| Abstract: |
My paper presents a corpus-based case study aimed at designing an electronic dictionary of Italian collocations and focussing on a small set of nouns belonging to the semantic feld of 'paura/fear'. In the paper, all of the steps involved in data retrieval, automatic and non-automatic evaluation, collocation selection and lexicographic organization are explained in detail. Lexicographic data are represented as a three-dimensional lexical framework displaying ontological, semantic and syntactic relations among lexemes. On the ontological level, paura as an entity is connected to contiguous emotions, but it also serves as a prototype for the category of the lexemes selected, shaping their syntactic and semantic behaviour. Collocations are formally categorized through a set of analytic parameters which enable a detailed lexical description as well as more finely grained dictionary search results. On the microstructural level, substantival collocation partners of the selected nouns are described in terms of thematic roles and semantic features, whereas adjectival collocation partners additionally have a set of principles derived from psychological studies applied to them. Finally, analysis of verbal collocation partners focusses on the interplay of the grammatical function and the thematic role of the noun they cooccur with, and on verbal Aktionsart. The intended exicographic description rests upon a coherent cross- reference network linking the prototype to the other lemmas, collocation partners to each other, and collocations belonging to the narrower semantic feld of paura, as well as collocations belonging to other semantic felds (e.g., the semantic feld of emotions). At the same time, according to an open source principle, corpus-based lexical data can be deductively expanded using framework information. |