Consensus and recommendations

• It seems that individual methods of comparison and different matrices for comparing the phenomena covered and devices used by the approaches observed are necessary for the different levels of linguistic description. Once these comparison criteria have been established, it is possible to determine commonalities and differences between the approaches. Using the comparison grid, the next step must be to point to the set of shared basic descriptive devices and to the set of core phenomena catered for in the different approaches; in a similar way, the comparison must point out those areas of description, where either there is no possibility to relate different approaches, or only a few approaches are at all available which take the respective phenomenon into account. The areas where commonalities are observed are those which lend themselves to the formulation of proposals for consensual guidelines.
• In the areas, where no consensus can be achieved, at least guidelines for choosing among alternative, although mutually exclusive approaches should be given, such as to give the user a possibility of choosing on a basis of a definition of his individual needs. This latter point will be performed on the basis of the experience of the participants, since it is difficult to give operational criteria.

The production of recommendations is thus seen as a successive filtering process, and the survey phase is a first input to the comparison of alternative approaches. The survey itself can evidently make use of previous surveys carried out in different frameworks.