Corpus characteristics

A comparable Italian/German corpus was created, in so far as both, corpus composition and annotation, are based on the same principles for both languages:

• Around 20 verbs of Italian have been selected, mainly to cover different semantic fields and a variety of syntactic constructions, and to have examples of a medium degree of polysemy. On that basis, around 20 German verbs were selected, which are known to be (partial) translation equivalents of the Italian verbs. Again, emphasis was on a medium degree of polysemy, and on rich documentation in the source corpora. In five cases, two German verbs have been selected which are partial equivalents of one single Italian item, to facilitate experiments in translation equivalent selection and word sense disambiguation. A table summarizing the selected verbs is appended, for both languages (cf. appendix).
• The corpora serving as raw material are mainly journalistic: for Italian, the following local or national sources have been used: Il Corriere della Sera, La Repubblica, La Stampa, Il Sole 24 Ore, Unione Sarda and Panorama. For German, Frankfurter Rundschau, die Tageszeitung, and Stuttgarter Zeitung are the main sources: again local or national newspapers.
• For each Italian verb around 50 sentences have been selected from the data. For each German verb, it was aimed at the same number; in a few cases, somewhat less material was available, which was only accepted when two German verbs shared an Italian equivalent.
• For lexical semantic annotation, basically verbs and their arguments noun groups and prepositional groups are of interest. Consequently, emphasis was laid on non-sentential readings, and readings with sentential complements were only used as auxiliary information, to complete the syntactic description, wherever necessary.