Formalização da variação fonológica na aquisição da nasal velar em inglês pelo ORTO Ajuste Paramétrico

Optimality Theory (OT), (Prince; Smolensky, 2004 [1993]), originated from classic generative theory, has established a consistent formal proposal for the representation of linguistic structures through the mapping of mental representations (input) and surface representations (outputs), evaluated by universal constrictions instead of rules, in a parallel way instead of a serial one, as linguistic analyses had been done so far. Regardless of its strong potential to the formalization of languages, OT is uncapable of explaining variation - an unquestionably attested characteristic of language systems -, since it does not provide the mechanisms for the generation of two or more variable outputs. OT and Harmonic Grammar (HG) derived models (Legendre; Miyata; Smolensky, 1990a, b; Smolensky; Legendre, 2006), associated to learning algorithms, arise in linguistic science in order to account for such gap left by OT. The algorithms work as computational simulators of foreign language acquisition, appearing to be efficient regarding the tasks they must perform. In this article, we present a new alternative of harmonic learning algorithm, the ORTO Ajuste Paramétrico (Dornelles Filho, 2014), whose application is very similar to GLA-HG (Boersma; Hayes, 2001). We exemplify its functionalities through data of variable language acquisition of the velar nasal in English by Portuguese speakers. ORTO has proved to be adequate to the formalization of variation in a certain developmental stage of English acquisition as a foreign language.