This research proceeds from the perspective that various types of face-to-face interaction are convention-based or rule-governed. This approach to interaction research follows a two-step process. First, once actions are transcribed using Theme coder software, exploratory analyses of these actions are designed to yield hypotheses concerning one or more conventions being used by participants in an interaction. These empirically based hypotheses, termed structures, include among other things: (a) a set of appropriate actions, termed elements, for each participant in the interaction; and (b) a set of rules, describing the sequences in which these actions may appropriately occur. By identifying patterned sequences of transcribed actions, Theme provides invaluable information relevant to the formulation of structures; the information most directly relevant to hypothesizing structures will be reviewed. In this phase of the research, Theme patterns are not in themselves the results; rather, these patterns provide information necessary for the formulation of structures.

Second, once these structures - that is, the 'rules of the game' - have been hypothesized, it becomes possible to examine interaction strategy: the way in which the participants jointly construct specific instances of interaction, within the constraints provided by the rules. Obvious examples of structure and strategy would be chess or basketball, in which many different games and game outcomes (strategy) may occur within the constraints of a constant set of rules (structure). It is clear that productive analyses of strategy in these games is dependent on prior knowledge of their rules - particularly those points at which participants choose between various rule-defined options. Similarly, the analysis of interaction strategy only becomes meaningful once the structures describing the elements and rules relevant to the interaction have been hypothesized.

In part, interaction strategy can be described as sequences of elements, performed by the respective participants, at those points in the interaction at which the structure provides choices between alternative elements. These choices are a major constituent of the joint construction of an interaction by the participants. Theme can be used to analyze sequences of elements, uncovering patterns of choice. In addition to these choice patterns, analyses of interaction strategy can be used to examine, for example: (a) the degree of influence of each participant upon the other as the interaction proceeds; and (b) whether a choice of action at a specific point can be described as a Markov process, or as something more complex. Thus, the Markovian nature of this process is treated as an empirical issue.

This presentation will include examples of interaction structures and accompanying results, showing the degree of fit of these structures to the transcribed data. Based on these structures, examples of strategy analyses, derived from Theme analysis of element sequences, will be presented. These will illustrate the fluctuating influence of each participant upon the other's choice of elements as the interaction proceeds, and evaluate this influence as a Markov process.

Paper presented at Measuring Behavior 2002 , 4^th International Conference on Methods and Techniques in Behavioral Research, 27-30 August 2002, Amsterdam, The Netherlands

© 2002 Noldus Information Technology bv