LESSON OBSERVATION: THE CHALLENGE OF SEEING BEYOND OBSERVABLE BEHAVIOURS

: The article focuses on the problems which trainee teachers face in reflecting on observed lessons taught by experienced teachers. Observing and discussing live and recorded lessons is part of a practical module in a Bulgarian state university run together with a course in ELT methodology, with the aim of helping trainee-teachers link theory to practice in preparation for their school-based internship. The discussion is based on the results of 74 BA and MA trainee-teachers’ written assignments of a whole video lesson observation (with the task of identifying its stages, procedures, and the rationale behind them), and a follow-up collective feedback aimed at revealing the sources of identified problems. A checklist including targeted observation foci was used for collecting quantitative data related to the elements of the lesson which were identified and correctly interpreted by the participants. Data analysis revealed no significant differences between the results of BA and MA trainees, who experienced similar difficulties in recognizing teaching techniques and interpreting teacher’s actions in terms of ELT theory, perceiving the lesson as a coherent entity, and understanding the logic of its organisation. The post observation feedback provided some insights into the problems, suggesting that they might stem from insufficient theoretical preparation and the lasting impact of trainees’ native educational culture with its traditional models of teaching; factors which might lead to conscious rejection of practices perceived as strange or inappropriate for the local teaching context. The implications of the study are linked to possible measures for overcoming the challenges and enhancing trainees’ development of professional expertise.


Introduction
Professional expertise in the field of foreign language teaching can be acquired in different ways depending on a number of factors, such as the options for teacher training available to candidates, educational traditions in a country, culture-specific preferences, locally established or adopted theories and ideas about teaching and learning, identified learner needs, and personal preferences.Teacher learning, which is a prerequisite for developing expertise, is nowadays seen as a socialization into the professional thinking and practices of a community of practice (Tsui, 2009), people who share common concerns, problems, interests and goals.Unlike traditional models for acquiring professional expertise, the craft, applied science, and reflective models (Wallace, 1991), modern conceptualizations of professionalism are related to constructing new knowledge and theory through participating in specific social contexts and engaging in particular types of activities and processes (Burns & Richards, 2009).Becoming part of such contexts, integrating within them and understanding the principles they follow and the values they subscribe by, is of primary importance for novice teachers.
Traditionally, one of the activities which precede novice teachers' immersion into the context of their future work is lesson observation.Observing a dynamic phenomenon, such as a classroom at work led by an expert professional, and trying to decipher his or her actions, recognise ideas, approaches and techniques which are studied in the methodology course, is the first step to socialization into the profession.The need to observe teachers in action before attempting to teach arises from the nature of the teaching profession and the high risks and costs involved in it.It also stems from the need to bridge the gap between theory and practice in pre-service teacher education.At the beginning of the training process, observation of experienced teachers both on video recordings in real classrooms serves mainly as a source of discussion and visual support for the theoretical input.However, this presumably 'safe' stage of teacher training poses unexpected and often under-researched challenges, which if neglected can lead to serious problems and even failures in trainees' professional realization and further development.The present article seeks to contribute to the research on the role of lesson observation by focusing on trainees' ability to make sense of observed behaviours in best practices for teaching ELT captured in a video lesson.The study explores different challenges in observation related to identifying and interpreting lesson elements and aspects of teaching, and suggests possible reasons for their occurrence.

Literature review
There is abundant research on observation, especially on the types and purposes of observation.Hopkins (2002) differentiates four categories of observation: open, focused, structured and systematic.According to Flick (1998), it can be structured and systematic, or unstructured and unsystematic; participant or non-participant; overt or covert; in natural settings or in unnatural settings; self-observation or observation of others.Irrespective of its type, however, lesson observation enables teacher trainees to gain data and learn directly from what they see -the here-and-now context of the lesson.It is a rich source of authentic information about the teacher's and learners' behaviour in class, the language they use, the activities and procedures, and the way they are organized to make the lesson a coherent and meaningful entity which provides conditions for student learning.
Traditionally, the purposes of lesson observation are conceived within the established frameworks for acquiring professional expertise, such as Wallace's (1991) craft, applied science and reflective models.In the craft model, in which trainees observe and imitate an experienced teacher under whose supervision they work, observation provides an option to see what the teacher does, without necessarily understanding the reason behind teachers' actions and decisions.Observing lessons within the frame of this model, trainees diligently describe what they see in terms of teacher and learners' behaviours and language, so that they can be later reproduced in a similar context.Unfortunately, this leads to uncritical repetition of teaching acts and behaviours, and eventually to reproducing traditional ways of teaching, thus cancelling any options for innovation and change.The applied science model in observation overcomes the craft model's disregard of research as natural source of language teaching and learning theories and practical ideas.In it trainees, while observing, have to recognise the ideas, techniques and behaviours from the theoretical researchbased course and see the rationale for their use in the lesson, a task which is difficult without the help of an instructor.The reflective model requires observers to think back to what they have seen in class, and assess it through the prism of their received knowledge about teaching and their experience as language learners.
More recently, Diaz Maggioli (2012) identified four traditions in teachers' professional preparation, calling them look -and-learn, read-and-learn, thinkand-learn, and participate and learn.The first three of them are not very different from Wallace's craft, applied science and reflective models.The fourth tradition, participate and learn has a lot in common with the sociocultural turn described by Tsui (2009), with its origin in the constructivist philosophy of education through support provided by more knowledgeable individuals while participating in social activities.This approach involves live lesson or video observations and discussion, learning about theories, key concepts and principles in language teaching and learning, micro-teaching, and feedback, all exercised as a cycle of activities rather than single experiences, so that learning can take place.
At the same time, Kumaravadivelu (2012) in his book Language teacher education for a global society: a modular model for knowing, analyzing, recognizing, doing and seeing, proposed a new model of language teacher education which is "cyclical, process-based, transformation-oriented, and holistic, an alternative to current traditional linear, product-based, transmission-oriented, discrete courses" (ibid., pp.ix-x).His model is organized around teachers' strategic thinking, and is based on three main principles: particularity, practicality, and possibility, emphasizing local needs and lived experiences understood through observing teaching, reflection-on-action, and theorising teachers' own practice.Kumaravadivelu's modular approach to language teacher education integrates five modules: knowing, analyzing, recognizing, doing, and seeing.Knowing is the dynamic process of building professional, procedural and personal knowledge rather than mastery of previously accumulated, static knowledge.Analyzing is linked to teachers' understanding of learner needs, motivation, and autonomy in a globalized world.Recognizing focuses on the identities, beliefs, and value systems that form the 'teaching self' a teacher brings to the classroom which determines their teaching behavior and shapes learning outcomes.Doing is the process of becoming an effective language teacher who is open to the transformation from "traditional" language teacher to "critical" language teacher, one who can teach learners to communicate in a globalized world (Kumaravadivelu, 2012, p. 78).The last module of the model, seeing bears particular relevance to the present discussion.It looks at classroom activities from the teachers', learners', and observers' points of view, but goes beyond the visible phenomena.The three types of seeing are: seeing-in -looking at what is already there; seeing-as -connecting and comparing past experiences, images, and actions with new ones, and seeing-that -applying relevant knowledge as a tool to understanding what is seen.This understanding is important for reducing the inherent subjectivity of observers' interpretation of what they see, as "we often observe what we want to see" (Wragg, 1999, p. vii).
As the overview of the major theories in teacher professional preparation shows, observation is part of any significant modern teacher education paradigm.In order to facilitate trainees in recognising and understanding processes in class, teacher trainers use what Lieberman (2009) calls "lesson study", involving learners in discussion and analysis of observed lessons.An important presumption, that trainees have to be aware of in lesson observation, is that "… classroom teaching is perhaps the most complex, most challenging, and most demanding, subtle, nuanced, and frightening activity that our species has ever invented" (Shulman, 2004, p. 258).Kincheloe (2004) suggested that a checklist approach to observation, which is typically used in competency-based models for assessing teacher effectiveness, would greatly reduce the complex holistic phenomenon of a lesson to a simple sequence of activities and techniques.Still, targeted training in perception is needed to help direct trainees' attention to the significant classroom events and situations that are relevant to the discussion and learning points (Santagata, 2009;Santagata & Angelici, 2010).Blomberg et al. (2011) point out that one of the key aims of university-based teacher education should be to systematically foster the development of professional vision, which is a knowledge-guided process (Palmeri, Wong & Gauthier, 2004) that draws on both generic knowledge and domain-specific knowledge.Professional vision includes two main subprocesses: noticing and knowledgebased reasoning (Sherin, 2001(Sherin, , 2007;;Sherin & Russ, 2015).Noticing refers to selective attention to the significant components of teaching and learning in the complex classroom environment which is knowledge-guided.Knowledgebased reasoning refers to the ability to process and interpret the noticed classroom events based on trainees' professional knowledge about teaching and learning (Borko, 2004).Sherin and van Es (2009) differentiate among three levels of knowledge-based reasoning: (1) description: the ability to identify, differentiate, and classify teaching and learning components; (2) explanation: the ability to link the observed classroom situation to professional knowledge; and (3) prediction: the ability to forecast learning-related consequences based on professional knowledge.The development of professional vision could be done through observing live and video lessons, or fragments of lessons, which have long been known to effectively activate observers' knowledge (Ellis, 1986;Ramani, 1987;Goldman, Pea, Barron & Denny, 2007).
More recently, a number of studies and research projects provided comprehensive reviews of different aspects of digital video use in the field of language teacher education (Cherrington & Loveridge, 2014;Calandra et al., 2018;Hockly, 2018;Major & Watson, 2018;Mann et al., 2019;Huttner, 2019).The reason for the increased integration of video resources in teacher education lies in the specific affordances they offer, such as the provision of a lasting record of teaching practice, allowing repeated viewing, selection of focus, and reduced demands of remembering the observed lesson (Huttner, 2019).In addition to this, "digital video is versatile: it allows more possibilities for context-sensitive noticing, reflection, editing, sharing, repackaging and tagging, especially in combination with screencapture software" (Mann et al., 2019, p. 7).
There are abundant videos for teacher training, such as Shaping the Way We Teach English: Successful Practices Around the World, or the DVDs, which are part of teacher training guides, e.g.Scrivener's Learning Teaching (2011), or Harmer's The Practice of Teaching English (2015), to mention just a few of these widely used resources.These video materials provide both subject specific and generic aspects of instruction and thus have the potential to activate knowledge of both these aspects.The digitalization of all spheres of education allowed for the creation of collections of good practices which demonstrate how video is currently used in language teacher education.An example of such collection is the Video in Language Teacher Education project, funded by the British Council ELT Research Partnership Awards Scheme, and its two outputs: the ViLTE Project website, which contains useful information, vignettes, transcripts of interviews, links, research team profiles, and the viLTE video case studies (see Mann et al., 2019).
While broadening trainees' horizons by exposing them to various teaching and learning contexts, such video materials can pose challenges due to the extent of their difference from trainees' native educational milieu, and thus potentially lead to misinterpretation of teachers' behaviours, intentions and decisions.In this line of argumentation, the best video materials would be the ones learners can recognise and identify with (Brophy, 2004).Moreover, exposure to video materials which reflect local contexts is easily provided by the ease with which lessons or fragments of teaching can be recorded even with a mobile phone, provided the ethical conditions of filming students are observed (Schuck & Kearney, 2006).These video materials, together with live lesson observation can easily trigger trainees' recall of their previous experience of teaching and learning, and activate their preliminary knowledge and expectations about the nature of teaching.This context-specific pre-training knowledge, in the form of existing conceptual schemata or mental constructs, serves as a foundation for developing trainees' professional vision.However, it can come into conflict with new ideas and practices seen in samples of teaching from different contexts, especially if they are misunderstood due to lack of experience.In outlining the differences between novice and expert teachers, Blomberg et al. (2014) mention that novice teachers' reflections on the observed are frequently overly judgmental and focus on descriptions, rather than abstractions and interpretations which integrate prior knowledge.One of the aims of the present study would be investigate if this trend is observed in our data.

The study
The study whose results are discussed in this article is based on a whole lesson observation which was assigned to the trainees as a final task in the course of lesson observation.Prior to this final assignment, trainees had been involved in observation sessions usually presented as structured and guided activities, divided into separate observation tasks with a clear goal and instructions for the observers.Video fragments and lessons were watched more than once when necessary, with the instructor helping trainees in their noticing and reflection on the targeted items.In pre-planned observation visits to schools, the focus of the observation is usually decided on in advance and negotiated with the teacher.In some sessions, however the choice of what to focus on was left to the trainees.Giving them freedom in what to observe can lead to important insights about the elements of teaching which they find salient and important.For their final assignment, all trainees had to watch the same lesson -a recording of Lindsay Warwick's lesson (from J. Scrivener's book Learning Teaching supplementary DVD).They had to identify and name the lesson stages, to describe the procedures and the rationale behind them, and to formulate the lesson's overall aims.In their work they had to demonstrate an ability to transfer the acquired theoretical knowledge and apply it to the concrete context of the observed lesson.

Participants
The study, which was held in the academic year 2021/22, included 74 university students (28 BA and 46 MA) enrolled in a teacher training module at a state Bulgarian university.The module is optional for BA students, but is traditionally chosen by most of them, due to the possibility of immediate employment as teachers of English upon their graduation.The optional module is the same for the MA students of English studies, and has exactly the same number of hours for each discipline.The MA students in the Pedagogy of foreign language teaching have more specialized subjects related to acquiring a teaching qualification, but the hours for observation are the same as in the optional module for the other participants.The exact number of trainees was divided into three groups as follows: • 28 BA students of English studies and Applied linguistics studying for a teaching qualification.None of them had any previous teaching experience by the time of the study; • 31 MA students (Pedagogy of foreign language teaching), 7 of which with previous teaching experience as teachers of English; • 15 MA students (English studies -Linguistics and translation, and English for business and tourism) with an optional module in teaching qualification, 5 of which with previous teaching experience as teachers of English.
All 12 students who had already had some teaching experience by the time of the study worked as substitute teachers or were in temporary employment, usually in the villages or small towns they lived in.

Research questions
The analysis of students' assignments was meant to provide answers to the following research questions: • Can trainees identify the main stages in the lesson, the transitions between the stages, and the reason behind organising the stages in a particular way?
• Can trainees identify and name specific activities and techniques (e.g.substitution drills, concept checking questions, controlled practice), studied in the methodology course and previously focused on in isolation?
• Can trainees see the lesson as a coherent whole consisting of a series of activities working towards achieving common aims?
• What are the main difficulties trainees experience in "seeing beyond" the surface of observable behaviours, and what might be the reason for them?
• Is previous teaching experience a factor in the particular tasks achievement?

Methodology
The obtained data, consisting of trainees' written observation records were processed for identifying trainees' challenges in recognizing and understanding key elements in the lesson, and the logic of the lesson development as a whole.The selected items present aspects of teaching which are explicitly observable in the video and which have been discussed in the lectures and seminars in methodology, as well as being studied in isolation as teaching techniques in both video clips and live lessons at school.For the purposes of clarity and establishing a quantitative dimension of student performance, a checklist was devised, which included the main foci of observation -the items which tap trainees' knowledge-based reasoning (Blomberg et al., 2011).The numbers were calculated depending on whether trainees have mentioned the respective item in their description of the lesson or not, and whether they have commented on it and interpreted it correctly, in line with the input they have received in the methodology course.The data indicate students' task achievement for each category in per cent (Table 1), as well as the mean and the standard deviation in the respondents' data (Table 2).The mean as an indicator of the typical value in each category of the assignment was used as a yardstick for all observations, and the standard deviation indicated how far trainees' individual responses deviated from the mean.

Results
Table 1 below presents the quantitative results of analyzing trainees' assignments in the format of the checklist used for their identification.The figures indicate the percentage of BA and MA respondents who identified and correctly interpreted the respective element of the lesson.For example, all 28 BA students managed to identify the introductory stage of the lesson and correctly interpret its main aim to introduce the topic of the lesson.However, the first pre-listening stage was identified by 61% of the trainees, and only 54% of them understood the role of the focus questions based on a picture.

Discussion
The initial observation of the figures in Table 1 and Table 2 reveal three significant trends which are relatively consistent throughout all items in the checklist.
The first trend is that the percentages of correctly indentified and interpreted items are relatively uniform among BA and MA students, which is not surprising, bearing in mind that the theoretical course, the instruction practices and the academic hours allotted for observation in both BA and MA programmes are the same.The overall values for the mean and standard deviation in BA and MA data do not show significant differences.The same refers to the coefficient of variation which is consistently low for the different sets of data -0.29 for identified items and 0.30 for their interpretation in BA group, and slightly lower at 0.21 for identified, and 0.24 for interpreted items in the MA group.This suggests that MA trainees have slightly better results, although the difference with BA trainees' achievement is not significant.The fact that 12 of the MA trainees had had some previous teaching experience prior to their studies at university, did not make any significant changes to the results.A possible reason for this is that trainees' teaching experience had taken place in contexts in which traditional models of teaching, such as Grammar translation, were prevalent, and they didn't have exposure to newer ideas and ways of teaching.
The second trend emerging from that data is the decreased percentage for the correct interpretation of the target items compared to their identification.This means that trainees could notice and identify observable phenomena, but found it more difficult to move beyond description and interpret the target items' significance in isolation, and as part of a whole entity that the lesson is.This finding is in line with Blomberg et al. (2014), who claim that novice teachers' reflections on the observed tend to focus on descriptions, rather than abstractions and interpretations which integrate prior knowledge.This also explains some trainees' failure to interpret teacher's actions in terms of the theory studied in the methodology course.Although the course in lesson observation ran simultaneously with the course in ELT methodology, trainees found it difficult to relate the theoretical ideas with their practical implementation, especially in contexts different from the ones they had previously experienced as language learners in Bulgaria.As Huttner points out, "background knowledge and experience create 'lenses' through which teaching performances are viewed, specific aspects noticed and interpreted" (Huttner, 2019, p. 479).Our data suggest that this happens even after different activities, techniques and teacher behaviours were observed and discussed in isolation in the training sessions, raising the question whether a single exposure to a teaching technique and its discussion is enough to make it part of trainees' received knowledge about teaching.
The third trend, which is clearly visible in Table 2 in the comparison of mean values for BA and MA trainees, is that both groups had more challenges in identifying and correctly interpreting the role of specific teaching techniques in comparison to identifying and interpreting the stages of the lesson.Thus, the means for BA trainees are 87 and 76 for identifying and interpreting stages vs. 60.09 and 52.18 for techniques.For MA trainees these values are 83.54 and 74.38 for stages, vs. 69.18,and 61.64 for techniques, respectively.As it was already mentioned, prior to the assignment teaching techniques were mainly focused on in isolation, both in theory and in practical examples provided through video clips from various contexts around the world, rather than through live demonstration in Bulgarian classrooms.In local context of teacher training, live school-based observations have more general aims, and instead of focusing on different elements or techniques in teaching, they "suggest a holistic view of the lesson and aim to develop trainees' ability to summarise and comment on the lesson as a coherent structure, paying attention not only to its integral parts, but also to its effectiveness as a whole" (Ivanova & velikova, 2012, p. 247).This suggests that if trainees are deprived of the opportunity to observe the use of these techniques in their local context, they might not recognise them and correctly interpret their role in a different educational context, such as the small private language class in the UK observed for the assignment.
The main challenges identified through data analysis could also be grouped into three broad categories which are closely interrelated: The first challenge is related to interpreting teacher's actions in terms of the theory studied in the methodology course.In addition to being an important tool for bridging the gap between theory and practice in teacher training courses, evidence from observation "has invariably been used to inform current conceptualisations of what makes for effective teaching and learning along with providing the basis on which judgements about the performance and competence of teachers are made" (O'Leary, 2012).Understanding this evidence, more specifically the logic behind teacher's and learners' observed behaviours, turned out to be a major challenge -the mean for trainees' overall achievement in this category is 63.92 for BA and 67.44 for MA respectively, both figures being lower than the ones for identifying lesson elements (73.4 for BA and 75.6 for MA).
As it was already pointed out, trainees' inability to make conceptualizations and judgements is exacerbated when teacher and student observed behaviours in the lesson differ substantially from the common practices and the roles typically enacted by teachers and learners in Bulgarian educational contexts.This increases the importance of the theoretical input in which different lesson elements are discussed, leading to a deeper understanding of teaching as a purposeful activity aimed at achieving clear pre-planned learning outcomes.In supervised training sessions, such as seminars in lesson observation, trainees are constantly reminded of the need to reflect on what they see in terms of what they have already studied in the theoretical course.However, when they work on their own on assignments, some of them fail to make this back-reference, offering instead superficial, often naïve descriptions or commentaries, such as: "…the teacher tried to make the lesson more entertaining by using games and role plays", where the purpose of the activity was to use a variety of techniques to pre-teach key vocabulary; or "the teacher let students speak to see what they think about the topic", instead of the real purpose of the activity, which was personalisation -giving students the chance to use the new language to talk about themselves.
The second challenge is in perceiving the lesson as a coherent entity consisting of logically sequenced stages.As it was already pointed out, trainees' written accounts of the observed lesson were more descriptive than interpretive.Trainees tended to describe the sequence of activities in a great detail, often using general or vague language instead of special terminology.This preoccupation with detail did not allow trainees to penetrate beyond the surface of observed behaviours and see the lesson as a coherent whole with pre-defined teaching and learning aims.Reading some of their accounts created an impression of an incoherent compilation of random actions with no clear purpose.The failure to perceive the lesson coherence is closely related to difficulties in identifying the stages of the lesson and the logic of their sequencing, e.g.seeing how a stage or activity prepares for, or leads to the next one, going from simple to complex, grading activities in terms of difficulty, providing variety or a change of focus.The most challenging for the trainees turned out to be the grammar practice part of the lesson with its carefully staged activities, evolving from strictly controlled drilling to a personalized communicative activity.The data show that the overall score of BA and MA trainees for perceiving the lesson as a coherent entity is exactly the same -43%, with MA trainees having slightly higher score for understanding the nature of lesson coherence -41%, compared to 32% of BA.
One possible reason for this could be that all 12 trainees with some previous teaching experience were from the MA group.However, the fact that less than half of all respondents commented of this aspect of the assignment shows that this area of professional knowledge needs further work based on reflection on practical examples of complete teaching events, such as whole lessons, rather than isolated segments which illustrate single activities and techniques.
Among the lesson stages which were easily identified and interpreted by both groups of trainees were the beginning and the closing stages of the lesson, the second listening -for detail, and the fill-in-the-gaps controlled practice sub-stage.However, in these stages too, in line with the pervasive trend for more problems in correctly interpreting the meaning of the observed actions, the trainees recognized the type of the activity, e.g.gap-filling, sentence completion, drill, etc, but not the logic of sequencing the stages.In the case of grammar practice, it was increasing the challenge of the activity by using simple substitution of the verb form before substitution with a tense change, or moving from a grammar exercise with a single focus -'fill in the gap with the correct verb form', to the more complex one -'fill in the gap and then change the sentence so that it is true for you'.Only a few students mentioned how the controlled practice activities build up to the freer communicative task with elements of personalization.The other activity, which caused problems to most trainees, in addition to the communicative one near the end, was the pre-listening at the beginning of the lesson, which aimed to activate students' background knowledge through focus questions based on a picture.In it, few trainees identified the true purpose of the focus questions, and even fewer noticed that the same questions were used later in the lesson to check the results of the gist listening.
The third type of difficulty consisted in recognizing teaching techniques and the reasons for their use.In the methodology course trainees were introduced to a number of teaching techniques and approaches which are not widely used in Bulgarian educational contexts.These involve: a variety of drills, concept checking questions, different ways of presenting the meaning of vocabulary or grammar items, skimming and scanning in reading, situational presentation, test-teach-test, personal anecdotes as a way of contextualising language items, timelines, using fingers and gestures, different ways of monitoring, use of pair work for checking answers prior to answering teacher's questions, etc.As indicated in the checklist in Table 1, a lot of these techniques were targeted in the assigned observation.However, data analysis showed that instead of using the correct terms for these items, trainees tended to use general words, such as exercises or tasks, games, stories, examples, etc.The failure to recognise the studied techniques in the teacher's behaviour almost always entailed misunderstanding of the rationale for their use.The data from the checklist show that trainees did not seem to have difficulties with identifying certain features, such as gestures and finger use, and error correction (96 and 82% for BA trainees), modeling pronunciation and substitution drills (87% for MA trainees for both).However, other feature posed more challenges: concept check questions and praise being the lowest in score (36% for both for BA trainees, 50% for both for MA trainees).Other challenging elements include eliciting techniques (43% for BA trainees), focus questions (with 50% for both groups), and feedback (54 and 61% for BA and MA respectively).Our data also suggest that a failure to recognise a technique results in a failure to interpret its meaning, e.g. the same values for both identification and correct interpretation for concept check questions and eliciting (36 and 43 for BA trainees respectively for recognizing the feature and for interpreting its role in the lesson).Describing teacher's behaviors some trainees failed to understand their intention, e.g. a trainee wrote: "The teacher uses a lot of gestures and even her fingers to attract students' attention and make the lesson more lively", while the purpose for using gestures really was to supplement verbal instructions, and to clarify time reference in explaining past simple tense used for hypothetical events in second conditional.Finger use for indicating the relevant element in a structure was also largely misunderstood by some trainees.Among other examples of misunderstanding is the use of focus questions to channel students' attention, activate their mental schemata and elicit predictions before the listening stage.
Only 46% of BA trainees and 43% of MA mentioned some of these functions.
Error correction also posed difficulties, when it was delayed rather than on-thespot, the type of correction trainees typically observe in Bulgarian classrooms.Data suggest that if certain behaviours are perceived as strange or unusual, e.g.excessive use of gestures, squatting while monitoring, pervasive pair and group work, students' drilling, etc., they are less likely to be identified and interpreted adequately.What is more, they are less likely to become part of trainee teachers' personal teaching repertoire.

Conclusion, implications and recommendation for supporting trainees
The present study aimed at revealing trainees' problems in observation which are not easily detectable in routine oral post-observation discussions.The study revealed that, when asked to describe the procedures in a lesson, trainees can produce detailed accounts which, however, are predominantly descriptive rather than being reflective, an observation which confirms previous research by Blomberg et al (2014).
The analysed data threw light on some context-specific trends, such as the similar results for BA and MA trainees when the course of tuition, instruction practices and the amount of exposure to observation are comparable.Our data also suggest that trainees have more difficulties in identifying and correctly interpreting specific teaching techniques than in identifying the stages of the lesson.However, correctly identified stages do not always mean that trainees perceive the lesson as a coherent entity consisting of logically sequenced stages.
Our findings imply that the reasons for the identified problems are multiple, and even an open discussion with the trainees cannot always shed light on the sources for the experienced challenges.Any attempt at addressing identified challenges and supporting trainees requires further research into the reasons for the difficulties.Among these reasons might be insufficient theoretical preparation -a lot of trainees rely only on the instruction received in lectures and seminars, avoiding or neglecting required readings from teacher training guides and research articles.Another reason could be the lasting impact of trainees' native educational culture with its traditional models of teaching, which might lead to conscious rejection of practices perceived as strange or inappropriate for our local teaching context.
One possible recommendation for teacher trainers would be to increase the exposure of trainees to good practices of English language teaching which extend beyond local contexts, supplemented by explicit discussion of their advantages over traditional, dated or unsupported by research local teaching practices.
As Hennessy (2014) suggests, bringing videoed vignettes and interviews with teachers into training sessions can help break down the divide between 'foreign' and 'local'.This should be supported by suggestions for adapting and customising 'foreign' ways of teaching to local contexts as part of a consistent focus on the ways in which teaching ideas and models from internationallyaccepted ELT courses can be adapted and used as sources for practical decisions in local contexts.University lecturers should work in close collaboration with local mentors who can demonstrate lesson models and techniques consistent with the ones promoted in the university methodology course.
Finally, one of the most important ways of supporting trainees in developing a professional vision and expertise is encouraging reflection on observed teaching, so that they find it easier to see beyond the surface of observable events and behaviours.

Table 1 .
BA and MA respondents' data from the checklist of lesson elements in per cent

Table 2 .
Mean and standard deviation in the respondents' data