Communicative interaction in face-to-face situations involves subtle processes of emotionalisation, that is, it is characterized by a continuous variation of positive and negative departures from a neutral mood. The aim of our research is to show, first, that these small-scale changes of emotionalisation do occur in normal as well as in pathological discourse, and, second, that they contribute to an individual's actual and global well-being. Specifically, with regard to communicative exchanges with psychotic index persons, we attempt to show that these mood variations and how the psychotic individual is able to cope with them is relevant for the course of a given mental illness.

In order to grasp variation of emotionalisation we concentrate on three types of variables:

• physiological measurement of vessel volume changes registered by photoplethysmographic equipment [2],
• behavioral measurement of changes of head movement behavior [1], and
• behavioral measurement of vocal changes during communicative interaction [3].

We conceive of these variables on the one hand as being expressive of a given emotional-activational state of the index person and on the other as being instrumental in guiding the partner's and the index person's own subsequent communicative behavior. Furthermore, we have evidence that in turn communicative reactions are fed back and may modulate (i.e. neutralize) a given state of emotionalisation.

Within this theoretical framework we are confronted with a number of methodological problems which, as a rule, are not encountered in experimental settings. However, problems concerning the control of artefacts, the coordination of data from multiple behavioral channels, the definition of relevant units and magnitude of variation are specifically discussed in the contributions of Jossen et al. (physiological measurement), Altorfer et al. (head movement measurement) and Zimmermann et al. (vocal measurement). Here, the discussion is restricted to a more fundamental problem pertaining invariably to each of these domains. It is the question of how, in the absence of an experimental manipulation which normally delimits baseline from intervention-dependent data, meaningful segments entering a comparison across situations may be created.

Answering this question is indispensable for working with data from natural settings. To come to grasp with the definition of classes of equivalent events in natural communicative interaction, we adopt the following approach: The background of all of our analyses is formed by a minimal model of communicative exchange. Within this model three utterances of two participants A and B are assumed to be functionally related in the following way. A's first utterance A1 is an offer to be interpreted by B; B's utterance B1 contains an interpretation of A1, and A's second utterance A2 is based on an evaluation of the expected with the observed B1; in case of actual discrepancy, A2 (ideally) contains a correction. Within this framework, critical communicative events (deviations from or violations of communicative standards, e.g. B's interruption of A; B's silence after A's offer etc.) can be defined independently of any other variable (esp. independent of the physiological and expressive ones we assume to be affected by communicative violations).

The application of the minimal model of communicative exchange to the analysis of specific research questions has proved to be fruitful, as will be demonstrated by a review of some pertinent results.

References

1. Altorfer, A.; Jossen, S.; Käsermann, M.L.; Foppa, N.; Zimmermann, H. (1998). Measurement and analysis of head movement behavior during conversation. This volume.
2. Jossen, S.; Altorfer, A.; Käsermann, M.L.; Foppa, N.; Zimmermann, H. (1998). Computer-based analysis of physiological data in natural situations. This volume.
3. Zimmermann, H.; Käsermann, M.L.; Altorfer, A.; Jossen, S.; Foppa, N. (1998). Measurement and analysis of vocal aspects of communicative interaction. This volume.