Mid-vowel alternations in the Brazilian Portuguese verb

The underlying system of consonants and vowels in Brazilian Portuguese (henceforth BP), together with the lexical and word-level phonological rules and the interactions between them, has been studied in great detail (see e.g. Harris 1974; Lopez 1979; Redenbarger 1981; Quicoli 1990). The detailed knowledge we possess in this area of BP phonology makes the language an excellent test case for evaluating theoretical innovations, especially since genuine descriptive and explanatory progress can only be achieved by virtue of improved theoretical models. The discussion in this paper will focus on the rules of Vowel Harmony (henceforth VH), Truncation, Neutralisation and Vowel Lowering in verbs (henceforth VL), for which a non-linear analysis will be proposed. One of the purposes of this paper is to show that a treatment of the BP vowel alternations in non-linear terms yields an elegant and explanatory description of a substantial part of the BP phonological component. Another goal is to present the theoretically relevant aspects of the BP facts. It will be argued that aperture distinctions in BP are advantageously represented by a single phonetic parameter, which will not only dispense with the features [high] and [low], but also with [ATR] or [tense], which have been frequently used to distinguish the two series of BP mid vowels. Also, further evidence will be provided in favour of the hypothesis advanced by a number of phonologists (see Clements 1991a; Odden 1991) that aperture features should be represented as constituents in the vocalic feature tree.