• No results found

3.5 Data collection tools

3.5.3 Classroom observation

Although questionnaires were the initial and main data collection tool, I was aware that they often have limitations in the number and quality of responses between teachers' beliefs and practices (Grotjahn, 1991; Richardson et al., 1991; Schultz, 1970). Merriam (2009) suggests that the benefit of observation over interviews is that it allows the documentation of behaviour as it occurs first hand.

Grammar pedagogy is ‘the performance of teaching together with the theories, beliefs, policies and controversies that inform and shape it’ (Alexander, 2000, p. 540). Alexander (2000) identifies the curriculum with policies and dismisses its value when it is in this form. Instead, Alexander brings ‘curriculum’ and ‘pedagogy’ together, arguing that it is pedagogy that should be the focus of any analysis of practice in education. The teacher is representative of a particular community of practice, and characterising individuals in terms of their participation is characteristic of a sociocultural perspective (Cobb and Yackel, 1996). Children in Ireland are participants in very different types of learning activities and these activities are culturally organised at the Irish societal level.

In seeking teachers’ beliefs, Pajares (1992) has highlighted that their beliefs do not lend themselves to empirical investigation, because of the problem in defining them. This is because beliefs are based on evaluation and judgement, whereas knowledge is based on objective facts. Beliefs underlie both what teachers declare as declarative knowledge and what they actually adopt in practice seen as procedural knowledge. I chose to observe

lessons to enquire beyond stated beliefs (Basturkmen et al., 2004) and to witness the

theories in action (Argyris and Schon, 1974). This was in line with the recommendations of Pajares (1992) and Borg and Burns (2008) that studies of belief should include some degree of observation that enables researchers to compare espoused beliefs to practice, expanding the study of cognition to ‘the study of what teachers know, think, and believe and how these relate to what teachers do’ (Borg and Burns, 2008, p. 457). The value of observing teachers in their natural contexts (the classroom) and discussing their practices has been highlighted by Borg (2005b). In fact Borg (2003b, p. 105) expresses scepticism about whether language teachers’ cognition can be usefully studied without reference to what happens in the classroom. Direct evidence of behaviour is witnessed and it allows a large amount of descriptive data to be collected.

Observations may reveal beliefs which are embedded in context and practice, tacit or even unconscious, and may reveal competing, inconsistent, transient beliefs, or even beliefs which are in the process of change (Richardson et al., 1991, p. 578), whereas interviews offer beliefs which are more decontextualised and usually propositional.

However, both Pajares (1992, p. 27) and Silverman (1993, p. 106) highlight the importance of linking interviews to observations when examining belief. Calderhead (1996, p. 711) states that, ‘observation alone is of limited value, for the cognitive acts under

investigations are normally covert and beyond immediate access to the researcher’. It is true that ‘beliefs in use’ can only ever be inferred, and it is important to be clear that different types of ‘belief’ are examined through interview and through observation.

Because teachers develop a complex personal framework of values and beliefs in the course of their lives and because they bring these to the classroom they generate cultural models (Gee, 1999). Cultural models are theories that help to make sense of the world and experiences in it which are rooted in our socially and culturally defined practices. I was searching for the cultural models (Gee, 1999) which serve to inform

teachers’ beliefs and subsequent grammar practice. As these are often held unconsciously, it may be also be necessary to infer them from people’s behaviour.

As there could have been a significant issue of reactivity through the presence of the researcher in the classroom, it was necessary for me to be introduced to the class and to give the children an understanding of why I was present. I had suggested that the teachers call me by my first name when introducing me to the children and to explain that I am a person who was spending some time in Irish lessons in different classrooms to see how Irish was being learned. Labov (1972) referred to the ‘observer’s paradox’, which

highlights the fact that although the aim of most observational research is to collect data as unobtrusively as possible, the presence of an observer can actually influence the linguistic behaviour of those being observed. As previously stated, I was very conscious of the Hawthorne effect (Landsberger, 1950), whereby people modify their behaviour because they are being watched. I had found that in the pilot observation, because I was not directly visible to the pupils in the cloakroom at the back of the classroom, it seemed the class forgot that I was there. I therefore deduced that it is ideal to be ‘hidden’ in the classroom. I had not asked teachers how they actually taught grammar (being aware that the

communicative approach is the one prevalent in Ireland) so observation offered me the chance to see their grammar teaching unmitigated by the reflection in interviews and questionnaires.

The three classroom observations served both to elucidate and validate or otherwise data generated via the questionnaires and interview but also provided qualitative data on the context of the grammar teaching and provided other unforeseen data which may have arisen incidentally. The impression that triangulation implies a realist ontology enabling a closer approximation to ‘truth’ is questionable because it may be argued that different methods reveal different aspects of a research subject, providing ‘different versions or

“levels” of answer’ (Mason, 1996, p. 149) or revealing ‘situated’ actions and accounts which cannot be simply decontextualised through ‘triangulation’ (Silverman, 1993, p. 157).