A Computational Study of Cross-situational Lexical Learning of Brazilian Portuguese
Pablo Picasso Feliciano de Faria · 2015
In this paper, a particular algorithm for lexical acquisition -taken as a problem of learning the mapping from words to meanings -is evaluated.The algorithm in Siskind (1996) is adapted to handle more complex input data, including data of Brazilian Portuguese.In particular, the input data in the present study covers a broader grammatical knowledge, showing both polysemy and higher inflectional and agreement morphology.Results indicate that these properties create difficulties to the learner and that more substantial developments to the algorithm are needed in order to increase its cross-linguistic capabilities.