ELF-sensitive teaching from the perspective of Polish trainee teachers

The advance of English as a lingua franca (ELF) is believed to carry implications for English language teaching and language assessment. The present contribution is an attitudinal study carried out among trainee teachers of English in a Polish university setting. The study sets out to investigate whether trainee teachers have a positive attitude towards ELF-sensitive teaching and whether their receptiveness (or lack thereof) to the concept in question is reflected in how they approach correcting language forms regarded as regular features of ELF. The findings show that there are elements of ELF pedagogy that respondents seem to be enthusiastic about – they readily acknowledge the importance of accommodation skills and they want students in the classroom to be exposed to many different non-standard English varieties. As regards the correction of non-standard English, respondents display a norm-driven approach, especially when teaching a student who they need to help pass an examination in the near future. In the conclusion of the paper, it is stated that respondents react positively to some aspects of ELF-sensitive teaching, but they show strong attachment to native-speaker norms, accuracy and the traditional no-tion of error , which is reflected in their approach to correcting non-standard language items.

The advance of English as a lingua franca (ELF) is believed to carry implications for English language teaching and language assessment. The present contribution is an attitudinal study carried out among trainee teachers of English in a Polish university setting. The study sets out to investigate whether trainee teachers have a positive attitude towards ELF-sensitive teaching and whether their receptiveness (or lack thereof) to the concept in question is reflected in how they approach correcting language forms regarded as regular features of ELF. The findings show that there are elements of ELF pedagogy that respondents seem to be enthusiastic about -they readily acknowledge the importance of accommodation skills and they want students in the classroom to be exposed to many different non-standard English varieties. As regards the correction of non-standard English, respondents display a norm-driven approach, especially when teaching a student who they need to help pass an examination in the near future. In the conclusion of the paper, it is stated that respondents react positively to some aspects of ELF-sensitive teaching, but they show strong attachment to native-speaker norms, accuracy and the traditional notion of error, which is reflected in their approach to correcting nonstandard language items.

Literature review
Research on ELF (English as a lingua franca) is extensive and encompasses a number of areas of interest. As the focus of the present paper is the involvement of ELF in ELT (English language teaching), the literature review section briefly discusses some of the strands of research related to ELF from the perspective of language pedagogy.
Early research on ELF was primarily concerned with the analysis of ELF-based communication at a range of linguistic levels. Empirical investigations, many of which based on the first ELF corpus, the Vienna-Oxford International Corpus of English (VOICE) (Seidlhofer 2001), identified language features that regularly occur in ELF interaction. Initially, it was assumed that the identification of language regularities -characteristic forms of ELF communication -may constitute a basis for the emergence and codification of a separate variety of English in the fullness of time. However, as more research has become available, the data have shown that apart from what is regular and stable across lingua franca interactions, ELF communication is characterized by fluidity and flexibility. Although the initial endeavour of codification has thus been questioned, the findings of these early studies carry implications for language pedagogy. From a language teaching perspective, it is suggested that forms which are emerging as systematic and frequent in ELF interaction should be considered ELF variants rather than errors (Jenkins, Cogo and Dewey 2011: 289) and it is recommended that teachers do not correct them (Jenkins 2005: 67). What is expected, in turn, in order to make the ELT classroom more ELF-oriented, is a gradual shift in orientation from a focus on correctness and accuracy to that of appropriateness and intelligibility (Jenkins 2007;Seidlhofer 2004Seidlhofer , 2011. The analysis of the implications that the advance of English as a means of international and intercultural communication carries for ELT and language assessment is another branch of ELF research. A number of terms have been offered in the literature to describe teaching practices that are based on the ELF perspective: ELF-aware , ELF-oriented (Takahashi 2014), ELF-informed (Vettorel 2016) and, more recently, ELF-sensitive teaching (Sekanina 2020). 1 The point of departure of many studies oriented towards ELF teaching is that for the majority of learners of English in the Expanding Circle (Kachru 1985) their future interlocutors are likely to be NNSEs, 2 who significantly outnumber NSEs. This demographic shift is believed to carry implications for what happens at the classroom level. Some recommendations as to what an ELF classroom should look like include: exposing students to a wide selection of native and non-native varieties of English (Matsuda 2012), training students in the use of accommodation skills (Watterson 2008), liberating students from the focus on native-like pronunciation (Jenkins 2000(Jenkins , 2002 and using teaching materials that portray NNSE characters communicating with other NNSEs in non-Anglophone countries (Vettorel and Lopriore 2013).
ELF-oriented pedagogy is unlikely to be put into practice until it is acknowledged in the domain of assessment and testing. Hall (2014: 379) defines testing as "an activity which perhaps more than any other dictates what is taught". As explained by Jenkins (2006: 42), both teachers and learners are unlikely to take kindly to the principles of ELF unless they are reflected in the targets set by the language testing community. For this reason, a critical examination of the nature and purpose of English language testing has been called for (Jenkins 2006, Hall 2014. What has been suggested, instead of a close concentration on accuracy, is for examination boards to refrain from penalizing the use of language features identified as recurrent in ELF interaction (as has already been mentioned above), while rewarding the successful use of accommodation strategies and penalizing their absence (Jenkins 2006: 49). Also, tests are re-1 These terms are used in the present paper interchangeably, but the most appropriate one seems to be ELF-sensitive teaching. As put by Sekanina (2020: 6), the term "is not linked to any proposed approach or framework of a pedagogy of ELF" and so it denotes a flexible, context-sensitive approach to ELT. 2 The concepts of NSE (native speaker of English) and NNSE (non-native speaker of English) are recognized as imprecise, controversial and generally problematic (see Davies 1991). They are used in the present paper, however, as they are still commonly referred to in the ELT literature. commended to include multiple native and non-native varieties of English for the assessment of receptive skills (Hall 2014: 384).
An ongoing line of research has investigated how different groups of respondents in the pedagogic environment have reacted to ELF. ELF-oriented attitudinal studies have been conducted, among others, on learners of English in a school setting (Ranta 2010, Szymańska-Tworek 2013, trainee teachers (Erling and Bartlett 2006) and in-service teachers (Grazzi and Lopriore 2020). Although the results of these studies reveal some differences between respondents in different countries, the strong overall tendency in most of the studies is preference for conformity to standard native-speaker norms. A powerful instrument in fostering awareness of ELF and initiating attitudinal shift is inclusion of ELF-oriented training as part of teacher education . Of particular interest are studies reporting on how pre-service and in-service teachers attending such training programmes re-evaluate their position and develop more flexible attitudes towards ELF (e.g., Vettorel and Corrizzato 2016). As pointed out by Vettorel and Corrizzato (2016: 506), teacher education is a first step for ELF to make its way into everyday classroom practice and to have a long-lasting impact there.
The present contribution fits into the tradition of attitudinal studies by investigating how trainee teachers of English in the Polish educational context respond to aspects of ELF-sensitive teaching.

Inspiration for the study
The inspiration for the present research was Dewey (2012), whose study addresses the relationship between ELF-related beliefs and actual classroom practices among English language teachers. Dewey's research revealed a certain inconsistency in teachers' approaches to ELF. For instance, one of his respondents who acknowledged the relevance of ELF in teaching, turned out to be, in practical terms, far more inclined to correct nonstandard language than a respondent who disregarded the concept of ELF as "pie-in-the-sky". Based on this example, Dewey points out that teachers' declared receptiveness towards ELF as a concept may have little to do with how they approach the issues of correctness and acceptability. This complex interaction between the way teachers approach ELF in theory and in practice has served as an inspiration for the present study.

The aim of the study
The present contribution is an empirical study carried out in a Polish university setting. The paper sets out to investigate to what extent trainee teachers are supportive of aspects of ELFsensitive teaching and whether this support (or lack thereof) is reflected in how they approach correcting non-standard language items identified in language corpora as salient features occurring in ELF interactions. Three main questions that guided the present study were: (1) What is the trainee teachers' general reception of ELF-sensitive teaching? (2) What is the trainee teachers' approach to correcting nonstandard English? (3) Is the trainee teachers' receptiveness to ELF-sensitive teaching reflected in their approach to correcting non-standard English?

Respondent characteristics
The study took place in Poland -an Expanding Circle country. The subjects were post-graduate students enrolled in a Master's degree of the TEFL programme of the Philology Department (now Faculty of Humanities) of the University of Silesia in Katowice.
The study concentrates on a sample of 93 participants; in terms of gender ratio, there were 80 women and 13 men. As part of their M.A. degree programme, students were required to attend a variety of modules on the theory and practice of TEFL, including courses in ELT, SLA, linguistics, applied linguistics and psychology. The curriculum also included English language classes, academic writing and the history, culture and literature of English-speaking countries. The language of instruction in all of these courses is English. During practical English classes students typically practise one of the two varieties: Standard British English or Standard American English. All participants have had teaching experience gained through a student teaching practice (practicum), which is an obligatory component of teacher education coursework and includes both classroom observation and their own individual teaching in schools. What is more, more than two-thirds of the participants have had at least some professional experience as teachers of English -they teach in private language schools or give private lessons to individual students. For this reason, the traditional distinction between "pre-service" and "in-service" teachers is not applicable to this population.

Data collection tool: questionnaire (statements)
The aim of this part of the study was to examine if trainee teachers have a positive attitude towards ELF-sensitive teaching. In his research, Dewey provided respondents with three terms (English as a lingua franca, English as a global language and World Englishes) and asked if they are relevant to teaching. An earlier study of this author (Szymańska-Tworek 2016), conducted on respondents of the same profile as in the present contribution (students pursuing their M.A. degree in ELT at the same university), showed that some trainee teachers are not familiar with the term English as a lingua franca. For this reason, in the present study I decided to avoid asking directly about ELF; instead, respondents were provided with fourteen statements that are based on some of the recommendations found in the research literature as to what an ELF-informed classroom should look like. In this way, the statements inquire into (i) exposure of students to multiple native and non-native varieties of English (statements 1, 2, 3 and 4), (ii) training students in accommodation strategies (statements 5 and 6), (iii) the role of NSEs and NNSEs in the ELT classroom (statements 7 and 8), (iv) the importance of pronunciation and accent (statements 9, 10, 11, 12 and 13), and (v) exposure of trainee teachers to different varieties of English (statement 14). Respondents were asked to express how far they agreed with the statements on a 5-point scale from 1 = strongly disagree, to 5 = strongly agree. They were encouraged to think about their own teaching context(s) when responding to the statements.

Data collection tool: language evaluation task (sentences)
In the second part of the research respondents were provided with fifteen sentences that included non-standard language items selected from ELF corpora as characteristic features of ELF-based communication. This part of the study was also inspired by Dewey (2012), who asked his respondents to rate a number of utterances selected from ELF corpora in connection with the following areas: correctness, acceptability, intelligibility and importance to correct. In the present study, respondents were first required to decide how intelligible the sentences were on a 5-point scale from 1 = not at all intelligible, to 5 = very intelligible. Then, respondents were asked to decide how important it was to correct these sentences in the ELT classroom, using a 5-point scale from 1 = not at all important, to 5 = very important. In order to add a context so that respondents had a better sense of who they were (or were not) correcting, this task was divided into three parts. First, respondents were informed that the sentences to be ranked were produced by a student who was about to take her Matura examination. 3 In the second part of the task, participants were informed that the sentences were uttered by a student in a language school who worked as a travel guide and needed English to communicate with foreign tourists in the Polish city of Kraków. In the third part of the task, respondents were provided with information that the sentences were produced by a student in a language school who needed English to work as a babysitter in London. The instructions to the task differed only in that they described a different type of learner (Matura student, travel guide, babysitter), while the sentences were the same for all parts of the task. The sentences are presented and described in Table 1.

Table 1
Description of the sentences used in the evaluation task The sentences used in the evaluation task Features of ELF they contain 1) You remembered to feed the cat, isn't it?
2) The girl which sat beside him was his daughter.
3) Warsaw is Polish city. 4) I need to contact with my parents. 5) He like fast cars.
These sentences contain some of the lexicogrammatical features identified in VOICE (Seidlhofer 2004: 220) as regularities in ELF communication. These features include: -using non-standard forms in tag questions, -confusing the relative pronouns who and which, -omitting articles, -inserting redundant prepositions, -dropping the third person present tense -s, Seidlhofer (2004: 220) points out that all of the features "appear to be generally unproblematic and no obstacle to communicative success".
6) How long time did it take you to solve this problem? 7) I am interested to see the results of this study.
These include some of the characteristics of ELF lexicogrammar identified by Cogo and Dewey (2006: 65), namely: -increased explicitness, -preference for bare and/or full infinitive over the use of gerunds, Cogo and Dewey (2006: 64) observe that all lexicogrammatical features they identified are communicatively effective.
8) I am hating this awful weather! This sentence includes a grammatical construction -the use of the progressive on the so-called stative verbs -which Ranta (2006) identifies as a salient feature in an academic ELF speech corpus. Ranta (2006: 110) emphasizes that this type of use of the progressive does not seem to cause misunderstanding in any of the instances in the data. 9) She gave me an advice that I'll never forget. 10) I am here since two o'clock. 11) I know that even if I would practice the rest of my life, I would never be good enough. 12) She plays the piano beautiful. 13) Yesterday my mum did a delicious chocolate cake.
These sentences come from Erling and Bartlett (2006), who described some of the linguistic features characteristic of English produced by students studying English at the Freie Universität Berlin. These are, among others: -a loss of distinction between countable and uncountable nouns, -the use of present tense for present perfect meaning, -an extended use of the modal verb would for expressing condition, -variations in adverb use, -a loss of distinction between word pairs that have similar meanings, e.g., make/do. 14) He suffers from claustrophobicy so he never travels on underground trains. 15) The research examinated the effects of alcohol on long-term memory.
The last two sentences respondents were provided with contain lexical innovations identified by Pitzl, Breiteneder and Klimpfinger (2008) in their study that draws on a subcorpus of VOICE. The sentences contain examples of two categories of lexical innovations identified in the study, namely suffixation and backformation. Pitzl, Breiteneder and Klimpfinger (2008: 22) state that the lexical innovations identified in their data seem to be communicatively effective.

The questionnaire (statements)
The first part of the present study was intended to examine trainee teachers' reception of ELF-sensitive teaching by asking them to relate to fourteen statements. The data for this part of the research are presented in Table 2. 4,5 The data show that respondents reject the idea of exposing pupils 6 to nativized English and non-native English. Interestingly, the question about nativized varieties of English was assigned an even lower rating score (M = 2.06) than the question about non-native English (M = 2.27). This may be because examples of non-native English provided in parentheses in the questionnaire were German English and Russian English -presumably respondents showed more interest in these two variants of English because Germany and Russia are neighbouring countries to the respondents' native Poland. Most respondents (57 %) 7 acknowledge the need for pupils to be exposed to many different ENL (English as a native language) varieties of English, although nearly a half (45.2 %) state that exposing pupils exclusively to British and American English is sufficient to prepare them for international communication. The results suggest that respondents seem unenthusiastic about exposing pupils to non-ENL varieties. 64.5 % of respondents feel that pupils should listen to examples of non-understanding or miscommunication that was successfully overcome by the use of communication strategies and as many as 93.5 % of informants believe that teachers should train pupils on how to behave in case of miscommunication by showing them accommodation strategies. The positive response displayed by trainee teachers towards these two questions 6 When describing the results of the study, the terms students and pupils refer to learners in a school setting, while student teachers participating in the study are referred to as trainee teachers or respondents. 7 The original 5-point Likert scale, with "strongly disagree" and "strongly agree" at the extremes, has been collated into three categories and presented as percentages.
shows that most of them acknowledge the need for pupils to be trained in how to handle non-understanding and communication breakdowns.
The statements about whether pupils should be prepared for communication primarily with native-speakers of English and whether textbooks should present non-native speakers in non-Anglophone contexts received a variety of responses. In the case of both questions, responses are roughly equally distributed between positive, negative and neutral ones, while extreme forms of agreement and disagreement tend to be avoided. This could be interpreted as a sign of a certain flexibility displayed by respondents, whose decision in this respect presumably depends on whether the figures of NSE and NNSE are or are not relevant to particular groups of students in their own context.
The attitudes that respondents display towards pronunciation teaching seem quite traditional. Although most respondents (62.4 %) state that the ability to be an effective communicator is more important than having a native-like accent, 69.9 % disagree with the idea that correcting pupils' pronunciation is not important, even when it is intelligible. The majority of respondents (63.4 %) find it problematic when a pupil substitutes the voiceless dental fricative sound /θ/ with [t] or [f]. 8 The majority (66.6 %) declare it is important that teachers expend much effort to make pupils sound as native as possible. Although, of course, there is nothing inappropriate in teachers making an effort to teach pronunciation, the point made by ELF scholars is that classroom time spent on instilling native-like pronunciation in learners could be used more efficiently and productively, for instance on developing skills and competences which are shown by research to be crucial for international intel-8 It is worth noting here that the sound /θ/ is not included in Jenkins's Lingua Franca Core (2000) -a pronunciation syllabus of phonological and phonetic features identified as crucial for international intelligibility -as her research shows that the mastery of this sound is not necessary for mutual intelligibility and thus various substitutions are permissible. What is more, the sound /θ/ has been acknowledged as exceptionally difficult to master because it does not occur in the majority of the world's languages and is even missing in some ENL varieties (Jenkins 1998: 122). ligibility. The statement about pupils' preferences to speak L1-accented English obtained a more varied response -while a small majority (50.5 %) of respondents declare it would bother them if a pupil wanted to speak Polish-accented English, onefifth of respondents claim they would accept it.
The last statement in the questionnaire inquired about respondents' preferences regarding their own language education, and more specifically, whether university teachers should acquaint them with different native and non-native accents and varieties of English. 75.2 % of respondents reacted positively to this statement, which means that although respondents displayed scepticism about exposing pupils to non-ENL varieties, they seem enthusiastic about getting to know accents and varieties from across the three Kachruvian circles for their own language development.
Summing up this part of the research, there are elements of ELF pedagogy that respondents seem to be enthusiastic about. These are, first and foremost, training pupils on how to behave in case of miscommunication, exposing them to examples of non-understanding that was successfully overcome with the use of communication strategies and raising pupils' awareness of accommodation skills. What is more, respondents acknowledge that being able to communicate effectively is more important than having a native-like accent. Most respondents also feel that the ELT classroom is a place where pupils should be exposed to many different ENL varieties, such as Australian English or Canadian English. Last but not least, respondents react positively to the idea of being acquainted with native and non-native English in their own language education.
In contrast to the above, respondents have a circumspect approach to some principles of ELF-sensitive teaching. Trainee teachers are critical of recommendations that mainly concern two aspects of ELF pedagogy: the presence of non-native English in the classroom and issues connected with pronunciation teaching and accentedness. As far as non-native English is concerned, respondents decidedly reject the idea that pupils should be exposed to either Outer-or Expanding-Circle English. The statements concerning pronunciation also reveal respondents' traditional approaches to pronunciation teaching: the majority of respondents find it problematic if pupils substitute the sound /θ/ with /t/ or /f/ or if they want to speak L1-accented English, most respondents think it important to correct the pupils' pronunciation, even if it is intelligible, and finally, most respondents believe teachers need to make much effort to make pupils sound as native as possible.

The language evaluation task (sentences)
In the second part of the research, respondents were provided with sentences that include non-standard language forms identified as recurrent features in ELF interaction. First, respondents were asked if they found the sentences intelligible. The data are presented in Table 3. The data in Table 3 show that all the sentences were assigned high rating scores and the overall mean rating for intelligibility amounts to 4.03. It can be stated then that the respondents render the sentences highly intelligible. The respondents were then asked to decide how important it is to correct these sentences in the ELT classroom, when produced by three students: a Matura student, a student who works as a travel guide in Kraków and a student whose plan is to work as a babysitter in London. The results are presented in the following Table 4, Table 5 and Table 6 for each student respectively. Total -Importance to correct (Matura student) 4.02 0.98 Total -Importance to correct (travel guide) 3.72 1.18 The data show that although respondents find the sentences highly intelligible, they also think it is important to correct them, especially if they are uttered by a student who is to take an examination in the near future (the overall mean rating for Matura student is 4.02). The non-standard language is considered less important to correct when produced by a travel guide (M = 3.72) and a babysitter (M = 3.62). This indicates that respondents' decisions to correct the non-standard features is, at least to some extent, dictated by whether the language skills of a given student are to be verified in the process of testing or not. Respondents are most inclined to correct the Matura student and this attitude is not unexpected -after all, this is a student who is about to take a "high-stakes" state examination, where correctness is measured against traditional ENL rather than ELF language norms. Correcting non-native language items in the case of this student is, as was probably assumed by trainee teachers, responding to her immediate need, that is, helping her to pass the examination. What is notable is that respondents are slightly more relaxed about correcting the travel guide and the babysitter. This may be dictated by the fact that both of these students need English for real-life communication in highly multilingual and multicultural contexts rather than for passing an examination. In the instruction to the task, respondents were informed that the student who is a travel guide "needs English to communicate with people who come to visit Kraków from all over the world", while the other student "needs English to work as a babysitter in London". These results may indicate that, because of the washback effect, testing practices dictate what in fact happens at the classroom level -respondents display a highly norm-driven approach when correcting the Matura student, most likely in order to satisfy examination requirements.
However, what also needs to be noted is that although respondents report less concern about correcting the travel guide and the babysitter, the overall mean ratings for both of these students are nevertheless high (M = 3.72 for travel guide and M = 3.62 for babysitter). Respondents were provided with information that the travel guide needs English to communicate with tourists who come to Kraków from across the world. This description presents the student as a regular participant of ELF communication who uses English to tell the legend of the Wawel Dragon to listeners from countries from across the Kachruvian circles, such as, Japan, Germany, Singapore and the US. Unlike in the case of the Matura student who needs grammatical accuracy to get through examinations, it can be argued that the travel guide could benefit from language instruction that prioritizes communicative efficacy, intelligibility and accommodation, as well as the development of sociolinguistic, discourse, strategic and intercultural competence. Most respondents, however, insist on correcting the sentences provided in the study, also when they are uttered by the travel guide -13 out of 15 sentences were ranked by the majority of respondents as important and very important to correct.
As far as the third student -the babysitter -is concerned, the sentences produced by her were ranked as the least important to correct, although the overall mean ratings for her and the travel guide differ only negligibly. Respondents were informed that this student needs English to work as a babysitter in London. The capital of the UK, although situated in an Inner-Circle country, is one of the most cosmopolitan cities in the world where different varieties of English and different cultures coexist. During her stay in London, the student is likely to come into interaction with Britons, but also with Indians, Pakistanis, Nigerians, Jamaicans, Italians, Lithuanians, Turks and many other nations, all speaking English in their own way and with different levels of proficiency. From the perspective of ELF communication, the non-native language forms provided in the study have no negative effect on understanding. Again, however, most respondents insist on correcting the sentences -12 out of 15 sentences were ranked by the majority of respondents as important and very important to correct when uttered by the babysitter.
In order to examine if the sentences considered as the least intelligible were the ones rendered the most important to correct, the correlation between intelligibility and importance to correct was checked. The result is presented in Table 7. The data presented in Table 7 show that the correlation is not statistically significant, which means that respondents' decisions to correct a given sentence was not dictated by their perception of how intelligible that sentence is. For example, the sentence "He like fast cars", in which the third person present tense -s is omitted, is considered as intelligible and very intelligible by 83.8 % of respondents (M = 4.26). At the same time, the sentence is ranked as important and very important to correct by 86.1 % of respondents in the case of the Matura student, 69.9 % in the case of the travel guide and 67.8 % in the case of the babysitter. The third person present tense -s is a feature commonly tested in many standardized examinations and respondents' insistence on correcting the sentence when produced by the Matura student does not come as unexpected. However, as communication and intelligibility are not hindered by the omission of the -s morpheme, it can be surmised that respondents' insistence on correcting the travel guide -a regular ELF interactant -is motivated by reasons other than intelligibility.
One of the aims of the study was to examine whether the degree of respondents' receptiveness to ELF pedagogy (exemplified by their responses to the statements in the first part of this research) is reflected in their approach to correcting non-standard use of English. The results for the Matura student, travel guide and babysitter are presented respectively in Table 8, Table 9 and  Table 10. The above data show that there is a statistically significant negative correlation between respondents' receptiveness to ELF sensitive teaching and their approach to correcting non-standard English in the case of the Matura student (Table 8). This indicates that the more ELF-oriented respondents were, the less often the sentences uttered by the Matura student were perceived as in need of correction. The correlation is, however, weak. The correlation between trainee teachers' receptiveness to ELF pedagogy and their willingness to correct was not confirmed in the case of the travel guide (Table 9). A statistically significant negative correlation reappears in Table 10, which refers to correcting non-standard language produced by the babysitter -the more ELF-friendly respondents were, the less often they decided that the sentences required correction. Again, as in the case of the Matura student, the correlation is weak. The results show that respondents' receptiveness to ELF-informed teaching is reflected in their approach to correction only to some extent. There is a tendency that more ELF-friendly participants are less likely to correct forms characteristic of ELF interaction, but there are exceptions.

Conclusions
In the first part of the present study, respondents were asked to relate to recommendations concerning ELF teaching. While respondents readily acknowledged the importance of accommodation skills to sensitize pupils to communication breakdowns and how to deal with them, generally speaking, they did not respond to the research items in pro-ELF ways. Most respondents displayed scepticism towards two principles which lie at the very heart of the ELF paradigm: providing students with exposure to multiple varieties of English and liberating students from the need to focus on native-like pronunciation. Respondents want to be acquainted with native and non-native English in their own language development, but they see no place for Outer-and Expanding-Circle English in school education -the inclusion of nativized and non-native English in ELT is decidedly rejected. The attitudes that respondents display towards pronunciation teaching also show little alignment with the ELF perspectivemost respondents state that pupils' pronunciation should not deviate from native-speaker norms and teachers need to expend much effort to make pupils sound as native as possible. The picture emerging from these data is that respondents are not ELF-friendly, but they are not entirely negative towards ELF either. They reacted positively to some selected principles of ELFinformed teaching.
In the second part of the research respondents were provided with sentences that include non-standard language items selected from ELF corpora. Non-standard English was recognized by respondents as highly intelligible, and yet important to correct. Respondents' perception of how important it is to correct a given language form seems to be context-dependent -the sentences were considered in greater need for remediation if they were produced by a student who is expected to take an examination in English, whereas those uttered by students who need English for communication purposes in international contexts were less likely to get corrected. There seems to be a tendency for respondents to be slightly more relaxed about correction when they teach students whose immediate learning need is to take part in communication amongst linguistically diverse groups of people. These results also point to the washback effect that testing practices have on what happens in the classroomrespondents display a highly norm-driven approach when working with a student who they need to help pass an examination in the near future.
However, although respondents are slightly less inclined to correct non-standard language uttered by students who need English mostly for ELF talk, they nevertheless attach considerable weight to correcting those students. Also, respondents' decisions to correct the non-native language is dictated by reasons other than intelligibility -the non-standard forms were ranked as important and very important to correct even when they were regarded as highly intelligible. Even though respondents make some allowances for the context in which students are likely to use the language, their approach to correction is traditional, i.e., based on standardized native-speaker norms.
One of the aims of the present study, as inspired by Dewey (2012), was to examine if respondents' receptiveness to ELF pedagogy (or lack thereof) is reflected in their approach to nonstandard language. The results in this respect are not conclusive. There is a tendency that respondents who are in support of the ELF perspective are less likely to correct forms characteristic of ELF communication, but the tendency is not strong. This is what Dewey draws our attention to in his research -his study shows that even those teachers who are in full support of ELF conceptually may hold a conventional stance as far as correctness is concerned. In other words, as pointed out by Dewey, some teachers accept ELF in theory, but reject it in practice.
The same sentiment seems to be shared by the respondents in the present study. Even if some of them support ELF in an abstract, ideological way, they are against abandoning a NSE benchmark when correcting non-standard English. Respondents react positively to some aspects of ELF-sensitive teaching, but at the same time they show strong attachment to nativespeaker norms, accuracy and the traditional concept of error, which is reflected in their approaches to correcting non-standard uses of English. This preference for NSE standards may at least partly be related to language assessment practices, where traditionally correctness is measured against ENL rather than ELF language norms. Respondents may feel that adopting a less conventional approach to correcting of what they consider as "errors" is likely to put at risk a student's chance of successfully passing an examination.
In terms of future research, an interesting follow-up to this study would be to investigate whether trainee teachers develop a more flexible attitude to ELF and non-standard English after attending a course dedicated to the topic in question. The impact of ELF-oriented teacher education on teachers' beliefs and classroom practices is still an under-researched area, but the results of those research projects which have been carried out are promising. As remarked by Vettorel and Corrizzato (2016: 505), "[O]nce informed, teachers do acknowledge the importance of dealing with topics related to the current developments of English and their pedagogic implications".