Universidade Federal do Paraná : Master Degree
Thesis: Retroflexão gradiente nos róticos em coda no PB de Curitiba
-
This dissertation verifies the occurrence of a gradient allophonic variation on the production of rothics in coda position by speakers of Brazilian Portuguese from Curitiba. Previous data (Clemente 2005) indicated that it is possible to observe in this context a variety of R- sounds being produced by the same speaker at the same segmental environment. Gradient allophony was previously reported on phonetic literature by, e.g., Sproat & Fujimura (1993). Their study shows that /l/ sounds of American English exhibit a non-categorical allophony that lays beyond two categories: a light one, the alveolar lateral, which occurs especially in the beginning of words and a dark one, the velar lateral, occurring at the end of words. They have observed that the strength of the boundary following the laterals exerts influence over the allophony of the lateral, so that the greater the strength of the boundary, the stronger the tendency for speakers to produce laterals tending to the dark version. According to the authors, the /l/ is composed by two gestures, one related to the movement of the tongue dorsum and the other related to the tongue tip movement. The strength of boundary acts especially over the coordination between these two gestures.
So, following Sproat & Fujimura (1993), the hypothesis here is that rhotics in word-final coda exhibit a gradient allophony influenced by the strength of boundary. Between two categories, assumed here to be the tap and the retroflex approximant, there is a variety of R-sounds which will tend to one category or another depending on the strength of the boundary they are submitted to. The strengths will act over the degree of retroflexion of the tongue dorsum. Stronger boundaries may lead to varieties presenting a greater degree of retroflexion because it creates a time slot that allows the tongue dorsum, a slowly articulator, to reach its targets. Thus, to check this hypothesis, two boundary strengths were taken into account: a stronger one, elicited by a vocative, and a weaker one, in a declarative
sentence.
Thus, there were prosodic minimal pairs like “ Almir, apaga a lousa” x “Almir apaga a lousa”. (Almir, erase the board! X Almir erases the board.) Besides the strength of the boundaries, the vocalic environment adjacent to the rhotics was examined as well, in order to check any possible influence of the vowels on rhotics gradiency. That’s why the words in the experimental corpus contain three different vowels before and after the rhotic. Analyzed data indicates the existence of gradient allophony relying on the strength of boundaries but the vowel context seems not to influence the allophony. Greater strengths promote a longer time slot that allows the tongue dorsum to reach its target during the rhotic production. If there is an interaction between prosody and segment and this relation acts over the temporal patterns of the sounds, this phenomenon is not a merely mechanical implementation but it is a grammatical fact. As grammatical, it needs to be processed by the speaker’s phonology. But traditional models have primitives endowed with extrinsic time, which makes impossible the interpretation of a gradient allophony. The phonological primitive should have on its structure the variable time. It is the only reasonable way to promote that interaction. This is the reason why I argue in favor of the Articulatory Phonology and try to base my analysis on it.