Arthur Brakel will present a talk entitled, "Phonological Markedness and Levels of Analysis"
Abstract
My project, Portuguese Morphophonology: A Markedness Approach, attempts to provide a complete grammar of that language’s inflectional and derivational morphophonology, as well as that of nominal compounding. The analyses it provides are based on A) constraints on representation advocated by ‘natural’ phonologists, B) a theory of morpheme boundary affinities, C) notions of markedness (rarity, restriction, complexity, and opposition) I advanced in Phonological Markedness and Distinctive Features (Indiana, 1983).

Identifying Portuguese’s 26 phonemes requires 12 distinctive features. In representative Portuguese texts each of the 12 distinctive features (phonological marks) should be present in fewer phoneme tokens than in the tokens lacking that distinctive feature (e.g. there should be fewer [+nasal] tokens than [-nasal] tokens.) And that is the case for all features, save one: [+contoid], whose presence among the 18,455 tokens in the study’s morphophonemic segmental dataset outnumbers the phoneme tokens lacking that feature. However, in ‘systematic phonetic’ representations [+contoid] occurs in fewer tokens than [-contoid]. The issues this result brings up are: A) At what level of analysis are these markedness criteria valid evaluators of linguistic data? B) What would be the results of a feature count in Portuguese texts with extreme vowel reduction? C) Are my results a peculiarity of Portuguese? D) Are they an artifact of my analysis? E) Are they valid for Portuguese? F) In other languages’ texts do contoids’ and vocoids’ tokens occur in similar proportions?