Estudos preliminares para a modelização de um avatar para a LGP: os descritores fonológicos

The Portuguese Sign Language (LGP) lacks basic computational tools. Among the reasons for the inexistence of such resources is the fact that little is known about it, and there is no existing description that is detailed and systematic enough to allow for an adequate computational modeling. In the context of the development of an avatar (an artificial signer) for LGP, we have been systematically describing LGP’s phonologic system, set on the compilation of a computational word list that could be used to produce that and other useful computational resources for this language. In this paper, we put forward some elements aiming to contribute to a first phonological description of LGP with computational modeling in mind. We identify the contrastive parameters (based on minimal pairs) for the description of LGP among the phonological categories pointed out by the Hand -Tier Model. Besides these, we identify and describe facial expression, which are not taken into account as distinctive in the model, but essential in the case for LGP, as it enables the distinction of minimal pairs (which shows strong categorical relevance), and is of upmost relevance to a description which leads to a sufficiently complete computational representation model that intelligibility and validation by the end users of the system, the Portuguese deaf community.