O Rio Grande como fronteira linguística: um estudo do comportamento das vogais médias pretônicas

Abstract

The main objective of this study is to describe the linguistic variation present on the south and north of Rio Grande, observing the behavior of pre-stressed mid vowels. The phenomenon is shown in words like r[ε]lógio, for r[e]lógio and c[ͻ]lega, for c[o]lega, in Uberaba/MG, in Triângulo Mineiro, north of Rio Grande. The methodology adopted was that of Variationist Sociolinguistics, based on studies by Labov (2008) and Weinreich, Labov and Herzog (2006). For data collection, in Uberaba/MG, we interviewed 12 informants, selected by stratified random sampling. For the southern region of Rio Grande, we collected data in Igarapava/SP, Franca/SP and Ribeirão Preto/SP, by selecting television interviews, based on studies by Labov (2008). The dependent variable was: behavior of pretonic mid vowels, when there are low or mid-low vowels in stressed position. There were two variants: application of the vowel harmonization rule (pretonic performed as mid-low) and non-application of the rule (pretonic performed as mid-high, high, nasal or deleted). We used four linguistic independent variables (nature of target vowel, height of stressed vowel, distance between target vowel and stressed vowel, type of lexical item) and four extralinguistic variables (gender, education, age and geographic region of the informant). The orthographic transcription of the interviews was carried out, followed by the selection and coding of data. The sample was analyzed using the GoldVarbX statistical program. For Uberaba, considering all the contexts of pre-stresses mid vowel with low or mid-low vowel in stressed position, we obtained 749 occurrences, with 110 lowerings of the pre-stressed vowel. By the factor weights indicated by GoldVarbX, it was observed that the lowering of the pre-stressed vowel with low vowel in stressed position does not occur (weight 0.118). Thus, we performed a new sample cut, comparing only the mid-low vowel data in stressed position of the two regions under study. For the 179 occurrences in the northern region of Rio Grande, there were 108 records of application of the rule (factor weight 0.902). For the 95 occurrences in the southern region of Rio Grande, there were 6 records of application of the rule (relative weight 0.015). Furthermore, they were shown to favor the application of the rule: target vowel /o/, the word not being a noun. Social variables were considered not relevant by the statistical program. It was demonstrated, by the feature geometry, that there is a regressive spreading of features, so that the target vowel assimilates the opening feature of the trigger vowel, causing the behavior as a mid-low vowel. It was concluded that there are distinct variations in the two regions under study.