One of the first results obtained about the use of the modalities indicates that all of the modalities were used by the subjects during the various sessions. In addition, one observes an evolution in the use of multimodality during sessions:
During the first sessions one observes a homogeneous distribution in the use of the modalities which corresponds to a phase of an informal test of the system. This fact is corroborated by the data resulting from the autoconfrontation.
But some preferences in the use of the modalities quickly appear in the following sessions.
These preferences vary according to the subjects.
On a very global level (all sessions and all subjects mixed), one does not note specialisation of one modality to the realisation of one or several commands. Therefore, if there is specialisation, it can be located only at individual level, bringing the need of carrying out a specific analysis of the data for each subject.
As mentioned above, one observes in many subjects a tendency to preferential uses of modalities. These tendencies are specific (they appear only during only one session of interaction) or recurring (they appear in other sessions), are individualised (the same modality is used to carry out only one and same command), plural (the same modality is used to carry out several commands).
The activity graphs and the verbalisations of the subjects made it possible to identify several factors which are likely to direct the choice and the change of use of the modalities. We present only the four principal ones here:
Context of the activity in progress (recurring operational procedure)
We could identify that the tendencies to preferential use generally result in sequences of recurring actions carried out by the subjects. These sequences are strongly related to the context of the activity in progress.
Change of goal in the activity
The changes of modality are often associated with phases of local reorientation of the activity.
Properties of the modalities (implementation)
During the evaluation of a multimodal device, it is advisable to distinguish well the inherent characteristics of a given modality from the implementation mode of this modality. The fact that a modality is little used does not mean that this modality is unsuited but that its technical realisation can be inadequate or constraining. Thus, during the experimentation the tactile modality sometimes was under employed by the subjects because of the constraint induced by holding the stylet. This last is perceived as an external appendix of the PDA, which one possibly ends up putting down to not having to hold what will contribute to limiting its use or to restrict it with specific c ses (problem with the other modalities to achieve particular actions such as closing error messages).
Many changes of modalities appeared in dysfunctioning situations in order to correct errors coming either due to the subject, the system or of the accomplice.