Standardisation by consensus: Requirements in the field of the lexicon

Above, a few properties of the consensual approach to standardisation have been sketched out; throughout its work, the CLWG has aimed at satisfying the requirements related with the approach of standardisation by consensus.

• Broad overview:
  • The survey phase was based on input from both the fields of computational lexicography and of corpus linguistics (corpus processing is seen as the most prominent application of both morphosyntactic and syntactic classification of lexical material. In this context, we would like to particularly thank the EAGLES Subgroups on morphosyntactic and on syntactic corpus annotation for the valuable input and the amount of effort offered as a contribution to the work summarised here).
  • The morphosyntax work, although highly language-specific, covers all of the European Union languages, plus several languages of Eastern and Central European countries.
• Critical mass:
  • Broad language coverage has been ensured (see above); contributions on the different languages have been provided by both industry and academia; the first proposals, documented in EAGLES (1996g), have been discussed in many LRE and LE projects, whose feedback was then integrated.
  • The comparison of individual specifications for the individual languages (i.e. EAGLES (1996a), EAGLES (1996b), EAGLES (1996d), EAGLES (1996c)) is eased by the fact that all specifications are strictly parallel (based on the same descriptive devices, the same presentation, etc.).
  • The availability of detailed documentation allows for easy-to-reproduce descriptions. This is enhanced by the fact that an integrated package is being prepared which will contain the morphosyntax specifications, the practical guidelines for their use in corpus tagging, and tagged sample texts of 50-60,000 wordforms, for both German and Italian (see below).
• Proof of relevance, for practical applications and for leading edge technology development:
  • The breadth of the fragment coverage and the depth of the description makes the proposals applicable outside toy situations.
  • In addition, more sophisticated applications can hook up their own, specific descriptions to the basic ones proposed by the CLWG.
  • The formalisation employed makes the proposals reinterpretable (the reinterpretation being even automatically verifiable) and thus more easily reusable than proposals for which such possibilities do not exist.
• Accessibility and testability: a pilot resource has been produced, to be distributed via the European Linguistic Resources Association (ELRA), containing 50-60,000 wordform mini-corpora of Italian and German newspaper text, annotated according to the proposals of the CLWG. The preparation of these corpora was carried out jointly with the ELSNET Language and Speech Resources Task Group. We most gratefully acknowledge the financial support provided by ELSNET for this task.