The goal of the project is to investigate the main types of semantic alignment that can be specified between WordNet.Br and Princeton WordNet verb databases where, in Brazilian Portuguese, there occur pronominal verbal constructions like "Ele se feriu com uma faca" (He hurt himself with a knife), "Maria e Paulo abraçaram-se" (Maria and Paulo hugged each other) "Ele se orgulha dela" (He is proud of her), "O ar perfuma-se com diferentes aromas" (The air got scented with different fragrances) "Aqui vive-se bem" (One lives well here), and non-pronominal verb constructions with patient subject like "Ele apanhou muito quando criança" (He got spanked a lot as a child). Accordingly, the project intends to study this semantic alignment task from two complementary perspectives. From the linguistic point of view, it will focus on the study of verb argument structure representation and realization, lexicalization patterns, diathesis alternations, and the meanings of those verb constructions; from the computational-linguistic view, i.e. from the wordnet building perspective, it will focus on the semantic alignment of the pronominal and affected-subject verb constructions in Brazilian Portuguese with their semantically equivalent ones in English.
News published in Agência FAPESP Newsletter about the scholarship: