Increasingly, researchers have recognised the complex interactive nature of coaching, highlighting its multifaceted, integrated and dynamic character (e.g. Saury and Durand, 1998, d’Arripe-Longueville et al., 2001 and Jones and Wallace, 2005). It is a portrayal of coaching, whilst not totally dismissing its developmental disposition, gives growing credence to the non-processual social pressures to which it is also subject (Puddifoot 2000). Hence, the assumptions of unambiguous linearity and sequence associated with traditional processual conceptualisations of coaching are questioned, with the coach being seen not only as a ‘mountain guide’ through athletic peaks and troughs, but also someone capable of changing the landscape itself (Mayer-Kress 2001).

Although such an appreciation is welcome, accompanying investigative methodologies have not been forthcoming. Indeed, the growing acknowledgement of the co-operative complexity of coaching, containing elements of initiation, reaction and exchange within a temporal process, means that traditional research methods often miss the details and nuances on which much of coaching actually rests. Consequently, if Jones and colleagues’ belief (Jones et al., 2002 and Jones and Wallace, 2006) that coaching is not just delivered through a series of given, linked episodes but is a dynamic unfolding social activity is taken as conceptually sound, research methodologies permitting an effective in-depth exploration of it must be sought. The aim of this chapter is to provide a rationale for, and description of, a method that takes into account the social complexities of coaching, by scrutinising the meaning and interpretation that an individual gives to a task before deciding on an appropriate course of action. The objective is to discuss and illustrate a means to record and subsequently interrogate coaches’ understanding of their actions in relation to why they coach as they do within their collective contexts.

The value of the chapter is two-fold. Firstly, it builds upon Puddifoot’s (2000) critique of what counts as a ‘social process’, arguing that a more intricate and complex appreciation of the concept would be advantageous for a further insightful conceptualisation of coaching. Far from being some underlying regular activity over an unspecified period of time where a predicted linage or causal connection takes place (Puddifoot 2000), the act of coaching is alternatively posited to be founded on the interaction or negotiation between coach and athlete in context (Cushion & Jones 2006). In this respect, coaching is reconceptualised as a ‘dynamic social process’, de-emphasising considerations of unproblematic chronology and the assumption that an identified linkage in one context would ever be repeated in another (Puddifoot 2000). It is a view gaining increasing credence within the field of pedagogy in general (e.g. Sonsino & Moore 2001) as well as sports coaching (d’Arripe-Longueville et al., 2001, Bowes and Jones, 2006 and Hauw and Durand, 2007). Here, it has been claimed that optimum learning occurs at or near the so-called ‘edge of chaos’; a state which lies neither in a zone of complete stability nor total flux (Sonsino & Moore 2001). The trick for coaches then, is to keep the coaching context hovering in this ever-evolving zone of dynamism through managing the micro-interactions among the agents who comprise it (Bowes & Jones 2006).

Our claim to originality here is tempered, as we are, to a considerable extent, following in the footsteps of others (e.g. d’Arripe-Longueville et al., 2001, Hauw and Durand, 2005 and Hauw and Durand, 2007). Consequently, the second and principal significance of this chapter lies not so much in breaking virgin ground, but in developing and more firmly locating complexity-related studies, and the methodologies they employ, within the field of coaching research. In doing so, we hope to bring an innovative means of investigation to a wider audience in order to stimulate further study and understanding of coaches’ immediate and temporal actions, strategies and responses. In terms of structure, we firstly present a brief critique of traditional methods employed to explore coaching, thus making the case that more sensitive means are required for a greater nuanced understanding of the activity. The method we put forward, course of action analysis, inclusive of both situated action (Suchman 1987) and activity theory (Leont’ev 1978), is then defined and discussed as one example through which this can be achieved. Finally, a conclusion summarises the main points made whilst further outlining the implications of course of action analysis for future coach education programmes and coaching practice.

Without re-presenting the case made in earlier work, suffice to say that traditional methods used in coaching research have largely followed a positivistic approach, with reductionism and generalisability at its heart (e.g. Jones & Turner 2006). This agenda has been reflected in the use of methods such as questionnaires, systemic observation and structured interviewing, giving rise to a conceptualisation of coaching through particular constructs (e.g. Jowett & Ntoumanis 2004) and various models (e.g. Côté et al 1995). In many ways, this approach represents an attempt to identify ‘effective’ practice from ‘good’ coaches; to extract knowledge and actions considered tried, tested and, therefore, repeatable by others. It is a conceptualisation of coaching as a knowable sequence of decisions over which coaches have command, and reflects the assumption that the coaching process is primarily governed by a set of measurable, achievable goals with resources being freely available to realise those goals.

Recently, Jones and Wallace (2005) argued that a fundamental problem with this approach, and the subsequently espoused coaching knowledge, is that the phenomenon of coaching itself is insufficiently understood to effectively prescribe a model of ‘best practice’. They go on to discuss that, in utilising a reductionist perspective, behaviour is explained in a mechanistic manner and, hence, is viewed as causal and predictable; the price paid being over-precision in guiding coaching practice. Furthermore, Cushion (2007), in his critique of Lyle (2002), argues that plotting a series of hierarchical relationships and proposed interactions in a model of coaching does little to illuminate the complex actions and their precursors, which actually underpin the activity (Jones et al. 2006). This is because such an approach takes little account of social, moral and cultural influences on individual behaviour, thus disenfranchising the person (McFee 2005). The problem, therefore, with adopting a positivistic approach, which still holds considerable sway in some quarters (e.g. Voight 2007), is that it leaves many coaches frustrated because of the extent to which its representations are divorced from reality. As Jones and Wallace (2005) alluded, the direction given in coaching models is not viewed as applicable as it ignores the tensions and social dilemmas that characterise coaching practice. What we are often left with then is an unbridgeable gap between the best intentions and goals that drive and inspire coaches, and their capacity to attain them – a situation which Jones and Wallace, 2005 and Jones and Wallace, 2006 described as a ‘pathos’ within the coaching theory to practice domain.

Coaching has also come to be recognised as a personal venture, negotiated between various stakeholders within a specific context (Jones et al 2004, Purdy et al in press). Such a position further problematises many of the processual assumptions and discourses that underpin our understanding of coaching and how we go about researching it. In depicting the activity as being reliant on patterns of interactions between individuals, this evolving conceptualisation has placed coaching close to the tenets of complexity theory inclusive of both predictable and random results, and where small differences in initial conditions can result in large changes in outcomes (Puddifoot, 2000, Mayer-Kress, 2001 and Bowes and Jones, 2006). To date, however, a definitive framework to adequately explore this complexity in coaching has proved elusive.

While we have outlined the principal limitations of the positivistic approach, the works of Côte et al., 1995 and Lyle, 2002 in particular have provided an impetus for recognising coaching as a sophisticated interpersonal process. Hence, although criticised for oversimplifying a complex practice (e.g. Cushion, 2007 and Jones, 2007), it could be argued that such work has served as a necessary precursor for a subsequent more discerning approach (Lyle 2006). If the initial signposts were there for us to read, much of the credit for leading us in this critical, complex direction goes to the work of Saury and Durand, 1998, Sève and Durand, 1999 and d’Arripe-Longueville et al., 2001. It is to this body of work that we now turn to further make the case for a more sensitive methodological approach to study the phenomena of sports coaching.

One of the first studies that began to shed some light on the complexity of the coaching process in situ was that of Saury and Durand (1998). Having critiqued Côté et al’s (1995) model for portraying coaching tasks as being specifiable in advance, Saury and Durand (1998) adopted a cognitive ergonomic (i.e. designed along human need) perspective where the task was presented as a set of constraints facing individuals (e.g. the varying abilities of athletes, the facilities at hand, and/or the immediate objectives of particular actors). The interaction of these constraints generated a set of complex, contradictory, and ill-defined problems. Given this unpredictability, the approach embraced by the coaches studied by Saury and Durand (1998) was based on past experience and was characterised by negotiation and (to varying degrees) shared responsibility with athletes. The study also demonstrated that elite coaches draw upon knowledge far beyond the procedural, rule-based information conveyed in training manuals. Rather, it extended to practical and personal knowledge tied to experience; highly context-dependent knowledge which, because it is often characterised by automation, is difficult to articulate (Saury & Durand 1998). Similarly, Sève and Durand (1999) in their investigation of the temporal and contextual organisation of the working behaviours of three elite coaches, found that such practitioners’ activities were not planned in advance but were reactive and heavily typified by flexibility in response to ever-changing, unforeseeable situational characteristics (Bowes & Jones 2006).

Applying this conceptualisation of coaching to the context of competition, d’Arripe-Longueville et al (2001) explored coach-athlete interaction during elite archery contests. Defining situated actions as actions taken in the context of specific, concrete circumstances where cognition is seen as inherently linked to experience and ecological constraints, they identified coaching per se as being collectively constructed. This was particularly so in relation to the shared control that was evident between the actors present. Here, they suggested that so-called ‘common’ goals were not entirely predetermined, but were collectively constructed through interaction between the agents. Coordination between team members was considered as emerging from mutual constraints, a process which did not follow a general procedure or pre-defined plan. Saury and Durand (1998) similarly proposed coaching as a set of interacting constraints: that coaching tasks comprise a series of routines and cognitive anticipation in relation to flexible plans, whilst works by Sève et al., 2003, Hauw and Durand, 2005 and Hauw and Durand, 2007 amongst others, have also positioned coaching along such complex lines. Therefore, although grounded in experience, coaches’ practice is increasingly being viewed as very flexible and responsive to unforeseeable situational characteristics (Bowes and Jones, 2006 and Sève and Durand, 1999). Coaches’ subsequent planning then, is not considered as being related to pre-set applied goals, but rather is dependent on ‘continuous step-by-step tuning to the context’ (Saury & Durand 1998 p 264).

Such a conceptualisation of coaching echoes that of Jones and Wallace, 2005 and Jones and Wallace, 2006, who recently posited that the quantifiable processual goals often set by coaches are doomed to failure given the need for variable and constant adaptation to context; where physical, social and personal factors militate against any sequential, unproblematic progression. The implications of such work are that coaching expertise requires constant adjustment to evolving constraints (albeit within certain parameters) imposed by the situation, the actors, and the broad pedagogical objectives. In this respect, coaching becomes a contested, negotiated practice between those involved within the social setting (Cassidy et al., 2004 and Jones, 2007). Although many of the studies we have considered here utilised small samples, which preclude their findings’ general application, we believe that the uncovered evidence points to potential commonalities and, hence, to a very fruitful area of investigation.

In an attempt to accommodate this increased appreciation of complexity as an important facet in explaining coaching interaction, Bowes and Jones (2006) utilised the notion of relational schemas and complexity theory. Here, relational schemas provided a coherent explanation for coaches’ possession of an experience-based, constantly evolving framework for expected patterns of interaction (which regulate their behaviours). Complexity theory meanwhile was used to depict the coaching environment as problematic and non-linear, therefore aligning it with propositions of coaching theorists (e.g. Sève & Durand 1999). Within this framework, the pedagogical process is viewed as complex and dynamic, with controlled yet chaotic interaction being the catalyst for learning (Sonsino & Moore 2001). Teaching moments are derived rather than pre-ordained, with the coach’s role being essentially to manage the various agendas and goals held by those present for the greater good within a context characterised by continuous transformation (Sonsino and Moore, 2001 and Jones and Wallace, 2005). Consequently, studying coach-athlete interaction as a dynamic process involving measured reflection within flexible plans necessitates an exploration and understanding of situated action, the meanings of such action, and the constructed cognition underpinning such actions within a given social (coaching) system. It is to a considered discussion of course of action analysis as a means through which we can realise this goal that we now turn.

To fully understand an activity, its history and growth should be taken into consideration. In studying development, Thelen and Smith (1996) argued that although descriptions of events are useful if not essential to understanding, they also hold the potential of being flawed explanations. This is because the move from description to explanation requires only a subtle shift. As evidence of this, descriptions of coaching have tended to become ‘how to’ models leading to a superficial acceptance or adoption of causality (Jones & Wallace 2005). This has resulted in some neophyte coaches unproblematically replicating the behaviours of their more experienced counterparts with little appreciation of social or individual factors that may have affected the original development, resulting in an often ineffective interaction and confusion for the coach (Jones et al 2003).

An alternative position argued by Thelen and Smith (1996) is that of viewing development as a process that possesses non-linear flexibility while still acknowledging a (target) goal state. This conceptualisation of a self-modifying process moves away from the proposed division between cognition and behaviour, where the boundary between action and response is difficult to define (Varela et al 1991). Instead, such a stance contends that it is impossible to separate the products from the processes because the processes used change the user (Minsky 1986). Research involving reflection (e.g. Schön, 1983 and Kolb, 1984) supports the impact of this evaluative or self-modifying process, where the implications for the act of coaching look more like the complex picture given elsewhere, and less like a knowable coaching process (Bowes & Jones 2006). Anecdotally, this also seems to match the reality of coaching, where athletes’ perceived need is gleaned through exposure to various competition-related situations as opposed to repetitive coach instruction. It is a position which suggests the importance of participation in less controlled activity in order to develop the skills needed to negotiate the complexities of (game and social) interaction, thus questioning the assumed causal link between prescribed coaching acts and defined outcomes.

Course of action analysis (Theureau 1992) is suggested as a means to better engage with the complexity of non-linear phenomena such as coaching as it seeks to describe and analyse the action of agents from their point of view in relation to the characteristics of the situation. Epistemologically then, without wishing to apply definitive and, hence, restrictive labels, course of action analysis lies within the interpretive realm, inclusive of interactive and subjective methods.

Grounded in both activity theory (Leont’ev 1978) and situated action (Suchman 1987), course of action analysis acknowledges three dimensions of the ‘human system’; the task, individual differences and, importantly, the timescale of behavioural acts. It is a perspective that increases the focus or the unit of analysis to include more than just the individual actor or the environment, to the interaction that occurs between these elements over time. Underlying the notion of course of action is the premise that cognition utilises many sources including the social, cultural and material characteristics of the context in which it occurs. It is a sentiment that echoes that of Simon (1977), whose work highlighted the dynamic and indistinct quality of real life problems, which are both impacted upon by current knowledge and generated from interaction with the environment.

Course of action analysis fundamentally argues for the importance of participation in structuring thought (Varela et al 1991). Here, it emphasises the dynamic quality of experience, suggesting agents’ interactions with their environment are developed in line with particular contexts and are not predefined (Varela et al 1991). Such experiences are not solely individually enacted but collectively manufactured, meaning that individual events are interwoven with culturally bound artefacts, symbols or actions.

It is a position that draws heavily from three key ideas of situated approaches. Firstly, that cognition is inseparable from the activity where it is produced, thus actors can mobilise and build their own knowledge from the emergent coupling of activity and situation (Guillou & Durny 2007). Secondly, that activity is a construction of meaning which does not pre-exist activity but emerges with it (Theureau 1992). Finally, that activity is a dynamic totality, with no distinction between action and interpretation (Varela et al 1991). Such a perspective holds the potential to examine what recent research has termed the seemingly intuitive, unplanned actions of coaches; actions which accommodate the ‘unforeseeable contingencies of particular situations’ (d’Arrippe-Longueville et al 2001 p 277). This last notion, that activity is a dynamic totality, is rooted in Simon’s (Newell & Simon 1972) concept of ‘satisficing’, where problem-solving and decision-making rest not on a painstaking search for optimal idealistic solutions as the dynamic context does not allow such opportunity. Rather, they rely on a judgement regarding what is good enough, that satisfies the needs of the complex real world. Consequently, a course of action constitutes a dynamic entity comprising actions, communication, interpretations and feelings that are meaningful for the actor in situ. Furthermore, the action (described from the actor’s point of view) is presented and analysed in relation to the overall extrinsic characteristics of the situation (d’Arripe-Longueville et al 2001).