The number and characteristics of the informants vary according to the analyst's objective and phonological specialty. These parameters also modify the type and method of data collection. For example, in this study I am looking at vowel quality and consequently I put a control on the sex and age of the subjects. Sociolinguistic studies regarding Portuguese phonology, such as those performed as part of the «Competências Básicas de Português» (Lemle & Naro 1977; cf. Guy 1981a, 1981b; Votre 1979, 1981), do not investigate vowel quality specifically. However, they do consider geographic location, education, economic status, race, sex, etc.). On the other end of the scale, generative theoreticians work from an abstract concept of a «standard language» and often perform their analyses without defining specific informants (Duarte & Teixera 1979). These disparate methods of data collection exemplify the need for LB phonologists to be open minded about other's procedures in order to utilize their data.

From an articulatory point of view, the first formant frequency inversely correlates to vowel height (a lower F1 frequency compares to a higher vowel). The second formant frequency roughly corresponds to the front-back articulation of the vowel where a more fronted vowel exhibits a higher F2 formant.

Klatt (1976) offers a more complete review of the segmentation of sound sequences in connected speech. His parameters have been applied to the segmentation of BP by Simões (1987).

Disner (1988) provides a detailed explanation of how to apply an ANOVA to the study of vowel quality. Hatch and Farhady (1982) offer a description of ANOVAs for applied linguistics. I might add, parenthetically, that hispanists in both language and literature have been slow to accept the «validity» of analyzing statistic significance. Nearly all research in second language acquisition, ESL, and most components of sociolinguistics is more developed in testing the significance of their results.

See also Wright (1988) for an acoustic description of vowel space and Portuguese nasality.

The total variance, SST (sum of squares total) has two components: SSB (sum of squares between) which represents the variance between groups and SSW (sum of squares within) which represents the variance within the groups, obtained by subtracting the SSB from the SST The SSB is due to the variance in Factor A (Velocity), Factor B (Oral vs. Nasal), or the interaction of Factors A and B. The MS (mean squares) indicates how much of the variance can be attributed to each factor and is determined by dividing the sum of squares by the degrees of freedom (df.) which is the number in each group minus 1. The Fratio (MSB ÷ MSW) indicates whether the between-group variances MS_a, MS_b, MS_ab) are significantly larger than the within-group variances (MSW).

Head (1987), which examines the origin and phonetic nature of the «r caipira» exemplifies how research can integrate several levels of LB phonology. He incorporates evidence from linguistic atlases, articulatory descriptions, phonetic parameters, acoustic and articulatory distinctive features, and dialect studies, all to develop the phonological processes involved in the pronunciation of caipira «r».

Research for this article was aided by a leave from the University of California, Berkeley, and by a travel grant from its Center for Latin American Studies. Parts thereof were presented in a paper, «O Português do Brasil: Língua Padrão e Língua Popular» (Eighth Symposium on Portuguese Traditions, University of California, Los Angeles, 1985), and at lectures elsewhere. I thank Lyris Wiedemann for valuable comments on a early version of this paper.

The attributed quotations are from adult speakers from the cities of Belo Horizonte, Campinas, Porto Alegre, and São Paulo. Educated informants were university-trained middle-class professionals. Uneducated informants were working-class persons engaged in occupations requiring few or no specialized skills. The label «colloquial» covers the range of intimate and casual styles, and the label «formal» comprehends the consultative and formal styles, as described in Joos 1961. Some examples were taken from tape-recordings and others were jotted down during or shortly after informal conversations, by phone or in face-to-face situations. Since this article is intended as an overview of the main vernacular features found in the colloquial speech of educated persons, the data has not been treated quantitatively.

Literary use of stylized versions of the rural vernacular (e. g., Pires 1935a, 1935b or Silveira 1920, 1931, 1937, 1945) has met with scant success.