Abstract
It is commonly agreed that learner target language output and spoken interaction are essential to communicative language learning. This video-based classroom observation study of five lower secondary schools in Norway investigated how second language (L2) English and third language (L3) French teachers responded to their students when they spoke the target language in class. Using categories and functions from classroom interaction research and corrective feedback research, L2 English and L3 French teachers’ response patterns to learner target language output were identified and compared. Findings indicated that L2 English teachers taught in a meaning- and fluency-oriented context, frequently ignoring errors and providing ample content-related responses, while L3 French teachers taught in a form-and-accuracy context, using varied strategies to correct errors and rarely engaging with the content of the learners’ utterances. The article discusses some consequences of this lack of context shifts within subjects and recommends an increased focus on contextual differences to facilitate mutual transfer of successful practices between contexts.
Keywords
I Introduction
It is commonly agreed that learner target language output is essential for communicative language learning and the development of oral skills. According to Swain (2005), learner output serves at least three functions: It makes learners notice what they actually produce and what they intend to produce; it helps them test their hypotheses about how the target language (TL) works; and it pushes them to reflect on language structures that are not yet fully acquired. For all these functions to be fulfilled, it is paramount that learners’ output be part of interactions with other users of the language and that learners receive some kind of feedback or response to their utterances. In the language classroom, this response usually comes from the teacher in what is often called a ‘follow-up move’.
The well-known Initiation–Response–Follow-up (IRF) pattern (Sinclair & Brazil, 1982) can account for most classroom discourse, regardless of subject (Walsh & Sert, 2019): the teacher initiates an interaction in the first turn (often by asking a question); the learner responds in the second turn; and the teacher follows up on this response in the third turn, often with an evaluative comment. This pattern has also been referred to as Initiation–Response–Feedback (Sinclair & Coulthard, 1975) and Initiation–Reply–Evaluation (Mehan, 1979). However, the third turn can comprise non-evaluative acts, such as asking for justification or clarification, or providing a comment (Nassaji & Wells, 2000; Sinclair & Coulthard, 1975).
Evaluative third turns – particularly those involving corrective feedback (CF) – have received particular attention in second and foreign language research (Waring, 2008). A smaller body of research examined other types of third turns, often in the form of micro-level analyses of selected interactional exchanges in specific classrooms, using a conversational analytical approach. Such studies advance the field’s understanding of interactional patterns that can occur in language classrooms, but they rarely offer a broader picture of the classroom interaction environment, and they do not examine variation across educational contexts and instructional settings. However, awareness about contextual variables and their importance has been increasing in SLA research (Valeo, 2021). Recent CF research indicates that the CF types that teachers provide and the effect that they exert, vary with factors such as second vs. foreign language instruction, learner TL proficiency, and learner age (Brown, 2016). These factors influence not only what kind of CF is provided and its effect, but also TL classroom interactional patterns in general, including amounts and types of learner output and teacher (or peer) responses. However, to our knowledge, no study to date has systematically investigated how TL classroom discourse unfolds differently across instructional settings. During a time of increasing pressure from educational policy makers to answer the somewhat naïve question of ‘what works’, gaining more knowledge about the characteristics of different types of language instructional settings and how these characteristics affect the language classroom’s perceived affordances is essential in research.
The present study sets out to investigate TL classroom discourse in two different, yet similar, settings: second language (L2) English and third language (L3) French classes in lower secondary schools in Norway. 1 It examines how much and what kind of learner output is produced in these classrooms, as well as how teachers approach this output, be it through CF or other types of verbal response. The aim is to identify the characteristics of L2 English and L3 French classroom discourse, as well as contribute to an understanding of how instruction in these two settings differs, and what they can learn from each other.
II The educational context: L2 English and L3 French in Norway
English is the first foreign language (L2) that students encounter in Norwegian schools. The subject is mandatory from first to 11th grade (ages 6 to 16), and out-of-school exposure to English is considerable (Brevik & Hellekjær, 2018). Studies indicate that Norwegian students have high English proficiency compared with their peers in other countries (Bonnet, 2004; Education First, 2021). However, Hellekjær (2010b) pointed out that while Norwegian students have well-developed Basic Interpersonal Communication Skills (BICS) in English, they have limited Cognitive Academic Language Proficiency (CALP) (Cummins, 1984). As a result, many university students struggle with academic English (Busby, 2021; Hellekjær, 2005, 2010a).
Lower secondary schools in Norway must offer at least one second foreign language (L3) as an optional subject starting in eighth grade (age 13). The most frequently offered languages are French, German and Spanish. To qualify for studies in higher education, students must study an L3 sometime during secondary education. About 75% of eighth-graders choose to study an L3, and about 12%–14% study French (Doetjes, 2018). The presence of French in Norwegian society is low. Unless the learners deliberately seek French input, e.g. through web-based resources, they will rarely encounter French outside school.
National policy documents state that the overarching approach to L2 and L3 teaching is communicative. Subject curricula are inspired by the Common European Framework of Reference for Languages (Council of Europe, 2001) and promote practical language use and communicative competence (Ministry of Education and Research, 2004; Norwegian Directorate for Education and Training, 2020). They include a focus on language as well as content. Thus, the underlying theoretical principles are similar for L2 English and L3 French teaching.
The roles of these languages in Norwegian society, however, are very different. According to Rindal (2014), the status of English in Norway comes close to that of an official second language, mirroring the situation in Sweden (Bardel, Erickson & Österberg, 2019) and Denmark (Fernández & Andersen, 2019). French has no prominent role in Norwegian society and represents a truly foreign language for most students.
The terms second language and foreign language can be understood in different ways (Nayar, 1997; Sato & Storch, 2020; Valeo, 2021). Nayar (1997) distinguished three usages of the concept English as a second language (ESL); the first referred to English for immigrants in English-speaking countries, the second to English in former British colonies, and the third to the Scandinavian context, in which English plays an important role, although it has never had the role of an official language. The term ‘foreign language’ (FL) can refer to a language taught in a country where it has no official or important role, or to a language taught in a context that offers limited exposure to the TL. Nayar (1997) asserted that the frequently used acronyms SLA (second language acquisition) and FLL (foreign language learning) underscore the view that a SL is acquired (through exposure in naturalistic settings) and a FL is learned (in classrooms). In line with Storch and Sato (2020, p. 54), I will use exposure (understood as availability of TL input and possibilities for practice) rather than official status to distinguish between SL and FL settings: a SL context is ‘a context where learners have plentiful opportunity to be exposed to the L2 and an authentic need to communicate in the L2 inside and outside the language classroom’, whereas a FL context is an ‘environment where learners lack sufficient exposure to the target language’. Thus, L2 English in Norway is taught in a SL context and L3 French in a FL context.
III Literature review
TL use varies considerably across instructional settings (Ellis & Shintani, 2014). Previous empirical research from Norway indicated that TL is used extensively in secondary school L2 English classrooms (Brevik & Rindal, 2020) and somewhat less in primary schools (Krulatz, Neokleous & Henningsen, 2016). With L3 instruction, TL use is very limited (Askland, 2018; Vold & Brkan, 2020). None of the studies on TL use in Norwegian classrooms examined how teachers respond to learner TL output when first produced, and this is the present study’s aim. Two research strands are particularly relevant in this respect: classroom interaction research and CF research.
1 Interactional patterns in second and foreign language classroom discourse
The triadic IRF structure has been identified as the dominant pattern in second and foreign language classrooms (Thoms, 2012). The teacher is responsible for classroom interaction and, thus, the large majority of initiations (Walsh & Sert, 2019). According to Walsh (2011), this explains why teachers generally speak more than students: for each student reply, the teacher has two turns: initiation and follow-up.
Cullen (2002) distinguished between ‘evaluative’ and ‘discoursal follow-up moves’, affirming that both could support learning. Evaluative follow-up moves comprise positive or negative feedback, i.e. praise or error correction, while in discoursal follow-up moves, the teacher elaborates on learner contributions and uses them to extend the discussion and encourage further participation from the entire class. Although discoursal follow-up moves may include an implicit correction of the learner’s utterance, the primary focus is on meaning (Cullen, 2002). Based on transcripts from a secondary school lesson of English as a foreign language (EFL) in Tanzania, Cullen (2002) identified several strategies that the teacher used to follow up on learner contributions, such as reformulations, repetitions, elaborations, and personal comments.
Walsh (2002, 2011) coined the term ‘shaping learner contribution’ to describe the different manners in which a language teacher can respond to learner output. Shaping learner contribution involves ‘taking a learner response and doing something with it rather than simply accepting it’ (Walsh, 2011, p. 168). It encompasses repair, 2 but also strategies such as scaffolding, paraphrasing, reiterating, summarizing and extension, checking confirmation, and asking for clarification, justification and elaboration. Through scaffolding, teachers mediate the development of learners’ oral skills by providing relevant vocabulary or structures whenever the learner hesitates (Walsh, 2002). When asking for clarification, justification and elaboration, teachers elicit more learner output. Walsh (2011) stressed the fact that each of these strategies may serve several functions. As shown in Walsh and Li (2013), reformulations may serve not only to correct errors implicitly, but also to ensure that other students understand as well. Such double functions are essential in whole-class discourse, in which teachers ‘perform for all of the students rather than for only one of them’ (Daskin, 2015, p. 35).
Based on an analysis of classroom recordings of eight teachers’ EFL lessons, Walsh (2002) concluded that strategies for shaping learner contributions constituted an important tool for creating spaces for learning in the language classroom, but that they also could obstruct learner participation when used inappropriately. In a classroom video study of two EFL lessons in China, Walsh and Li (2013) found that teachers facilitated learner participation by reformulating, seeking clarification, pushing for more information, and ignoring errors that did not impede communication. They concluded that shaping learner contribution ‘has major import for the co-construction of meaning and for enhancing learning opportunities’ (Walsh & Li, 2013, p. 262).
Further studies that have addressed the shaping of learner contributions have confirmed and extended the concept. Using audio- and video-recorded data from six lessons in a Turkish university EFL class, Daskin (2015) demonstrated how a teacher created space for learning through many of the techniques that Walsh (2002) identified, as well as through strategies such as translation and writing on the board.
In a study of repair based on a varied database of language lessons from around the world, Seedhouse (2004) distinguished four classroom contexts depending on the interaction’s pedagogical focus. Two of these are of particular importance to the current study: the form-and-accuracy context and the meaning-and-fluency context. Seedhouse (2004) found that these contexts triggered different kinds of teacher response, and that repair types and foci would change as teachers shifted from one context to another. The impact of classroom context on teacher feedback has been demonstrated in several further studies, e.g. Park (2014). Using video- and audio-recorded data from ESL and EFL classrooms in the U.S. and Korea, respectively, Park (2014) specifically examined third-turn repeats, finding that their role differed depending on the interaction’s pedagogical focus. In meaning-and-fluency contexts, third-turn repeats were used to encourage learners to elaborate on their responses, while in form-and-accuracy contexts, they confirmed to the rest of the class that the learner’s response was accurate.
Specific discourse markers that often occur during the third turn have also been examined. Girgin and Brandt (2020) examined how minimal responses (such as uh-huh) could create space for learning and encourage learners to continue their turn. Waring (2008) studied positive assessment markers (such as good, excellent) and found that these tended to obstruct learner output because they marked the end of the interaction. Sert (2013) studied teachers’ responses to learners’ claims of insufficient knowledge (‘I don’t know’ responses) and found that these often took the form of ‘epistemic status checks’ (‘You don’t know?’) before the turn was allocated to another learner. Non-verbal teacher responses – such as wait-time (Alsaadi & Atar, 2019; Yaqubi & Pourhaji Rokni, 2013), facial expressions and gestures (Sert, 2013; Wang & Loewen, 2016) – have also been studied.
Although the IRF pattern still seems to be dominant, Walsh (2011) pointed out that it was derived from data collected in traditional classrooms in the 1960s, and that the pattern might not adequately describe contemporary classrooms’ communication structure. In a study of interactional exchange in a secondary French classroom in the UK, Crichton (2013) observed that this was particularly true for what she called ‘conversation-type sequences’: exchanges which can occur in classrooms where the TL is the language of communication and, thus, used for relational, social and instructional purposes. Conversation-type sequences resemble spontaneous real-world communication, but they do not have an overt pedagogical function, although they may be used to develop learners’ communicative competence (Crichton, 2013). Most of the exchanges in Crichton’s study could be described using the IRF structure, but alternative analytical frameworks were necessary to capture conversation-type sequences adequately.
2 Corrective feedback in second and foreign language classrooms
A substantial body of research has been conducted on oral CF in language learning, and several taxonomies of CF strategies have been elaborated following Lyster and Ranta’s (1997) seminal study, which identified six different CF types. Sheen and Ellis’ (2011) taxonomy is frequently used in recent research and will also be used in the current study. It builds on Lyster and Ranta’s (1997) original classification, but is set up slightly differently. It divides CF moves into categories along two dimensions: input-providing vs. output-prompting strategies and implicit vs. explicit strategies. Input-providing strategies provide the learner with the correct form, either as an explicit correction (with or without a metalinguistic comment) or a recast, which can be conversational (meaning-focused and used to resolve a communication problem) or didactic (correction-focused). Conversational recasts are classified with implicit strategies and didactic recasts with explicit strategies. Output-prompting moves push the learner to self-correct through clarification requests or repetitions (implicit strategies) or paralinguistic signals, metalinguistic clues, and elicitations (explicit strategies).
Lyster, Saito and Sato (2013) challenged the dichotomous distinction between implicit and explicit strategies, pointing out that recasts appear more or less salient depending on the teacher’s use of stress or intonation. Placing the different CF types on a salience continuum, conversational recasts and clarification requests were placed at the implicit end of the scale and strategies involving metalinguistic comments at the other end. In contrast to Sheen and Ellis’ (2011) classification, didactic recasts were placed further towards the implicit end.
Previous empirical studies include descriptive studies related to instructional practices or teacher and learner attitudes and effect studies related to different CF types and their impact on learning. Brown (2016), synthesizing results from descriptive classroom observation studies of CF practices, investigated factors that might affect teachers’ choices. Meaning-oriented vs. form-focused instruction proved to be an influential factor; teachers provided a greater proportion of prompts in form-focused instructions than in meaning-focused contexts. In a review of oral CF practices in SL and FL settings, Mayo and Milla (2021) argued that the difference between meaning-oriented and form-focused instruction often amounted to a difference between SL and FL settings. 3 They found that FL teachers tended to focus more on form than SL teachers, and thus, they used more explicit CF strategies.
Recasts seem to be frequent in many instructional settings (Brown, 2016; Llinares & Lyster, 2014; Milla, 2017). Comparative empirical studies have shown, however, that the nature of recasts is different in FL and SL settings: in FL classrooms, recasts are didactic and explicit, while in SL settings, conversational and implicit recasts dominate (Llinares & Lyster, 2014; Milla, 2017). Llinares and Lyster (2014) found that recasts in Canadian French SL classrooms were often indistinguishable from non-corrective repetitions and followed by topic-continuation moves and praise, thus constituting a very implicit type of CF.
The findings from these comparative studies were corroborated by Ha and Murray’s (2020) study on Vietnamese primary EFL classrooms. The EFL teachers frequently used didactic recasts but rarely conversational recasts. The recasts were often partial, i.e. reformulating an isolated part of the learner’s utterance instead of the entire learner turn. Isolated recasts are more salient than full, integrated recasts that form part of the communication flow (Ha, Murray & Riazi, 2021a). Ha and Murray (2020) asserted that isolated recasts are very similar to explicit correction. Sheen (2006) also argued that short, partial and repeated recasts were more explicit and led to more learner uptake than full conversational recasts.
Teachers’ preference for recasts was confirmed in recent studies on in-service (Ha & Murray, 2020; Kamiya, 2016; Milla, 2017) and pre-service teachers (Kartchava et al., 2020). Researchers explained this preference by the fact that recasts do not disrupt the communicative flow and are face-saving; they save the learner from the potentially negative experience of being corrected in front of others (Kamiya, 2016; Li, 2017; Roothooft, 2014; Roothooft & Breeze, 2016). However, FL teachers do not seem concerned about students’ affective responses to explicit CF (Ellis, 2017; Ha, Nguyen & Hung, 2021b). In form-focused classrooms, teachers and learners see CF as a natural part of classroom instruction (Ha et al., 2021a, 2021b; Li, 2010; Sheen, 2004). This shared view might explain why FL teachers tend to correct more errors than SL teachers (Mayo & Milla, 2021; Milla, 2017). Overall, teachers’ fear of ‘humiliating’ learners through overt correction seems to be exaggerated. Studies from a broad range of contexts have found that learners want to have their errors corrected (Ha et al., 2021a, 2021b; Jean & Simard, 2011; Kartchava et al., 2020; Roothooft & Breeze, 2016; Yan & Beilei, 2019).
The differences in CF preferences between FL and SL teachers might not be an inherent difference between FL and SL instruction but rather a difference in lesson orientation or task type. In a comparative study of CF provision in four different classroom settings, Sheen (2004) found that one of the SL classrooms shared many characteristics with the FL classroom, in particular an orientation towards language form. The results for this classroom were more similar to the FL classroom than to the other two SL classrooms. In a video-based study of teacher feedback in nine university-level Spanish-as-a-foreign-language classrooms, Gurzynski-Weiss and Révész (2012) found that teachers corrected more errors overall when students were engaged in form-focused activities compared to meaning-based tasks. They concluded, in line with Seedhouse (2004), that the communicative orientation of classroom discourse contexts largely affected teacher CF patterns.
Brown (2016) found that TL exerted no notable impact on teachers’ CF practices. However, Simard and Jean (2011) suggested that the teaching traditions of different languages might affect teachers’ focus on forms. Comparing the amounts and types of form-focused instruction in English and French SL classrooms, they found that more time was devoted to form-focused instruction in the French classrooms, although not to CF in particular. They explained the difference by pointing to the historical importance given to accuracy in French SL classrooms.
Students’ proficiency level also seems to influence teachers’ CF choices. Brown’s (2016) meta-study showed that higher proficiency students received a greater proportion of recasts than did lower proficiency students. This aligns with research findings that showed that high-proficiency students benefit more from recasts than low-proficiency students (Ammar & Spada, 2006; Mifka-Profozic, 2020).
The effect of different CF types on different learner groups is, however, difficult to determine. An extensive body of research has examined the effectiveness of oral CF practices, and several meta-studies have reviewed the results (Li, 2010; Lyster & Saito, 2010; Mackey & Goo, 2007; Nassaji, 2016; Russell & Spada, 2006), which often pointed in different directions. Although some of the variations might be explained by methodological differences, Nassaji (2016) stated that the relationship between CF and uptake is complex and may vary depending on individual and contextual factors. After a period of investigating the effects of different CF types on specific target structures, yielding inconclusive results, researchers have begun to take more interest in how feedback types work together and with other instructional variables in different instructional settings. Moreover, although classroom intervention studies or experimental studies conducted in laboratory settings inform acquisition research, their direct pedagogical implications are limited, since laboratory settings and intervention-controlled classrooms are very different from real-life classrooms (Nassaji, 2016; Thoms, 2012). After all, CF strategies rarely occur in isolation in real-life teaching situations: teachers often combine different CF types and other types of feedback in the same turn (Lyster et al., 2013). The current study follows in this holistic vein by investigating not only CF but also any kind of teacher response to learner utterances.
To summarize, it seems clear that classroom interactional patterns vary based on the pedagogical aims of the specific activity, class or programme. While recent research on repair and CF increasingly has focused on contextual factors and how they affect practices and outcomes (e.g. Gurzynski-Weiss & Révész, 2012; Seedhouse, 2004), there is also a need of studies that investigate how teachers overall respond to learners’ spoken TL output across settings and how the classroom’s communicative orientation affects teacher response strategies generally. CF has elicited considerable research attention compared with other teacher response types (Ellis, 2017), but in communicative language teaching research, we should be interested not only in how learners’ spoken output is corrected, but also more generally in how learners are met when they speak the TL. This article seeks to fill this gap by identifying recurrent patterns in teacher responses to learner TL output across two different settings. The following research questions guided the investigation:
Research question 1: (1a) How much and (1b) what kind of learner TL output is produced in L2 English and L3 French classrooms?
Research question 2: How do teachers respond to this output?
IV Methodology
The present study was conducted within the larger Linking Instruction and Student Experiences (LISE) project at the University of Oslo, Norway (LISE, 2017). From 2015–2018, LISE collected video recordings of instruction in six subjects in seven lower secondary schools in Oslo and surrounding areas. Surveys tapping into students’ perceptions of the instruction were also conducted. For the present study, I used data from 10th grade English and French lessons. Five schools were filmed in both these subjects in 10th grade and are included in this study.
Invitations to participate were sent to the schools’ administrators, and upon approval, information meetings were held with teachers. Participating teachers, students and their parents all signed written consent forms. The Norwegian Centre for Research Data approved the study.
1 Participants and data collection
Table 1 provides school and teacher background data as well as an overview of the video material. The participating schools’ demographic (demogr) and socio-economic status (SES) varied, as did the teachers’ levels of formal subject competence and years of teaching experience (YTE). The English classes had more students than the French classes. The learners were proficient users of English, as they began learning the language during their early school years and had extensive out-of-school exposure to it. They were beginners in French, as they were in their third year of French studies, with only two-to-three teaching hours a week and limited out-of-school exposure.
Overview of school and teacher backgrounds and lessons recorded.
Notes. * Small = less than 10 students, medium = 10–20 students, large = more than 20 students. ECTS = European Credit Transfer and Accumulation System.
Two wall-mounted cameras recorded lessons: one faced the class, and the other faced the board. The teacher wore one microphone, and a second one was placed in the middle of the room to capture students’ voices. This set-up did not fully capture learner-to-learner interaction unless the teacher was standing nearby, but it was sufficient for studying teachers’ responses to learner output in whole-class and in group/pair discussions. A member of the research team was present in order to monitor the recordings and administer the survey, but did not intervene in the instruction.
We used a validated Norwegian version (Klette, Blikstad-Balas & Roe, 2017) of the Tripod Student Perception Survey developed by Ferguson (2010). This survey taps into students’ perceptions of the instructional quality of the classroom, assessing content and pedagogy as well as relations. The students responded to items on a five-point Likert scale. They filled in one questionnaire for each subject teacher. For the current study, I analysed three items that directly addressed interactional patterns and error correction (see below).
Previous studies suggest that four consecutive lessons provide sufficient information to obtain an overview of teaching practice (Klette et al., 2017). Accordingly, we recorded four consecutive lessons in each subject at each school. The lessons’ duration varied between schools, ranging from 40–70 minutes. Altogether, 41 lessons are included in the data set, totalling approximately 38 hours of teaching.
2 Data analysis
This study’s research approach is qualitative video-based classroom observation. Teachers’ responses to learners’ output are grouped into different categories based on their content and functions. Quantifications are included to identify response patterns in the two subjects; see Maxwell’s (2010) recommendations for the use of quantifications in qualitative research.
The unit of analysis is teacher responses to learners’ TL utterances; thus teachers’ initiation moves and learners’ responses to teachers’ feedback or follow-up moves are not part of the analysis, but are included in the examples to provide context to the turns examined.
Data analysis took place in six phases, from preparation of the material via coding and counting to reliability testing and triangulation. Each of the phases will be presented below.
Phase 1: Preparing for analysis and transcribing the material
I watched the videos to get an overview of the material and decide on the necessary amount of transcription. Considering that the English lessons were conducted primarily in the TL (Brevik & Rindal, 2020), I opted for transcriptions of entire lessons in this subject. In the French lessons, TL use was very limited (Vold & Brkan, 2020), and therefore only relevant (i.e. French-medium) parts were transcribed. Research assistants transcribed the English lessons. I verified these transcriptions and transcribed the French TL utterances. Orthographic transcription was viewed as sufficient for this study’s aim. In the examples given below, stress is marked when relevant, and pronunciation features of interest to the analysis are explained, or rendered through IPA symbols. Transcription conventions are provided in Appendix 1.
Phase 2: Coding learner TL turns
This study addresses teachers’ responses to learner TL use, and thus focuses – in line with Lyster and Ranta (1997) – exclusively on teacher-student interaction. I included all audible learner TL turns that captured the teacher’s attention. According to van Lier (1988, p. 103, cited in Lyster & Ranta, 1997, p. 59), ‘one of the prerequisites for a turn to count as a turn is prominence, i.e. it must be attended to by the other participants’, in this case the teacher. The teacher could attend to a turn by giving a verbal response or by standing next to a student group listening and observing. Similarly to Lyster & Ranta (1997), I excluded turns that comprised only one word such as ‘oui/yes’, ‘merci/thanks’ and ‘ok’. I also excluded turns uttered in chorus, such as ‘Bonjour Madame!’ in the beginning of class.
The learner turns were classified as being either linguistically accurate (target-like utterances) or not (non-target-like utterances). The distinction between the two was relevant because the nature of the teacher’s response is likely to vary depending on whether or not the learner made errors. 4 A similar procedure was used in Lyster and Ranta (1997) and Gurzynski-Weiss and Révész (2012), though the question of what counts as an error was not discussed in any detail. However, the answer to this question is far from straightforward, particularly for oral communication and regardless of whether pronunciation, grammar, or lexical choice is concerned. Regarding pronunciation, I did not try to distinguish between features such as voiced and unvoiced fricatives: first, the audio quality was insufficient to determine such nuances in pronunciation, and second, deviations from the standard in this respect are not necessarily viewed as errors (compare the lingua franca core; Jenkins, 2002). As for grammar, I considered utterances in which the finite verb was not correctly conjugated as erroneous, even though such forms can be heard in native speakers’ everyday language (e.g. he don’t). Such use would traditionally be viewed as non-target-like in a school setting. Regarding lexical choice, I distinguished between words that clearly were inadequate in the context and words that were not optimal concerning register and style, treating only the first ones as non-target-like. Extract 1, in which the learner used the word ‘land’ for ‘country/state’, illustrates such non-target-like use.
Extract 1: T: Did the exercises change the way you feel about death penalty? L: Yes. I didn’t know so much about it, and I didn’t know it was allowed in so many lands . . .
An example which was deemed target-like despite a questionable lexical choice occurred during a task in which the students described each other in French. A boy said about his male classmate: Tu es jeune et très mignon. The teacher laughed (in a friendly manner), probably because male classmates do not usually describe each other using the word ‘mignon’ (cute).
In addition to utterances that include non-target-like pronunciation, grammar or word choice, as explained above, turns in which the learners hesitated to the point where the teacher came to their assistance (either on the teacher or learner’s initiative) were counted as being non-target-like.
Phase 3: Procedures to answer research questions 1a and 1b
TL learner turns were counted, and the number of words for each was registered using the word count in Microsoft Word. 5 This information was used to calculate average turn length. 6 To answer research question 1b, I read the transcripts to identify the topics discussed when learners used the TL.
Phase 4: Procedures to answer research question 2
I scrutinized all learner utterances to see what kind of teacher responses they triggered. For responses other than CF moves, the process of defining and selecting categories is best described as abductive. Such responses were previously analysed mainly through data-driven approaches such as conversation analysis, which rejects the use of pre-determined categories (Walsh, 2002). Thus, no pre-conceived taxonomy was available. I let categories emerge from the data, but unavoidably had in mind categories previously used in the literature, such as Cullen’s (2002) list of discoursal follow-up moves and conversation analytical categories such as scaffolding, elaboration, extension, and confirmation checks (see Walsh, 2002, 2011), of which several overlap with categories from Cullen’s (2002) list. An overview of the final categories that were used can be found in Table 2.
Teacher responses to target-like turns in L2 English (n = 916) and L3 French (n = 95) lessons.
The analysis of CF moves was more theory-driven. I used Sheen and Ellis’ (2011) taxonomy, with three small adjustments. First, since the distinction between implicit and explicit strategies is more of a continuum than a dichotomy and perhaps more dependent on contextual variables than on CF type itself (see discussion above), I did not categorize CF types as either implicit or explicit. Instead, the degree of explicitness will be commented upon in Section V. Second, although paralinguistic signals also play a role in teachers’ oral CF, I did not include them in the analysis since the current study focuses on verbal responses. Third, all instances of metalinguistic clues in the data functioned as elicitations, and instead of distinguishing between metalinguistic clues on the one hand and elicitations on the other, I divided elicitations into occurrences with or without metalinguistic information, in parallel with the input-providing strategies explicit correction with and without metalinguistic information.
In instances of multiple follow-up moves (i.e. several response moves for one turn), I coded the extract with all relevant categories (examples are provided in Section V). This approach differed from the one used for multiple CF moves in Choi and Li (2012), Ha and Murray (2020), and Lyster and Ranta (1997), who coded such instances with the dominant CF type.
The qualitative analysis software NVivo12 was used to sort the teacher responses into categories. As new categories emerged from the data, I revisited the previously coded material to check whether modifications were appropriate. In Tables 2 and 3, for each of the categories, I indicated the source from which I borrowed its name.
Teacher responses to non-target-like turns in L2 English (n = 193) and L3 French (n = 139) lessons.
Notes. * Cullen (2002) described several functions of repetitions, including repetitions with a rising intonation that serve to indicate that an error has been made and elicit a reformulation from the learner. However, this use of repetition as an output-prompting CF move did not occur in this study’s material.
Phase 5: Testing reliability
Tables 2 and 3, which present the categories with descriptions and examples, were given to a research colleague. We first coded one lesson together to ensure a similar understanding of the codes. The colleague then independently coded 20% of the data. NVivo’s reliability test indicated high agreement. The average agreement rate ranged from 93% agreement for the code Content summary and extension at the low end, to 100% for some codes, such as No teacher comment and Repetition at the high end. Regarding CF strategies, explicit correction and recasts had the lowest agreement rate (94% and 95%, respectively), which was due to the similarities between partial, didactic recasts and explicit correction (see Ha & Murray, 2020), leading to instances where one coder coded a CF move as a recast while the other coded it as explicit correction.
Phase 6: Triangulating the results
After finalizing the video-based analyses, I analysed the learners’ responses to the three selected items from the student survey to see whether these would confirm the video observations related to research question 2. The relevant items were the following:
a) When students respond to questions from the English/French teacher, s/he asks them to justify their answers or elaborate on them in more depth.
b) In the English/French class, we learn to correct our errors.
c) The English/French teacher gives us time to explain our thoughts and ideas.
V Findings and analyses
Research questions 1a and 1b: Amount and type of learner TL output
The English part of the corpus comprised 1,109 TL learner turns altogether (an average of 57 per hour). The vast majority of turns (83%) were viewed as target-like and 17% non-target-like. The average length was 15.4 words: 12.8 for target-like turns and 28 for non-target-like turns.
In the French data, there were 234 TL learner turns altogether (an average of 13 per hour), of which 41% were viewed as target-like, and 59% as non-target-like. The average length was 5.6 words: 4.5 for target-like turns and 6.6 for non-target-like turns.
The students thus spoke considerably more in L2 English than in L3 French during lessons, both in terms of number and length of turns. In both subjects, target-like turns typically were shorter than non-target-like turns.
In the English classes, a range of serious ethical, social and political issues were discussed: teenage pregnancy (RM); the US presidential elections (SH); the death penalty (UM); world cultural heritage and global challenges (UH); and the pros and cons of being young (SM). Two language-oriented topics were touched upon briefly: formal vs. informal language (RM) and how to express what you like and dislike (SM). In addition, literary genres (fables and short stories) were addressed in one class (SM).
In the French classes, many of the TL learner turns occurred during description tasks in which the learners were expected to use recently learnt vocabulary. The learners described pictures (RM), peers (SH), or Norway (UH). In UM, the learners were asked to formulate correct sentences to practise recently learnt grammatical forms. Learners also produced TL output in translations (SH), read-aloud exercises (SH, RM, SM), and listening or reading comprehension tasks, in which they answered questions to audio files or written texts about an exchange student (SH).
Research question 2: Teacher response to TL learner output
Tables 2 and 3 provide an overview of the strategies used to respond to TL learner output in the English and French lessons for target-like and non-target-like utterances, respectively. In Table 3, all Sheen and Ellis’ (2011) categories of CF are represented, apart from repetition, which was non-existent as a CF move in the data, mirroring findings in Choi and Li (2012).
The tables also provide illustrative examples and frequencies. The number of references for teacher responses does not correspond to the number of learner turns for obvious reasons: A learner utterance may receive several types of teacher responses within the same teacher turn (Nassaji & Wells, 2000), and one feedback move may apply to several turns, e.g. when student pairs are reading aloud dialogues, in which case each line counts as a turn, but the teacher provides feedback only at the end.
Most learner turns and subsequent teacher responses were analysable within the IRF model. Some exchanges did not fit this pattern, instead falling into two overlapping categories: conversation-type sequences (see above) and sequences in which the learner was responsible for the initiation. These were more frequent in the English than in the French lessons (see Tables 2–3). They were not analysed in further detail, and the remainder of this section focuses on teachers’ third-turn follow-up moves.
a Teacher response strategies in the English classes
In the English classes, the most frequent teacher responses to target-like turns were content-related non-evaluative remarks: the teachers summarized or elaborated on what had been said (24%). Many turns also received minimal responses (14%), i.e. discourse markers such as right, OK used in-between turns to signal that the learner should continue, or at the end of a learner’s turn to close the sequence without further response. 7 Repetition and elaboration questions each occurred with 12% of the turns. These strategies were often combined in a response to the same turn, mirroring Park’s (2014) finding that third-turn repeats’ in meaning-and-fluency contexts are used to encourage learners to elaborate on their response. Extract 2 from a lesson on global challenges, illustrates the combination of the two.
Extract 2: T: OK. I want to hear . . . the challenge that you think is the most important . . . your most important challenge? L: We wrote global warming. T: Global warming, mhm. And why is that a challenge?
Confirmation checks (7%) were used when the teachers wanted to be sure that they had understood correctly or when they could not hear properly. Two of these were epistemic status checks in response to a learner’s claim of incomplete knowledge (see Sert & Walsh, 2013, who discussed whether these should be viewed as confirmation checks at all). Specific content feedback and general praise were given to only 5% and 4% of the learner utterances, respectively. Even rarer was teacher reformulation of learner utterances (3%) and elaboration on the language used (1%).
Regarding non-target-like turns in the English classes, teachers often ignored errors, as 63% of these turns received no CF or scaffolding. As was the case for target-like utterances, the teachers summarized or elaborated on the content (32%), provided minimal response (16%), or asked elaboration questions (10%), regardless of whether or not errors were corrected. Content feedback and praise were provided to 6% and 5% of non-target-like turns, respectively. Scaffolding following learner hesitation or explicit request for help occurred in 11% of non-target-like turns. When CF was used, recasts (11%) or clarification requests (8%) were most frequent. Other types of CF were hardly ever used, and the types involving metalinguistic comments were absent.
All recasts were didactic in the sense that there was no communication breakdown (Sheen & Ellis, 2011), but they were not made salient through features such as stress and intonation. On the contrary, they were integrated into the conversation, which the teacher moved forward through comments or questions, as in Extracts 3 and 4. In the first example, the teacher integrated a correction of the past participle of the verb drive into a confirmation of and comment on the content. In the second example, taken from a lesson in which the class was talking about the US presidential elections, the teacher used a non-salient recast (‘Democrats’) and added a short comment (‘every single time’) before immediately moving on with the conversation by asking a new question.
Extract 3: L: In some states they actually, if you are over sixteen you are allowed to drive a car. So I think everyone could have droven the car. T: Yeah, everyone could have driven the car. Even though Jim Styles provided it, everyone could have driven it, and everyone could have given the order that he would provide the car. Extract 4: T: Who always wins California? L: The Democratics. T: Democrats, every single time. What about Texas, [student name]?
These extracts illustrate the typical use of recasts in the English data. The recasts are immediately followed by a topic continuation move and seem very similar to the non-corrective repetition in Extract 2 above, a pattern similar to Llinares and Lyster’s (2014) findings. These are integrated recasts (Ha et al. 2021a), and although they may be correction-focused, and thus didactic in Sheen and Ellis’ (2011) terminology, they are not particularly salient or explicit. Only on one occasion in the English data (Extract 5) did a teacher make a recast salient through features such as stress and intonation: Extract 5: L: We don’t know so much about the country. It’s a dictator. T: It’s a dictator, so it’s a dictator
The CF strategies used in the English classrooms thus lean towards the implicit side of Lyster et al.’s (2013) continuum and have the advantage of not disrupting the communicative flow. As such, they are more in line with feedback provided in authentic dialogues.
b Teacher response strategies in the French classes 8
In the French lessons, repetition was the most frequent form of teacher response to target-like turns (39%), followed by minimal response (19%). More rarely, the teacher provided general praise (9%) or further information about the language used (8%). Comments and elaborations on content were even less frequent (7%), as were elaboration questions to push for further output (7%). Content feedback and reformulations each occurred for 4% of the turns.
For non-target-like turns, explicit corrections were the most frequent type of teacher response (28% altogether), with (7%) or without (21%) metalinguistic comments or explanations. 9 Recasts were provided to 25% of the turns and elicitations to 22% altogether; 14% with and 7% without metalinguistic clues. Scaffolding was provided to 20% of the turns.
Recasts were – unlike in the English classes – most often isolated, including only one word or a chunk, made salient through stress or repetition (see Sheen, 2006), and not immediately followed by topic continuation moves: Extract 6: L: Le temps en Norvège est froid et il pleut beaucoup ((froid is incorrectly pronounced with final d)) T: L: froid? ((pronounced correctly)) T: froid, oui. Il fait froid et il pleut beaucoup.
When scaffolding was provided, the teacher gave explicit vocabulary help following students’ request for assistance (see the example in Table 3). The teachers seemed to come to the learners’ rescue when they were genuinely stuck or explicitly asked for help, and they did not complete learner turns to save time. However, a few instances of scaffolding might have lingered on the borderline of what Walsh (2002) termed ‘teacher turn completion’. In such cases, the teacher, by completing the learner’s turn, deprives the learner of an opportunity to produce output, instead of helping the learner speak more. In Extract 7, the teacher, upon the learner’s hesitation, recast the first part of the sentence, in which the learner incorrectly used an adverb (bien) instead of an adjective (bon). She also provided the learner with the necessary vocabulary (ami), which was coded as scaffolding. However, the learner might have known this word, but hesitated due to uncertainty about the adverb, in which case the teacher’s intended scaffolding is really turn completion: Extract 7: T: C’est ton meilleur ami? [Student name] est ton meilleur ami? L: Eh . . . il est . . . un . . . très bien . . . ((hesitates)) T: oui, un très bon ami.
Teachers often used a combination of CF types, as previous studies pointed out (Lyster et al., 2013; Milla, 2017), and as illustrated by the following feedback episode. In Extract 8, the teacher used an elicitative question in line 2 before offering scaffolding in terms of vocabulary help in line 4. She continued with two elicitative completion moves
10
(lines 4 and 6), before adding an explicit correction with a metalinguistic comment and a recast in line 8. This extract is a typical example of naturally occurring CF: Extract 8: 1 L1: Je ne suis plus avec Elisabeth. 2 T: Yes, I am no longer with, but I am no longer together with, how do we say that? 3 L1: Ah, ok, être . . . 4 T: sortir avec. And then we have . . .? 5 L2: sors? ((incorrectly pronounced with final ‘s’)) 6 T: Yes, je ne . . . 7 L2: Je ne sors plus avec ((sors is still pronounced incorrectly with the final ‘s’)) 8 T: Yes, not /sᴐrs/, but /sᴐr/. The s is silent. Je ne sors plus avec Elisabeth ((pronounces ‘sors’ without the final ‘s’)).
Responses to non-target-like turns other than CF and scaffolding were rare. Teachers provided praise to 8% of the turns, minimal responses to 7%, content elaboration to 5%, and elaboration questions and content feedback to 3% each. Non-corrective repetition was provided to 2% of the turns.
Only 14% of the French non-target-like turns were left uncorrected, but the fact that an utterance received some kind of correction does not mean that all errors were corrected. There were often multiple errors in a turn, and the teacher could address one error while ignoring others.
c The survey
The three selected items were viewed as relevant because they tapped into students’ experiences with error correction (b) and teacher requests for justification (a) and explanation (c). The ‘justify’ and ‘explain’ items corroborate the findings from the video observations: For these items, the English teachers received a higher score than the French teachers (Figure 1a and 1c). The correction item does not lend support to the observations: Students reported that their English teachers corrected their errors nearly as often as their French teachers did (Figure 1b). The survey did not distinguish between written and oral activities, and the students might have had both written and oral production in mind when answering, which might explain this slight discrepancy between survey results and observations.
Students’ responses to items a, b, c, on a Likert scale from 1 (never) to 5 (always). (a) When students respond to questions from the English/French teacher, s/he asks them to justify their answers or elaborate on them in more depth. (b) In the English/French class, we learn to correct our errors. (c) The English/French teacher gives us time to explain our thoughts and ideas.
VI Discussion
Despite the curricula for L2 English and L3 French both focusing on content as well as language, the above findings show that in practice, the two subjects differ considerably when it comes to lesson orientation. Below, each research question is discussed separately, before the pedagogical implications are addressed.
Research question 1: Amounts and types of learner TL output
The differences between L2 English and L3 French classrooms when it comes to frequency and length of learner TL utterances and the proportions of target-like vs. non-target-like utterances corroborate previous research which showed that low-proficiency learners in FL contexts speak less and make more errors than learners in SL classrooms with a higher proficiency level (Azkarai & Oliver, 2019). This is an expected finding that reflects natural parts of the learning process. However, the very low number of TL turns in French classrooms suggests that the communicative approach advocated by the L3 curriculum has not been fully implemented. This finding is supported by previous observations from L3 Spanish classrooms in the same educational context, which also showed limited TL use (Askland, 2018). Compared with previous literature from other parts of the world, even the number of learner TL turns in L2 English lessons (57 per hour) might at first glance seem surprisingly low. However, this variation is easily explained when contextual differences are taken into account. Many previous studies have focused on immersion settings (e.g. Lyster & Ranta (1997), who reported 179 learner turns per hour 11 ) or on language support programmes for immigrants with different first languages (L1s) (see Choi & Li, 2012, who reported 337 learner turns per hour). These are very different ESL settings than the one investigated in the current study. The varying findings regarding the frequency of learner turns underscore the importance of conducting research in varied educational contexts. ESL immersion settings are very well represented in the international research literature (Ha et al., 2021b), but findings from these settings cannot be transferred to other SL or FL settings.
The topics dealt with in the two subjects revealed a systematic and striking difference between the content of L2 English and L3 French classroom discourses. With the exception of the comprehension tasks in school SH, in which learners summarized content using the TL, learner TL output in the French classrooms was always structured and form-focused: the main aim was to practise recently learnt grammar and vocabulary, and learners’ messages or personal opinions were not relevant. The English classrooms’ output was almost exclusively communicative in that the aim was to produce meaningful contributions and share personal perspectives. The topics discussed were likely to ignite debate, and learners participated actively and regularly. As mentioned above, there were only two brief form-focused sessions in the L2 English classrooms. To summarize, the French classes were characterized by form-and-accuracy contexts, while the English classes were characterized by meaning-and-fluency contexts.
These observations confirm a tendency described by Storch and Sato (2020, p. 64): in SL contexts, ‘communicative tasks deploying authentic topics are familiar to the students’, while in FL contexts, ‘discussion topics tend to be based on the textbooks, are decontextualized and rarely authentic’. From a policy perspective, these findings are nevertheless unexpected, as the curriculum explicitly emphasizes the need for communicative input and output.
Research question 2: teachers’ response strategies for the two subjects
Some clear patterns emerged from the observations presented above, indicating considerable differences between L2 English and L3 French regarding how teachers responded to TL output and thus, shaped learner contributions (Walsh, 2002).
First, L2 English and L3 French teachers responded differently to learners’ target-like utterances. The English teachers followed up on learners’ contributions by summarising, elaborating, and commenting on the content of learners’ turns. They also pushed for more learner output through elaboration questions that asked for clarification or justification. These techniques have all been shown to create space for learning and facilitate learner participation (Walsh & Li, 2013). The French teachers’ most frequent response to target-like turns was confirming repetition. These repetitions were useful for the other learners in that they confirmed and stressed appropriate TL structures (Park, 2014), but they also served to close the sequence. Praise also served as a sequence-closure at some occasions (Waring, 2008). Strategies that extended the interaction or pushed for more output were rare. Minimal responses sometimes served this function, as they did in the English classes. The use of minimal responses to target-like turns was similar in both contexts: they were used relatively often, sometimes in-between turns to elicit more learner output, as shown in the French example in Table 3, and sometimes to close the sequence, as illustrated by the examples in Table 2.
Second, regardless of whether utterances were target-like or non-target-like, a sharp contrast existed regarding content-related comments. In the English classes, teachers frequently summarized, elaborated, and gave personal comments also to non-target-like turns. In the French lessons, content-related comments were rare, and the teachers’ frequent use of confirming repetitions for target-like utterances and linguistically oriented CF for non-target-like utterances, did not encourage content elaboration. In view of the curriculum’s focus on content along with language, this could be considered an unexpected finding. A possible interpretation could have been that L3 French teachers in Norwegian lower secondary schools work with content components in the L1, and thus separately from the language learning per se. However, previous research suggested that the content component of the curriculum has a minor role in lower secondary French classrooms (Vold & Brkan, 2020).
Third, the extent to which the teachers corrected errors differed. While only 37% of non-target-like turns in the English classes received some kind of correction or scaffolding, 86% of non-target-like turns in the French classes did. This pattern supports previous research that stated that overall, error correction is considerably more frequent in form-focused than meaning-and-fluency contexts (Ellis, 2017; Gurzynski-Weiss & Révész, 2012; Mayo & Milla,2021). However, despite the frequent corrections in the French classes, the results cannot be said to support Walsh and Sert’s (2019, p. 6) statement: ‘error correction occupies a considerable amount of teachers’ time’. The amount of learner TL output in the French lessons was so limited that these corrections did not occupy much time.
Fourth, when error correction occurred in the English classes, implicit strategies were the preferred choice, whereas, in the French classes, teachers used a broader range of CF types, including both implicit and explicit strategies. Recasts were frequent in both settings but of a different nature: the L2 English teachers used integrated and implicit recasts, while the L3 French teachers used isolated and explicit recasts (Ha & Murray, 2020; Sheen, 2006). This pattern supports previous research that stated that implicit correction types, such as conversational recasts and clarification requests, play a more important role in meaning-and-fluency contexts than in form-focused contexts (Daskin, 2015; Mayo & Milla, 2021). The data also align with Brown’s conclusion that prompts are offered more often in form-focused than in meaning-and-fluency contexts, although this was the case only for elicitations, not clarification requests. Moreover, strategies that involved metalinguistic comments were used in the French classes but not in the English classes. Again, this is in line with previous observations that FL classrooms are more form-focused than SL classrooms (Ellis, 2017; Mayo & Milla, 2021) and French instruction historically has a strong tendency to focus on language issues (Simard & Jean, 2011). The French teachers’ frequent use of explicit strategies and metalinguistic comments underscore previous studies’ assertions that teachers and learners in form-focused classrooms are used to providing and receiving CF, and they see it as a natural part of instruction (Ellis, 2017; Ha et al., 2021b; Sheen, 2004).
In both contexts (L2 English and L3 French), teachers often used a combination of response strategies and feedback moves to a single learner turn. According to Lyster et al. (2013), a pattern of combined moves is typical of naturally occurring feedback, which partly explains why experimental effect studies comparing isolated feedback types might not yield productive results for practicing teachers.
Summary of discussion
This study set out to examine differences between L2 English and L3 French teaching in five lower secondary schools in Norway regarding amount and type of learner TL output and teachers’ response to that output. The findings revealed considerable differences between the two subjects: In the English classes, learners produced large amounts of communicative output, while in the French classes, they produced only limited amounts of structured output. The English teachers often ignored errors, instead responding to learner output with non-evaluative, content-oriented comments (discoursal follow-up moves in Cullen’s (2002) terms), while the French teachers often provided CF, but rarely addressed the content of the learners’ utterances.
The findings in this study support previous research stating that interactional patterns vary between meaning-and-fluency contexts on one hand and form-focused contexts on the other. However, while previous studies documented how teachers successfully shifted between these different contexts within the same lesson (Daskin, 2015; Walsh, 2002), the present study demonstrated that each of the two language subjects under investigation belonged to one of these contexts: L2 English instruction was conducted in a meaning-and-fluency context, while L3 French instruction was conducted in a form-focused context. The investigation of teacher follow-up practices revealed that the two languages were taught in two distinctly different manners even though instruction took place in the same educational context (Norwegian lower secondary education), even in the same schools, and that the theoretical principles underlying the subject curricula were identical for both subjects.
VII Pedagogical implications
A pedagogical focus that is either form- or meaning-oriented has consequences for learning outcomes, as well as for learner motivation. The critique of exclusively form-focused instruction is well-known: learners do not get to practise communicative skills, and there is a lack of authenticity and real-life communication that might influence learner motivation negatively. In terms of interactional practices identified in this study, what the L3 subject can pick up from L2 teaching is, first, an emphasis on more meaningful and engaging topics to select for classroom or group discussion, and second, an emphasis on feedback oriented towards how to get meaning across, in addition to accuracy-oriented feedback that many of the teachers already provide. For instance, adding more summarising, extensions and elaborations to the follow-up on learner utterances would scaffold and consolidate topic-relevant language structures and vocabulary.
An exclusive focus on meaning also has its disadvantages. Nassaji (2016, p. 538) affirmed that ‘teaching approaches that are purely communicative in nature and focus only on message are not adequate for language development’ and that ‘learners . . . need opportunities for attention to form’. When learners’ errors are not corrected, there is a risk of fossilization, as learners do not get the opportunity to polish their skills and achieve a higher proficiency level. Therefore, such a lack of focus on accuracy and on improving learners’ spoken production in L2 English lessons can be viewed as unfortunate. Although learners in Norway generally master English well, scholars have noted that learners’ skills are more developed for everyday language than for academic English (Busby, 2021; Hellekjær, 2010b), and that in working life, Norwegians often make mistakes that make them come across as less professional (Hellekjær, 2010b). Thus, there are good reasons to provide CF even when no communication breakdown occurs, particularly in language teaching at intermediate and higher proficiency levels.
Teachers should strive to find a balance between evaluative, accuracy-oriented feedback on one hand, and discoursal, content-oriented feedback on the other. Excessive use of evaluative feedback may impede development of communicative classroom dialogue, while excessive use of discoursal feedback will not help learners notice and repair their errors (Cullen, 2002). Judging from the present study’s data, it seems that this balance is difficult to strike, and that teachers easily fall into either a meaning-oriented or form-focused classroom context. Similarly, Spenader, Wesely and Glynn (2020) found that teacher students had difficulties balancing language- and content-oriented learning goals.
Some limitations in the present study should be mentioned. First, it does not cover all situations in which learners receive teacher feedback on their spoken TL output. Such feedback might be provided outside the classroom. Second, it included a limited number of classrooms (five for each subject), and more large-scale studies are needed to establish statistically significant differences between the two investigated contexts. Third, the article provided an overview of how teachers respond to learners’ TL output in L2 vs. L3 classrooms, but did not offer detailed analyses of how interactional patterns unfold between speakers. In-depth analyses of extracts from the corpus could provide a fuller picture of how teachers encourage, scaffold and evaluate learner TL talk.
VIII Conclusions
The present study contributes to research on classroom interactional practices and CF focusing on naturally occurring feedback and how feedback practices relate to other instructional variables and contextual factors. The study brought to the fore some striking differences between L2 and L3 instruction and left little doubt that the communicative orientations of L2 vs. L3 classrooms in Norway are very different. Language instruction within the same educational context might belong to different worlds, depending on the subject taught and its status in the relevant context.
Considering that these language subjects’ status and role are so different, one might ask whether it is fair to set the same expectations for them regarding instructional practices such as TL use and teacher feedback moves. Vold and Brkan (2020) asserted that an overly communicative approach to language teaching might lead to unrealistic and overambitious goals for L3 teaching, and that a need exists for an approach that takes into account the characteristics of a FL setting and the realities of the teaching context when it comes to learners’ and teachers’ TL proficiency level, number of teaching hours allocated to L3 learning, and limited out-of-school TL exposure. Such an approach could include systematic work to steadily increase TL use without aiming for a TL-only approach, more real-life and authentic tasks while still keeping them small and simple, and techniques to combine linguistic learning with learning how to express meaning – preferably based on learners’ needs and desires for discussion topics.
An increased awareness of differences between language subjects within the same context is of great importance to policy makers, as well as for teacher education. Many teacher education institutions offer only joint language didactics courses, which group together future teachers of different languages with different statuses (Vold, 2014). Future research should examine how pre- and in-service language teacher education courses can address different language subjects’ specific needs and how these fields can collaborate and draw on each other to develop new and improved practices in all fields.
Footnotes
Appendix 1
Acknowledgements
I am very grateful to postdoctoral research fellow Camilla Gudmundsdatter Magnusson for fruitful discussions about teacher response types and for her agreeing to reliability code parts of the data. I am also thankful to the anonymous reviewers for their constructive feedback.
Declaration of conflicting interests
Funding
The author received no financial support for the research, authorship, and/or publication of this article.