The mentor–tutor partnership in Turkish special education initial teacher training: an exploration of collaboration and agency

ABSTRACT This paper explores the perspectives of the tutors and mentors involved in Turkish special education departments` initial teacher training, and how their understanding of and degree of agency shapes the nature of the partnership in it. The data collected from four partnership settings identified three sub-themes: expertise; experience; and expectations. The collective understanding within the teacher training environments shaped the behaviour of individuals. However, rather than working in collaboration with schools, the dominant role in the partnership is either taken by the tutors, and the mentor is largely ignored, or tutors leave the dominant role to the mentors and participate in the system superficially. Exploring the collaboration in Turkish teacher training programmes seems to be an important area to improve the overall quality of teacher training. Looking at ways in which mentors and tutors exert agency offers an opportunity to strengthen levels of collaboration.


Introduction
It is widely recognised that teachers make a significant difference to the educational outcomes for young people (Barber & Mourshed, 2007;Mourshed, Chijioke, & Barber, 2010;Muijs, 2006). Unsurprisingly, initial teacher training (ITT) has therefore attracted attention from governments as a means of securing improvements in the quality of teaching and learning in schools. This raises important questions about how best to train and educate teachers, so that they are able to impact positively on the lives of the young people they teach. Learning from expert supervision in a real environment is regarded as crucial in the process of becoming a teacher (Brouwer & Korthagen, 2005;Hagger & McIntyre, 2006;Jackson & Burch, 2019;Yuan, 2017).
The focus of this paper is to explore the perspectives of the university-based tutors and school-based mentors involved in the training of special educational needs (SEN) trainee teachers, and how their understanding and degree of agency shaped the nature of the partnership in the SEN departments` ITT in Turkey. In Turkey, special educational needs teachers receive specialised teacher training, and the SEN training partnerships are worthy of study for two reasons. Firstly, students with SEN often require particular teaching techniques, materials that will appeal to different sensory modalities, and both one-to-one and small group sessions. Hence the training of SEN teachers may show differences from the training of teachers for typically developing children. Secondly, the area of special education, especially in Turkey, is less well researched compared to other subjects. Research into SEN teacher training is therefore an area that will have both practical and theoretical implications.

Context
In Turkey, teacher training programmes typically last four years, and each year is divided into two semesters, from September to January and from February to June. In the last year of the programme, according to government regulations (the Ministry of National Education (MoNE), 2018), trainee teachers have a school-based experience, which they are expected to attend for six hours a week (either one full day or two half days per week). In the training school, the trainees have to prepare teaching plans, lesson materials, individual education plan/programmes for students and perform teachings in different subjects at least four times each semester. Furthermore, the MoNE (2018) has assigned specific roles to those involved in ITT, outlining the expectations for the schoolbased mentor, university-based tutor and trainee (see Figure 1). As part of the training, tutors, mentors and trainees are supposed to plan activities in the classroom together, and whilst the trainees are practising in the classroom, the tutor and mentor are supposed to observe and evaluate the student in collaboration. Trainees who successfully complete these duties are deemed to have passed the ITT programme.
Due to the centralised structure of the Turkish education system, the teaching practice programme has the same standards for all teacher training departments. Therefore, special education departments must also follow these rules. This study reveals the current situation of the special education departments, which has to comply with the general rules.

Literature review
It is important to investigate this area because the Turkish literature (Alptekin & Vural, 2014;Bural & Avsaroglu, 2012;Özen, Ergenekon, & Batu, 2009) highlights problems that arise when the relationship between school and university-based colleagues in special education is poorly defined or poorly understood, and other international studies show that there are issues caused by fuzzy and contested understandings of mentoring (Kemmis, Heikkinen, Fransson, Aspfors, & Edwards-Groves, 2014). In this literature review, studies that highlight the importance of the relationship within an ITT partnership will be discussed, as well as models of partnership and the role of agency in examining how individuals within a partnership work together.
University-based tutors and school-based mentors have a significant role to play in trainees' perceived efficacy because not only are they supervisors for trainees, but they are also often seen as role models (Hagger, Burn, Mutton, & Brindley, 2008), shaping the trainees` perceptions of what a professional teacher should be; from this, their teacher identity can start to develop in their mind, and this can increase their self-efficacy (Jones, Kelsey, & Brown, 2014). Further, tutors and mentors are supposed to be in regular communication to ensure effective coordination between the practices in the school (Grau et al., 2017;Jackson & Burch, 2019;Oates & Bignell, 2019), and to inform each other about the ongoing learning processes at both university and school (Brouwer & Korthagen, 2005;Grau et al., 2017;Jackson & Burch, 2019;Paulson, 2014). They are also encouraged to share their expectations, ideas, beliefs, future plans, classroom priorities and pupils' performances with each other (Jones et al., 2014;Walkington, 2007) for creating a successful partnership in ITT.
Internationally, there have been studies exploring effective features of ITT. In most cases, the level of collaboration was seen to be weak. For example, in the Republic of Ireland, Meegan, Dunning, Belton, and Woods (2013) examined tutors' experiences and mentors' perceptions about physical education practicums. They found that tutors and mentors did have a surface level of collaboration, but some mentors were unwilling to participate in the programme because they were not paid for participating. Ibrahim (2013) studied trainees` supervision in the United Arab Emirates (UAE) and found that tutors and mentors did not have a proper collaborative approach and although tutors visited the school two or three times during the ITT programme, there was little attempt to communicate with mentors. These studies reveal what Kemmis et al. (2014) refer to as highly situated 'practice architectures' and highlight there can be a serious disconnection between mentors and tutors, making it hard for both sides to understand the role and expectations of each other. This matters, because various studies, such as Bardak (2015) in Northern Cyprus (where students have a system equivalent to Turkey's Council of Higher Education [CoHE]), Lillejord & Børte (2016) in Norway, Mokoena (2017) in South Africa and Grau et al. (2017) in Chile, show that weak collaborative partnerships create tensions between tutors and mentors. This in turn impacts on the professional learning experience of the trainee teachers (Lillejord & Børte, 2016).
In the Turkish context, there is no study which directly investigates the tutor-mentor collaboration in the ITT system for training special educational needs teachers. Although Alptekin and Vural (2014), Bural and Avsaroglu (2012), Ergül, Baydik, and Demir (2013) and Özen et al. (2009) addressed mentor-tutor collaboration and indicated that tutors and mentors lack a clear sense of effective collaborative partnership during the ITT programme, their studies do not directly focus on mentor-tutor collaboration and their data were only collected from trainees.
A common feature of the studies mentioned is the weak levels of collaboration relating to the practical aspects of teaching training. The studies, based in different educational contexts and with different systems of ITT, all highlight problems that occur where issues arise in the collaboration between tutors and mentors (solid arrow in Figure 1). A key issue in any form of collaboration is the nature of the agency exerted by both sides; human behaviours and activity are affected by people's sense of agency, which in turn affect the way collaborative partnerships work.

Models of collaboration
Partnership between universities and schools in ITT is now commonplace, but as noted earlier, this does pose several challenges in ensuring this partnership is effective. One of the issues highlighted by Furlong, Barton, Miles, Whitty, and Whiting (2000) is the model of partnership that is envisaged and enacted. They suggest that there are two main partnership models: complementary and collaborative partnerships (see Table 1) in teacher training programmes. In the complementary model, the school and the university are separate and have complementary duties, but they tend to lack a strong collaborative ethos and dialogue between the two sides. In this model, the school-based mentor has the primary responsibility for supporting the trainee teacher and oversees the school-based training and assessment. Basically, the mentor is the person tasked with conducting the system and the tutor generally relinquishes much of the training to the mentor in the school. Essentially there is little dialogue expected between the school and university provision unless trainees are in trouble (see Table 1). The other model is the 'collaborative model', which is based on shared responsibilities between the tutors and mentors (see Table 1). Trainees implement what they have learned from the university element of the programme and are expected to critique the learnt skills and knowledge within the school, which occurs in the light of collaboration between tutors and mentors. In Turkey, the Ministry of National Education [MoNE] (2018) clearly states that trainees` activity planning, regular observation and assessment in schools are carried out collaboratively between mentors and tutors, and trainees collaborate with them to achieve ITT aims. The expectations from the authorities are that the schools and universities should adopt a collaborative rather than a complementary partnership.

Agency theory
Despite clarity about the need for partnership and different models of partnership, the quality of partnership work is shaped by agency. Those involved in the partnership have to implement it, so, to a greater or lesser degree, exert some form of control over how the partnership operates. Understanding how people exert agency within this collaborative partnership is extremely important because people act intentionally in order to make a change in their environment (Philpott & Oates, 2017;Priestley, 2015). Tutors and mentors take responsibilities in ITT, and they have their own free thoughts about making things happen, but these thoughts are the result of an ongoing process of their agency; it is a response to past experiences and perceptions of current constraints/opportunities and future intended actions ( Figure 2) (Biesta, Priestley, & Robinson, 2015).
However, in this study, the behaviours that happen need to be explained collectively by considering both environments in the partnership. Because the ecological model does not capture the intersection between two environments -university and school -we propose an adaptation ( Figure 3) derived from Biesta et al. (2015).
Models of partnership involve differing degrees of agency from those working within these systems. In fact, these partnership models are also shaped according to the agencies of the ITT actors. In other words, they affect each other, so they are also the results of agentic behaviours. A collaborative model would imply a more evenly distributed model of agency and collective agency, whereas a complementary model would require different levels of agentic commitment from the actors in the partnership. This means it is more likely to experience complementary partnerships in environments with actors with different agentic levels, while individuals with similar levels of agencies are more likely to have collaborative partnerships in ITT environments.
The role of collective agency is an important issue that needs to be considered. In ITT, the various actors are expected to communicate and collaborate with each other and share ideas and expectations about supporting trainee teachers. However, tutors and  mentors each inhabit their own unique 'mini-worlds'; these work environments are dynamic, shaping and being shaped by those who are within that sphere. And then there is the dynamic interaction between these separate spheres of work, i.e. the work worlds of the university tutors and school-based colleagues (Bandura, 2000(Bandura, , 2006Philpott & Oates, 2017). This means that partnerships between two different work environments are complex, especially where the two environments have, for example, differing priorities. Within each environment there are likely to be established ways of working, mutual forms of communication, a sense of joint action, interpersonal harmony and coordination, generating a sense of collective agency (Bandura, 2000(Bandura, , 2006. These collective understandings direct the behaviour of the tutors/mentors within their working environment. Although the actors in each sphere may have a degree of personal agency, they are also likely to take their department/school`s norms into consideration as well, either consciously or unconsciously. Adopting behaviours contrary to the written or unwritten rules of a group could potentially leave tutors and mentors feeling isolated within their own worlds of work. Consequently, examining the role of collective agency will help make sense of tutors` and mentors` behaviours within the partnerships.
To sum up, this paper explores two issues: firstly, the nature of the partnerships in different Turkish teacher training programmes for those training to be teachers of children with special educational needs; and, secondly, this paper draws on agency theory to make sense of how these partnerships work in practice.

Methodology
This study adopted a qualitative, interpretivist approach. This was because the focus was on the experiences and perceptions of those involved in the teacher training programmes; therefore, any data collected had to allow participants to share their lived experiences of taking part in an ITT partnership.

Sample
In Turkey there are more than 20 special education teacher training departments in universities; four of them were selected purposively on the basis of the institutions` different opening dates, the numbers of staff involved in each programme (see Table 2 some details lack the precision to help preserve the anonymity of the institutions) and the geographical spread of the institutions. In total, the data were collected from 26 participants: 13 tutors and 13 mentors from these four partnership settings. The number of staff was limited, so gender or experience criteria were not considered for the sampling frame in this study.

Data collection
Two forms of data were collected. One was a simple form to collect background information outlining participants' experience and the other was a semi-structured interview. In the background form, the questions were designed to gather information about the main actors' roles in ITT. Semi-structured interviews explored the tutors` and mentors` experiences, personal histories and perspectives about the system, and examined their beliefs, attitudes and perceptions of their role within it. Semi-structured interviews are more flexible, and in this study helped to generate rich information and to understand internal and external reasons for particular forms of behaviour (Kvale & Brinkmann, 2009).

Ethics
The study gained ethical approval from the University of Reading ethics committee, and procedures were followed that adhered to the BERA (2018) guidelines. All participants gave consent for their involvement and were provided with pseudonyms. All institutions have been anonymised and are referred to as Cherry, Elm, Maple and Pine Universities. Also, no individual schools have been named.

Data analysis and authenticity and credibility
A number of themes -professional, structural, material and partnership -were identified in the data, but this paper reports on one: partnership. The theme of partnership was identified in the work of Ball, Maguire, and Braun (2012) and Biesta et al. (2015), but in both cases the studies focused on the perspective of the school environment. In this study the focus is on the two environments, both attempting to work in partnership.
Within qualitative research, validity is commonly associated with notions of authenticity, credibility, plausibility and rigour. For example, authenticity is the ability of the study to express the data cogency, credibility and plausibility, from the participant's point of view and eyes (Cohen, Manion, & Morrison, 2011). As regards credibility, the collected data had to accurately reflect the views of tutors and mentors interviewed (Thomson, 2011).
The coding process was inductive, and initially adopted an in vivo approach, using the words of the participants that were relevant to issues relating to the theme of partnership (Saldana, 2016). These were then clustered together to identify initial categories within the wider theme of partnership. This process was conducted by two researchers independently and resulted in similar ideas being observed; any differences were discussed and resolved. Finally, a second cycle of coding was conducted to seek a conceptual link between the working categories. From this process emerged the categories of expertise, experience and expectations (see Appendix 1). These themes are interlinked and at times do overlap. These categories help to show a cycle of actions from the past to the future within the context of how a partnership can work. Dividing the ongoing process of actions into categories also gave an explicit sense for the understanding of mentor/tutor agency and the issues which affect their agency.

Findings
The findings will be presented in two main sections, from the perspectives of the mentors and tutors. Each section will set out the findings under the key themes of expertise, experience and expectations. These emerged inductively through the process of data analysis and seemed to capture key issues that explained the challenges noted in the interviews. These three themes are defined later.
The competence that individuals have achieved as a result of the knowledge, skills and experience they have acquired individually and/or collectively in their ecological environment in the period from past to present can be defined as expertise.
Experience is a phenomenon that knowledge and skill acquired in the past and present of individuals by communicating and interacting with each other within their physical environment containing values and rules.
Expectation is shaped by the prior experiences that give shape to the practices that are taking place in the present in an ecological environment. This process moulds individuals` feelings and perceptions, along with their beliefs concerning practices, about what they expect to experience in the social environment in the future.
This will allow for clear issues within each of the four partnership models to be identified. Individuals' behaviour towards a goal is shaped by their past experience, present constraints and future aspirations and expectations (Biesta et al., 2015). Therefore, discussing the collected data in terms of 'expertise', 'experience' and 'expectation' within this past-present-future model of agency will provide a more meaningful way to interpret the issues in this study.

Expertise
In each of these four partnerships, mentors came from different subject backgrounds and had varying degrees of teaching experience and/or working with trainee teachers (see Table 3). One issue seemed to be that trainees valued the expertise of their tutors more than their mentors. Mentors also respected the tutors` knowledge. 'They (tutors) know better than us' (Busra, Pine). Furthermore, in the Elm partnership, Hasan said that there was a conflict between their MoNE`s SEN pupil evaluation forms and what tutors gave to trainees. It seemed that trainees preferred to listen their tutors rather than their mentors. The same thing happened in the Cherry partnership as well. Ayse: I gave them some directions and forms thinking about my own bachelor's degree ITT but the trainees told me that I am wrong, that their tutor said that they must practise in a different way. Because of this inconsistency, they did not listen to me.
Although a dispute over which forms to use may seem trivial to some extent, it highlights a tension between what mentors feel is appropriate to use and what tutors think. At the micro-level in the partnership the expertise of the mentors is questioned.
Another issue seems to arise from mentors' subject backgrounds. In Turkey, teachers of students with SEN are supposed to have received specialist training. However, in many partnerships mentors without specialist SEN training were employed. For example, in the Cherry partnership, although both mentors involved in the study were SEN specialists, the other eight possible mentors were not trained as SEN specialists (and were not selected by the headteachers to be involved in this study). The majority of the mentors interviewed in the Elm and Pine partnerships were also not SEN trained. For this reason, some tutors were sceptical about their competence (Elm-Maple and Pine). So, this seems to be a point of tension between mentors and tutors and seems to inhibit collaboration. However, Table 3 shows that many of these mentors had considerable experience in SEN teaching, but due to their subject background their expertise was not valued by the tutors.

Experience
In general, mentors at Maple and Pine were confident in the tutors` practices while mentors at Cherry and Elm were more critical of tutors` practices in ITT. Fatma at Elm said: 'I do not know what tutors do . . . because they do not come to the school, and we do not communicate.' In these settings, Cherry and Elm, tutors did not go to school to collaborate with mentors and managed the trainees from a distance. Ayse (Cherry) was opened to sharing ideas, information, feedback and experiences, and wanted to work collaboratively, but this was handicapped by the tutors' unwillingness to visit the school regularly and had their own way of practice, which is mostly based on tutors` directions.
In contrast, Mustafa (Elm) was happy with the current system. He did say that tutors did not visit the schools, and his answers also showed that he was unaware of what collaboration is required under the ITT regulations. It seemed that the tutors at Cherry and Elm were unwilling to work collaboratively, even though Ministry of National Education [MoNE] (2018) practice regulations (article 6-5-c) states that tutors must observe and inspect trainees' studies regularly in conjunction with the mentor.
In contrast to Cherry and Elm partnerships, the mentors at Maple and Pine stated that tutors regularly observed trainees in schools. While mentors at Maple stated that there was a collaborative relationship with tutors, mentors at Pine were less directly involved in the ITT, and tutors took all the control at school.
In the Maple partnership, the tutors had given advice and support on how to provide feedback to trainees. Hatice had taken this advice and felt she provided better feedback to the trainees. Muhammet and Seyma at Maple had a similar perspective and claimed that they had a good relationship with the tutor. Huseyin said that 'tutors generally come to observe trainees` sessions. When they come to our class, we share our experience and knowledge . . . and [they] inform us about new materials and techniques.' The comments made by the mentors show that they were happy with the current programme and claimed that they conducted the programme in collaboration with the tutors. However, neither mentors at Maple nor other mentors in the other partnerships were fully aware of their responsibilities under the government's ITT regulations. They drew mainly on their own ITT, professional experience and collective behaviours at schools.

Expectations
Mentors were generally keen to collaborate, but each partnership has its own practices. For example, Zehra and Esma at Pine stated that they talked with a tutor once or twice per year, but did not want any more significant communication because it entailed additional work for them, and they felt their main responsibilities were towards their pupils rather than trainee teachers. So, their expectation was not of involvement in the system; instead, they felt that tutors were best placed to support the trainees.
Most mentors generally saw that the tutors were the most significant element in developing the trainees' practice, and the role of the mentor was secondary. However, mentors at Cherry and Elm were critical of the tutors for not going into schools during the placements and felt that 'tutors do not appreciate what we (mentors) do in a training class' (Ali-Cherry). Ayse (Cherry) said: 'I might not know what to do clearly, but the tutors did not attempt to talk to us about these issues.' Ali (Cherry) continued, 'Tutors might be more active; they only come at the beginning and end of the year to give and collect training folders.' When tutors did visit the school they may only talk to the head teacher, as happened in the Elm partnership. Consequently, many mentors, such as those in the Cherry and Elm partnerships, complained about the tutors` lack of collaborative evaluation, observation and other practices, which is contrary to the MoNE`s main principals: cooperative evaluation; content; diversity. There is a clear expectation from MoNE that activities will be planned, run and evaluated together by tutors and mentors. The failure to achieve cooperative working can potentially hinder the development of trainees' practice, and fails to comply with the government's regulations.

Expertise
In all settings, tutors generally thought that tutors and mentors were important actors in ITT, but that tutors should have the final say on whether trainees are capable to become teachers, as the tutor's role was more important than that of the mentors.
Although tutors had different background subjects and years of experience too (see Table 4), they tended to talk negatively about the mentor`s role, background and practice. Most of the tutors at Elm, Maple and Pine felt mentors struggled to perform their role. For example, Turgut at Elm said: 'There is a conflict between what we say to our trainees and what the mentors tell trainees to implement . . . mentors (certified teachers) are incapable to provide guidance due to not having SEN undergraduate background' (see Table 3). This was a common view among tutors.
Most tutors at Maple stated that mentors did not know the regulations, and did not attempt to learn their roles, because the special education department at the universities was leading the implementation of the ITT, and mentors comply with these departments` practices through trainees. Hence, although mentors at Maple thought that they had good collaboration in the programme, it was clear that the tutors were keen not to involve them in the practice programme in any great depth, because 'these schools are different from our organization, and mentors are not open to criticism and do not develop themselves (Ozay)'. This was a commonly held view in Maple and can be explained by the process used in the university to induct and train new tutors; although this training was thorough, the emphasis was on a master-apprentice model, which seems to have created a common mindset amongst the university staff towards mentors.

Experience
Due to the shortage of staff, tutors at both Cherry and Elm struggled to develop a strong partnership model. Elm also had some other issues; lack of mentors` expertise and tutors not visiting the training schools for observation as a departmental routine. This meant tutors tended to be based in their university departments, and unlike tutors at Maple and Pine, did not regularly observe in schools during ITT. However, tutors at Pine were not keen on involving mentors fully in the ITT programme, and similarly some of Maple`s tutors tended to adopt a 'light-touch' towards mentors: We generally share some knowledge with mentors, but we do not get involved in their business . . . . We do not talk with mentors too much, because if we discuss things in detail, various problems might arise between us and mentors. (Narin-Maple) On the other hand, tutors at Maple stated that they considered mentors` role to be important because '[w]hen we leave the school, the trainees stay with the mentors in the class (Tayfur)'. Overall, the tutors generally believed that even if the mentors' role is important, they were unwilling to share some information, and generally did not fully respect the role they played in a trainee's development.

Expectations
The tutors in all partnerships had a similar understanding about their high importance in ITT. They seemed unaware of, or preferred to ignore, any serious criticisms from mentors about tutors` lack of collaboration. Turgut (at Elm) further added: 'We do not have an observation culture in our department, so we do not go to the training school to observe trainees.' Aybike (Elm) also supported these claims: 'We have a newly employed tutor, and she wants to change all ITT system, but in this department, no one can change anything . . . we do not go there to check our trainees.' Article 6-5-c of the government regulations, however, states that trainees' practicum must be observed and inspected by tutors and mentors regularly (Ministry of National Education [MoNE], 2018). Additionally, Anil at Elm claimed that he did not have much time to observe trainees in the training school, but added that 'trainees should take the teaching role independently and should integrate with the class. They can learn teaching better when they are left alone.' This reflects a 'sink or swim' approach towards teacher training, in which the tutor has little input. Aybike at Elm also commented on her limited time, which meant that she could not visit the training school regularly. It seems these tutors had some external constraints on their ability to engage with trainees in schools, which also manifested itself in a lack of willingness to work on building strong relations with mentors.
There were similar challenges in the Cherry partnership. Selcen said that there was an insufficient number of tutors in the department, 'so the visits (to schools) might not happen regularly'. Azra at Cherry also said there was no expectation to develop collaboration with mentors.
Lastly, tutors at Pine indicated that they preferred not to collaborate with mentors, as well as complaining about a lack of time to carry out their role. They also actively encouraged trainees to ignore advice from their mentors. Aydan said: 'We say to our trainees, "Do what your teacher says in the class, but ignore their advice. Just carry out what we planned."' Ata was a practice coordinator who had the authority to design the ITT and said that 'the trainees should be trained by someone [tutor] who can look at the ITT from a broader perspective with a theoretical infrastructure, not by someone [mentor] who are from training schools'. This understanding shapes the department`s practicum design. Moreover, tutors stated that in Pine department, the tutors are the only participants who deliver the system and the mentors should just accept trainees into their classrooms. For these reasons, the collaboration between the participants could be perceived not to be working properly: the dominant role in school is taken by the tutors, even if they are not there to observe and provide feedback.

Overall
Each actor, tutor or mentor has their own point of view, but their practices and ideas tend to align with colleagues in their own working environments collectively. Although the government regulations require tutors and mentors to have a collaborative partnership in ITT, the collective understanding within the special education department teacher training environments shapes the behaviour of individuals. However, rather than working in collaboration with schools, the dominant role in the partnership is either taken by the tutors, and the mentor is largely ignored, or tutors leave the dominant role to the teacher and participate in the system superficially due to the reasons given earlier. It can be said, as will be discussed in more depth later, that none of the partnerships are 'collaborative', and they can therefore be categorised under 'complementary partnership'. Within this model, the Maple partnership tended to be a stronger model than the other three partnerships. There was a clear sense of collective agency within the group of tutors, and although they were still wary of too much interaction with mentors, there was at least an acknowledgement of the important role mentors play in supporting trainee teachers. In contrast there was little sense of any collective agency on behalf of the mentors. It seemed that they didn`t have much choice in the system. Due to this, they mostly adopted a passive role in the partnership.

Discussion and conclusion
Previous research into ITT conducted in Turkish special education departments considered the perspectives of the trainees. In contrast, this study approached the teaching practice exploring how the notions of `expertise`, `experience` and `expectations` shape mentors` and tutors` level of agency. Many questions about mentor-tutor collaboration are raised and offer scope for further studies on a range of different issues, rather than focusing mostly on trainee teachers` opinions in ITT.
Adopting a model of ITT based upon a school-university partnership is widely seen as the ideal formula for training teachers effectively. The principal actors would be the tutors and mentors, working together to achieve common goals (CoHE [Council of Higher Education], 1998; Ministry of National Education [MoNE], 1998[MoNE], , 2018, thus improving trainees' pedagogical practices (Bulunuz & Bulunuz, 2016;Burn, 2007;Oates & Bignell, 2019). However, the findings of this study suggest there are problems within the institutions involved in this study, and that the current partnership models could be improved. A significant issue appears to be that power in the system is predominantly centred in one environment (see Table 5). This seems to create a number of obstacles to collaboration and the quality of training.
These findings suggest a possible need to reconsider how we might define partnership in ITT. Furlong et al. (2000) focused on only the school-dominant roles under the complementary partnership, but the dominant responsibilities in the Maple and Pine partnerships are undertaken by tutors, i.e. the universities; they tend to exclude mentors. For example, when tutors visit the school and observe, assess and give feedback to the trainees on their practices (see Table 6), mentors do not participate. In this model, the tutors undertake the main role of supervising trainees in school and university, and there is little dialogue with the mentors. Interestingly, tutors seem to be using mentors` classes as a laboratory and their pupils as subjects and appear to exclude mentors in this programme due to the tutors` perceptions about mentors` lack of expertise.
This model can still be explained as a complementary partnership because the responsibility is still taken predominantly by one environment in the system. Thus, in a complementary model, involving two different professional contexts, it would appear that the gulf in expertise, experience and expectations mean that agency is exerted more strongly by one context than the other -so the agency is not shared or promoted. In particular, it seems perceptions of expertise have a significant effect on how partnership is enacted by participants. Where mentors believed that tutors understand the ITT requirements better and are more knowledgeable, partly due to the high regard in which academics in Turkish culture are held (Sunar et al., 2015), the mentors seem to let tutors dictate the programme.
Practices in three partnerships studied here are partly a result of collective agency. Collective agency itself results from the exercise of personal agency; individuals are able to exert some degree of personal agency and the level of each participant's personal agency can affect the measurement of the collective agency (Bandura, 2000). However, in these settings, mentors seem to relinquish any degree of personal agency, preferring to entrust themselves to an agent (tutors) rather than making decisions in an area where they feel they lack expertise in comparison to the academics. This decision to trust in the agency of others may be related to issues of status, but it can also appear as if mentors abdicate 'responsibility' for any outcomes, whether those are positive or negative. As long as mentors have complied with externally imposed expectations there is a sense that they have fulfilled their role. This approach seems to occur where there is a lack of mentoring knowledge, which directly influences mentors' perceptions of their agency. If mentors were more confident in their knowledge about their role and responsibilities, their level of professional agency may increase and there would be less reason for mentors to relinquish so much power to the tutors. This might then allow mentors to work towards stronger mentor-tutor collaboration, as the best way to develop ITT. This would require both individuals` [mentors/tutors] personal capability, expertise and experiences within their environment to be understood clearly for improving the capabilities of the group (Bandura, 2000). This is because the capability of a whole group, how the participants interact and coordinate with one another would affect the group's collective actions and therefore their collective agency.
In the universities in this study, tutors appeared to take control of the system. This seems to be because mentors were seen to lack expertise, and to be insufficiently trained because some had transferred into SEN teaching from other subject areas (see also Alptekin & Vural, 2014;Ergül et al., 2013). As a result, tutors did not appear to value the experience of mentors and actively undermined them in places. Furthermore, the tutors in the Elm, Maple and Pine partnerships lacked trust in the mentors' ability to offer trainees appropriate guidance. Tutors in the Elm partnership actively instructed their trainees to ignore the mentors' instructions and documents, despite the tutors visiting trainees in schools infrequently. This situation in Elm appears to reflect an embedded culture, which was unquestioned by the tutors. In such a situation it is hard for someone to feel they have the personal agency to change a collective culture, especially where that culture seems to make it easier for the tutors to fulfil their role: it brings consensus to the department, and saves time for busy tutors, who have limited resources. This resource problem also seems to limit tutors' sense of agency, forcing them to organise the programme around environmental factors. Although environmental factors appear to reinforce the situation, the perceived weaknesses in the expertise of the mentors and the collective practices of the tutors have generally lowered expectations within the partnerships in these institutions. In order to improve teachers` mentoring quality, since 2018, the authorities have made mentor training compulsory for teachers to work with trainee teachers. This may help to improve mentors' expertise and to get more experience. Nevertheless, the low expectations shown in this study may make it hard for mentors to exert much agency and improve collaboration. Greater success would be achieved if tutors engaged with mentors more actively in the ITT programme and had greater trust in their capabilities.
Another obstacle to developing a positive partnership is how tutors and mentors understand their roles. This is not a unique issue. As Kemmis et al. (2014) point out, there are different conceptions of mentoring between parties involved in teaching and teacher training globally. However, in this study, the data show tutors within university departments generally had a shared understanding of their role and how they intended to work with mentors (although this varied by university department). This shared understanding appears to affect those who are new to the environment (see Table 4), as new tutors seem to adopt practices that align with existing approaches. Similarly, mentors within schools appear to have a shared understanding of their role. Yet the findings suggest there is little dialogue across these settings to examine what each side could contribute to the training process. A combination of the collective understandings and the common societal beliefs seem to shape the behaviour of the individuals and limits people's sense of agency. Promoting more free-ranging thinking, where individuals began to develop their own sense of self, and feel able to challenge the departments` norms, appears necessary to change the existing cultures in these partnerships.
This research would suggest particular changes are necessary within the Turkish ITT system, but the findings may influence practices in other teacher training systems. This study suggests there is a need for a greater sense of shared agency, so that mentors and tutors work more as a team. The voices of both stakeholders should be heard, to develop and enhance mutual trust and communication between participants (Greany & Brown, 2015). Developing a more effective collaborative model would seem to be the first step in creating a partnership, which can then work together to address the challenges involved in running a teacher training programme. Essentially this would require a change in culture, but would need to start with greater clarity amongst the various stakeholderstutors, mentors, school headteacher, head of SEN department, school and department practice coordinators (see Figure 1) -about the expectations held by different partners and their role in the ITT programme, as well as discussing and acknowledging the experience and expertise of all these partners. By clarifying roles and what each partner can contribute to the training programme, there should be a greater sense of commitment to supporting trainees' development. This would then need to be augmented by clearer and more consistent lines of communication between schools and university departments, in order to have a more coordinated programme for the trainee teachers, which would hopefully be of benefit to these trainees and to the future students they will teach.
Based on the findings of this study, exploring the mentor-tutor collaboration and the levels of agency within this relationship seems to be an important area to develop to improve the overall quality of teacher training and offers an opportunity to strengthen levels of collaboration. It may be that other teacher training systems experience similar challenges and this study could provide a means for understanding how the lack of agency within a system may affect the level of collaboration and effective partnership. This study may also encourage future researchers to explore this area further, by focusing on the agentic relationships within teacher training programmes, in order to find effective ways of bringing about stronger collaborative partnerships.

Disclosure statement
No potential conflict of interest was reported by the author(s).