This paper discusses the hypothesis that the outputs of derived pre-stressed high vowels [i] and [u], in Brazilian Portuguese, are acoustically different from the outputs of the non-derived high vowels, although both are perceptually equivalent. The sample totalizes 1152 tokens extracted from a controlled corpus, recorded at the Phonetic Laboratory of Rio de Janeiro Federal University (UFRJ), with eight university graduate students, four men and four women, from Rio de Janeiro, using acoustic

more »

... asurements (PRAAT) and multivariational analysis (GOLDVARB X) in order to detect the conditioning factors related to the first and second formant values. The results reveal differences between underlying and derived vowels.