Head movement analysis during conversation: relations to verbal behavior and psychophysiological data

A. Altorfer, M.L. Käsermann and S. Jossen

Department of Psychiatric Neurophysiology, Psychiatric Institutions, University of Bern, Bern, Switzerland.

Head movement behavior during conversation is recorded and analyzed by using a direct methodology that allows tracking of head positions over time with high accuracy [1]. This procedure results in a classification of head movement patterns according to their similarity with defined phenomenal criteria (e.g. head rotations to partner or head rising). As part of a nonverbal communication system head movements are usually embedded in a social context which influences their quantitative and qualitative features. Therefore, investigations concerning head movement behavior have to take into consideration various contextual variables. From a methodological viewpoint, a software package for recording, extracting, and analyzing head movement patterns should offer links to other sets of variables which are believed to yield important information for a functional interpretation of behavioral patterns.

For the realization of this claim, two types of integration of contextual variables into the data of head movement patterns are proposed: 1. The segmentation of verbal interaction (on the levels of speech acts, utterances, pauses between speech acts and utterances, etc.) is related to head movement behavior (and vice versa) (Module "VarRelative") and 2. localized head movement patterns are related to concomitant psychophysiological variables (and vice versa) (Module "HMP/VVC"). The basis of these connecting software modules is the time that is recorded for each variable (channel) as common reference. Thereupon, lists of time borders of relevant events are brought together and evaluated for the respective key variable. This means that head movement patterns are located e.g. in turns or pauses of verbal interaction (occurrence time and category of head movement pattern) or that concomitant physiological changes (e.g. vessel volume changes measured with photoelectric finger pulse volume; see [2]) are indicated during the appearance of classified head movement patterns.

Figure 1. Integration of software-modules "VarRelative" for the relation between verbal interaction and head movement patterns (HPM) and "HMP/VVC" for the relation beween head movement patterns and wesel volume changes(VVC).

For each module empirical examples are given to show the functional value of these linking procedures. For the connection between head movement patterns and the segmentation of verbal interaction, data are presented which show that the temporal occurrence of head movements differ in communicative situations of pauses depending on their pause categorization (i.e. with/without succeeding change of speaking turn). Concerning the inclusion of physiological variables, it could be shown that strong head rotations are correlated with physiological changes that are usually indicative either for stress or orienting responses. Stress responses are found in movement patterns which include only rotations away from the partner for more than 40 degrees, whereas orienting responses are found in all other head rotations of more than 40 degrees in both rotational directions (towards and away from the partner). During head rotations of more than 40 degrees towards the partners, there was no reaction in vessel volume changes.


  1. Altorfer, A.; Jossen, S.; Würmle O.; Käsermann, M.L.; Foppa, K.; Zimmermann H. (2000). Measurement and meaning of head movement behavior in everyday face-to-face communicative interaction. Behavior Research Methods, Instruments & Computers, 32, 17-32.
  2. Jossen, S.; Käsermann, M.L.; Altorfer, A.; Foppa, K.; Zimmermann, H.; Hirsbrunner, H.P. (2000). The study of emotional processes in communication: II. Peripheral blood flow as an indicator of emotionalization. Behavior Research Methods, Instruments & Computers, 32, 47-55.

Paper presented at Measuring Behavior 2000, 3rd International Conference on Methods and Techniques in Behavioral Research, 15-18 August 2000, Nijmegen, The Netherlands

© 2000 Noldus Information Technology b.v.