Video feedback and Foreign Language Anxiety in online pronunciation tasks

Despite many studies about video feedback in both face-to-face and online settings, little research has been carried out exploring how this technique is perceived by students learning the pronunciation of specific sounds in a foreign language. Adopting grounded theory as the methodology and a dialogic approach as the conceptual framework, the present study shows that anxious students welcome video feedback. The design of a learning activity for students to practise a specific problematic pronunciation target in English, carried out in an e-learning environment, more specifically in an online English language course, is described. The results show three aspects of teacher's corrective video feedback, perceived as more relevant: the Emotional input of feedback, referred to the feelings around the feedback delivery which foster dialogue, closeness, motivation and empathy; Enhanced understanding, related to the clarity, the usability and personalization of the feedback; and feedback engagement, which are the conditions favouring agentic engagement that involves the students sharing responsibility for making feedback processes effective. Implications related to video feedback practices are also discussed.

specific feedback is included. We will then identify strengths of video feedback in second language teaching and discuss implications related to video feedback practices.

Literature review
Feedback here is understood as "a process through which learners make sense of information from various sources and use it to enhance their work or learning strategies" (Carless & Boud, 2018: 1315, all of which framed within a dialogic approach where dialogue is understood as a relationship where participants think and reason together in order to build knowledge together. In addition, students take on a proactive role in the process since they seek feedback and implement it to better their learning process (Dann, 2019;Winstone et al., 2017). This dialogic conceptualization of feedback could be inconsistent with corrective feedback, which is one of the most common types of feedback in language learning (Afitska, 2015). Even though this type of feedback could be considered to be not very formative feedback, in the specific context of Foreign Language Acquisition, corrective feedback adopts some particularities that could bring it closer to a formative approach, that is, that promotes the improvement of learning.
Feedback in this research will take the form of corrective feedback (CF) in an online context. Our research will adopt Lightbown and Spada (1999)'s definition of CF: any indication to the learners that their use of the target language is incorrect. Following Lyster and Ranta (1997: 46-48), there are six different corrective feedback types used in a language learning environment which were first reported by and have been accepted by other scholars. These are: Clarification request (the teacher asks students to repeat what they said); Repetition (the teacher emphasizes the student's grammatical error by changing his/her tone of voice); Explicit correction (the teacher gives the correct form to the student with a grammatical explanation); Elicitation (the teacher asks the student to correct and complete the sentence); Metalinguistic feedback (the teacher gives a hint or a clue without specifically pointing out the mistake); Recast (the teacher repeats the student's utterance in the correct form without pointing out the student's error). These types of CF easily fit into the kind of feedback which is interactive and dialogic since providing feedback could be the first step to start a conversation between learner and teacher in order to improve a specific performance.
Previous research on CF, concludes, for instance, that anxious students prefer recast and metalinguistic CF (Martin and Álvarez, 2017;Zhang & Rahimi, 2014). In addition, Rassaei (2015) found high-anxiety learners benefited more from recasts rather than metalinguistic corrective feedback. Roothooft (2014) found some indications that elicitation could be a good technique for fostering uptake and repair of past tense errors. All three examples show us that carefully designed corrective feedback should be fostered by teachers when planning feedback delivery, more specifically to anxious students. Knowing what kind of CF your students prefer is key when designing feedback delivery.

Video feedback and Foreign Language Anxiety
Feedback itself can also be a source of negative reactions in the student's learning process (Krashen, 1982). However, feedback can be positively perceived through video feedback, which could in turn help students with a high-anxiety profile feel more relaxed. Following is what previous research has found out about how video feedback is perceived.
First, it facilitates feedback personalization (Henderson & Phillips, 2015;Öztürk, 2016;Turner & West, 2013;Wood, 2021). For instance, Turner and West (2013) carried out a mixed methodology study in which the majority of students preferred video feedback since it was more personalised and enhanced understanding as well as promoted feedback engagement.
Secondly, Ryan et al. (2019) found digital recordings for feedback provision were described as effective, but more precisely as detailed, personalised and usable. An extensive review of the literature which included empirical studies investigating the employment of web-based technology to enhance EFL learners' speaking performance (Cong-Lem, 2018), found that web-based audio-and video-based technology helped to reduce L2 learners' learning anxiety while making them become more active and motivated language learners. In addition, the role of the language instructor in web-based technology implementation was also found to be crucial, e.g. monitoring the L2 learners' learning progress and providing instructive feedback (Estaji & Farahanynia, 2019).
Third, the teacher's presence, be it through their voice or image, adds an affective component to feedback delivery, which is valued positively by students. Borup et al. (2014) found that video feedback contained the teacher's emotional input in the feedback, thus making it sound more "human". It was also perceived as more conversational and interactional as well as more connected to the teacher and personalised (Espasa et al., 2019;Mahoney et al., 2019).
In conclusion, video feedback seems to be a technological tool that, when pedagogically-driven in online environments, can be accepted since the perception of video feedback by students is clearly positive when learning pronunciation, even better valued than other modes of feedback such as written comments. In addition, video feedback could also be beneficial to students with high anxiety especially combined with specific corrective feedback for this specific group of students.

Pronunciation as the arena for exploring Foreign Language Anxiety
FLA is specific to a particular situation and is associated with learning a foreign language or with having to communicate using it. FLA may especially affect word pronunciation, word stress, weak forms, rhythm, linking and assimilation (Szyszka, 2017). Saito and Akiyama (2017) found that communicatively-oriented and interactive tasks carried out online and synchronously improved the students' vocabulary richness and grammatical accuracy as well as fluency but not necessarily accuracy in pronunciation. The authors recommend contextualised explicit instruction (e.g. drills) in order to improve segmental and suprasegmental aspects of pronunciation since in situ corrective feedback may not be enough in order to change the student's L2 phonetic knowledge, which is a way to teach how to pronounce sounds.
In particular, some challenging aspects for Spanish and Catalan speakers are initial s + C clusters as well as the difference between /b/ vs /v/ since s + C tends to be pronounced with an extra vowel before the initial cluster as in "school" > "eschool" and /v/ becomes /b/, leading to communicative breakdown. Our research has included two different pronunciation problematic targets, according to literature (Calvo, 2013;DeKeyser, 2005), which are the "ed" for the past and the vowel in "sir" (open-mid central unrounded vowel), which may also be sources of communicative breakdown and are hard to acquire by Catalan and Spanish speakers.
Against this background, focusing on a group of online students who reported Foreign Language Anxiety, this small-scale research aims to throw important light on their perception of video feedback by answering the following research question: what is the perception of video feedback delivery in an online pronunciation activity targeting complex sounds for Catalan/Spanish upper-intermediate online students of English as a Foreign Language? A further exploratory question is whether there are differences between anxious and non-anxious students.

Participants and research context
The participants were enrolled in an asynchronous online upper-intermediate English language course (B2.2) at the Institut Obert de Catalunya, based on continuous assessment with optional synchronous online speaking classes. Students sit for an exam at the end of the course and receive the B2 certificate of English upon passing the exam. Students at this level can understand the main ideas of a text; understand spoken language, live or broadcast; and understand texts with a broad reading vocabulary and large degree of autonomy (Council of Europe, 2001).
Students were invited to participate in the research through the publication of a message on the notice board in the Learning Management System (LMS) of the course. A total of 88 responded to the initial questionnaire, thus confirming their consent to participating in the research. Most of the participants were women (66%) and all the students were older than 25 with previous online learning experience. In the second stage of our research, three studies were carried out in which an optional activity was offered to practise a specific aspect of the pronunciation of English which was key to the students' pronunciation competence. Each study was carried out in different semesters: Study 1 (first semester in 2018-2019; from October to January); Study 2 (second semester 2018-2019; from February to June); and Study 3 (second semester 2019-2020; from February to June). Each study approximately took four weeks. Six students completed the optional activity for Study 1, three in Study 2 and 18 in Study 3.
These were the actual participants in the research presented in this article. Although the three studies were consecutive, no student repeated the experience.
Due to the optional nature of the activity, participation was low but enough to collect data to answer the research question for our present research. Even though the teacher and staff encouraged participation as much as possible, no attempts were made to recruit a minimum number of participants. In contrast with how participants are recruited in experimental studies, they were neither remunerated nor compensated. They need to be strongly motivated to participate in a study that continues over a period of time and requires extra effort (Khatamian-Far, 2018). It can be argued that student participation was largely self-directed.

Data collection and analysis
Prior to the optional activities and in order to identify anxious students as well as their preferences for Corrective Feedback strategies, two questionnaires were employed, which were joined in an online questionnaire (Google Form) in order to facilitate its administration. Both of them were translated into Catalan, which is the official language of the institution, and underwent a series of procedures such as back translation in order to guarantee its linguistic equivalence. The two questionnaires were: Foreign Language Classroom Anxiety Scale (FLCAS; adapted from Horwitz et al., 1986). It was used in the initial stage of the study as a measure of the students' specific anxiety reaction towards the learning of a foreign language in an online setting. This scale consists of 33 items worded as a 5-point Likert-type scale, ranging from "strongly agree" (5 points) to "strongly disagree". It was designed to assess the degree to which learners feel anxious about learning during English class, including the dimensions of Comfortableness in using English inside and outside the classroom; Communication apprehension; Speech anxiety; Fear of failing the class; and Negative attitudes toward learning English. In our current sample, the items pertaining to Communication apprehension, Speech anxiety, and Negative attitudes toward learning English were best described by a single anxiety factor, whose alpha for internal consistency was 0.861. The Comfortableness with the foreign language factor was 0.647. Inasmuch as the internal consistency of this factor was marginal, we included these data for heuristic value only given their substantive importance. The results should be interpreted with caution.
Corrective Feedback Belief Scale (CFBS) (adapted from Fukuda, 2004). This instrument collects the beliefs students have on the feedback received when speaking in a foreign language. There were 22 items which were aimed at the exploration of students' judgments about the giving and receiving of spoken error correction, frequency of giving and receiving spoken error correction, time of spoken error correction, types of errors which need to be corrected, types of spoken error correction as well as sources for providing spoken error correction, and finally specific CF methods. Each of the above-mentioned items in the questionnaire were designed based on a 5-point-Likert-scale. The alpha reliabilities for CFBS in the present study were 0.749. This scale was administered in the initial stage of the research to the 88 students who expressed willingness to participate in this study.

Procedure
In order to identify anxious and non-anxious participants in the whole group (N = 88), a cluster analysis was performed. More specifically, in this study we applied the two-step clustering and hierarchical clustering with squared Euclidean distances. The respondents were clustered based on the variable Anxiety, yielding two clearly distinct groups (F = 130.72; p = 0.000). Finally, 58 participants were grouped into the first cluster group, characterised by low scores (M = 30.1, SD = 4.8), and 30 (34%) into the second cluster group, characterised by high scores (M = 75.7, SD = 11.7).
Significant correlations allowed us to design a Corrective Feedback Strategy for anxious and non-anxious students, which consisted of what corrective feedback method is used and when it is applied. As far as methods of CF are concerned, anxious students showed a significant correlation with Elicitation (r = 0.3, p < 0.05) whilst non-anxious students preferred Explicit correction (r = 0.28, p < 0.05).
As previously mentioned, out of the total of students who answered the initial questionnaire, six were recruited for Study 1, three for Study 2 and 18 for Study 3. Students were classified into anxious vs. non-anxious by means of the two-step clustering and hierarchical clustering with squared Euclidean distances. Out of the total number of participants, 15 were classified in the anxious profile (one in Study 1, three in Study 2 and 11 in Study 3). Analyses were made using the Statistical Package for Social Sciences (version 20.0 for Windows). This specific result allowed us to continue with our feedback which was carefully designed for a specific anxiety profile.

Pronunciation targets
After answering the questionnaire, students were invited to do the pronunciation activity in a forum specifically created in the LMS of the course. The two pronunciation targets selected for the study are the "ed" inflection (Studies 1, 2 and 3) and the vowel in "sir" (Studies 2 and 3). Following is the presentation of the linguistic material employed in the activity in each of the studies.

Study 1: the "ed" inflection
In study 1 the pronunciation target was the "ed" inflection. The participants were first asked to watch a video with the explanation about how to pronounce the "ed" in English. Then students had to narrate a series of events in the life of a person (Mr. Brown), using the prompts in the task (see Appendix 1). While narrating, students had to video record themselves. Finally, students had to post the video in the forum, for which the course tutor provided feedback in video format and posted it to the forum as a reply (Fig. 1).
The video feedback was created using Edpuzzle (https:// edpuz zle. com/) and included the student's production with a brief introduction to the task. Also, correction feedback was embedded throughout the video, and further instructions at the end of the video were included, all of which recorded in audio format by the tutor. See Fig. 2 for an example.
Students were asked to do the same task twice (Try 1 and Try 2) and received feedback for each one of them. The feedback for Try 1 was created using Edpuzzle as already mentioned. The student's video was played and stopped whenever a target sound was pronounced, at which point corrective feedback was delivered by the tutor while all of it was being recorded. In order to provide feedback for Try 2, a screencast session showing the tutor's face was recorded (https:// scree ncast-o-matic. com/) giving corrective feedback for each pronunciation target without playing the student's video.

Studies 2 and 3: the "ed" inflection and the vowel in "sir"
In Studies 2 and 3 the pronunciation targets were the "ed" inflection as well as the vowel in "sir". Participants were invited to do the task (Mr. Brown) but this time without giving them any theory previously in order to obtain a more realistic picture of the student's starting point. The vowel in "sir" had not been included in Study 1 (see Appendix 1), during which the tutor detected the need for including this specific vowel in future iterations of the study since it proved to be a difficult sound target for the students and a source of communicative misunderstanding.
In both studies, average preparation time of the video feedback was approximately 10 min, which was significantly reduced after a few times of repeating the same procedure. All the feedback provision was recorded by means of screencasting software (https:// www. apowe rsoft. es), showing both the student's video and the task simultaneously on the screen. The tutor's face was not shown in any of the recordings. Corrective feedback for each pronunciation target was provided whenever it was pronounced by the student. At which point, the tutor would stop the student's video and give corrective feedback.
Just as in Study 1, students were asked to do the same task twice (Try 1 and Try 2) and they received video feedback from their tutor for both tries. Figure 3 summarises the procedure of the pronunciation activity.
Once the students completed the activity, they were contacted for an interview with the course tutor. The semi-structured interview was adopted in order to allow for instances when further exploration of the student's reaction to video feedback was required.

Thematic analysis
Prior to the analyses, carried out using the qualitative analysis software package Atlas 8.4, the audio files were transcribed carefully. Constructivist grounded theory was the methodological orientation of the study in order to account for educational phenomena using qualitative data (Chong & Reinders, 2020). The researcher analyzes data by constant comparison, initially of data with data, progressing to comparisons between their interpretations translated into codes and categories (Mills et al., 2006). Based on previous literature, associated categories were identified. Codification yielded three main themes. A previous draft was shared with specialists in the field and improved. Coding categories were verified independently by two researchers to increase reliability. Specifically, three stages of coding were conducted to identify emergent categories (themes), focused coding to merge and re-categorise the categories identified in the initial coding stage, and axial coding to identify the relationships between categories and enrich the properties (themes) of each category. The authors analysed the data using the qualitative data analysis software Atlas.ti. This labour-intensive process concluded with a coding system that reflected how participants in this research perceived video feedback in an asynchronous online pronunciation activity (See Table 1).

Findings
In order to answer the research questions, we analysed the contents of the interviews by means of Thematic Analysis. The analysis was carried out employing a coding system which consists of nine categories, which were grouped into three main themes: Emotional input, Enhanced understanding and Feedback engagement.
The Thematic analysis (see Fig. 1) reveals that the students' perception of the feedback received during the pronunciation activity in the form of video feedback can be associated with the following three main aspects (relative frequency in %): Enhanced understanding: 40.48%, Feedback engagement: 38.53% and Emotional input: 20.9%. The frequency of the categories within each theme shows that Emotional input is associated more often with Caring and motivating 53.48% followed by Conversational and interactional 25.58% and Closeness 20.93%. It also shows that Enhanced understanding is associated more often with Clear and detailed 49.39%, then followed by Personalised corrective feedback 27.71% and Usable 19.27%. Finally, Feedback engagement is associated more often with Paying attention to feedback 58.22%, then followed by Proactive recipience 31.64% and Committed to change and development 10.12%.
Therefore, if we take the first category in each theme, the feedback received during the activity is mostly Caring and motivating (from Emotional Input), Clear and detailed (Enhanced understanding) and helps you Pay attention to feedback (Feedback In addition, frequency of the themes varies as a function of anxiety. Anxious students' comments display the following pattern: Enhancement understanding 47.11%, then Feedback engagement 34.61% and Emotional input 18.26% whereas in non-Anxious comments the pattern is Feedback engagement 42.57%, then Enhancement understanding 33.66% and Emotional input 23.76%. Emotional input is the least present theme of the three in both groups (anxious vs non-anxious students).
Following are some comments made by participants associated with each of the key themes.

Emotional input
The three categories allowing us to identify what subjects thought about video feedback in the research were: the Conversational and interactional aspects of feedback; the perception of Closeness; and Caring and motivating on behalf of the tutor. Each participant is coded with the number of the subject participating in the research and the initial letters of their name and surname. Also, in brackets the number of the study is included.
The participants considered video feedback had more of a human touch than other feedback formats such as audio or written:

Enhanced understanding
The categories defined for this theme allowed us to identify some of the aspects of the video feedback that were positively perceived by the participants such as a better comprehension of the feedback itself as well as other positive aspects of corrective feedback provision that may have been anxiety inhibitors rather than triggers. This is especially interesting for our research since we expect students to identify the value of feedback personalization when designing the activity. The subjects' comments about this were codified in three categories: Personalised corrective feedback; Clear and detailed; and Usable.
The participants positively valued the fact that video feedback helped them understand what had to be improved: Well, very good. It is very good to use the video in order to highlight errors because I could better identify them, which is hard to do when speaking. Did I say it like that? Do you understand? Sometimes you were correcting me and I was certain that I was already saying it like that. [S8 EM (Study 2)] I think that in this task, which was a pronunciation task, it is obvious that the video was great. Also, in the video you were pointing out each area with the cursor. You could see when I was saying something. I think that the video is great. [S7 JH (Study

2)]
The students liked the fact that each mistake was corrected after task completion in the video feedback. Being able to speak without being interrupted was key for this student but also being able to see the mistakes along with their corrective feedback after task completion was also positive. For instance, this student found video feedback usable:

(…) It is a good way. And it's good that you speak for a bit and then being told what you need to correct and what not. [S1 II (Study 1)]
Making sense of information within a dialogic approach is seen in S8 EM (Study 2) when she said that thanks to video feedback she could understand where she had made a mistake. Interestingly, she did not take feedback at face value but took some time to accept it since she was convinced that she was already pronouncing correctly. After a while she accepts her mistake and change is triggered.

Feedback engagement
Proactive recipience; Paying attention to the feedback; and Committed to change and development are the three categories employed to identify the conditions favouring engagement in the comments about the video feedback.
Reflecting on how video feedback contributes to improving one's learning, several students expressed this tool helped them avoid making the same mistakes in the future: Interestingly, all of the participants viewed video feedback favourably but when asked about format preferences, differences emerged. Unfortunately, there does not seem to be a clear pattern for us to affirm that a specific anxiety profile prefers one or the other mode of feedback provision, which leaves an open door for a future line of research.

Discussion
The objective of this study was to explore the perception of video feedback delivery in an online pronunciation activity targeting complex sounds for Catalan/Spanish upperintermediate students of English as a Foreign Language. The analysis reveals some interesting results we would like to share.
The participants' perception of video feedback revolves around three key themes: the Emotional input, Enhanced understanding and Feedback engagement. It has also been observed that in each theme there is a category which has been most cited: Emotional input > Caring and motivating; Enhanced understanding > Clear and detailed; and Feedback engagement > Paying attention.
Taken together, we can conclude that the video feedback employed in the activity has generally been characterised by participants as clear and detailed in order to enhance understanding, delivered in a caring and motivating manner and helping students pay attention to what needed to be improved. Such three characteristics should be taken into account when designing future feedback delivery. Following is a more detailed look at each of the key themes.
First, Emotional input refers to the feelings perceived around the feedback delivery from the teacher, which is associated with dialogue, closeness, and being caring and motivating. As previously observed by Borup et al. (2014), the participants in our research also considered that video feedback had more of a 'human touch' than other feedback formats such as audio or written. This aspect seems especially relevant to anxious students in online foreign language oral tasks since a welcoming environment may help them feel more confident and more willing to practice pronunciation so as to improve their communicative competences in a foreign language (Szyszka, 2017). Thus, we would be increasing the value of technology to facilitate the student's learning in the feedback process, so that learning becomes learner-centred and not only the transmission of information (Winstone & Carless, 2019).
Enhanced understanding was another aspect associated with video feedback which was positively perceived by the participants. It facilitates feedback personalization, which is highly valued by students since it allows for more personalised corrective feedback (Dann, 2019). As previously observed (Cong-Lem, 2018;Henderson & Phillips, 2015;Henderson et al., 2021;Jones et al., 2012;Ryan et al., 2019), the CF in the form of individualised video recordings of the lecturer commenting each assignment had a positive effect on how students perceived the feedback, thus describing it as clear, detailed and usable. This is especially interesting for our research since our students also identified the positive value of feedback personalization as an anxiety inhibitor. In addition, video feedback by means of screencast software was useful to deliver in-depth explanatory feedback (Desouki, 2016;Estaji & Farahanynia, 2019;Thompson & Lee, 2012).
Thirdly, regarding Feedback engagement, participants provide evidence of the proactive recipience of personalised feedback, paying attention to the feedback, and showing commitment with learning, all of which are the conditions favouring agentic engagement that involves students sharing responsibility for making feedback processes effective (Winstone et al., 2017).
As far as methods of CF are concerned, anxious students expressed a major preference for Elicitation whilst non-anxious students preferred Explicit correction. Such results confirm the fact that feedback preferences vary as a function of degree of anxiety (Abedi et al., 2015, Martin andÁlvarez, 2017;Rassaei, 2015;Zhang & Rahimi, 2014), which needs to be considered when designing feedback provision in order to provide personalised feedback. On the other hand, due to the nature of an exploratory study, we should cautiously take into account the positive effect of Elicitation in relation to repair of past tense errors (Roothooft, 2014). However, following Lyster (2005)'s warning, the effects of such interactional moves of corrective feedback on L2 learning require assessment in carefully designed, experimental classroom studies before one can make claims for their effectiveness.
In general terms, our results indicate the need to consider individual differences in terms of anxiety when learning a foreign language and students' beliefs on CF to help them progress towards achieving their learning goals in an interactive online environment. As expected, the pronunciation of these sounds was a challenge for the students (Calvo, 2013;DeKeyser, 2005). Students who reported high levels of anxiety as well as those with low levels of anxiety show evidence that receiving CF reduces anxiety. The small sample allowed us to examine how personalised video feedback can affect how tasks which require some extra effort are realised. By and large, the participants' experience with video feedback was positive despite their anxiety profile. Thus, students positively and warmly welcomed video feedback, in line with Henderson and Phillips (2015)'s findings.
This analysis is especially interesting in students who perceive themselves as more anxious when realizing online pronunciation tasks. The substantial data obtained from the initial questionnaires as well as from the interviews from the subsample of students who participated in the exploratory study have provided context and, in some cases, have accounted for cognitive and emotional motivational factors concerning video feedback. Such factors are perceived as learning triggers of complex pronunciation targets like the ones presented in this research.
We should also highlight the fact that the supplementary pronunciation activity offered in and specially designed for the online course has helped students practise the two pronunciation targets which are challenging for students of English as a foreign language (Saito & Akiyam, 2017;Szyszka, 2017). The technique employed in the design of feedback provision is screencasting where audiovisual information is employed to correct and clarify some aspects of the student's performance. Here is where dialogic feedback is key since through dialogue, albeit asynchronous, the tutor can make the student see where they must improve, thus giving the student a chance to perform better, which fits nicely within the dialogic approach to feedback where participants think and reason together (Dann, 2019). One must bear in mind that communication here takes place asynchronously between tutor and student, thus somewhat resembling synchronous communication, which is why we would call this interaction "quasi-dialogic" (Winstone & Carless, 2019).

Conclusions
From the data we can conclude that video feedback is positively perceived by students in an e-learning context and has been associated with three dimensions by means of a thematic analysis: Emotional Input, Enhanced understanding, and Feedback engagement. If we take the most mentioned category in each theme, we can thus describe video feedback as feedback which enhances understanding, which is caring and motivating and which is clear and detailed. As for its delivery, teachers can create videos where both the student's performance and the teacher's feedback are included. Students can then view it as many times as they need to, thus helping them to progress in their learning process and feel that their teacher is closer, which is especially useful in an online learning context. Hearing the teacher or seeing his face is something that written feedback fails to offer but online students need.
The limitations of our study are that it has few participants and general conclusions cannot be made from such a small sample. Future lines of research could include investigating interactions between gender, task type and feedback preferences.
Feedback provision can fail to be clear, thus preventing any uptake from taking place, which in turn can have a negative impact on the student's perception of their own learning progress. According to the results, this study reinforces the importance of using feedback purposefully in the design of the activities and in the form of video feedback, carefully personalised and paying special attention to the learning needs and CF preferences of the students.

Fig. 2
Example of a video with feedback incorporated using Edpuzzle for S2 JN in study 1. Note: Each dot represents a piece of personalised feedback about the pronunciation target except for the first dot which is an introduction to the task made by the tutor Task 1 Students do the task and video record themselves (Mr Brown) without giving students any theory about the pronunciation targets (in Study 1 students are given theory before the task) The course tutor corrects Task 1 and provides corrective feedback embedded in the student's output (in Study 1: Edpuzzle; in Studies 2 and 3: screencasting software). Instructs student to repeat the task.

Task 2 Student video-records himself and repeats the task
The course tutor provides corrective feedback for task 2 and general feedback about the activity.