The present document reflects the results obtained after several phases and many cycles of work.
The first version (December 1993) of the document contained the survey of the various systems, their comparison and a first sketch of the consensual nucleus of morphosyntactic specifications. This preliminary proposal was applied to a number of European languages: Italian, German, Spanish, English, Dutch and French. This phase, which was developed in a timespan of a few months, corresponding to different versions of the present document, generated much feedback, especially at the language-specific level.
A revision of the preliminary proposal in the light of the feedback coming from the first cycle of applications then took place, which caused changes mainly in the tables at Level 2b. This was carried out in March 1994 and June 1994.
The MULTEXT experience, in the multilingual framework of a concrete project, was an important test-bed for the proposal which introduced some modifications in the common proposal itself (October 1994).
More than 20 European and national projects made extensive use of EAGLES specifications, actively interacting with EAGLES: LRE DELIS, RENOS, CRATER, MECOLB, MULTEXT, COPERNICUS MULTEXT-East and TELRI, MLAP-PAROLE, ESPRIT-ELSNET, French GRACE, German Textcorpora und Erschließungswerkzeuge, LE-SPARKLE, ELRA, EUROWORDNET and PAROLE.
Some small revisions were made in May 1995.
Within the MLAP-PAROLE project, the applications to some EU languages not yet covered, i.e. Catalan, Irish and Swedish, were contributed and added to the present document, together with the experience gained within the project itself.
Furthermore, comments coming from experts in the field who evaluated these recommendations were incorporated in the present version of the document.
Finally, as already mentioned above, the experience gained within the validation phase, where guidelines providing definitions and explicit criteria for the application of the specifications in the different languages were carried out, has been taken into account.
Experience teaches that the process of consensus building, in order to arrive at a stable and broadly accepted standard, is of necessity a slow process: the proposal must undergo many phases of discussions between experts in the field and many cycles of applications for testing, evaluating and refining the specifications. It is essential that the interaction between proposals and applications is truly two-way, as happened for this document. The EAGLES subgroup responsible for this document will therefore be grateful for any relevant comment generated by the circulation of the present proposal. The feedback will be taken into account and incorporated in further revisions.