Categorização Fonológica e Representação Mental: uma análise da alternância entre [Ks] e [s] à luz de modelos de uso

• Ano de defesa: 2009
• Autor(a) principal: Maria Mendes Cantoni
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Modelo de Exemplares; [ks]; Fonologia do português brasileiro; Fonologia de Uso; modelos de uso; Lingüística; Fonética; Lingua portuguesa Consoantes; Representação mental; Fonologia lexical; Língua portuguesa Fonologia; Língua portuguesa Ortografia e silabação
• Link de acesso: http://hdl.handle.net/1843/ALDR-7RAK3L

This work has the aim of evaluating patterns of variation in sound sequences of BrazilianPortuguese. Focus is given to the observed alternance of [ks] and [s], as in sintaxe [si'taksi]~[si'tasi]. Regarding this variation, three groups of words were identified: (1) words in which [ks] and [s] alternate; (2) words in which only [ks] is realized or (3) only [s] isrealized, with no alternance. The relations between words in each group were evaluated considering a broader discussion on mental representations and storage dynamics of sounds in the worlds languages. This work also evaluated the issue of whether [ks] can be best accounted for as a sequence of sounds or as a single sound. Studies on Brazilian Portuguese sound system, all based on a formalistic point of view, claim that [ks] is a consonantal cluster,but none of them recognizes the alternance of [ks] and [s] that can be attested in specific words. In this work, it is claimed that unirepresentational models cannot offer a straightforward analysis of the phenomenon studied here, for they cannot explain lexical conditioning that operates in the alternance of [ks] with [s], neither the functional interactions established between [ks] and other sounds of the language, nor the motivations for the alternance. This work claims that the representational models presented by Cognitive Grammar (LANGACKER, 1987), Usage-Based Phonology (BYBEE, 2001) andPierrehumberts Exemplar Model (2000) can provide a more complete account of the alternance of [ks] with [s]. A comparison is developed on the one side between [ks] and Brazilian Portuguese heterosyllabic consonantal clusters which are not broken by epenthesis, and on the other side between [ks] and Brazilian Portuguese complex consonants. It was attested that [ks], but not heterosyllabic consonantal clusters: (1) can occur in final coda, a position in which [ks] has turned out very productive in the past years; (2) doesnt undergo epenthesis, which was not attested in the experiment. Voicing assimilation of [s] in words with [ks] reduced from a CVC structure was detected, but not in words with [ks] in a plainCC structure. Among the words containing [ks], only a few presented the possibility of alternance with [s], suggesting a case of lexical diffusion. Based on experimental results, multirepresentational analyses of the alternance between [ks] and [s] are presented. It is suggested that this pattern of alternance consists of a case where similar structures displaydifferent mental representations. Results of the experiments add proof to some assumptions of usage-based models, such as (1) token frequency has an impact over sound changes (more frequent words are affected by reductive changes first); (2) fine phonetic details are present in mental representations; (3) linguistic use promotes online restructuring of representations; (4) grammar management is specific and individual.