the view of English as a language of instruction in schools, but rather favored the possibility of taking English as an elective course.In addition, students exhibited negative attitudes towards having course books written in English and towards the use of English when speaking to each other.According to this author, this finding may be due to negative attitudes towards the English language and/or the fact that students may not feel the need for using English in their daily lives yet, and so they prefer delaying English until adulthood when they may need English in professional contexts.Finally, Busse (2017) surveyed secondary school students' attitudes towards learning English and other European languages in four European countries (Bulgaria, Germany, the Netherlands and Spain).She showed that even though all the students were highly aware of the global status of English, contextual factors at the school and socialrelational factors (specifically, the relationship to the teacher) played an important role in shaping students' attitudes towards English and other FLs.

The Present Study
As described above, previous research shows that the characteristics of individuals' L2 learning environments can have an impact on their levels of classroom-related anxiety, competence beliefs, motivation and attitudes towards L2 learning.In the present study, we examine whether the above-mentioned learner factors are influenced by the distinctive characteristics of two secondary school learning contexts, one in Denmark and one in Spain.The two learning environments differ in relation to the social penetration of English.Denmark, like other Nordic countries, can be characterized as an English-rich environment with a large presence of English in the media and in the linguistic landscape.This presence may facilitate both implicit learning, that is, learning without awareness, and incidental learning, that is, learning without intention but with or without awareness (Rieder, 2003).For example, English television programs and films are typically seen in the original language with L1 subtitles.In fact, the status of English may be changing from a foreign to a second language in countries like the Netherlands or the Nordic countries (de Bot, 2014).In contrast, Spain can be characterized as an English-poorer environment given the lower presence of English in society, for example, with English audiovisuals being dubbed into the country's home language (see Muñoz et al., 2018;Muñoz & Cadierno, 2021).
Previous research comparing the acquisition of English by children and adolescents in these two contexts has supported the difference in status of English in the two countries and its effect on L2 learning.Muñoz et al. (2018) compared the receptive English grammar skills of two groups of 7 and 9-year-old Danish children at the beginning of L2 English instruction and two groups of Spanish children of the same ages after several years of instruction.They found that the greater out-of-school contact with English as well as cognate linguistic distance explained why Danish children's receptive knowledge of English prior to school instruction was largely similar to that of Spanish children after several years of instruction.Similarly, in a study with a sub-sample of adolescents included in the present study, Muñoz and Cadierno (2021) found that the Danish group, in an English-richer environment and with a shorter linguistic distance from English, attained significantly higher levels of L2 proficiency on a listening comprehension test and a grammaticality judgment test, but not on a metalinguistic knowledge test, a measure of explicit and declarative knowledge, in which the Spanish group had an advantage.The authors interpreted this result as indicating that the Danish learners may have resorted in their test performance to both implicit and explicit knowledge acquired both outside and inside the classroom.In contrast, for the Spanish learners the most important source of knowledge may have been classroom instruction.In addition, an out-of-school exposure questionnaire showed that the Danish group engaged longer in out-of-school activities than the Spanish group, and differences were significant in relation to watching audiovisual material without subtitles, speaking and writing in English.
In sum, the results of these two studies suggest that differences in the particular contexts in which the L2 learning takes place have an influence on learners' acquisition of L2 English.An aspect that was not examined in our two previous studies, however, is the role that the L2 learning environment may have on a set of learner factors, which is addressed in the present study.

AIMS AND RESEARCH QUESTIONS
This study extends our previous research by exploring (a) whether differences in two learning environments (Denmark and Spain) have a differential influence on a set of learner factors, specifically, FLCA, ECB, motivation and attitudes to English language learning by Danish and Spanish learners; and (b) whether the L2 learning environment and learner factors have a differential influence on the L2 proficiency of the two groups of learners, as measured by a test that is highly related to classroom instruction and which may better reflect explicit learning (grammaticality judgment test; see N. Ellis, 2005) and a test that may gauge out-of-the classroom learning as well (i.e., frequent exposure to English-spoken television) and which may also capture implicit and/or incidental learning (listening test).
Specifically, the study addressed the following research questions (RQs): • RQ1: To what extent does the L2 learning environment have an influence on Danish and Spanish learners' foreign language learning anxiety (FLCA), English competence beliefs (ECB), motivation, and attitudes towards English?
• RQ2: To what extent do the L2 learning environment and the above-mentioned learner factors (i.e., FLCA, ECB, motivation and attitudes towards English) impact learners' English language proficiency, as measured by a grammaticality judgment test (GJT) and a listening comprehension test (LisT)?

METHOD Participants
The participants were two groups of 14-15-year-olds, one from a school in Denmark and one from a school in Spain.
The Danish group consisted of 56 learners (31 female and 25 male), who were finishing the 8 th grade whereas the Spanish group consisted of 112 learners at the end of the 9 th grade (54 female and 56 male participants; two participants did not report gender) and who were Catalan-Spanish bilinguals.Both schools were semi-private and were similar in terms of socio-economic background.Apart from being selected on the basis of their similarity, the schools were convenience samples.School principals agreed to participate in the study and informed consent was obtained from the parents.The EU's General Data Protection Regulation rules were followed in terms of data collection, processing and storage.The Danish students had had six years of English instruction (360 hours) and their average starting age was 9-10 years.In both schools, teachers had the required national degree in education and possessed a high level of English proficiency.They typically used English as the vehicle of instruction and communication in class.Both schools used a communicative approach to teaching English, including a focus on form, but with notable differences.In the Danish school, a topic-based syllabus was used.Topics were presented through a variety of texts, typically discussed in class and summarized in writing; grammar was mainly practiced through web exercises at home and later discussed in class.In the Spanish school, a structural syllabus was used.Linguistic structures were practiced in writing through exercises, essay writing, and oral communicative activities.

Questionnaires
Participants completed a background questionnaire including questions about their history of English learning (e.g., age of onset, extracurricular tuition, number of weekly classes) and a questionnaire that assessed the learner factors included in the study developed on the basis of previous studies (e.g., Cadierno et al., 2022;Horwitz et al., 1986;Karahan, 2007).Both questionnaires were presented to the students in their L1, Danish or Catalan.Previous results had indicated that the Danish students had more frequent contact with English outside the classroom than the Spanish students (see Muñoz & Cadierno, 2021). 2 The present questionnaire consisted of 20 items divided as follows: FLCA: 8 items, ECB: 5 items, Ideal L2 self: 4 items, Attitudes towards the presence of English in school (AES): 3 items.For FLCA, ideal L2 self, and AES, participants had to mark the extent to which they agreed with the statements on a 7-point Likert scale.Similarly, for ECB, participants had to evaluate the statements presented to them on a 7point Likert scale (see Appendix A).The questionnaire was pilot tested with 35 students of the same age to make sure that the statements and questions were correctly understood.Native speakers of Spanish and Danish were consulted in order to ensure that the translations of the questionnaire were equivalent in the two languages.
The reliability of the different affective factors was calculated using Cronbach's alpha.The factors had acceptable values of reliability (greater than .06;see Loewenthal, 1996;Nunnally, 1967).For FLCA, Cronbach's alpha was .897for the Danish sample and .890for the Spanish sample.For ECB, Cronbach's alpha was .928for the Danish sample, .920for the Spanish sample.For ideal L2 self, Cronbach's alpha was .872for the Danish sample and .811for the Spanish sample.Finally, for AES, Cronbach's alpha was .699for the Danish sample and .654for the Spanish sample.

Language Tests
Two tests were selected to measure learners' English proficiency: a grammaticality judgment test (GJT) and a listening comprehension test (ListT).The GJT, which was adapted from previous studies contained grammatical constructions that were balanced to include syntactic and morphological rules of early, intermediate, and late acquisition (see R. Ellis, 2009).The original set of constructions was adapted to this age group after piloting with 10 learners of the same age in each context.Each grammatical structure was represented by two grammatical and two ungrammatical sentences, with a total of 52 sentences (see Appendix B).The reliability of the language test adapted for this study, namely, the GJT, was calculated using Cronbach's alpha.The GJT had acceptable values of reliability: Cronbach's alpha was .946for the Danish sample and .796for the Spanish sample.The learners' accuracy scores were calculated for all the test items (range of scores: 0 to 52 points).
To measure learners' listening comprehension skills, we used the Listening section of the Oxford Placement Test (OPT; Allan, 2006).The ListT was scored following the OPT test instructions (range of scores: 0 to 100).
Data collection took place in two separate sessions one week apart.In the first session, which lasted about 40 min, the participants completed the background questionnaire, the ListT and the affective factors questionnaire.In the second session, which took approximately 20 min, they took the GJT.

Data Analysis
In order to address the first research question, we entered the average raw scores of the FLCA, ECB, ideal L2 self, and AES items as dependent variables, with learning environment as independent factor, in individual generalized linear models (GLM), using Stata (Release 14) (Statacorp, 2015).To check for any potential effects of gender, we also included gender as an independent variable.To address the second research question, we converted the raw Likert scores to z-scores for comparability purposes.We further inverted the z-scores for the variable ideal L2 self to approximate a normal distribution.This was done to implement mixed effects generalized linear models (Linck & Cunnings, 2015).To arrive at the best fitting models, we entered the learner factors and learning environment as fixed factors.We then proceeded by manual backward elimination, that is, by repeatedly eliminating nonsignificant interaction terms and then non-significant main effects.The choice of variables to be considered for elimination from one model to the next was based on significance levels.During the model-fitting process, nonsignificance was assessed by p < .10 to avoid being too conservative, as is typically done in the case of modelfitting in exploratory studies (Fisher, 1925; see also Cadierno et al., 2022).Lower-order terms and main effects that were part of a significant interaction effect were kept regardless of significance level.When presenting the results of the mixed models below, however, we maintain the conventional criterion for significance of p < .05.

Role of Learning Environment in Relation to Learner Factors
The aim of the first research question was to examine whether the L2 learning environment had an influence on a set of learner factors.The scores of the learner factors by learning environment and gender are shown in Table 1 as well as the significance tests of the individual GLMs.

Table 1. Means, Standard Deviations, and Significance Levels from Generalized Linear Model of Learner Factors by Learning Environment and Gender
Note.GLM = generalized linear model; FLCA = foreign language classroom anxiety; ECB = English competence beliefs; L2 = second language; AES = attitudes towards the presence of English in school.
One of the factors showed a significant difference between the two learning environments.FLCA was more pronounced for the Spanish learners (M = 3.46) than the Danish learners (M = 3.01), F(1, 162) = 5.81, p = .017,ηp 2 = .04,a small to medium effect size (Cohen, 1988).For that factor, there was also a significant main effect of gender in that female students had higher levels of FLCA (M = 3.69) than male students (M = 2.90), F(1, 162) = 16.56,p = .001,ηp 2 = .09,a medium to large effect size (Cohen, 1988).There were no other significant main effects and no significant interactions between any of the factors.However, in the case of ECB, there was a tendency toward a significant effect of gender.Male students tended to have higher ECB scores (M = 4.98) than female students (M = 4.71), F(1, 159) = 3.40, p = .067,ηp 2 = .02.

Role of Environmental and Learner Factors in Danish and Spanish Learners' L2 English Proficiency
The second research question aimed at examining the role of the environmental and learner factors (FLCA, ECB, ideal L2 self, and AES) in relation to our language outcome measures (GJT and ListT).As a preliminary analysis, given that the GLMs presented above had shown some effect of gender, we first tested whether gender had a significant influence on outcome scores.We therefore conducted twofactor GLMs on the two outcome measures including learning environment and gender as factors.The means for the GJT and ListT by learning environment and gender are shown in Table 2.There was a main effect of learning environment for both outcomes (FGJT (1,157)  = .35,both large effect sizes according to Cohen, 1988).Overall, the Danish students obtained higher mean scores on both outcome measures (GJT: 42.09 vs. 37.37; ListT: 80.73 vs. 70.1).There was no significant effect of gender (FGJT (1,157) = 0.50, p = .824,ηp 2 = .00and FListT(1,162) = 0.80, p = .373,ηp 2 = .01),nor of the interaction between gender and learning environment (FGJT (1,157) = 1.30, p = .256,ηp 2 = .01and FListT (1,162) = 2.14, p < 0.146, ηp 2 = .01)in this nor the previous analysis (see above).We therefore did not include gender in the mixed models to follow.
Tables 3 and 4 present the final fitted models for the GJT and ListT scores, respectively.In the following sections, we report the results in terms of estimates and their significance levels as obtained from the fitted models.
In relation to learning environment, the models confirmed the significant main effects for both outcome measures from the GLMs above: Danish participants scored significantly higher than Spanish participants.As shown in Table 2, for the GJT the difference between them was 4.72 points in raw scores in favor of the Danish participants (see Table 3: estimate: 3.66, z = 5.41, p < .001);for the ListT, the difference was 10.63 in raw scores (see  Note.N = 152.FLCA = foreign language classroom anxiety; ECB = English competence beliefs; L2 = second language; AES = attitudes towards the presence of English in school; SE = standard error; LL = lower limit; UL = upper limit.Note.N = 158.FLCA = foreign language classroom anxiety; ECB = English competence beliefs; AES = attitudes towards the presence of English in school; SE = standard error; LL = lower limit; UL = upper limit. As for the learner factors, for FLCA, there was no significant main effect for the GJT (estimate: 1.68, z = 1.32, p = .186),but there was a tendency towards a main effect (estimate: 1.56, z =1.65, p = 0.098) for the ListT.There was also a significant interaction between FLCA and learning environment for the GJT (estimate: -2.28, z = -3.85,p < .001).This interaction was not significant for the ListT.For ECB, there was a significant main effect for the GJT (estimate: 3.29, z = 4.76, p < .001)and a tendency towards statistical significance for the ListT (estimate: 2.08, z = 1.75, p < .081).However, there was also a significant triple interaction between ECB, FLCA, and learning environment for both the GJT (estimate: -1.94, z = -4.46,p < .001)and the ListT (estimate: -2.29, z = -2.09,p < .036).
The differential impact of ECB depending on the learning environment and FCLA is illustrated in Figure 1.On both the GJT and ListT, for Danish students with high scores on the ECB items (the dashed line), and low levels of FLCA (a z-score of -1 on the x-axis), high ECB scores corresponded roughly to high GJT and ListT scores, that is, a strong belief in one's self-competence in English corresponded to high proficiency scores.However, Danish students with high ECB who also scored high on the FLCA items (a z-score of 1 on the x-axis) tended to obtain lower proficiency scores regardless of their strong beliefs in their own competences.In contrast, for Danish students with low ECB (i.e., the solid line), their proficiency scores were lower than the students with high ECB overall.Thus, the level of FLCA had little estimated impact on their proficiency scores.As can be seen by the confidence intervals shown on the slope of the different levels of FLCA for Danish students with low ECB, the estimated points may overlap, indicating uncertainty whether the slope in fact rises or is flat.
In contrast, for the Spanish sample, there was less of a differential impact of FLCA in relation to both outcome measures.As can be seen in the right-most panels of Figure 1, the slopes of ECB remained essentially flat regardless of the increases in FLCA (plotted on the x-axis), with students with high ECB (the dashed lines) tending to have higher scores than students with low ECB (the solid lines).

Figure 1. Estimated Means for the Grammaticality Judgment Test and the Listening Comprehension Test as a Function of Learning Environment and Foreign Language Classroom Anxiety at Low and High Levels of English Competence Beliefs
Note.The English competence belief scores were split in low and high scores at the mean (z-score = 0).

Figure 2. Mean Scores on the Grammaticality Judgment Test and the Listening Comprehension Test as a Function of Learning Environment and Attitudes Towards the Presence of English in School
Note.AES = Attitudes towards the presence of English in school.These scores were split in low and high scores at the mean (z-score = 0).
For the factor AES, there was a tendency toward a significant main effect for the GJT (estimate: 0.39, z = 1.78, p = .076);for the ListT, the direction of the relationship was also positive, but there was no main effect (estimate: 1.00, z < 1.38, p = .167).For both outcome measures, there were significant interactions with learning environment (GJT: estimate: -1.45, z = -5.12,p < .001;ListT: estimate: -2.31, z = -2.07,p = .039).The interactions are illustrated in raw scores in Figure 2. On both the GJT and ListT, Spanish students who endorsed the factor AES more, that is, with more positive attitudes towards the presence of English in school, also obtained higher scores whereas this difference was less pronounced for the Danish students.For the GJT, a difference of 3.9 for the Spanish versus 0.5 for the Danish students: For the ListT, a difference of 2.4 for the Spanish versus 0.4 for the Danish students.
In the case of the factor ideal L2 self, it was kept in the fitted model for the GJT, but had been eliminated for ListT because it was not significant.For the GJT, there was no main effect of ideal L2 self (estimate: 0.23, z = 0.77, p = .442),but there was a significant interaction with learning environment (estimate: -1.05, z = -2.69,p = .007).This interaction is illustrated in Figure 3.The Spanish students who endorsed the factor ideal L2 self more, also obtained higher outcome scores whereas this difference was less pronounced for the Danish students; a difference of 4.3 for the Spanish versus 3.3 for the Danish students.

Figure 3. Mean Scores on the Grammaticality Judgment Test as a Function of Learning Environment and Ideal L2 Self Scores
Note.The ideal L2 self scores were split in low and high scores at the mean (z-score = 0).

DISCUSSION
This study examined the potential influence of L2 learning environment on a set of learner factors and the influence of learning environment and learner factors on L2 achievement.
The first research question examined whether differences in the learning environment of Danish and Spanish learners of L2 English had an influence on learners' level of FLCA, ECB, motivation, and attitudes towards L2 English.The results showed significant differences between the two groups in relation to one individual factor only, namely, FLCA.Specifically, Spanish learners experienced a significantly higher level of FLCA than Danish learners.This finding may be in part attributed to the differential amount of extramural contact with English that is observed in the two learning environments, although cultural differences may have also played a role.As found in a previous study that included a sub-sample of the current participants, (Muñoz & Cadierno, 2021), Danish learners experienced a higher level of contact with English outside the school when compared with Spanish learners.Arguably, Spanish learners' lower level of contact with extramural English may lead them to feel more anxious when having to speak English in the classroom.This explanation is consistent with the findings in Sundqvist (2009) where Swedish learners who spent more time in out-of-classroom English activities reported feeling less anxious about speaking English than those who spent less time on such activities.Our findings add further support to previous research showing that the learners' background can impact their level of FLCA (e.g., Woodrow, 2006).Additionally, the shorter linguistic distance between Danish and English may also have affected Danish learners' perception of difficulty and hence diminished their initial level of anxiety in the learning process.
The analysis also found two interesting results concerning the relationship between those learner factors and gender.Firstly, females were found to have significantly higher levels of FLCA anxiety than males regardless of the learning environment, which is in line with some previous results (Fenyvesi et al., 2020;Steiner, 2021).However, research has yielded mixed findings, as shown by the meta-analysis by Botes et al. (2020), with some studies finding higher levels of FLCA for females (e.g., Park & French, 2013) or for males (e.g., Alidoost et al., 2013), with other studies reporting no significant difference (e.g., Aida, 1994).In a more recent meta-analysis, Piniel & Zólyomi (2022) found a tendency for females to experience higher FLCA, but gender-related differences were not significant.Secondly, a tendency towards a significant effect for gender was obtained for ECB.Males tended to be more confident about their English competence than girls.As was the case for FLCA, previous studies on ECB and gender have produced mixed results, with some studies finding higher English competence beliefs for girls (e.g., Heinzmann, 2009) or for boys (e.g., Heinzmann, 2013).Our findings are, nevertheless, in line with our previous research conducted in Denmark with younger students (Fenyvesi et al., 2020), which found that primary school boys, especially those who were older (aged 9-10 years), had higher ECB than girls.
The second research question asked whether the L2 learning environment and the learner factors FLCA, ECB, motivation, and attitudes had an influence on the two groups' performance on the English proficiency tests.The results of the study showed that the Danish students obtained significantly higher scores in both tests when compared to Spanish students.Notwithstanding the possible influence of other factors, such as differences in instruction, this finding is in line with the finding of our previous study with younger Spanish-L1 and Danish-L1 learners (Muñoz et al., 2018), in which the influence of teaching on Danish-L1 learners was negligible (some 13 hours).The findings show a consistent pattern: Danish learners, in an English-richer environment and with an L1 that is typologically closer to English, attain higher levels of L2 proficiency than their age-matched Spanish counterparts.Our findings are also consistent with other studies conducted in similar English-richer environments, such as Flanders (De Wilde & Eyckmans, 2017) and with typologically similar L1s as shown in studies with Dutch (De Wilde et al., 2020;Puimège & Peters, 2019).
From a theoretical perspective, our results are in line with usage-based accounts of language learning that view L1 and L2 acquisition as being rooted in concrete usage events and taking place without learners being conscious of it, that is, largely implicit (e.g., N. Ellis & Cadierno, 2009;N. Ellis & Wulff, 2008;Langacker, 1987;Tomasello, 2003).Danish learners, who reside in an English-richer environment, benefit from their frequent contact with extramural English, which may facilitate their process of implicit and/or incidental L2 English learning.The Danish learners' superior performance in the listening test can be thus explained by their frequent exposure to spoken English through a variety of media (e.g., English subtitled television), which may also have reinforced the learning of the linguistic constructions in the GJT.
When examining the relation between L2 learning environment and learner factors with respect to their performance on the GJT and the ListT, three interesting results were obtained.First, a triple interaction was found between FLCA, ECB, and learning environment for the two outcome measures.For the Danish learner group, students with high levels of ECB, that is, students with strong beliefs in their English competence, but with high levels of FLCA obtained relatively low outcome scores.Thus, the correspondence between high ECB and proficiency was overridden by high FLCA.This finding corroborates similar results obtained with a younger group of Danish primary school children (Cadierno et al., 2022;Fenyvesi et al., 2020).In contrast, for the Spanish learner group, their proficiency scores were less influenced by FLCA.This inter-group difference may suggest that the two learner groups draw from different sources when evaluating their English competence beliefs.In the case of the Danish learners, given the high degree of contact with extramural English 2 (see Muñoz & Cadierno, 2021), their ECB scores are likely to reflect their perception of their English abilities both in relation to their performance inside, but also outside of the classroom where there is less corrective feedback regarding their English competence.Danish students with high FLCA, who presumably speak less in the classroom may be able to maintain high confidence in their English competence because the many out-of-school contexts provide less feedback on their English abilities.In contrast, for the Spanish students, given their lower degree of contact with extramural English, the evaluation of their English abilities is likely to mainly reflect their classroom experience, which includes more corrective feedback.This explanation is in line with previous theoretical frameworks and L2 research that have emphasized the close link between the students' self-beliefs and the specific learning environment (e.g., Bandura, 1986;Clément et al., 1994;Onorato & Turner, 2004) as well as with previous research that has shown a negative link between self-perceived L2 competence and FLCA (e.g., Cheng, 2002;Horwitz et al., 1986).
The second interesting result concerns the significant interaction between learning environment and AES in relation to both outcome measures.The Spanish students who endorsed this factor obtained higher scores in both tests whereas this difference was less pronounced for the Danish students.In other words, AES had a more predominant role in the Spanish/Catalan context than in the Danish context.The difference between the two groups can again be related to the aforementioned difference in the penetration of English in the two L2 learning environments.We can hypothesize that endorsing a higher presence of English in the Danish academic context is more in tune with the high social presence of English in society.As a result, AES is a less discriminating factor in relation to the outcome measures as compared to Spain, where there is less contact with English.Thus, in Spain as opposed to Denmark, being in favor of more English in the academic context may be related to higher motivation towards learning English and therefore coincide with higher English proficiency levels.
Finally, a significant interaction was found between learning environment and ideal L2 self for GJT only.For the Spanish students, those who endorsed this factor the most obtained the highest scores in the GJT, which is in line with previous studies that have found a positive influence of the ideal L2 self in relation to objective measures of L2 proficiency (e.g., Dörnyei & Chan, 2013).However, this difference was less pronounced for the Danish students.In other words, as it was the case with AES, ideal L2 self had a more predominant role in the Spanish/Catalan context than in the Danish one.There are two possible interrelated explanations for our findings.
The first one is related to the aforementioned difference in the presence of English in the two contexts.Spanish students in general are less likely to project themselves as future speakers of the language when compared to Danish students whose frequent contact with English out of the school may facilitate the process of imagining themselves as future speakers of English.This explanation supports our hypothesis on the basis of Dörnyei's (2009b) conditions that can enhance or hinder the motivational impact of the ideal L2 self.Specifically, the higher degree of contact with extramural English in the Danish context may prime the development of a stronger ideal L2 self when compared to the Spanish learners.Additionally, the status of the learners' L1 may also play a role, with Spanish being spoken much more widely in the world than Danish.The perceived usefulness of English for their future lives (e.g., in relation to the job market) is likely to be higher for the Danish than for the Spanish students.That is, projecting themselves as future speakers of English is more likely to be the norm for the Danish students, which would explain why the role of this factor is less predominant in predicting outcomes in the Danish context than in the Spanish one.
Finally, one possible explanation for the absence of interaction for the LisT could relate to the greater effort required to acquire the type of knowledge elicited by the GJT compared to listening skills.This would seem to provide evidence for the stronger relation between L2 self and intended learning effort reported in the literature (e.g., Csizér & Kormos, 2009;Taguchi et al., 2009).

CONCLUSION
The aims of the study were to explore whether differences in L2 learning environment had a distinctive influence on a set of learner factors: FLCA, ECB, motivation, and attitudes, and whether those different learning environments and individual factors had a differential influence on the L2 proficiency of the two groups of learners.Additionally, the role of gender was also examined.
The results of the study showed that the learning environment had an influence on FLCA.The Spanish learners in this study, with less contact with extramural English, experienced more anxiety when having to speak in English in the classroom.Additionally, gender impacted FLCA, with females being more anxious than males, and also ECB, with males tending to have a higher degree of confidence in their English abilities than females.
Furthermore, the study showed a varying impact of the different factors on the L2 English proficiency of the two learner groups.Specifically, AES, that is, the learners' attitudes towards the presence of English in the academic context, and the learners' ideal L2 self had a more prominent role in the Spanish context than in the Danish one.Furthermore, the significant interaction found between FLCA and ECB had a differential role in the L2 proficiency of the two learner groups.Only for the Danish students, FLCA trumped over ECB.This difference was attributed to a possible difference in the sources of learners' evaluation of the English competence, with the Danish learners' evaluation reflecting their English competence inside and outside the classroom, and the Spanish learners' evaluation being mainly based on their classroom experience and performance.
Our findings on the complex interaction between FLCA and ECB in the two learning contexts provide a new insight into L2 research on individual differences, namely, that the type of learning context (in this case with higher or lower presence of ambient English and lower or higher linguistic distance) seems to modulate the role of learner factors in L2 learning.Additionally, a methodological implication for future research is that whereas the questions in FLCA questionnaires are formulated in relation to the use of English in the classroom, the ECB questions are not.Therefore, learning contexts with or without frequent opportunities for contact with English outside the classroom may not show the same dynamic relations between the two predictive factors.
All in all, the results of this exploratory study point to a crucial role of the L2 learning environment and, thus, provide empirical support for a context-dependent view of the expression of learners' individual characteristics in relation to L2 learning.Additionally, our findings are in line with recent calls made in the literature about the important role of context when examining learners' individual factors (e.g., Dörnyei, 2009a).As stated by Dewaele (2007), research designs that incorporate single independent variables risk presenting a distorted picture of a complex reality.Research designs such as the present one that include several factors are more likely to capture the complex interactions among variables.
The study has several limitations, general caveats concerning cross-cultural questionnaire research aside (e.g., Smith, 2020).Data on learner factors and L2 English proficiency were collected simultaneously; thus, it is not possible to predict L2 performance on the basis of the learners' characteristics.Future research designs with different time frames for the collection of learner factors and proficiency data would allow the examination of the predictive role of learners' characteristics in the development of L2 proficiency.Future studies could also include learner groups in other learning contexts in order to examine the role of learning environment in shaping individual factors and L2 proficiency.Another limitation is the small size of the sample, that precludes generalizing our results to the schools' larger environment.Further research with stratified samples would be needed in order to make such generalizations possible.However, at the micro-level of two schools, the findings have provided evidence of how different factors combine and influence language learning and, ultimately, of the dynamic nature of language learning.

Table 2 .
Raw Scores for the Grammaticality Judgment Test and the Listening Comprehension Test by Learning Environment and Gender Note.GJT = grammaticality judgment test; ListT = listening comprehension test.

Table 3 .
Results of Fitted Mixed Effects Repeated Measures Model for Grammaticality Judgment Test Scores (With Reference Levels in Parentheses)

Table 4 .
Results of Fitted Mixed Effects Repeated Measures Model for Listening Comprehension Task Scores (With   Reference Levels in Parentheses)