Flexibility, Configurability and Optimality in UNL Deconversion via Multiparadigm Programming

The fulfillment of the UNL vision is primarily conditioned on the successful deployment of deconverters, each translating from the UNL into a target language. According to current practice, developing deconverters ultimately means configuring DeCo, the deconversion engine provided by the UNDL Foundation. However, DeCo has a number of limitations that hinder productivity and might even preclude quality deconversion. This paper discusses some of these shortcomings and introduces an alternative deconversion model – Manati, which is the result of work on UNL-mediated PortugueseBrazilian Sign Language human-aided machine translation. With Manati we attempt to exemplify how multiparadigm – namely, constraint, object-oriented and higher-order – programming can be drawn upon not only to specify an open-architecture, optimum-searching deconversion engine but also and above all to rationalize its configuration into deconverters for target languages.