Description of the guidelines for the syntactico-semantic annotation of a corpus in Spanish

The aim of the SenSem project1 is to build a databank that reflects the syntactic and seman-tic behavior of Spanish verbs. This databank will eventually consist of a verbal lexicon linked to a significant number of examples from corpus. These examples are being manu-ally analyzed following the guidelines pre-sented here. 1 The SenSem project The final aim of the SenSem project is to create a da-tabank (a lexicon linked to manually analyzed corpus examples) that reflects the syntactic and semantic be-havior of the verbs selected. As the initial phase of the project, a reference corpus for Spanish annotated with syntactico-semantic information is being constructed. A major problem in the SenSem project has been to bridge the gap between traditional grammatical con-cepts and the actual phenomena found in a corpus from real language. The final goal of the guidelines used is to bring the theoretical insights to the annota-tion of the actual examples found in corpus and to provide annotators with procedures as objective as possible to deal with phenomena found in corpus. Each sentence is linked to the verb sense it exem-plifies. Each verb sense is in turn associated with its Aktionsart and its argument structure in the form of a semantic role list. So sentences inherit this information from the sense