Insights into Teacher Beliefs and Practice in Primary-School EFL in France

: Teacher beliefs affect choices of methods, representations of learning, and classroom practice, and are important in understanding primary EFL teaching in France, where language teaching has been a compulsory subject entrusted to generalist class teachers for 20 years. This quantitative study explores questionnaire data from 254 primary teachers, associating teacher beliefs and classroom practice. With respect to views of language teaching and learning, the study reveals a three-way division of teachers between grammar-oriented teaching (PPP), communicative-language teaching (CLT), and ‘sceptical’ teachers. The PPP ( n = 72) group employed the smallest range of teaching activities and rarely taught older pupils. The CLT group ( n = 60) tended to have higher English proﬁciency and more in-service training and offered the widest range of oral activities. The sceptical group ( n = 85) took no strong theoretical position, had lower English proﬁciency, and focused on listening and speaking skills. We found no correlation between teacher age and language learning beliefs or teaching practices. However, teachers who offered a wider range of activities in any of the ﬁve competences tended to have more in-service training and higher English proﬁciency. Further correlations were found between oral language teaching and technology integration, and written language teaching and teaching experience. The paper concludes with links to previous teacher cognition research and suggestions for teacher education.


Introduction
considers that the introduction of modern languages in primary schools is "possibly the world's biggest policy development in education". Copland et al. (2014) suggest that this trend is linked to the assumption made by education stakeholders that 'earlier is better,' despite research in second language acquisition increasingly questioning the Critical Period Hypothesis (Muñoz 2008;Kihlstedt 2019;Singleton and Leśniewska 2021). One major challenge to this policy development relates to the teacher education that has accompanied the introduction of languages, including English, as a compulsory subject at primary school. Copland et al. (2014) report a lack of consideration of who will teach English to young learners that has resulted in a global tendency for one of two difficulties: (a) teachers may have sufficient language proficiency and pedagogical training but lack specific competence with young learners, or (b) they may be (pre-)primary education specialists without adequate language skills or language-specific training. The present study focuses on young English language learners in French primary schools, where institutional policy over the past decade has switched reliance on visiting specialists (corresponding to (a) above) to generalist class teachers (b). It seems important to investigate the practices and beliefs of teachers undertaking this new responsibility.
As part of a collaboration on technology-mediated primary EFL involving academics and education authorities, the present study was conducted within a two-year project financed by the French Ministry of Education: RAVEL (Ressources pour l'Apprentissage en classe Virtuelle et l'Enseignement des Langues; learning resources for technology-mediated language teaching). This project aimed to compile a systematic overview of research on a specific topic relating to digital technologies in education; combine research contributions on a specific topic with feedback from practitioners, teacher educators and inspectors; create learning scenarios that incorporate the use of digital technologies; and disseminate project results as open-access resources to create a sustainable knowledge base on the topic. We began with a survey of generalist primary teachers currently teaching English, which investigated teacher beliefs and teaching practices in relation to a range of contextual factors including language proficiency, training, experience and classroom equipment. Our paper opens with a review of research on teacher beliefs regarding languages and technology, followed by a short presentation of the French primary school context.

Teaching Beliefs in Language Education
There is strong evidence to suggest that teaching beliefs are powerful predictors of teachers' classroom behaviours (Pajares 1992) and that teachers' lesson preparation and classroom practice are more influenced by their teaching beliefs than by their disciplinary knowledge (Williams and Burden 1997). In the field of language education, Borg (2003, p. 81) defines teacher cognition as the "unobservable cognitive dimension of teaching" which refers to what teachers think, know, and believe, as well as the relationship between these mental constructs and what teachers do in the language classroom. He describes four main areas to be considered: (a) 'schooling' or teachers' prior experiences in the education system as learners coupled with their own educational background that together inform their initial perceptions of teacher training whilst also continuing to exert an influence throughout their career, (b) professional preparation and training, (c) contextual factors that play an important role in determining the extent to which teachers are able to implement teaching which is congruent with their cognitions, and (d) classroom practice, which can influence cognitions unconsciously and/or consciously through reflective activities. As authors including Farrell (2006); Löfström and Poom-Valickis (2013) and Lin (2013) underline, teacher beliefs, or the personal pedagogical theories held by teachers, affect their choice of methods, representations of how learners learn, decision-making in the classroom and, thus, their actions and responses in the classroom. That is to say, teachers' cognitions affect the activities they offer, how they implement them, and therefore the learning opportunities made available to pupils. Teacher cognition is "often personal, leading to a vast variety of beliefs held by individual teachers" (Reynolds et al. 2021, p. 2); teachers are also emotionally invested in these beliefs, rendering them more difficult to define and measure.
Studies of English language teachers' beliefs have primarily examined secondary school teachers (Cirocki and Farrell 2019;van den Broek et al. 2018) and university lecturers (Farrell and Guz 2019;Farrell and Yang 2019). Others have investigated teacher cognition with respect to specific language skills (Bai and Yuan 2019;Borg and Burns 2008;Ngo 2018;Sato and Oyanedel 2019). Copland et al. (2014) note that few studies have yet examined language teacher beliefs in the primary sector despite research showing that primary and secondary school teachers often hold differing views. Pre-primary teachers in an interview study by Jacoby and Lesaux (2019) cited a number of essential factors in teaching very young learners, including appropriate support for learners to develop social-emotional skills, and mother tongue use to help both L1 and L2 language and literacy development. These in-service teachers also attached great importance to the classroom environment and resources to encourage what they termed 'natural' learning. Reynolds et al. (2021) also explored pre-primary EFL teacher beliefs among pre-service teachers using writing prompts to collect data. Their participants stressed the importance of content-rich and ageappropriate activities to accommodate the specific characteristics of very young learners, develop positive emotions towards the target language, and exploit the inquisitive nature of young learners by using their interests to create learning opportunities. Participants prioritised listening and speaking skills over reading and writing skills and the authors noted that teacher beliefs in their study were coherent with early childhood education research and linked to learning through play and non-teacher-centred approaches. Very few participants in the study, however, described beliefs relating to learning to teach, the teacher education programme or about 'self': their own self-efficacy or teacher emotions. When participants did evoke the latter category, beliefs focused on how to successfully achieve professionalism by improving their own personal qualities to become qualified English teachers, including their level of English.
Two studies of teachers of older primary learners highlight the often negative impact on teacher beliefs and practice of contextual factors such as institutional norms and expectations. Bai and Yuan (2019, p. 141) investigated attitudes to teaching pronunciation among Hong Kong English teachers and found "a wide range of personal and contextual obstacles, which not only created a gap between their teaching beliefs and practices, but [also] reduced their self-perceived efficacy in pronunciation teaching". Moodie and Feryok (2015) observed and interviewed four primary EFL teachers in Korea over an 18-month period and identified a number of organisational obstacles, which meant that none of the participants "consistently and effectively addressed communicative competence in their students" (Moodie and Feryok 2015, p. 466). They noted, however, that the teachers were sustained by enjoyment in their own learning of English: "positive emotions associated with learning English endured through teaching assignments that threatened it" (Moodie and Feryok 2015, p. 466).
Research on teaching beliefs and technology integration in general primary education shows an impact for teachers' views of technology on its uptake in the classroom: unsurprisingly, positive attitudes towards technology lead to greater integration and negative attitudes discourage it (Prestridge 2012;Jimoyiannis and Komis 2007;Ward 2020). Technological barriers including limited access to technology have also been linked to teachers' beliefs about the role of technology (Ertmer et al. 1999;Mama and Hennessy 2010). Niederhauser and Stoddart (2001) report that when presented with a large palette of different technologies which afford different activities and approaches, teachers will choose those technologies that help accommodate their own perspectives on teaching and learning. Sacré et al. (2021) conducted a systematic literature review regarding technology integration in the young English learner classroom (5-13 year-olds). Results revealed that empirical research could be categorised under six technology types: virtual exchange programmes, digital narration, mobile technologies, virtual reality, games and simulations, robots, and interactive whiteboards. Without demonstrating specific learning gains, the literature provides examples of technology affording access to authentic communicative situations and suggests positive effects on pupils' oral skills and motivation. The systematic literature review agrees with Whyte and Cutrim Schmid's claim that "the field of technology-mediated language education with young learners [is] somewhat bereft of both theoretical underpinning and empirical findings" (Whyte and Cutrim Schmid 2018, p. 338). This finding provides impetus for the present investigation of teachers' technology use in primary EFL and its relation to their beliefs and practice.

Language Education in French Primary Schools
Modern languages became a compulsory subject in French primary schools in 2000. To improve language education, the Ministry of Education focused on the major European foreign languages (MFL) while also developing bilingual education in some regional languages. Teaching guidelines in France are published in the form of official legal documents known as national programmes, which set out learning objectives without the level of pedagogical specificity found in national curricula in other countries. For languages, the programmes are based on the Common European Framework of Reference for Languages and the European Language Portfolio (Council of Europe 2001) which divide language competences into five areas: listening, speaking, reading, writing, and interaction. Modern languages are introduced in Cycle 2 (learners aged 5 to 8) as a specific discipline in the overall curriculum, to be delivered in two 45-min lessons per week, for an annual total of 54 h. The goal is to lay the groundwork for the initial development of students' plurilingual competence and prioritises speaking skills. By the end of Cycle 3 (learners aged 9-11), pupils are expected to have reached a CEFR A1 proficiency level in all five language competences. Table 1 shows the different aims of the two cycles. Cycle 2 (GS 1 , CP, CE1; 5-8 Years) Cycle 3 (CE2, CM1, CM2; 9-11 Years) Understanding oral language-listen and understand simple oral messages read by the teacher relating to everyday situations. Continuous oral production-using a model, recite, describe, read, or retell. Participate in a conversation-participate in simple exchanges to be heard and understood in diverse situations relating to everyday life. Discover cultural aspects related to the regional or modern language-identify major cultural landmarks in the pupils' everyday environment.
Listen and understand-Employ your auditive short and long-term memory to memorise common words and expressions; use auditory and visual cues to understand the meaning of unknown lexical items. Read and understand-use the context, illustrations and personal knowledge to understand a text; recognise isolated words in a short text; recognise the phoneme-grapheme relationship specific to the language. Continuous speaking-memorise and reproduce statements; express yourself modifying pace and volume; participate in simple exchanges.
Writing-write words and expressions for which the spelling and syntax have been memorised; use simple structures to write a sentence.
React and interact-ask simple questions; employ a range of ritualised conversations.
We can note the very broad nature of these descriptors and the frequency of the adjectives 'short' and 'simple', which allow for a wide variety of interpretations (see Valax 2011 for an interesting critique). The descriptors are intended to be methodologically neutral, and although an action-oriented approach (Piccardo and North 2019) is generally promoted, both grammar-oriented and communicative approaches are compatible. Some teachers favour a structural, or grammar-based method sometimes referred to as PPP (presentation-practice-production; cf. Anderson 2016) while others take a meaning-based approach, as in communicative or task-based language teaching (Ellis 2018). The national programmes also include details of evaluation, specifying that assessment should be separate from learning activities; only vocabulary already used in teaching should be included in assessment; and skills should only be assessed when learners are ready.
Pre-service teacher education in France is currently conducted at Master's level in Higher National Institutes of Teaching and Education (Instituts nationaux supérieurs du professorat et de l'éducation) which are associated with both universities and local education authorities. Students with an undergraduate degree in any subject complete a two-year course in primary education, which includes education studies, technology training, and specific modules in the different disciplines of the primary curriculum. Students also have school placements where they observe mentor teachers and teach pupils. They take a national competitive exam in addition to this Master's in Education in order to become qualified teachers. The time devoted to language teaching is necessarily limited, with perhaps 40 h each for language development and pedagogical preparation over the 2-year programme. Since English is by far the most commonly taught language in French primary schools, students who may be stronger in other European languages are nevertheless expected to learn to teach English. Once in post, teachers can request a place on in-service courses in any disciplinary area including English language teaching; some teachers have the advantage of an undergraduate degree in English studies.

Research Questions
The present study thus has a two-fold aim. We seek to contribute a growing body of research on the relationship between teachers' beliefs and classroom practice with a study of primary language teachers. On a more practical level, we consider that understanding teachers' pedagogical beliefs and practices is crucial to the success of our teacher education project. We thus developed a preliminary questionnaire to identify potential teacher profiles, which might helpfully inform workshops to co-design the pedagogical scenarios in the later phases of our project. Following the previous research reviewed above, the survey covers four areas: (a) teachers' reported pedagogical practices, (b) their use of technology in the classroom, (c) their beliefs about second language teaching and learning, and (d) background information about professional development and teaching contexts.
Our study explores three research questions: 1.
What do French primary teachers believe about language learning and teaching in general, and teaching young learners English in particular? Do they have specific views about technology integration in this area? 2.
What links can be identified between such beliefs and reported classroom practice? Are there particular groupings of beliefs/practices that allow the identification of particular teacher profiles? 3.
Are there connections between different aspects of teacher beliefs, age, experience, language proficiency, specific teaching and institutional contexts, and beliefs and practices in primary EFL?

Methodology
In this section, we describe the questionnaire designed to conduct our review of teaching beliefs and practices in primary EFL and its dissemination. We then give an overview of the in-service primary EFL teachers who participated in the study before detailing the data analysis methods used to explore the questionnaire results.

Instruments
The questionnaire was elaborated in a focus group (Finch and Lewis 2003) in November 2020, comprising a small working group of 9 researchers and 12 local education authority staff (language and technology coordinators, inspectors). It was created in LimeSurvey (Limesurvey GmbH n.d.), validated by Université Clermont Auvergne's ethics committee, and pre-tested with 8 primary teachers and coordinators. The final version of the questionnaire included 45 closed and open questions structured in 4 parts after an introductory consent section: language profiles (CEFR self-assessment grid: Council of Europe 2021). The lists of sample language activities and tools were drawn up by the language and technology coordinators for languages, and open questions were provided to allow teachers to make additions or include explanations, as well as sample materials from their recent teaching.
The full questionnaire took some 20 minutes to complete and was administered by the local education authorities in March 2020, sent via the digital technologies delegation to each primary school's headteacher in the regions of France involved in the project.

Participants
A total of 254 in-service primary EFL teachers answered the questionnaire, 91% were women. 2 Ages and professional experience varied, but the majority (80%) had more than 10 years' teaching experience and 76% were aged 36-55. Classes from pre-school to the end of primary school were represented, however, the largest proportion of teachers taught at the highest levels of primary school to pupils generally aged 9-11 years old (30-38%). A total of 27% of respondents had previous English language training but only 17% noted an extended stay in an English-speaking country. Similarly, a minority of respondents had received training in the use of digital technology (22% C2i; 21% C2i2e) 3 and almost none (0.5%) had received specific training in technology-enhanced English teaching.

Data Analysis
Following the descriptive analysis above, two types of statistical analysis were employed to investigate the relationship between teachers' responses to different sections of the questionnaire. Here we report at times on the full cohort of 254 respondents, and sometimes on a smaller dataset (n = 217) due to incomplete responses.
Principal component analyses (PCA) were used to examine variables related to (a) teachers' perceived pedagogical practices, (b) the integration and appropriation of digital technologies, and (c) L2 teaching beliefs as covered in the questionnaire. This exploratory factor analysis was selected with the goal of reducing the number of variables in the absence of particular expectations about the number of factors involved (Loewen and Gonulal 2015). The factorability of each analysis was verified using the respondent ratio (number of respondents divided by number of variables; min = 10 and max = 80), and testing for multicollinearity (0.1 < d < 0.7), followed by Bartlett's test of sphericity (for the main analysis, χ 2 = 639.27, df = 210, p < 0.001), and Kaiser-Meyer-Olkin (KMO) measure (0.67 < KMO < 0.82). Kaiser's criterion and scree plots were used to choose the number of factors retained, with the cumulative percentage of variance ranging from 50% to 77%. Cluster analyses (Ascending Hierarchical Classification; AHC) were then used on the same themes to identify classes or groups sharing similar views and practices. Finally, bivariate analyses (Chi-squared tests) were performed to identify significant correlations between these clusters and other teacher characteristics.

Results
The main focus of our study is teacher beliefs about language learning and teaching, and how these relate, on the one hand, to contextual variables such as age, education, and views of the role of technology, for example, and on the other to reported classroom practice. In this section, we analyse questionnaire responses in three areas: teacher beliefs about language teaching and learning; correlations between beliefs and contextual variables including teacher background, technological factors, and reported teaching activities; and finally the relationship between the teaching activities proposed and other contextual factors.

Teacher Beliefs about Language Teaching and Learning
Our analysis of teacher beliefs about language learning and teaching is based on responses to 21 statements translated in summary form in Table 2 above. The PCA analysis looks for patterns in all responses taken together: eight principal components, or main factors, emerged from the data, accounting for 61% of observed variance. These are shown in Table 3, alongside a second, AHC analysis, which examined patterns in individual teachers' responses: it revealed three clusters of teachers with distinct views of language education, and which corresponded closely to the PCA results.
The strongest correlation among the 21 items on teacher beliefs is a class of 5 items focusing on language Structure (T1, T3, T5, T6, L3). Teachers who agreed with these items believe in the importance of direct teaching and correction, using a presentationpractice-production model based on an explicit grammar syllabus and involving immediate attention to errors. A second class, also involving an explicit grammar syllabus (T3), was associated with two error items (T8, T9) and Motivation (L5). Another Error class showed a correlation between two items about the source of learner errors (T7, L8). We might consider these classes as fairly traditional or conservative approaches to language teaching predating the communicative turn of the latter part of the 20th century, but which are still alive in many of today's textbooks and classrooms.  Regarding communicative language teaching, our data also show three partially overlapping classes, all of which contrast with the previous three. The Implicit class is most distinct: these items relate to communicative language teaching from the earliest stages of classroom learning, communicative activities without systematic pre-teaching of forms, and the inclusion of unknown elements in teaching materials (T2, T4, T10). However, this class did not intersect two other overlapping classes: an Interactive class (L6, L10, L2), and a Natural class (L1, L2, T8). The remaining two classes emerging from this initial analysis are less well-defined in both statistical and conceptual terms. Teachers' views on Aptitude were related to no other items or classes, while the last Immersion class links early start (L7) and L2 only (L9). These classes seem theoretically compatible with either Structure/Error classes or Implicit/Interactive groupings, though we might have expected Immersion and Natural classes to intersect (i.e., that responses to items in one class might correlate with those in the other).
During our exploration of these data, we also ran separate PCA calculations for both the learning and the teaching sets. The learning analysis revealed four classes accounting for 56% of observed variance. The first was an Interactive class including the same three learning items (L6, L2, L10) which emerged from the combined analysis. However, none of the other correlations overlapped with the classes in Table 3. The teaching item analysis also yielded four classes accounting for 58% of variance; three overlapped with the combined analysis shown in Table 3. This analysis provides support for the Implicit class, since the first group included the same items identified (T2, T10, T4) but also included negative correlations for two Structure class items (T1, T3). This result increases our confidence in the Structure versus Implicit opposition. The second teaching class is close to the Structure one (T1, T5, T6) but also includes a negative correlation for T8 (not interrupting to correct). The third class in this analysis confirms the Error class (T11, T7).
Turning from our analysis of items to consider individual teachers' responses, we performed a cluster analysis on the slightly smaller population of 217 individuals who responded to all questionnaire items. The first cluster is composed of PPP teachers (n = 72/217 or 33%). The second is composed of teachers who espoused Communicative language teaching (CLT; 60/217 = 27%). The largest cluster (85/217 or 40%) we label 'Sceptical', since it was composed of teachers who generally disagreed with propositions. Table 3 shows that the PPP cluster is identical to the Structure class: all five items previously identified are associated with this cluster and not the others, and in four cases the association is strong. Other items strongly associated with the PPP cluster concern motivation, aptitude, and errors. Two other negative associations, with communicative teaching and the inclusion of unknown elements in teaching materials, serve to further distinguish these respondents from the CLT cluster.
The CLT cluster contains the Implicit class (strong associations for CLT, incidental learning, learning without presentation) and the Interactive class including expression, group work, and interaction (although interaction (L2) is also associated with PPP teachers). Confirmation that this cluster can be considered in opposition to the PPP cluster is provided by negative values for Structure items T1, T3, and T5. The CLT cluster is also strongly associated with error correction outwith communicative activities (i.e., not interrupting and post-remediation). Finally, both PPP and CLT clusters agreed on the importance of imitation (L1) and early start (L7) in addition to interaction (L2), while the item concerning whole-class teaching (T11) did not discriminate across clusters.
Taking these results together, the following key findings regarding the groups (classes or clusters) of teachers emerge: • Imitation-interaction: the majority of the primary teachers who responded to our questionnaire (60%) took an identifiable position with respect to their beliefs about language learning and teaching, corresponding to either a PPP or a CLT approach. However, both groups agreed on three items: they all view language learning essentially as a process of imitation, best begun early, and in which interaction plays a central role; • PPP or structured teaching: this subcategory of teachers prefers to follow a grammatical syllabus and provide a correct model for learners, using a presentation-practiceproduction approach and correcting learner errors immediately, even during communicative activities. They see errors as due to L1 transfer and likely to spread in small-group work (though no preference for whole-class activities is expressed); • CLT or implicit teaching: These teachers embrace communicative language teaching principles without necessarily presenting language items in advance or restricting materials to elements already taught explicitly. They see no need to present and practice grammar rules one by one, as indicated by negative correlations with these items, which are favoured in the Structured class. These teachers are in favour of teaching in the target language and want learners to express themselves in small group interactions with their peers. Errors should not be corrected during communicative activities but rather tackled via specific remedial exercises; • Sceptical teachers: a sizable minority of teachers (40%) positioned themselves outside the PPP/CLT dichotomy, adopting something of a "none of the above" position. It may be that our questionnaire items were too simple, admitting of many different interpretations of a particular statement, and this might discourage the purist. A second explanation is that, as generalist primary teachers with somewhat limited specific training, some respondents may not have a fully articulated view of language education and are adopting a somewhat eclectic approach to their language teaching.
Using this three-way division of teachers across CLT, PPP and Sceptical orientations with respect to language teaching and learning, we can now ask whether particular sets of beliefs are associated with other contextual variables (Section 6.2) or particular teaching practices (Section 6.3).

Correlations between Teacher Beliefs and Contextual Variables
A first set of links between teacher beliefs and contextual variables is shown in Table 4. Responses to items relating to teaching level (the last three years of primary school), teachers' own education, their proficiency in English, planning of English teaching, and practices and views with respect to technology in EFL are listed. Significant correlations (p < 0.05) with teacher beliefs (PPP, CLT and Sceptical teachers) are highlighted, and cells with none are left blank for clarity.
With regard to language proficiency and education, Table 4 shows that CLT teachers score consistently higher than the Sceptics, and PPP teachers significantly lower. A total of 77% of CLT teachers place themselves at least at CEFR level B (PPP 44%, Sceptics 68%). A total of 50% have additional training (PPP 13%, Sceptics 23%), and 27% have spent at least six months in a target community (PPP 10%, Sceptics 15%). CLT teachers were much more likely to have more advanced classes, accounting for approximately one half of CM1 and CM2 teaching, compared to around one quarter to one third for PPP teachers and over a third for the Sceptical group. There were no correlations with age, initial teacher education, or length of teaching experience.
Our data reveal that the use of digital tools for teaching English is relatively undeveloped among our respondents. Only half of the teachers regularly or frequently use online resources and 40% use videos and audio documents from textbooks. The Ministry of Education's resources are not widely used and synchronous communication via telecollaboration is very rare. Teachers cited lack of equipment and lack of training to explain the lack of digital uptake, and Table 4 shows some differences in views depending on teachers' beliefs. CLT teachers identified the main challenge in implementing technology-mediated English teaching as their own lack of competence (33%). The Sceptical group were divided over this variable (32%) and lack of time (34%), while the PPP teachers incriminated equipment (37%) followed by time (31%). Many more CLT teachers suggested other reasons for difficulties (24% compared to 5-9%). A second set of correlations concern the specific language competences targeted by respondents in their EFL classes. In the section of the questionnaire concerned with pedagogical practices, respondents were asked to choose from a closed list the teaching activities they generally used with their learners for each of the five language competences (e.g., for listening comprehension: Simon says, bingo, audio-recordings on tablets). Using the same statistical procedures, our analysis revealed the existence of three main categories in participants' responses: 26% of respondents reported a narrow range of activities, 28% used a wider range of activities but no reading or writing, while the remaining 46% used a wide range of activities covering all five oral and written competences. In Table 5, links between the same teacher variables examined earlier and these three categories of EFL teaching are shown.
We found a significant correlation between teacher beliefs and the reported range of teaching activities used with pupils, although the correspondences were never greater than 50%. A total of 46% of Sceptical teachers reported using oral but not written activities (PPP 29%, CLT 25%). A total of 47% of PPP teachers reported the narrowest range of activities (Sceptics 40%, CLT 13%), while the largest proportion of CLT teachers (37%) used the widest range (Sceptical 35%, PPP 28%). We consider teaching activities in more detail below (Section 6.3). Otherwise Table 5 shows that our respondents were more likely to use a range of oral and written activities with older pupils (56% in Years 4 and 5 as against 46% in Year 3) and if they had taken an in-service course in EFL teaching. A wide range of oral or oral and written activities was also associated with planning lessons in units, regular internet use, more frequent use of an interactive whiteboard and a class computer, and use of speakers to play audio or video material.
To summarise the findings from this analysis, we note that • CLT teachers were a relatively small but well-defined group, with strong, coherent views based on an implicit approach to language teaching. They represent one quarter of our sample. These teachers tended to have higher English proficiency and more in-service training; they were more likely to use the widest range of teaching activities and were more likely to teach higher-level classes. They saw a range of reasons for difficulties with technology, not least their own digital skills. • PPP teachers account for one third of respondents; they scored lower than both CLT teachers and the largest, Sceptical group on English proficiency and training. They reported the smallest range of teaching activities and were the least likely to teach older pupils. Their teaching/learning beliefs seem to demonstrate something of a fixed mindset and a desire for control (Dweck 2015): they focus on grammar and error correction, and see language aptitude as important. Barriers to technology integration for PPP teachers were essentially lack of equipment or time.

•
The largest group of respondents belong to a Sceptical group, which tended to disagree with the majority of our propositions concerning teaching and learning languages and therefore took no strong theoretical position. These teachers had average English proficiency (A2-B1), were more likely to teach mainly listening and speaking skills, and felt they lacked both time and skills for effective use of technology. It is interesting that such a large proportion of our sample (40%) should appear so relatively undecided on teaching/learning theory, given that our sample responded voluntarily to our questionnaire and so might be expected to be especially motivated with respect to language teaching.

Classroom Teaching Activities in Relation to Other Contextual Variables
In the final section of our analysis, we return to the connection between teacher beliefs and other contextual variables on one hand, and teaching practice on the other. We first consider activities involving listening, speaking, or interaction (Section 6.3.1), and then those which concern reading and writing (Section 6.3.2). Our analysis showed that respondents fell into three categories with respect to each of the five language competences: some teachers declared that they used very few if any of the activities suggested in our questionnaire (Group 1), others claimed to use only a subset of these activities (Group 2), and a third reported using all the activity types proposed (Group 3). 5 The full table of results with a breakdown of correlations for Groups 1, 2, and 3 with teacher beliefs and contextual variables is available in the Supplementary Materials.

Listening-Speaking-Interaction
When we consider all oral/aural/interactive activities together, the majority of respondents fell into Groups 2 and 3: 40% offered only the simpler listening and speaking activities, without interaction (Group 2), while 45% of teachers used a wide range of listening, speaking, and interactive activities (Group 3). Correlations with many other variables were significant: Group 3 had more training (44% versus 19% for Group 2) and higher English proficiency (40% B2 or higher versus 19%). They were most likely to plan lessons in units (85% versus 79%) and to work with older pupils (49-54% versus 30-32%). They tended to have better internet access (72% versus 53%) and used speakers more regularly (70% versus 53%); they also reported fewer difficulties with technical equipment (24% versus 30%) and greater self-efficacy with respect to technology use (only 22% cited lack of know-how versus 38%). Their pedagogical orientation was most frequently CLT (42% versus 29%).
Similar correlations are observable for each of the three separate language skills. For listening, Group 2 comprised 58% of teachers using a limited range of simple activities (pointing and naming objects, flashcards, ritual question-answers) while Group 3 included 19% of respondents who reported a wider range of more complex activities (reading stories, playing games, individual listening with tablets, and phonetic awareness activities). These groups correlated with teacher proficiency, pedagogical planning, and technology-related difficulties. Group 3 teachers had higher proficiency (22% level C, against 12% for Group 2). Similarly, 90% of Group 3 teachers planned lessons in units, (versus 78%); conversely 17% of Group 2 respondents designed their English teaching around unrelated learning activities (versus 6% for Group 3). Finally, the most common technological difficulty cited by Group 3 teachers was lack of time (34% versus 22%), while in Group 2 it was lack of know-how (38% versus 22%).
With speaking activities, Group 3 was the largest (43%): these teachers offered many activities (songs, poems, pronunciation practice, group work), while Group 2 (40%) tended to use only ritual question-and-answer activities. Compared to Group 1, the 17% of respondents who offered few or no speaking activities, Groups 2 and 3 were much more likely to have been teaching longer (60-61% over 10 years, compared with 40%), to teach higher levels and to plan by lesson as opposed to individual activities (71-84% against 58%). A total of 32% of Group 3 teachers estimated their level at B2 or beyond, compared with 21% for Group 2 and 5% for Group 1. Group 3 were twice as likely to have university training in English or language teaching and in-service training (22% compared with 5-13%), to have spent several months in an English-speaking country (40% versus 11-12%) and to espouse CLT principles (40% versus 5-25%). Regarding technology, this group tended to have more access to the internet (69% versus 44-57%) and to use speakers regularly (68% versus 42-56%); these teachers were also the least likely to cite insufficient competence as a challenge in this domain (20% versus 23-38%).
Concerning the teaching of interaction, Group 1 (40%, n = 98/247) reported minimal activities, while the others were evenly divided into Group 2 (memorisation only) and Group 3 (all activities). Group 3 teachers were much more likely to have had specialised pre-or in-service training (45% against 19-22%). A similar percentage taught upper levels (45-48% compared with 25-42%). Regarding technology, they used speakers much more regularly (72% versus 42-62%) and in methodological terms 39% of Group 3 teachers espoused CLT (versus 20-25%) while 40% of Group 2 respondents belonged to the PPP group.

Reading-Writing
Turning to reading and writing, fewer teachers reported working on these competences with their pupils, as we might expect in earlier stages of primary language education (Jacoby and Lesaux 2019). Considering reading and writing activities together, our analyses again revealed one group using a wide range of activities (Group 3: 48/126 = 38%) a second using a smaller number of simpler activities (Group 2: 34%); the remainder reported few reading-writing activities (Group 1). Group 3 teachers were more experienced (91% had taught for more than 10 years, versus 70-74% for the other groups) and trained at a time when teacher training colleges were largely separate from the university system (85% versus 65-75%). Group 2 teachers were more likely to be recently trained (ESPE/INSPE 22% versus 5-10%). Group 3 teachers tended to be more proficient (39% B2 or above, against 13-29%) and almost one in two had in-service training (49% versus 26-31%). Finally, we found a correlation with class level: Group 1 generally taught younger levels (CE2 64% against 29-55%) and conversely Group 3 tended to teach upper levels (CM1-2 66-71% against 35-61%).
Regarding respondents' practice of reading with pupils, the use of a wide range of activities (reading comprehension activities, picture matching, hangman) occurred only with a small Group 3 (34/168 = 20%), while a larger Group 2 (42%) used fewer, simpler activities such as comprehension questions and matching text and images. Group 3 teachers tended to have much more teaching experience (60% over 20 years, compared to 29-42%), and to have more in-service training (53% versus 25-33%). While over 80% of all teachers who responded here planned lessons in units, the Group 3 teachers were four times as likely to integrate English into other projects (12% versus 3%) and least likely to offer isolated activities (3% versus 14-17%). These teachers overwhelmingly taught the final year of primary school (73% compared with 35-54%).
Finally, concerning writing, 38% (49/130) of respondents belonged to Group 3, offering a wide range of activities, while 20% fell into Group 2, but the only correlation here was with class level. Group 3 was much more likely to teach the oldest pupils (73% versus 44-45%) and less likely to teach younger classes (CE2 32% versus 55-60%).
To summarise these results, we can note the following key points: • For each language competence, we found a pattern where some teachers used a wide range of activity types, some a more limited, basic range, and some none at all. A wider repertoire of teaching activities was often associated with higher English proficiency, more training and teaching experience, and pedagogical planning at the level of unit or project rather than activity. • There was no correlation between teacher age and teaching activities proposed, and little correlation with pre-service training. The influence of teacher education on teaching activities is seen at the in-service level, such that teachers who had taken courses in EFL teaching since graduation tended to offer a wider range of activities; • Upper levels of primary education tended to be offered a wider range of activities related to reading and writing, and interaction; for listening and speaking this variety in activities was available at all levels.

•
Correlations between teaching activities and both technological environment and teacher beliefs were apparent only with respect to oral/aural/interaction and not reading/writing competences. Teachers who used a broad range of activities in interaction tended to have better access to technology and reported more regular use and perhaps higher self-efficacy; they were also more likely to espouse CLT principles than their peers.

Discussion
Returning to the research questions which guided our study, we are now in a position to provide some overall answers. Regarding the question of French primary teachers' beliefs about language learning and teaching, the approximately 250 primary teachers in our survey fell into 2 broad categories: 60% consider languages to be learned through imitation and interaction, and favour an early start to instruction. These teachers then take one of two positions on teaching: either they favour a structured approach using a grammar syllabus with a strong focus on error correction, or they adopt communicative language teaching principles based on learner exchanges in small groups and a more flexible approach to learning materials and activities. Both these subgroups are distinguished from a third group of teachers representing 40% of our sample, who expressed disagreement with all the above-mentioned principles. As noted earlier, we were struck by the size of this last sceptical or agnostic group given the fact that participants responded to the questionnaire voluntarily and might therefore be expected to be more interested and involved in English language teaching than average. We speculate that these teachers may be working from a somewhat atheoretical, eclectic perspective, and/or might take a more nuanced position on the necessarily very direct, simple propositions in our questionnaire. The very broad descriptors of the official programmes which offer little guidance on language education theory may also play a role Valax (2011).
With respect to our second research question concerning potential links between teacher beliefs about language education and their reported classroom practice, we found a number of correlations which suggest that teachers' beliefs affect the types of activities they are likely to offer their learners, as Borg (2003) and others have suggested. When we considered all five language competences together, we found that the teachers who offered the widest range of activities were more likely to espouse CLT principles, while a majority of those who reported a limited, basic activity set were PPP teachers. Between these two extremes, teachers who proposed mainly listening and speaking activities (to the exclusion of reading and writing) were most likely to fall into the third group of sceptical practitioners. However, the representation of each attitudinal profile was only between one third and one half of all teachers reporting a particular pattern of teaching activities. This suggests a large degree of variation in reported practice. Thus, while there is a clear relationship between what teachers say they believe about language learning and teaching and what they say they do in the classroom, the correspondence is far from one-to-one.
Our third research question addressed connections between these teacher beliefs about primary EFL and reported practices, on the one hand, and other contextual variables such as teacher age, education, experience, and technological factors on the other. As we have seen, we found no correlation between teacher age and initial education (also a proxy for age) and either beliefs about language learning and teaching or reported classroom practice. However, our data do show an effect for in-service training and English proficiency. Across the board, teachers who offered a wider range of activities in any of the five competences tended to have more in-service training and higher proficiency: more of these teachers had taken specialised courses, spent time in English-speaking countries, and reported a language level of at least B1, often B2 and above. Regarding pedagogical organisation, the teachers who used a greater variety of activities also generally taught upper-level classes and planned their teaching as lessons in a teaching unit, rather than activity by activity.
We also found correlations between reading and writing activities and experience: that teachers who offered a wide range of reading/writing activities were overwhelmingly the most experienced teachers (over 10 years' classroom experience); there were no interactions here with either teacher beliefs or technological environment. Our survey reveals limited use of digital tools and resources overall, with only around half of teachers using online resources with any regularity and some 40% using the video and audio resources included in textbooks. Use of individual devices (e.g., tablets) and videoconferencing are extremely rare. The teachers who used technology most (e.g., internet and speakers) offered the widest range of oral activities, and tended to report fewer problems with equipment, and more frequent use of the internet and speakers. They were more likely to cite lack of time as an obstacle to technology use rather than technical problems, their own self-efficacy, or pupil proficiency. Those who blamed their own technical skills for lack of use were most likely to be CLT teachers, while deficient equipment was most frequently incriminated by those with a PPP profile. Previous research (Ertmer et al. 1999;Mama and Hennessy 2010) has shown that despite limited access to technology, some teachers still try to exploit the available resources in a pedagogically effective way while others make no attempt to use them-the essential difference lying in a differential appreciation of technology's role in teaching and learning. Here we might suggest that a more flexible pedagogical approach may also support a can-do attitude with respect to technology integration.

Conclusions
In conclusion, we consider some limitations of this research alongside implications for teacher education and perspectives for ongoing research. One question our analysis of correlations among teacher beliefs, classroom practice, and contextual variables is unable to answer concerns directionality or causality. We do not know, for example, whether teachers' reported preference for CLT is a consequence of their higher proficiency in English and additional in-service training, which might increase confidence with more challenging, open-ended activities, or whether this belief might have caused them to seek opportunities to improve their own language and pedagogical skills. Regarding technology, as just noted, do teachers report limited use and low self-efficacy because their classrooms are poorly equipped, or are classrooms where technology is present and well-used the result of bottom-up efforts by practitioners with particular views about the development of speaking/listening skills? Our project addresses this issue with semistructured interviews of a number of questionnaire respondents where teachers are invited to expand on their answers. A second limitation is common to all survey data in its reliance on self-reporting. Teachers may have felt a need to conform to certain expectations, particularly since our project involves institutional actors. Respondents may have answered based on their perception of priorities in the national programmes or teacher education instead of reporting genuine beliefs (Di Santo et al. 2017). To examine the relationship between self-reporting and in-class practices (Farrell and Guz 2019) our project also works with volunteers among the respondents to develop teaching materials and observe their implementation with pupils.
Nevertheless, the findings with respect to teacher beliefs about language learning and teaching revealed in our study appear to be particularly robust. The data from our sample of over 200 practising teachers from 3 separate educational authorities in France reveal 3 distinct teacher profiles: CLT, PPP, and 'Sceptics'. In our ongoing project we are working with these profiles in designing templates for the co-construction of pedagogical scenarios with our teacher participants. It seems clear that teachers fitting each of these three profiles are likely to have different strengths and weaknesses in their planning and implementation of English activities and may respond differently to various types of support and guidance. This last point raises the wider question of teacher education for EFL at primary level in France and no doubt a number of other European countries with similar teacher populations and institutional settings. Our figures suggest that around one third of those teaching English in primary schools today are structure-oriented teachers working with a rather restrictive view of language education, which has repercussions for both the range of activities offered to learners and the way these activities are implemented in class. Again, the issue of directionality seems relevant: do teachers espouse PPP beliefs because it offers a reassuringly manageable framework when confidence in their own language skills and pedagogical know-how is low? Or is their own lack of proficiency and confidence a result of a restricted view of what language competence entails and how languages are actually learned (as opposed to how they are often taught in school settings)? Moodie and Feryok (2015, p. 466) suggest that "persistent efforts to learn language" among their generalist primary EFL teachers in Korea "may become persistent efforts to learn to teach language". This question raises the intriguing possibility for pre-service teacher preparation of simply focusing on the language skills of future teachers, rather than providing theoretical background in an effort to change beliefs about language education. It may be that the best way to foster teacher development in this area is to support future EFL primary teachers in their own journey with the target language as a model for the language education dimension of their future careers. Funding: The digital thematic groups (GTnum), led by university research laboratories and other public operators, are coordinated by the office of support for digital innovation and applied research of the Direction of Digital Education (subdirection of digital transformation-TN2 office) of the Ministry of National Education, Youth and Sports (grant number 2103081117). The GTnum's objective is to make available to educational teams, in an accessible and open way, state-of-the-art research on some major themes related to digital education.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of Université Clermont Auvergne (protocol code IRB00011540-2021-49: date of approval 11 June 2021).

Informed Consent Statement:
Informed consent was obtained from all participants involved in the study.

Data Availability Statement:
The full questionnaire, data, and data analysis will be deposited in an open-access repository prior to article publication and information will be detailed here.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

1
The successive years of French pre-school and primary school education are referred to by two-letter acronyms: GS (grande section) is the last year of pre-school, CP (cours préparatoire) is the first year of primary school, followed by cours élémentaire (CE) 1 and 2, and cours moyen (CM). Cycle 3 also includes the first year of lower-secondary school. 2 Ministry of Education statistics put the general proportion of female teachers in French primary schools at 85.6% and their average age at 41.6 years (MENJS 2019). 3 The Computer and Internet Certificate (C2i) certifies the level acquired by a student in mastering multimedia tools and the Internet. The C2i2e certifies professional skills in the pedagogical use of digital technologies for teachers and trainers. The activities proposed in the questionnaire were selected after discussion with local EFL support staff based their experience of classroom observation and teacher education.