Young Musicians' Learning of Expressive Performance: The Importance of Dialogic Teaching and Modeling

Until recently little was known through systematic research about effective teaching methods to enhance children's expressiveness in music performance. A previous experimental study indicated that a dialogic teaching approach, consisting of questions and dialogue, improves pupils' expressive performance. Developing from this, a participatory action research study was conducted with the following objectives: (1) To explore how dialogic teaching and learning of expressiveness can be used in weekly individual instrumental lessons; (2) to investigate whether instrumental tutors find a dialogic teaching approach useful for facilitating pupils' learning of expressiveness; and (3) to explore what other complementing instructional modes tutors would like to employ. (4) To investigate pupils' views on their learning of expressiveness; and (5) pupils' views on the instructional strategies used for teaching expressiveness. Five instrumental music tutors participated in this research with two or three of their pupils (11 girls in total, aged 8–15, playing various instruments) for 4 months. Pupils played in informal performance sessions at the start, middle, and end of the project. Lessons and performances were video-recorded. Music diaries, questionnaires and video-stimulated recall interviews were used to collect information about participants' views. Participating tutors used mainly dialogic teaching, modeling, and playing along with pupils. Tutors thought that teaching and learning expressiveness is a complex process wherein “everything is intertwined”; several methods can be used within a dialogic teaching approach for working on various teaching aims. Aural modeling combined with dialogic teaching was seen as especially useful. Pupils' accounts indicate that they had learned to think about the musical character and how to convey this in performance. Tutors' questions had stimulated pupils' reflection and raised their awareness of the musical meaning, while teachers' modeling had helped to build up an aural picture of the music which had facilitated pupils' learning. The dialogic teaching approach supported by modeling had generated improved expressiveness in lessons and contributed to a growing sense of achievement, confidence, self-efficacy, and musical agency. These findings demonstrate the importance of dialogic teaching supported by modeling for meaningful instrumental music education as this can stimulate pupils' thinking, thus facilitating their learning and enhancing their expressiveness.


INTRODUCTION
Research has demonstrated that young children can be expressive in their songs and musical play (e.g., Moorhead and Pond, 1942;Moog, 1976;Tafuri, 2008). Consequently, one would expect them to have the ability to be expressive in their instrumental performance when they grow up. However, some studies have shown that the focus of instrumental lessons and practice in the early stages of Western classical music learning tends to be on technique and reading from notation and that there is limited attention for expressive communication (e.g., McPherson and Renwick, 2001;West and Rostvall, 2003;Karlsson and Juslin, 2008;Lisboa, 2008;McPherson et al., 2012). Some have suggested that the reason for limited instruction in this area might be that musicians' knowledge of expressivity is often intuitive (Lindström et al., 2003;Juslin et al., 2004). Therefore, this study is part of a project that set out to investigate instructional strategies for developing expressive performance to provide educators with tools for enhancing young musicians' expressivity.

Working Definitions
For this study we use the following working definitions: We understand learning as "a process that leads to change, which occurs as a result of experience, and increases the potential for improved performance and future learning" (Ambrose et al., 2010, p. 3). Teaching can be defined as guiding and facilitating the learning process to generate understanding. Teaching methods or instructional strategies can be described as actions used by tutors to facilitate student learning. Dialogic teaching refers to instruction that is characterized by asking questions and using dialogue rather than teacher presentation (Alexander, 2008(Alexander, , 2010. In this project we explore pupils' (aged 8-16) learning of expressive performance of works that are embedded in Western classical musical style. We interpret music as "that form of interhuman communication in which humanly organized non-verbal sound can, following culturally specific conventions, carry meaning relating to emotional, gestural, tactile, kinetic, spatial, and prosodic patterns of cognition" (Tagg, 2012, p. 44). Both the perceived and intended meaning of a musical work tends to be connotative and ambiguous (e.g., Cross, 2005), and affected by various factors, such as the socio-cultural circumstances, experiences and environment of composer, performers and listeners (e.g., Tagg, 2012). Although some contend that music can mean nothing beyond itself (e.g., Small, 1999), others maintain that music can be perceived as expressive of, or referring to, various extra-musical ideas such as affects (Cespedes-Guevara and Eerola, 2018) or emotions (e.g., Gabrielsson and Juslin, 1996;Gabrielsson and Lindström, 2010). Notation traditionally used for Western classical music cannot represent all the intuitive aspects of an expressive performance (e.g., Howat, 1995;Palmer, 1997). Because of the ambiguity of music's meaning (cf. Ashley, 2017) and the limitations of music notation, musicians must interpret a musical work to give a meaningful and expressive performance. Empirical research has demonstrated that musicians employ expressive devices, such as articulation, dynamics, tempo, timing, and ornamentation, to communicate their interpretation of the musical structure (e.g., Sloboda, 1983;Clarke, 1988;Palmer, 1996Palmer, , 1997 and character (e.g., Gabrielsson and Juslin, 1996;Gabrielsson, 1999;Timmers and Ashley, 2007). For this study we define an expressive music performance as a performance in which the musician communicates their interpretation of the compositional structure and musical character of a work convincingly to a listener (Meissner, 2018;. Expressiveness describes the expressive quality of a performance. We use musical character to refer to the affects, atmosphere, emotions, ideas, imagery, or motions that can be associated with a musical work (Shaffer, 1992(Shaffer, , 1995Brendel, 2011). A detailed discussion of expressiveness in music performance across styles and cultures can be found in Fabian et al. (2014).

BACKGROUND: TEACHING YOUNG MUSICIANS PERFORMANCE EXPRESSION
Several studies have investigated methods for improving tertiary students' performance expression. This body of research has shown that aural modeling, verbal explanation using metaphors and verbal explanation addressing concrete musical issues and technical aspects of playing (Woody, 2000(Woody, , 2002(Woody, , 2006 as well as constructive feedback (e.g., Hallam, 1998;Woody, 2001Woody, , 2003 can help to improve students' expressiveness (see Meissner, 2017 for an overview). However, research conducted with adults cannot be automatically extrapolated to children, as the acquisition of expressive performance skills by young musicians might be accomplished by other means, depending on their age (cf. Bonastre et al., 2017;Bonastre and Timmers, 2019), development or ability (Woody, 2006;Meissner, 2017).
A literature search shows that to date little is known through systematic research about effective teaching methods for developing children's expressiveness in music performance. Davidson et al. (2001) proposed that teachers should employ movements and gestures to help pupils experience and understand the direction and intention of the music they are learning. Likewise, Nijs (2017) suggested integrating movementbased technologies to provide children with visual feedback, emphasizing embodied experiences in music making.
Some small-scale studies explored methods for teaching children expressiveness through lesson observation and interviews. Brenner and Strand (2013) observed four tutors who used modeling, verbal instructions and task repetitions. These tutors tended to exaggerate dynamics and melodic contour, enlarge gestures, and overdo aural modeling in their work with children. Two tutors in McPhee's (2011) study with teenagers used various approaches for fostering expressiveness, including analysis, work on articulation, dynamics, technique and phrasing, marking in the score, and metaphors. She thought that several strategies were effective for facilitating expressiveness, provided students realized how their playing improved. McPhee proposed that it might be useful to give teenagers opportunities for making interpretative decisions. This is in line with findings by Davis (2011), who noticed that giving 10-year-old band players a "musical say, " regarding interpretation and the use of expressive devices, facilitated the development of ownership, agency, and expression. Additionally, Broomhead (2005) observed that providing students with problem-solving opportunities can facilitate the development of expressive performance in secondary school choir rehearsals.
A few experimental studies compared methods for improving teenagers' expressiveness. Broomhead et al. (2012Broomhead et al. ( , 2018 reported that boosting confidence during pre-performance routines facilitated teenagers' expressiveness in singing. Chester (2008) compared aural modeling, concrete verbal instruction, metaphors, and no instruction, but found no significant differences between these methods. Lesson duration had been short, and her sample consisted of a mixed ability group which might have influenced her findings. Vandewalker (2014) found that wind-band players' use of dynamics, tempo, and noteduration had changed more after instruction consisting of aural modeling and teaching using metaphors than after instruction consisting of reading instructions in a score or concrete verbal explanation. Chester and Vandewalker's studies were conducted with a short intervention and assessed pupils' expressivity by measuring their use of selected expressive tools. In Chester's study, this concerned measurement of articulation, dynamics and ritardando, while Vandewalker measured dynamics, tempo and note duration. The effect of instruction on overall expressiveness was not investigated, leaving scope for further research in this area.
In the aforementioned studies the use of aural modeling, metaphors, concrete verbal instruction, task repetitions, and movements were discussed or investigated. None of these studies explored the use of enquiry and discussion for teaching expressiveness in instrumental lessons. Nonetheless, questions and dialogue can be important meaning-making tools. Vygotsky (1986) proposed that understanding is constructed through spoken language and involvement in social events. Via enquiry and problem-posing teachers can invite their students "to participate actively in reshaping their own understanding of reality" Murakami, 2016, p. 13 referring to Freire, 1993). Thus, a dialogic teaching approach consisting of open questions and dialogue to stimulate and extend pupils' thinking (Alexander, 2010) fits within a constructivist learning theory. From a constructivist perspective, children need to construct their understanding of how to play expressively via problemsolving activities (Broomhead, 2005; Von Glasersfeld, 2012). As music notation cannot represent all the intuitive aspects of music, questions and problem-solving activities can facilitate the development of children's understanding of, and reflection on, musical character and structure. It can be helpful to consider the musical character using metaphors referring to emotions, imagery or ideas during the learning process (see Langer, 1957;Leech-Wilkinson and Prior, 2014).
In a previous exploratory project conducted by HM, nine instrumental teachers explored methods for facilitating pupils' (aged 9-15) expressiveness. Tutors in this study found that various methods can be employed for teaching children expressiveness, including enquiry and discussion, modeling, imagery, movements, listening to "own" recordings, and "projected performance". Results suggested that enquiry and discussion were useful for improving pupils' expressiveness (Meissner, 2017). This teaching strategy was confirmed to be effective in an experimental study, comparing the outcome of an experimental lesson using dialogic teaching with a control lesson focusing on accuracy and technique. Results demonstrated that a dialogic teaching approach is indeed effective for improving pupils' emotional expression and overall expressiveness . Several participants who had been taught via dialogic teaching explained in video-stimulated recall interviews that the questions relating to musical character and the use of expressive devices had been helpful for their understanding of the "musicality" of their pieces, as this had facilitated their reflection on the interpretation . As these studies have demonstrated that questions and dialogue regarding the musical character and expressive devices are effective for improving pupils' expressiveness, the following research questions arise: (1) How can dialogic teaching and learning of expressiveness be used in weekly individual instrumental lessons? (2) Do instrumental tutors find a dialogic teaching approach useful for facilitating pupils' learning of expressiveness? (3) What other instructional modes complementing this teaching approach would instrumental tutors like to employ for facilitating their pupils' learning of expressiveness? Several methods that were considered to be effective by teachers in the exploratory study, such as aural modeling, gestures, listening to "own" recordings, imagery or "projected performance" could complement a dialogic teaching approach of expressiveness and serve to illustrate the dialogue about the musical character and the use of expressive tools (Figure 1). Additionally, it would be important to explore pupils' views: (4) What are pupils' views on their learning of expressiveness? (5) What are pupils' views on instructional strategies used for teaching expressiveness?

METHODS AND MATERIALS Method
A participatory action research project (ARP) was organized to investigate the use of a dialogic teaching approach combined with supplementary instructional modes, for teaching and learning of expressiveness in weekly individual instrumental lessons, for pupils who were learning pieces from conventional music notation. In participatory action research (AR) practitioners explore an aspect of their work with the aim to improve it, or to understand it better (e.g., Altrichter et al., 2008). In a participatory paradigm, practitioner-researchers are part of the phenomenon under study, as they influence, and are being influenced, by the teaching-and-learning process they investigate (Cain, 2012, referring to Heron andReason, 1997). According to Cain (2012) participatory AR should fulfill the following conditions: it should include self-study; involve students; consider the influence of context; involve more than one action research cycle; and engage with, and contribute to, the development of theory (ibid., p. 409). We decided to organize this project in collaboration with other tutors, as AR is strengthened through cooperation with colleagues (Cain, 2012). From literature and experience we were aware FIGURE 1 | Theoretic model for teaching and learning expressive music performance, incorporating supplemental strategies in support of the dialogic teaching of expressiveness.
of potential pitfalls of AR in an educational setting: There is a risk that not the research but other events, like exam preparation, would be a priority during the project (Cain, 2013); colleagues might forget about the research or be concerned that their teaching might be criticized (Meissner, 2017). To avoid these difficulties, we ensured there was sufficient information about the aims and commitments involved prior to the start of the project. Additionally, HM ensured that she met all participating teachers on a weekly basis throughout the project, during informal moments, such as lunch breaks. These encounters also provided an opportunity for exchanging experiences and reflection. There are also important benefits to AR in educational settings: Collaboration among colleagues can be enjoyable in an otherwise isolated teaching environment (e.g., Hartwig, 2014); it fosters self-knowledge, growth and professional development for participating tutors and encourages a better understanding of pupils (Cain, 2012).
Several of HM's colleagues had expressed an interest in this research and were keen to explore teaching of expressiveness, as they wanted to help their pupils to improve their performance expression. These tutors were invited to participate. Thus, all tutors in this project collaborated to explore teaching expressiveness, and all were "researcher-teachers". For HM's colleagues the aim of the study was to explore teaching of expressiveness and to collaborate with others. For HM the aims were to reflect on her own teaching, to share ideas with, and learn from colleagues, and to systematically evaluate a dialogic teaching approach in combination with other instructional strategies in weekly teaching practice (cf. Altrichter et al., 2008, referring to Elliott, 1984. Thus, this study can be seen as a kind of "field

Nina13_R4
Recorder 4 Participating teachers were assigned a pseudonym while pupils were allocated a pseudonym followed by a code referring to their age, instrument and grade level. VSR-Interviewees in italic. experiment"; a "systematic and reflective trying-out" of strategies (Altrichter et al., 2008, p. 199) to increase our understanding of the teaching-and-learning of expressive music performance.

Participants
Five instrumental music teachers (one male, four female) took part in the study (Table 1): a trumpetist teaching brass instruments, a clarinetist/saxophonist, a pianist, a violin/viola teacher, and a recorder teacher (HM). All participating teachers had studied music at HE and had extensive teaching experience (≥10 years). Eleven girls (aged 8-15, at the level of Grade 1-5 1 ) participated in the project; two pupils for each tutor, while the pianist took part with three pupils. Participating pupils played French horn (1), Flugelhorn (1), clarinet (2), piano (3), violin (1), viola (1), and recorder (2). Teachers invited pupils who they thought needed help to improve their expressiveness 2 . Additionally, two recorder pupils who were not participating in the ARP were interviewed as they expressed views on learning expressiveness that were of interest for answering the research questions. All five tutors aimed to teach expressivity and had used various methods for this in the past ( Table 2). All tutors were open to changing their teaching approaches and looked forward to sharing ideas.

Research Ethics
Approval for this study was obtained through the standard University of Sheffield ethics review process. Participating teachers as well as pupils and their parents gave their written informed consent.

Procedure
As a rule, AR consists of at least two action Cycles containing the following steps: First the current situation is examined and a research question formulated; then an intervention is planned and implemented in an action cycle; the outcome is evaluated and a new intervention is planned (e.g., McNiff, 2010;Cain, 2012). The current ARP consisted of two Cycles of 4 or 5 weeks each (Figure 2). A meeting providing information about the aims and commitments of project participation was organized prior to the start of the project.
The project began with a workshop for participating teachers which provided an overview of previous research relating to performance expression, explaining methods through which this can be enhanced and taught. This workshop was followed by M1 wherein tutors discussed and agreed which methods they aimed to explore during the first Cycle. The first Cycle started and ended with a pupils' performance session and consisted of 5 weeks of one-to-one teaching. The first Cycle was followed by M2, in which updated objectives were agreed. M2 initiated Cycle 2, which continued until the end of term and consisted of 4-5 consecutive weeks of one-to-one teaching. At the end of Cycle 2 a third performance session was held, followed by M3.
Near the end of Cycle 1 we decided to make a video containing highlights of tutors' work to share ideas in M2. Each tutor had the opportunity to watch the extract from their lessons prior to the meeting, and tutors were asked for their consent to use this for the "highlights video, " which informed the discussion and decision making in M2. This procedure was repeated at the end of Cycle 2.
It was decided during Cycle 2 to organize a fourth performance session at the end of the term following the project, to see whether there was still an effect of the project on participants' practice. It was our aim to create a relaxed atmosphere during pupils' performance sessions and therefore we preferred not referring to these as concerts. Only participating pupils and tutors attended, as well as someone who filmed performances. No friends or parents were invited, as it seems likely that performance anxiety will increase with a larger audience (LeBlanc et al., 1997). Tutors were asked to choose easy repertoire for their students, especially for the first performance, to prevent high anxiety levels due to difficult task level (cf. Kenny and Ackermann, 2016;Papageorgi and Kopiez, 2018). However, difficulty levels of pieces varied per student and across performances (see Expression vs. Technique below).
Three months after Cycle 2 video-stimulated recall interviews (VSRI) were held with all tutors and several pupils to investigate participants' views on their teaching and learning during the project.

Material and Data Collection Instruments
Various data collection instruments were used to gather viewpoints from all participants and for triangulation of data. This provided a rich picture of the teaching-and-learning situation and an opportunity to see whether findings were consistent (Altrichter et al., 2008;Hartwig, 2014). The following methods for data collection were used: Research journal; Pupils' and tutors' questionnaires; Pupils' music diaries and tutors' notebooks; Notes and audio recordings of teacher meetings; Video-stimulated recall interviews; Video recordings of the "expression-phase" of lessons; Video recordings of pupils' playing in performance sessions.

Research Journal
As recommended by Altrichter et al. (2008), HM kept a handwritten research journal throughout the project to document notes from teacher meetings, informal conversations with teachers, observations during lessons with her own students, pupils' performances, and reflections. Regular journaling was useful as it served to develop ideas about the research and for preparing meetings. Afterwards all entries from the journal were transferred to a word document which was added to an NVIVO file for coding and thematic analysis (see Data Analysis).

Pupils' and Tutors' Questionnaires
To investigate tutors' views on teaching, and pupils' views on learning expressiveness, questionnaires were given to participants at the start and end of the project (Q1 and Q2). Pupils' questionnaires consisted of open-ended and scalar questions on a five-point rating scale. The response options in the scalar questions were completely labeled, and the scalar and multiple-choice items were explained to the youngest pupils (8-10-year-olds). The confidential nature of the survey was emphasized in information letters and verbal explanations. The student questionnaire was tested by two 9-years old TABLE 2 | Overview of methods used by teachers prior and during the action research project, and methods described as effective afterwards.

Teacher
Methods used prior to ARP Methods used during ARP Methods seen as most effective-end of ARP girls who indicated that questions on "shape" or "structure" of music were confusing. Therefore, these items were reworded, and it was explained to participants what "structure" meant (Appendix 1). The final pupils' questionnaire contained open-ended and scalar questions investigating students' views on the instructional strategies used by teachers (Appendix 2). Additionally, there were questions exploring pupils' views on their favorite piece and most enjoyable experience during the project, and there was space to leave feedback.
Tutors' questionnaires at the start of the project (Appendix 3) explored their views on, and experiences with teaching expression. Additionally, tutors could indicate how they had been taught expressiveness before and during their study at HE. The first questionnaire was given to tutors during the workshop at the start of the ARP. Questions regarding tutors' approach to teaching expression were completed at the start of the workshop before the presentation on teaching expressiveness. The remainder of the first questionnaire was completed after the workshop. Items in teachers' final questionnaire (Appendix 4) investigated their views on the usefulness of instructional strategies. All questionnaires were returned, except for one pupil's questionnaire at the end of the project.

Pupils' Music Diaries and Tutors' Notebooks
Basic music diaries were given to participants to explore their views and experiences during the project. Pupils were given a music diary to describe weekly what they had aimed for during practice and what they had learned during their lesson (Appendix 5). Additionally, there were questions on the performance sessions. A draft diary for pupils was evaluated by a teenager who recommended avoiding open questions on practice and performance as these can be difficult to answer. Based on her feedback the diaries for pupils were adjusted with fewer open questions while questions with tick boxes were added.
Teachers' notebooks provided space for short descriptions of methods used and observations during lessons (Appendix 6). These notebooks provided an opportunity for teachers to reflect on the teaching-and-learning process on a weekly basis. Most teachers used these notebooks, except for the pianist.

Notes and Audio Recordings of Teacher Meetings
To investigate tutors' experiences with, and their views on the instructional strategies, teacher meetings were audio recorded and HM made notes in the research journal during and immediately after meetings. Due to some technical problems the audio recording of the last teacher meeting was corrupted. Notes from the research journal provided information about the content of this meeting.

Video-Stimulated Recall Interviews
Three months after Cycle 2, VSRIs were held with all participating tutors and six pupils to collect detailed information about pupils' and tutors' evaluations of the instructional strategies used during the project. Semi-structured interviews (Appendix 7) seemed an appropriate method and a short videorecording (5-7 min) containing lesson extracts was used as stimulus. For each participant video-excerpts were chosen in which tutor and student were working on expressiveness, usually employing a method that had been selected during a teacher meeting. The selection was often guided by comments from tutors during the project or were identical to extracts used in the "highlights video". Participants viewed their video at the start of their interview via a computer screen and headphones. The aims of the interview were emphasized, i.e., we would not be evaluating tutors' teaching or pupils' performance but rather the methods that had been used. Participants were asked which instructional strategies they had found helpful, and whether any methods had been used which they had found difficult or not so helpful. Pupils' VSRIs lasted between 12 and 28 min (M = 21), while interviews with tutors took between 30 and 45 min (M = 36). All VSRIs were audio-recorded and transcribed in NVIVO.

Video Recordings of the "Expression-Phase" of Instrumental Lessons
Lessons were filmed to verify which methods tutors employed; the aim was not to analyze lessons in detail, but rather to verify which methods were used. Three teachers filmed the section of the lesson when they worked on expression (∼10-15 min), while two tutors preferred filming whole 40-minutes lessons rather than switching on the camera for work on expressiveness. Every week the "expression-phase" of the lesson was uploaded on a secure online file. Due to technical problems a couple of violin/viola lessons were not recorded. Notes from the violin teacher's diary provided information about these lessons. HM looked at all the video material every week and RT looked at some of this material to inform her interpretation. The tutors watched the "highlight videos" of their lessons before and during the meetings.

Data Analysis
Qualitative data from HM's research journal, questionnaires, pupils' music diaries, teachers' notebooks, and transcriptions from meetings and VSRIs were coded in NVIVO and analyzed with the 6-phase thematic analysis procedure as described by Clarke (2006, 2012). Because of the large data set, the second step of the procedure consisted of three parts. First, "nodes" were labeled in NVIVO for concepts that might become a theme (2a). Next, these nodes were assembled into "codes" (2b). Subsequently, codes were revisited throughout the data set (Braun and Clarke, 2012). These codes formed the basis for the themes that emerged in the third phase of the analysis process. The analysis was abductive (Aliseda, 2006) because our conclusions are intended to be the most plausible explanation of the collected data.

RESULTS
This results section starts with a description of tutors' decisions and actions during the project (see Tutors' Decisions and Actions). Next tutors' (Tutors' Views on Teaching Expressive Performance) and pupils' (Pupils' Views) views on teaching and learning of expressiveness will be reported. Subsequently, some issues that appeared to hinder expressiveness will be considered (Issues That Hinder the Development of Expressiveness). An overview of the main codes and themes can be found in Figure 3.

Tutors' Decisions and Actions
From their reports in the first questionnaire, which was handed out at the start of the workshop, it seems that none of the tutors had been in the habit of using open questions about the character of the music before the start of the project 3 . The teachers reported that they had used various methods for working on expressiveness: the brass, clarinet, and violin teachers had used mainly verbal teaching explaining technical aspects of playing or analysis of pieces; the piano, violin and recorder teachers had used modeling. Additionally, all tutors had used metaphors describing what the music should sound like ( Table 2).
During M1 the theoretic model (Figure 1) for teaching expressive performance was presented and discussed. The tutors decided unanimously to focus on questions and dialogue concerning the musical character supported by aural modeling for teaching expressiveness in Cycle 1. Video material revealed that the clarinet, brass and piano teacher experimented with asking questions related to the interpretation supported by modeling and gestures. Linda 4 , the violinist, focused mainly on accuracy and technique using verbal instructions, gestures and modeling (singing and playing). She was aware of her focus on technical issues but thought her pupils should learn the notes and techniques first, in order to play their pieces. Afterwards she acknowledged that her pupils' pieces had been too hard and therefore required more work on technique. Linda addressed performance expression by talking about dynamics and via singing to work on phrasing.
In M2 tutors discussed their experiences and shared ideas about methods they had used during Cycle 1. Watching the "highlights video" helped to evaluate the instructional strategies and to generate ideas for teaching. Teachers thought that asking questions and modeling had been effective and decided to continue asking questions concerning the interpretation and various other aspects of pieces, supported by modeling.
Additionally, some tutors noticed that playing along with pupils can be useful. Tutors decided to use these strategies more, and to complement these with other methods from the theoretic model whenever this would be appropriate. The video material showed that especially in the first week of Cycle 2, teachers explored ideas that had been suggested in M2. The piano and the clarinet teachers continued asking questions about the musical structure and character, reminding pupils of this in subsequent lessons. Additionally, Alicia, the pianist, used short verbal explanations containing metaphors supported by modeling and gestures. Caroline, the clarinetist, used open questions on all aspects of the music, including questions to evaluate playing; "How do you think that went? What can we improve?" She sometimes asked pupils to concentrate on the character instead of notes or technique. Her verbal instructions about rhythm and articulation were supported by modeling. Tim, the brass teacher, started using more modeling and playing along with pupils in Cycle 2, and occasionally asked pupils to project their performance. Tim mentioned that his pupils found it hard to imagine pitch and pulse, they had technical difficulties and were feeling insecure about their playing. Therefore, he decided to use modeling and playing along, as this increased their awareness of pitch, pulse and rhythm and helped these pupils to feel more confident (cf. Hallam, 1998). In the beginning of Cycle 2, Linda asked her pupils questions about the character of the music, expressive marking and tempo indications. In subsequent lessons Linda used mainly verbal explanations of technical issues, supported by modeling and playing along. In recorder lessons HM used questions concerning the musical character and how to convey this, supported by modeling and playing with pupils; sometimes playing along with them when they experienced difficulties with accuracy or technique, but mainly accompanying on bass recorder. Furthermore, her pupils were invited to imagine performing in a big space, and sometimes listened to recordings of their own playing.
Overall, video recordings of lessons revealed that the clarinetist adopted the greatest range of methods. Contrastingly, the violinist implemented fewer methods and varied less, as she primarily used concrete verbal instructions regarding technique and accuracy. She complemented this verbal instruction with modeling, singing, gestures, and some questions and discussion to work on expressiveness. Overall, the clarinetist and the recorder teacher employed more questions concerning the musical character and referred more to this throughout the project than the other tutors. It might be that they had more opportunity to do this than the others, as the pieces their pupils were studying were within their current level of ability or Zone of Proximal Development (Vygotsky, 1978).

Tutors' Views on Teaching Expressive Performance
Everything Is Intertwined Using the theoretic model for teaching expressiveness (Figure 1) as a starting point, participating tutors reflected on methods that can be used for teaching expressiveness. Our thinking about the connections between methods and teaching aims developed Frontiers in Education | www.frontiersin.org 9 February 2020 | Volume 5 | Article 11 throughout the project and HM adjusted the figure several times until everyone agreed (Figure 4). The tutors observed that "everything is intertwined" in instrumental teaching; it is not possible to differentiate which method is useful for a particular aspect of performance, as several methods can be used within a dialogic teaching approach for working on various teaching aims.
Modeling and playing along with pupils, for example, can help to develop technique and accuracy as well as phrasing; singing can be used to practice phrasing but also to work on accuracy. Furthermore, our teaching practice showed that all instructional strategies contained in this paradigm; modeling, playing with pupils, movements and gestures, and so on, can be used within a dialogic teaching approach. Asking questions and exchanging ideas in dialogue can be adopted for working on various aspects of playing; interpretation of musical character and structure, rhythm, pitch, performance directions, practice and evaluation of playing. Various methods can facilitate, support and illustrate the ideas that are explored in the pupil-tutor dialogue. Additionally, participating tutors thought that there can be no uniform approach but that methods should be tailored to the student and the situation. Tutors thought that instructional tools should be adapted to individual pupils because children have diverse strengths and weaknesses, and various factors affect the complex process of instrumental music learning. This is in line with recommendations by Hallam (1998Hallam ( , 2006 and tutors in Brenner and Strand's (2013) study, who mentioned that it is impossible to prescribe one ideal method for teaching all students since children develop at different rates.
Although participating tutors were convinced that "everything is intertwined" and teaching expressiveness is a complex procedure that should be adjusted to individual pupils and circumstances, a few methods stood out for them because of their usefulness, namely questions and dialogue; modeling; and to some extent playing with pupils: I think different methods are effective at different times and different students. However, the two basic ones for me are: 1. asking "what's the character of the piece?" 2. Modeling (Alicia, Q2).

Dialogic Teaching Approach Questions and dialogue concerning the musical character
Alicia's view that questions about the musical character combined with modeling are at the heart of teaching expressiveness, was shared by most teachers (4/5). Tutors used questions and modeling from the beginning stages of learning a piece, and Alicia commented that this was useful as it helped pupils to "land"; it helped them to realize what to aim for in their practice and playing. Open questions to stimulate pupils' thinking about the interpretation were used, and pupils responded with insightful ideas: Several tutors used authentic questions followed by a short tutor-pupil dialogue. Alicia and Caroline asked questions to facilitate thinking about expressive tools that can be used to convey the interpretation, and pupils were invited to explore this in their playing. They used an active listening approach, repeating pupils' answers, accepting their ideas, and asking further questions (Gordon and Burch, 2003;Hutchby, 2005). Furthermore, Caroline and Alicia returned to pupils' ideas in subsequent lessons. During the M2 tutors stressed that it is important to emphasize that there is no wrong answer to questions about the interpretation of musical character (cf. Timmers and Honing, 2002) and that it is important for pupils to reflect on the meaning of their music for them personally. This approach can be valuable for pupils' sense of identity as musician (Hallam, 2010) and for their development of agency, as it provides them with an opportunity to think of their own musical ideas (Wiggins, 2016).
Caroline and Alicia commented on the usefulness of asking pupils for their views and of reminding them of the character. The video-material shows how reminding a pupil of their chosen musical character immediately affected their playing, making it more expressive. Caroline talked about this in her interview: . . . I think certainly, talking about the character, expression, just seems to elevate their performances (. . . ) maybe it was just something I assumed they would automatically do once, once dynamics were there, once notes were there, I automatically kind of assumed it would, it would happen. (. . . ) I asked those questions about character and expression and suddenly it was there. Just the fact that I said the words, character and expression, I think that was a real light bulb moment for me (Caroline, VSRI). This demonstrates how Caroline had thought that executing all expressive markings contained in a piece should be sufficient for generating expressiveness. However, she observed that asking questions about the character had facilitated pupils' reflection on the interpretation and this had helped them to understand 6 [text] is used to clarify what the transcriber thought that the child meant or if the transcriber added a word. (text) is used for utterances that were difficult to understand. The brackets indicate that this is the most likely possibility of what the child said. (...) indicates that the authors have left something out, usually because it is impossible to hear what is being said, and occasionally because the complete text does not make sense or would be too long to cite. (( )) Transcriber's description of event. their music better, thus improving their expressiveness in lessons. Although she said in this extract that simply mentioning character and expression was enough, she later realized that this had been effective because she had previously asked her pupils about their views on the musical character and how this can be conveyed. When she reminded her pupils of the character afterwards, they concentrated on expressing the musical meaning, rather than focusing on notes or technique.
Although these tutors were convinced of the effectiveness of a dialogic teaching approach for pupils' learning of expressiveness, Alicia did observe that this was not always apparent in performance sessions, probably due to issues with performance anxiety (see Performing and Music Performance Anxiety).

Perceptive ideas
Pupils' ideas about the musical character were varied, ranging from metaphors like "happy" or "solemn, " to imaginative descriptions such as "summery, bird-songy, " "clowns at a circus, " "someone dying, losing hope." Like teachers in a previous project (Meissner, 2017) some tutors were impressed by pupils' perceptive answers. Alicia observed that pupils are not "empty bowls" but have insightful thoughts about their music. She had never realized that pupils could have such a profound understanding of the musical meaning of their pieces, even before studying or discussing these (cf. Davis, 2011). Linda remarked that pupils sometimes express ideas in an unexpected way or offer surprising views, suggesting that she might expect certain answers to her questions regarding the interpretation. It seems likely that tutors have their own views on interpretations, and it may be difficult to be open to diverging ideas from pupils.

Dialogic teaching is interactive
Some teachers observed that a dialogic teaching approach made pupils more active and more engaged in the learning process, which is in line with findings related to dialogic teaching in classrooms (Alexander, 2008). Linda reported that dialogic teaching engaged pupils more, and "they feel more responsible for their actions in the music." Likewise, Alicia noticed that asking questions makes "everything more interactive, " and that it "helps students to think more about the interpretation." Additionally, she reported that dialogic teaching had made her "listen to what pupils have to say more." According to Alicia her pupils were "more involved in what they were doing;" "more aware and active, " more "alive" because of her dialogic teaching approach. This was evident in Yasmine's behavior, as she started asking questions, which according to Alicia, she had never done before. As there was more tutor-student interaction wherein pupils expressed their ideas Alicia had gotten to know her students better (cf. Cain, 2012).

Turning verbal teaching into action can be problematic
Contrastingly, Tim thought that talking about the music was less effective than modeling or playing along with pupils. In M1 Tim observed that "verbal teaching" was appropriate for advanced students because "at advanced level you've got to think for yourself, " while modeling is probably best for teaching younger, less advanced pupils. According to Tim "younger students tend to need 'spoon feeding'." Additionally, he explained how he had been taught via aural modeling before he went to Music College, but that his teacher at HE had used mainly verbal teaching, which he had found frustrating at the time.
In the final questionnaire he reported that "too much talking could sometimes confuse/distract" and he wondered whether it "might be hard to turn verbal information into practice." Research findings confirm that too much talking is ineffective (West and Rostvall, 2003;Karlsson and Juslin, 2008) and that "concrete verbal teaching" is less effective than modeling (Ebie, 2004;Vandewalker, 2014) and takes more practice time (Woody, 2006). However, especially in Cycle 1 Tim used open questions in lessons to stimulate pupils' thinking about accuracy and interpretation rather than "concrete verbal teaching." His questions were authentic, and he did take pupils' replies seriously. It seems likely that Tim's views were influenced by his work with pupils who had technical difficulties and struggled with some essential aspects of playing; imagining pitch and awareness of pulse. It might well be that modeling and playing along with pupils is the best approach in such situations as it can assist pupils in building up an internal representation of the music, thus improving accuracy and building up confidence (Hallam, 1997(Hallam, , 1998. Even so, his pupils seemed to like the questions regarding character and Lucy mentioned the work on musical character as something that had been special for her learning (see Pupils' Views on Instructional Strategies for Working on Expressiveness).

Aural Modeling
All participating teachers reported that modeling had been helpful for teaching various aspects of music performance, and Linda thought it especially useful in combination with questions or verbal teaching to draw pupils' attention to a particular aspect of playing. Likewise, Caroline had started work on new pieces by listening to recordings with her pupils and asking questions about the character and various other aspects of the work. It might well be that modeling is especially effective when used in combination with questions and dialogue in which tutor and pupil evaluate what they hear. Teachers can help their pupils to focus on aspects of the modeled performance by asking questions.
Caroline had hardly used modeling in the past, as she initially thought that this would hinder the development of sight-reading skills. The idea that modeling might not be an appropriate tool was also apparent in previous research as some teachers thought that modeling might hinder pupils' thinking about their own interpretation (Meissner, 2017) and this might explain why modeling is not used frequently in instrumental teaching (Dickey, 1992, although see Bonastre and Timmers, 2019 for contrasting findings).

Playing With Pupils
Most teachers (4/5) found it useful to play along with pupils when they struggled with rhythm or pitch. Tim used this frequently in Cycle 2 and mentioned that this helped his pupils to continue playing, to maintain a steady pulse and to play with correct rhythm. However, he observed that as soon as he stopped playing, they "fell off the wagon." Video recordings of lessons suggest that playing along might be useful for supporting pupils in the early stages of learning (Hallam, 1998). Alicia who had played and sung with her pupils during the project said afterwards that playing along was not as effective as asking questions and modeling. She made a distinction between playing along with pupils and accompanying students. Alicia thought that accompanying provides support for pupils to play with direction and to make crescendos or diminuendos. Contrastingly, playing the same melody with pupils was helpful while it was happening, but the effect disappeared when pupils played on their own afterwards. This is in line with Tim's view that pupils tend to "fall off the wagon" when he stopped playing with them.

Pupils' Views on Their Learning of Expressiveness
In the questionnaire at the end of Cycle 2 pupils were asked for their views on their learning during the project. Two main themes came up in pupils' descriptions of their learning: they had learned (1) to think about the musical character and (2) how to express emotion or feeling in playing. Awareness of the musical meaning of their pieces, and the realization that they had to think about this had been an eye-opener: I think it's important to think, it's important to think about the mood you are trying to convey (Sophia14_Cl3, Q2).
For these pupils this had been new, and some described in their interview how this was different to their previous attitude to practice and playing: Well, before I was like concentrating on the notes, and I was. . . and I just did dynamics, but it never really occurred to me that [I] also need to play it like what it's supposed to mean (Rachel12_Va2, VSRI).
This illustrates how pupils tended to concentrate on "notes" and expressive markings before the project but had come to realize that there was more to their pieces than notes and "just doing dynamics." They had discovered that they should reflect on the musical character and aim to convey this. These accounts are similar to answers from participants in an earlier study ; in both studies students report that they have learned to think about the musical character and how to convey this in performance following teachers' questions.
Pupils' answers in the scalar questions regarding practice are in line with this finding, as a Wilcoxon Signed-Rank Test indicated that there was a significant difference between pupils' pre-and post-teaching scores for Thinking about the piece; what it feels like (p = 0.006, Z = −2.724, r = 0.59). Median values increased from 2 at the start to 3.5 at the end of the project, indicating that pupils had started considering the musical character more. Additionally, there was a moderate change after project teaching on I can express my feelings when I play pieces on my instrument (p = 0.059, Z = −1.890, r = 0.41, Md increased from 3 to 3.5) and I try to play with feeling (p = 0.071, Z = −1.807, r = 0.39, M increased from 3.27 to 4). No other changes in pupils' questionnaire answers were found that were significant or close to significant.

Pupils' Views on Instructional Strategies for Working on Expressiveness
Talking and thinking about the musical character Pupils' views on the usefulness of instructional strategies were similar to those of the teachers. Most pupils (9/10) indicated that they thought that questions about the musical character had been helpful, while one pupil indicated this had been "a bit" helpful. For Lucy talking about the musical character had been special. When HM asked her "what were moments that stood out for you, in terms of learning something about the music?, " she replied I think... we said more about character, more about like the moment ((hesitantly)) and the, mmm ((humming)) I don't know how to explain it, the atmosphere and how it should be played, and not how I played it. . . (Lucy14_FH4, VSRI).
Talking about the musical character and atmosphere of a work is the first thing Lucy mentioned when she was asked what had been special for her during the project. This contrasts with her teacher's view that "verbal teaching" is too complicated for intermediate students. Although she mentioned later that aural modeling had been especially helpful, which is in line with her teacher's view, she mentions talking about the character as the element in the project that stood out for her in terms of learning something new. Likewise, Yasmine explained why she had found it useful to talk about the interpretation: You're talking about how you play it more, rather than just playing it (. . . ). I'd just sort of played it (. . . ) thinking of that character (. . . ) you put yourself in their shoes and you're just able to do it better (Yasmine12_Pi3, VSRI).
It seems that both girls have difficulty describing how talking and reflecting made a difference to their playing. Lucy makes a distinction between "how I played it" and "how it should be played, " while Yasmine describes her default manner of playing as "just playing it." This suggests that both girls used to focus on the notation and how to play accurately, "just playing it, " without much further consideration. However, discussing the atmosphere, thinking about the character or imagining the emotion, had made a difference to them. This is also apparent in Rachel's description above (Pupils' Views on Instructional Strategies for Working on Expressiveness); she used to concentrate on notes and "just did dynamics" but had now realized that it is important to reflect on the interpretation.
Additionally, tutors' questions had helped pupils to reflect on ways to convey the musical character: I (. . . ) found that helpful because it... means I know what I could, kind of convey in like the music, like dynamics wise, and how I want it to feel, so if I want it to feel like, circus-like I would play it like an, like it's a clown kind of thing. So, with the dynamics I'd do like (maybe) loud then quiet, and yes, that was helpful (Sally13_Cl4, VSRI). This suggests that Sally had learned how to adjust expressive tools for playing expressively, and to think of her own interpretation of a work; she had been thinking about what she needed to do to make the piece sound "how I wanted it to feel".

Pupils' responses
Several pupils had responded to their teachers' questions with creative answers (see Perceptive Ideas). Most pupils responded immediately while a few pupils needed more thinking time. Caroline suggested giving hesitant pupils some options to choose from, as she had found this useful in her lessons with Sophia, who had indicated in the questionnaire that the questions had been "a bit" helpful. It might be that some pupils were concerned about giving a wrong answer, as suggested by some tutors, or they might have found it hard to describe the character. Nina mentioned that she found it difficult to describe the character verbally because she thought music expresses something that is more similar to emotions than a character: I just have trouble pin-pointing the exact... word I guess, like I can give a broad kind of generalization but it's all different I guess, you can't just call it one thing. (. . . ) There aren't really enough words to, you know, exact words to describe the specific emotion that you're feeling (Nina13_R4, Interview).
Nina's view on musical character is reminiscent of Langer's (1957) idea that music is expressive of constantly changing feelings and subtle moods which cannot be described adequately in words. Even though Nina found it difficult to describe the musical character verbally, she had found it helpful for her playing to think of a musical emotion that might be communicated by a composition.

Aural modeling: teacher playing for me
Most pupils 7 reported that aural modeling had been helpful for their learning. In interviews pupils explained that hearing their teacher play had been useful for building up an aural picture of the music: Well, I think before I play the piece myself, or even attempt sight reading, I like to hear it first, so hear you play it and hear it played at the speed it's meant to be played at, so that I have a picture in my mind of how it should be (Amelia15_R4, Interview).
For Sally hearing her teacher play was enjoyable, helpful for learning and it motivated her to improve her own playing: I liked listening to my teacher play it because it sounded amazing when she did it, it made me want to play like that (Sally13_Cl4, Q2).
Rachel 12_Va2 mentioned that she found it helpful to see her teacher's playing: 7 8 out of 10 girls indicated in the questionnaire that modeling had been helpful. Although Yasmine 12_Pi3A ticked that her teacher had not used modeling, she said in her interview that modeling had been helpful. Lucy 14_FH4A indicated in the questionnaire that modeling had been "a bit" helpful but in her interview, she said that her teacher's singing and playing for her and with her had been especially helpful. Matilda did not return the second questionnaire.
Well playing together and her showing (me) how to do it rather than like telling me. I prefer to be shown. Because then I can like, just see it properly (Rachel12_Va2, VSRI).
Rachel's account demonstrates that verbal explanation was not helpful if it was not demonstrated in playing. Rachel was the only pupil who mentioned that modeling was useful because of the visual information.
Pupils' views on modeling in this study seem different to those of students in the study by Lindström et al. (2003), who preferred teachers' use of metaphors or focusing on felt emotion over modeling. This difference might be due to their larger sample, different age and level of playing, or to the differing questions, as the questionnaire in the current study asked what had been helpful for pupils' learning, while Lindström and colleagues asked for tertiary students' preferred method for learning expressivity.

Teacher playing with me
Several pupils (7/10) indicated that they had found it helpful when their teacher played with them. However, the questionnaire item had not made a distinction between teachers playing the same melody or an accompanying part. Pippa explained in her interview that she found it helpful to talk about the musical character, to listen to modeling, or to be accompanied on a bass recorder, but she had not found it helpful when someone played along with her. She felt freedom to adjust her playing when she was accompanied on bass recorder, but not when HM had played the same melody along with her. Pippa's opinion is in line with Alicia's view that accompanying is more helpful than playing along. In clarinet and recorder lessons, pupils had experienced tutors' accompaniment on piano or bass recorder, as well as being supported by their teachers playing the same melody. In the string, brass and piano lessons tutors had not played accompanying parts in video-recorded sections of lessons.

Developing Self-Efficacy and Musical Agency
Pupils mentioned playing several pieces or "fun" pieces as the most enjoyable feature of the ARP. Their favorite pieces during the project had been compositions that appealed to them because of the character or mood. In answer to the question "Can you tell me what you liked about this [favorite] piece?" Sophia replied: I liked the mood and style of this piece, and that it is slightly more challenging than the other pieces, so it made it a more interesting process learning it. I also like the piano accompaniment to this piece and how the piano and clarinet parts are intertwined (Sophia14_Cl3, Q2).
Without exception, these pupils mention the musical character, mood or beauty as the reason why they like a particular piece. The difficulty level does not seem to affect their preference as much as the character of a composition. One girl even added that she liked a piece "even though it was hard" and Sophia mentioned that she liked a work because it was "slightly more challenging." It is important for tutors to realize that the musical content and quality might be more important for pupils' preference and motivation than the difficulty level of a work. Furthermore, several pupils observed that they had grown in confidence and had improved their playing, which had been enjoyable: . . . playing the pieces in the lessons, and then sort of each week it gets a bit better so it makes you quite happy because you can tell you're getting better at it (Yasmine12_Pi3, VSRI).
Additionally, some girls had realized that they can make pieces "their own": Playing with expression made it more interesting and relatable. You can make the piece yours (Sally13_Cl4, Q2).
Pupils' stories suggest that these girls experience a growing sense of achievement, self-efficacy 8 (Bandura, 1989) and musical agency 9 (Wiggins, 2016); they can learn pieces faster, they have learned to think about the interpretation and to work out their ideas in playing. The focus on expression and reflecting on their own interpretation had given them a "musical say" (Davis, 2011); a sense of ownership and musical agency. They were more aware of their ability to "make the piece their own, " and it seems likely that this had helped them to develop an identity as "musician in their own right" (Hallam, 2010). For some, the focus on interpretation and expression had reduced the pressure of playing without mistakes, or the feeling that they had to compete with peers.

Expression vs. Technique
Tutors' became increasingly aware of issues that may hinder learning of expressiveness. Several teachers (brass, violin and piano) observed that some pieces had been too challenging, or that insufficient practice or technical difficulties had hindered work on expressiveness. When pieces were too hard pupils needed more time and attention for "notes" and technique in lessons and practice. This is in line with findings in an earlier study (Meissner, 2017) and with Broomhead (2001) who proposed that technical skills are related to pupils' ability to perform expressively. Furthermore, Linda thought that the violin is a very technical instrument and that it is necessary to explain technical tools first in order to teach expressiveness. At some point in Cycle 2 she declared that expressive playing is dependent only on pupils' technical skills. Similarly, the violinist in a previous study had suggested that "a lack of expression is a lack of technique" (Meissner, 2017, p. 127). However, in her final questionnaire Linda wrote that the project had influenced her teaching practice because, It made me more aware of the issues involved in teaching expression. Also, I realized that it can be done despite basic technical problems (Linda, Q2). This suggests that she thought that expressiveness can be taught, even when pupils have technical problems, by discussing the musical character and via modeling. She still thought that technique is crucial for violinists' expressiveness, but she realized that both aspects of playing can be taught at the same time.
The video material of several lessons suggests that concentrating on technique can hinder expressiveness; while pupils concentrated on technique or reading from notation their playing was not very convincing, but as soon as their teacher asked them to focus on expression their expressiveness increased. This phenomenon was also observed by some teachers: . . . Saying, "don't worry about the technical side of this" (. . . ) takes the pressure of them as well and leads them to be able to play more expressively and with more character (Caroline, VSRI).
It seems therefore that difficulties with technique may hinder expressiveness but that concentrating on performance expression can help to improve technical fluency as well as expressiveness.

Performing and Music Performance Anxiety
An important theme that occurred throughout the data set was that feelings of anxiety can hinder pupils' expressiveness in lessons and performances. According to tutors, some pupils were afraid of making mistakes, while most participants were nervous during performance sessions. Near the end of the project it became apparent that performing in front of an audience had been a new experience for several pupils while others had only occasionally performed in public. Evidently, participants' inexperience with performing elicited feelings of arousal and anxiety. Although raised levels of arousal can hinder as well as elevate performances (e.g., Patston, 2014;Kenny and Ackermann, 2016;Papageorgi and Kopiez, 2018), pupils' responses suggested that most participants in this study, except for the youngest, felt very nervous during performances and that this had hindered their playing. Contrastingly, 9-year-old Pippa said that feeling nervous had served to focus during performances and had helped her not to "get overconfident." This phenomenon, that even performing for a small audience can raise arousal levels and induce anxiety, is in line with research by LeBlanc et al. (1997), who found that teenage wind-band players experienced increased arousal and anxiety when they had to play for a group of peers and researchers, compared to playing for one researcher. However, all participants in our study were in the performing situation together as equals; they all listened to and performed for each other, which was different to the study of LeBlanc et al. Several studies have revealed that high levels of anxiety before or during performances is reported more by female than male students (e.g., LeBlanc et al., 1997;Osborne and Kenny, 2008;Patston and Osborne, 2016), although Ryan (2005) found that this was not always the case for children under 12 in her study. As all participating pupils in this study were girls, it is perhaps not surprising that many in this group (the teenage girls) were affected by high anxiety levels. It seems, therefore, that tutors of female students should be particularly alert to signs of performance anxiety and appropriately proactive in coaching these pupils for performance.

Repeated performance experience is beneficial
All teachers and several pupils reported that performing regularly had been useful. It seems that feelings of anxiety had decreased and that the regular performing activities had increased pupils' confidence. Since Caroline had been aware of the potential impact of performance anxiety, she had chosen easy pieces for the first performance session, a slightly harder piece for the second and a challenging work for the third and fourth sessions. Caroline and her student Sally both reported that this gradual building up of pieces for performances had helped to increase confidence.
The finding that repeated performance experience within a short time frame reduces anxiety is in line with findings from a study by Boucher and Ryan (2011) with 3-4-yearolds. They found that second performances that were organized soon after a first performance elicited lower anxiety responses from young children than initial performances. It seems likely that it is easier for young children to become accustomed to performing than for teenagers, and it is important that tutors create frequent opportunities to practice performing in front of a friendly audience from the early stages of learning (Kenny and Ackermann, 2016;Yandell, 2018).

REFLECTIONS
As with all participatory AR, it is important to be aware of factors that could have influenced the research process. Firstly, the experience and former practice of participants might have influenced their assessments of the methods used during the project. Previously, Caroline had used mainly analysis and "adjectives to describe the music" rather than questions about the musical character and how this can be portrayed. She used to avoid modeling as she thought this would hinder pupils' ability to sight-read. Afterwards, Caroline reported that she had come to realize that expressiveness does not automatically occur when pupils play accurately and fluently, or when they hear expert modeling. Caroline was impressed with her pupils' progress during the project and she was convinced that this was the effect of modeling combined with questions about, and emphasis on, the musical character. Alicia mentioned at the end of the project that she had always assumed that modeling and talking about the interpretation of a work using metaphors would be sufficient for enhancing expressiveness. She thought that the dialogic teaching approach supported by modeling had been far more effective than modeling and describing the character for facilitating pupils' expressiveness. Tim had been taught via aural modeling prior to his study at HE and had found this helpful. Contrastingly, he initially felt frustrated with his tutor's verbal teaching at music college. These experiences might have influenced his ideas about the difficulty of using questions and dialogue for young students.
Secondly, pupils' technical problems, performance experience and level of confidence influenced the teaching-and-learning process. Tim's pupils had problems pitching their notes and feeling the pulse of the music. Therefore, Tim's use of modeling and playing along seemed especially helpful for their learning. Some tutors realized near the end of the project that their pupils' pieces had been too hard for their current level of playing, which FIGURE 5 | Potential effects of a dialogic teaching approach on various aspects of the teaching-and-learning process.
could easily have led to low confidence levels, thus affecting the teaching-and-learning process.
Thirdly, the musical style and challenges of instruments played (cf. McPherson et al., 2012) may have influenced the teaching-and-learning process. The violin teacher focused a lot on technique, even when working on expressiveness which was similar to the views of the violinist in a previous study (Meissner, 2017). Furthermore, the presentation of the musical score can influence the teaching-and-learning process. The recorder players studied several Baroque pieces from scores which provided ample opportunity to reflect on the musical character and phrasing as few expressive markings, if any, were included. Contrastingly, the other instrumentalists often studied pieces from Romantic or 20th century repertoire that included a great deal of expressive marking which often led to a focus on the score rather than interpretation.

DISCUSSION AND CONCLUSION
This ARP investigated how dialogic teaching of expressiveness can be used in weekly instrumental music tuition. It was our aim to explore whether tutors find a dialogic teaching approach useful for facilitating pupils' learning of expressiveness and to investigate what other instructional modes complementing this approach tutors would like to employ. Additionally, it was our aim to explore pupils' views on their learning of expressiveness and the instructional strategies used for developing expressiveness.
A salient finding from this study was tutors' observation that teaching and learning expressivity is a complex process wherein "everything is intertwined"; several methods can be used within a dialogic teaching approach for working on various teaching aims; improving expressiveness, accuracy, technique and practice. Tutors observed that increased accuracy and technical fluency as well as improved expressiveness contribute to the development of performance expression.
Overall, tutors and pupils thought that questions and discussion were useful for teaching and learning performance expression, especially when this was supported by aural modeling. Although it was sometimes hard to describe the musical character verbally, most pupils found tutors' questions helpful for reflecting on this. Participants' accounts suggest that open questions and dialogue can stimulate pupils' thinking, thus raising their awareness and understanding of the musical character and structure. Some tutors were impressed by pupils' perceptive ideas and the effect of reminding pupils of their chosen interpretation. Although this was not always evident in performance sessions, tutors thought that in lessons thinking of the musical character led to improved expressiveness. Pupils reported that thinking about the musical character had been a revelation, as they had never considered reflecting on the musical meaning before. This changed their approach from "just sort of playing it" or "just doing dynamics" to reflecting on and aiming to convey the musical character in performance (cf. . Some pupils had become aware of the fact that they could think about and convey their own interpretation, thus taking ownership of the music. This finding, that questions and dialogue are helpful for thinking and learning, is in line with Vygotsky (1978Vygotsky ( , 1986 who proposed that language is an important meaning-making tool. Questions and problem-solving had helped these pupils to develop their understanding of the music and their expressiveness in playing. Teachers in this project used open questions to facilitate thinking on various other aspects of playing and practice too; questions about pitch and rhythm, expressive marking, dynamics, and technical aspects of playing were employed to work on accuracy and technical fluency. Some tutors had asked pupils questions regarding their assessment of playing or planning of practice. It seems likely that inviting pupils to evaluate their own performance is more helpful than informing them of the teacher's evaluation. Questions regarding practice might assist pupils to start reflecting on this, which could be a useful step to developing self-regulated practice habits (cf. McPherson and Renwick, 2001;Hallam et al., 2012;Pike, 2017).
In line with research findings on dialogic teaching of academic subjects in classrooms (Alexander, 2008), some tutors observed that the dialogic teaching approach had made lessons more interactive; pupils were more engaged and felt more responsible for their learning when they were asked for their views. The pianist reported that this approach had made her listen more to her students and she felt she knew them better than before. Tutors thought that their pupils had grown in confidence and this is supported by reports from pupils. It seems therefore that the dialogic teaching approach had a positive effect on various aspects of the teaching-and-learning process ( Figure 5); questions and dialogue had stimulated pupils' thinking about the interpretation, thus contributing to their learning of expressiveness. Additionally, questions had been used to reflect on pitch, rhythm and technique, thus facilitating pupils' work on accuracy and technical fluency. Enquiry and exploration appear to facilitate process-oriented learning, which may enhance creative as well as accurate performance (e.g., Timmers et al., 2012). Moreover, asking questions about various aspects of the interpretation, playing and practice, and accepting pupils' ideas, allowing them space to explore interpretations and ways of practice, gave these pupils a sense of ownership, which can contribute to the development of self-efficacy and agency. When pupils have a "musical say" (Davis, 2011) in expressive decisions this can contribute to their growing sense of musical agency (Wiggins, 2016) and identity as musicians (Hallam, 2010). This process might move in both directions; dialogic teaching combined with modeling can affect agency and expressiveness and so on, and it seems likely that enhanced accuracy and expressiveness in turn would affect the dialogue and modeling.
Consistent with previous research (e.g., Sang, 1987;Woody, 2006) participants reported that aural modeling had been helpful. Modeling had been useful for building up an aural picture of the music, thus facilitating pupils' learning of expressiveness, accuracy and technique (cf. Dickey, 1992). Participating teachers tended to use modeling in combination with questions or short explanations about various aspects of playing, such as interpretation, articulation, dynamics, phrasing, pitch, rhythm or technique. Some scholars (Sloboda, 1996(Sloboda, , 2005Woody, 2000Woody, , 2003 suggested that there is so much detailed information contained in an expressive aural model that it might be too complex to be used for teaching children expressiveness. However, this study suggests that a dialogic teaching approach combined with modeling deals with this difficulty because teachers can help their students to reflect on what they hear in an aural model and how this relates to their interpretation. Although several studies have demonstrated that standalone modeling can be effective (Rosenthal, 1984;Sang, 1987;Woody, 2006), this is unlikely to stimulate students' thinking or to develop their understanding of the musical character and structure. It seems likely that modeling alone makes pupils dependent on the teacher (Broomhead, 2005), as repeated modeling might be required if students have not learned to reflect on the interpretation. The notion that standalone modeling or modeling combined with verbal instruction is insufficient for stimulating expressiveness is supported by accounts in the current study. The trumpetist had been taught mainly via aural modeling prior to his study at HE and had found it frustrating to be taught through verbal teaching afterwards. Additionally, the clarinetist observed that expressive modeling is not enough for improving students' performance expression, even if all expressive markings contained in a score are executed accurately. Furthermore, the pianist used to work on expressiveness via modeling and by labeling the musical character using metaphors. Although this had worked to some extent, she thought that asking questions combined with modeling was far more effective for facilitating pupils' understanding and expressiveness than describing the interpretation and modeling.
Participating teachers had also played along with students to support their playing, especially when pupils struggled with pitch, rhythm or technical fluency. Several teachers and pupils thought this had been helpful, although some thought it was less useful than asking questions, aural modeling or accompanying. It seems likely that playing along is helpful for pupils who are struggling or feeling insecure about their playing (Hallam, 1998). In such situations it can be helpful for children to get immediate feedback through their teacher's playing which contributes to their learning of pitch, rhythm and phrasing, thus increasing their confidence. Playing an accompanying part is likely to provide support while giving pupils some independence, as they can adjust their tempo and timing to the teacher's accompaniment. Additionally, accompanying might help pupils to experience tutors' phrasing and musical tension.
In this project tutors explored mainly dialogic teaching combined with modeling and playing along to facilitate pupils' learning of expressiveness. Although the other methods from the theoretic model (Figure 1) were not investigated systematically, several strategies were used by tutors, such as projected performance, listening to "own" recordings and gestures. Gestures were not mentioned by tutors, but the video recordings reveal that gestures were generally used to complement verbal explanations and modeling (cf. Simones et al., 2015).
Although the tutors thought that it is important to tailor methods to individual students (cf. Brenner and Strand, 2013) and situations, they acknowledged that they had often slipped into using one uniform approach for most pupils (cf. Carey et al., 2018). Therefore, it had been useful to watch examples of each other's teaching, to share ideas and discuss methods. For teachers, AR participation had been an enriching experience, in which they explored strategies for teaching expressiveness as well as other objectives. Sharing expertise and ideas were important contributors to this positive experience.
Our findings highlight the importance of regular performance experience for young musicians. Consistent with previous research (Boucher and Ryan, 2011), regular performances within a relatively short time frame had helped pupils to adjust to the feelings of nervousness and tutors reported that pupils felt more confident at the end of the project. It is difficult to determine whether this increased confidence was the result of the regular performing activities or the teaching or a combination of both. Nevertheless, it is important for pupils' development as musicians that they have regular opportunities for performing in front of a friendly audience from a young age, as performing needs to be practiced. Additionally, pupils are likely to benefit from preparing repertoire that is well within their technical ability, so that there is less reason for anxiety and a higher chance of success (Kenny and Ackermann, 2016;Papageorgi and Kopiez, 2018;Yandell, 2018).
There were some limitations to the study: There were more female than male participants; also, the project took place during only one semester, while it seems likely that children's learning of expressiveness is a long-term process (Kenny and Ackermann, 2016;Papageorgi and Kopiez, 2018). It would be worthwhile to continue exploring teaching and learning expressive performance in a long-term study. Additionally, it would be useful to conduct a detailed multimodal analysis of tutor-student interactions to explore how dialogue, speech, gestures and musical activities affect the teaching-and-learning process (cf. Rostvall and West, 2005;Kupers et al., 2018).
Although this study did not prove the effectiveness of methods, previous studies demonstrated that questions and dialogue can facilitate children's understanding of how to construct expressive performance . Subsequently, young musicians can expand their learning, understanding and expressiveness in various contexts; in group lessons and ensemble work pupils can ask each other for their views and model for one another, thus developing their ideas together. In our research, we have explored one aspect of interactive learning, that is promoting autonomy and reflective practice through dialogue integrated in a traditional teacher-student learning situation of Western classical music. This relates to initiatives advocating for informal, exploratory, collaborative and embodied learning (e.g., Schiavio et al., 2019). It remains to be examined how interactive learning can be further enhanced in such contexts, integrating various activities such as improvisation, listening, performance by ear (e.g., Green, 2017) and the adoption of a self-organizational perspective oriented toward rich musical experiences (Schiavio and van der Schyff, 2018;Schiavio, 2019).
Overall, participating tutors and pupils thought that a dialogic teaching approach supported by modeling is important for meaningful instrumental music education, as this can stimulate pupils' thinking, thus facilitating their learning and enhancing their expressiveness.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics reviewers at The University of Sheffield. Application reference number: 008981. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
HM designed method, discussed method and content of teacher meetings with RT, led the participatory action research project as teacher-researcher, collected and analyzed data, and wrote the first draft of the paper. RT co-designed method, watched video excerpts of lessons, gave comments on data analysis, and several drafts of the paper.

FUNDING
This research was supported by funding from the Society for Education, Music and Psychology Research (SEMPRE) and by a scholarship for HM's doctoral studies from the Arts and Humanities Research Council, Research Councils UK (grant number AH/K503289/1) with an additional stipend from The University of Sheffield. The University of Sheffield also funded the OA publication fee.